1 @chapter Filtering Introduction
2 @c man begin FILTERING INTRODUCTION
4 Filtering in FFmpeg is enabled through the libavfilter library.
6 In libavfilter, a filter can have multiple inputs and multiple
8 To illustrate the sorts of things that are possible, we consider the
13 input --> split ---------------------> overlay --> output
16 +-----> crop --> vflip -------+
19 This filtergraph splits the input stream in two streams, then sends one
20 stream through the crop filter and the vflip filter, before merging it
21 back with the other stream by overlaying it on top. You can use the
22 following command to achieve this:
25 ffmpeg -i INPUT -vf "split [main][tmp]; [tmp] crop=iw:ih/2:0:0, vflip [flip]; [main][flip] overlay=0:H/2" OUTPUT
28 The result will be that the top half of the video is mirrored
29 onto the bottom half of the output video.
31 Filters in the same linear chain are separated by commas, and distinct
32 linear chains of filters are separated by semicolons. In our example,
33 @var{crop,vflip} are in one linear chain, @var{split} and
34 @var{overlay} are separately in another. The points where the linear
35 chains join are labelled by names enclosed in square brackets. In the
36 example, the split filter generates two outputs that are associated to
37 the labels @var{[main]} and @var{[tmp]}.
39 The stream sent to the second output of @var{split}, labelled as
40 @var{[tmp]}, is processed through the @var{crop} filter, which crops
41 away the lower half part of the video, and then vertically flipped. The
42 @var{overlay} filter takes in input the first unchanged output of the
43 split filter (which was labelled as @var{[main]}), and overlay on its
44 lower half the output generated by the @var{crop,vflip} filterchain.
46 Some filters take in input a list of parameters: they are specified
47 after the filter name and an equal sign, and are separated from each other
50 There exist so-called @var{source filters} that do not have an
51 audio/video input, and @var{sink filters} that will not have audio/video
54 @c man end FILTERING INTRODUCTION
57 @c man begin GRAPH2DOT
59 The @file{graph2dot} program included in the FFmpeg @file{tools}
60 directory can be used to parse a filtergraph description and issue a
61 corresponding textual representation in the dot language.
68 to see how to use @file{graph2dot}.
70 You can then pass the dot description to the @file{dot} program (from
71 the graphviz suite of programs) and obtain a graphical representation
74 For example the sequence of commands:
76 echo @var{GRAPH_DESCRIPTION} | \
77 tools/graph2dot -o graph.tmp && \
78 dot -Tpng graph.tmp -o graph.png && \
82 can be used to create and display an image representing the graph
83 described by the @var{GRAPH_DESCRIPTION} string. Note that this string must be
84 a complete self-contained graph, with its inputs and outputs explicitly defined.
85 For example if your command line is of the form:
87 ffmpeg -i infile -vf scale=640:360 outfile
89 your @var{GRAPH_DESCRIPTION} string will need to be of the form:
91 nullsrc,scale=640:360,nullsink
93 you may also need to set the @var{nullsrc} parameters and add a @var{format}
94 filter in order to simulate a specific input file.
98 @chapter Filtergraph description
99 @c man begin FILTERGRAPH DESCRIPTION
101 A filtergraph is a directed graph of connected filters. It can contain
102 cycles, and there can be multiple links between a pair of
103 filters. Each link has one input pad on one side connecting it to one
104 filter from which it takes its input, and one output pad on the other
105 side connecting it to one filter accepting its output.
107 Each filter in a filtergraph is an instance of a filter class
108 registered in the application, which defines the features and the
109 number of input and output pads of the filter.
111 A filter with no input pads is called a "source", and a filter with no
112 output pads is called a "sink".
114 @anchor{Filtergraph syntax}
115 @section Filtergraph syntax
117 A filtergraph has a textual representation, which is recognized by the
118 @option{-filter}/@option{-vf}/@option{-af} and
119 @option{-filter_complex} options in @command{ffmpeg} and
120 @option{-vf}/@option{-af} in @command{ffplay}, and by the
121 @code{avfilter_graph_parse_ptr()} function defined in
122 @file{libavfilter/avfilter.h}.
124 A filterchain consists of a sequence of connected filters, each one
125 connected to the previous one in the sequence. A filterchain is
126 represented by a list of ","-separated filter descriptions.
128 A filtergraph consists of a sequence of filterchains. A sequence of
129 filterchains is represented by a list of ";"-separated filterchain
132 A filter is represented by a string of the form:
133 [@var{in_link_1}]...[@var{in_link_N}]@var{filter_name}@@@var{id}=@var{arguments}[@var{out_link_1}]...[@var{out_link_M}]
135 @var{filter_name} is the name of the filter class of which the
136 described filter is an instance of, and has to be the name of one of
137 the filter classes registered in the program optionally followed by "@@@var{id}".
138 The name of the filter class is optionally followed by a string
141 @var{arguments} is a string which contains the parameters used to
142 initialize the filter instance. It may have one of two forms:
146 A ':'-separated list of @var{key=value} pairs.
149 A ':'-separated list of @var{value}. In this case, the keys are assumed to be
150 the option names in the order they are declared. E.g. the @code{fade} filter
151 declares three options in this order -- @option{type}, @option{start_frame} and
152 @option{nb_frames}. Then the parameter list @var{in:0:30} means that the value
153 @var{in} is assigned to the option @option{type}, @var{0} to
154 @option{start_frame} and @var{30} to @option{nb_frames}.
157 A ':'-separated list of mixed direct @var{value} and long @var{key=value}
158 pairs. The direct @var{value} must precede the @var{key=value} pairs, and
159 follow the same constraints order of the previous point. The following
160 @var{key=value} pairs can be set in any preferred order.
164 If the option value itself is a list of items (e.g. the @code{format} filter
165 takes a list of pixel formats), the items in the list are usually separated by
168 The list of arguments can be quoted using the character @samp{'} as initial
169 and ending mark, and the character @samp{\} for escaping the characters
170 within the quoted text; otherwise the argument string is considered
171 terminated when the next special character (belonging to the set
172 @samp{[]=;,}) is encountered.
174 The name and arguments of the filter are optionally preceded and
175 followed by a list of link labels.
176 A link label allows one to name a link and associate it to a filter output
177 or input pad. The preceding labels @var{in_link_1}
178 ... @var{in_link_N}, are associated to the filter input pads,
179 the following labels @var{out_link_1} ... @var{out_link_M}, are
180 associated to the output pads.
182 When two link labels with the same name are found in the
183 filtergraph, a link between the corresponding input and output pad is
186 If an output pad is not labelled, it is linked by default to the first
187 unlabelled input pad of the next filter in the filterchain.
188 For example in the filterchain
190 nullsrc, split[L1], [L2]overlay, nullsink
192 the split filter instance has two output pads, and the overlay filter
193 instance two input pads. The first output pad of split is labelled
194 "L1", the first input pad of overlay is labelled "L2", and the second
195 output pad of split is linked to the second input pad of overlay,
196 which are both unlabelled.
198 In a filter description, if the input label of the first filter is not
199 specified, "in" is assumed; if the output label of the last filter is not
200 specified, "out" is assumed.
202 In a complete filterchain all the unlabelled filter input and output
203 pads must be connected. A filtergraph is considered valid if all the
204 filter input and output pads of all the filterchains are connected.
206 Libavfilter will automatically insert @ref{scale} filters where format
207 conversion is required. It is possible to specify swscale flags
208 for those automatically inserted scalers by prepending
209 @code{sws_flags=@var{flags};}
210 to the filtergraph description.
212 Here is a BNF description of the filtergraph syntax:
214 @var{NAME} ::= sequence of alphanumeric characters and '_'
215 @var{FILTER_NAME} ::= @var{NAME}["@@"@var{NAME}]
216 @var{LINKLABEL} ::= "[" @var{NAME} "]"
217 @var{LINKLABELS} ::= @var{LINKLABEL} [@var{LINKLABELS}]
218 @var{FILTER_ARGUMENTS} ::= sequence of chars (possibly quoted)
219 @var{FILTER} ::= [@var{LINKLABELS}] @var{FILTER_NAME} ["=" @var{FILTER_ARGUMENTS}] [@var{LINKLABELS}]
220 @var{FILTERCHAIN} ::= @var{FILTER} [,@var{FILTERCHAIN}]
221 @var{FILTERGRAPH} ::= [sws_flags=@var{flags};] @var{FILTERCHAIN} [;@var{FILTERGRAPH}]
224 @anchor{filtergraph escaping}
225 @section Notes on filtergraph escaping
227 Filtergraph description composition entails several levels of
228 escaping. See @ref{quoting_and_escaping,,the "Quoting and escaping"
229 section in the ffmpeg-utils(1) manual,ffmpeg-utils} for more
230 information about the employed escaping procedure.
232 A first level escaping affects the content of each filter option
233 value, which may contain the special character @code{:} used to
234 separate values, or one of the escaping characters @code{\'}.
236 A second level escaping affects the whole filter description, which
237 may contain the escaping characters @code{\'} or the special
238 characters @code{[],;} used by the filtergraph description.
240 Finally, when you specify a filtergraph on a shell commandline, you
241 need to perform a third level escaping for the shell special
242 characters contained within it.
244 For example, consider the following string to be embedded in
245 the @ref{drawtext} filter description @option{text} value:
247 this is a 'string': may contain one, or more, special characters
250 This string contains the @code{'} special escaping character, and the
251 @code{:} special character, so it needs to be escaped in this way:
253 text=this is a \'string\'\: may contain one, or more, special characters
256 A second level of escaping is required when embedding the filter
257 description in a filtergraph description, in order to escape all the
258 filtergraph special characters. Thus the example above becomes:
260 drawtext=text=this is a \\\'string\\\'\\: may contain one\, or more\, special characters
262 (note that in addition to the @code{\'} escaping special characters,
263 also @code{,} needs to be escaped).
265 Finally an additional level of escaping is needed when writing the
266 filtergraph description in a shell command, which depends on the
267 escaping rules of the adopted shell. For example, assuming that
268 @code{\} is special and needs to be escaped with another @code{\}, the
269 previous string will finally result in:
271 -vf "drawtext=text=this is a \\\\\\'string\\\\\\'\\\\: may contain one\\, or more\\, special characters"
274 @chapter Timeline editing
276 Some filters support a generic @option{enable} option. For the filters
277 supporting timeline editing, this option can be set to an expression which is
278 evaluated before sending a frame to the filter. If the evaluation is non-zero,
279 the filter will be enabled, otherwise the frame will be sent unchanged to the
280 next filter in the filtergraph.
282 The expression accepts the following values:
285 timestamp expressed in seconds, NAN if the input timestamp is unknown
288 sequential number of the input frame, starting from 0
291 the position in the file of the input frame, NAN if unknown
295 width and height of the input frame if video
298 Additionally, these filters support an @option{enable} command that can be used
299 to re-define the expression.
301 Like any other filtering option, the @option{enable} option follows the same
304 For example, to enable a blur filter (@ref{smartblur}) from 10 seconds to 3
305 minutes, and a @ref{curves} filter starting at 3 seconds:
307 smartblur = enable='between(t,10,3*60)',
308 curves = enable='gte(t,3)' : preset=cross_process
311 See @code{ffmpeg -filters} to view which filters have timeline support.
313 @c man end FILTERGRAPH DESCRIPTION
316 @chapter Changing options at runtime with a command
318 Some options can be changed during the operation of the filter using
319 a command. These options are marked 'T' on the output of
320 @command{ffmpeg} @option{-h filter=<name of filter>}.
321 The name of the command is the name of the option and the argument is
325 @chapter Options for filters with several inputs (framesync)
326 @c man begin OPTIONS FOR FILTERS WITH SEVERAL INPUTS
328 Some filters with several inputs support a common set of options.
329 These options can only be set by name, not with the short notation.
333 The action to take when EOF is encountered on the secondary input; it accepts
334 one of the following values:
338 Repeat the last frame (the default).
342 Pass the main input through.
346 If set to 1, force the output to terminate when the shortest input
347 terminates. Default value is 0.
350 If set to 1, force the filter to extend the last frame of secondary streams
351 until the end of the primary stream. A value of 0 disables this behavior.
355 @c man end OPTIONS FOR FILTERS WITH SEVERAL INPUTS
357 @chapter Audio Filters
358 @c man begin AUDIO FILTERS
360 When you configure your FFmpeg build, you can disable any of the
361 existing filters using @code{--disable-filters}.
362 The configure output will show the audio filters included in your
365 Below is a description of the currently available audio filters.
369 A compressor is mainly used to reduce the dynamic range of a signal.
370 Especially modern music is mostly compressed at a high ratio to
371 improve the overall loudness. It's done to get the highest attention
372 of a listener, "fatten" the sound and bring more "power" to the track.
373 If a signal is compressed too much it may sound dull or "dead"
374 afterwards or it may start to "pump" (which could be a powerful effect
375 but can also destroy a track completely).
376 The right compression is the key to reach a professional sound and is
377 the high art of mixing and mastering. Because of its complex settings
378 it may take a long time to get the right feeling for this kind of effect.
380 Compression is done by detecting the volume above a chosen level
381 @code{threshold} and dividing it by the factor set with @code{ratio}.
382 So if you set the threshold to -12dB and your signal reaches -6dB a ratio
383 of 2:1 will result in a signal at -9dB. Because an exact manipulation of
384 the signal would cause distortion of the waveform the reduction can be
385 levelled over the time. This is done by setting "Attack" and "Release".
386 @code{attack} determines how long the signal has to rise above the threshold
387 before any reduction will occur and @code{release} sets the time the signal
388 has to fall below the threshold to reduce the reduction again. Shorter signals
389 than the chosen attack time will be left untouched.
390 The overall reduction of the signal can be made up afterwards with the
391 @code{makeup} setting. So compressing the peaks of a signal about 6dB and
392 raising the makeup to this level results in a signal twice as loud than the
393 source. To gain a softer entry in the compression the @code{knee} flattens the
394 hard edge at the threshold in the range of the chosen decibels.
396 The filter accepts the following options:
400 Set input gain. Default is 1. Range is between 0.015625 and 64.
403 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
404 Default is @code{downward}.
407 If a signal of stream rises above this level it will affect the gain
409 By default it is 0.125. Range is between 0.00097563 and 1.
412 Set a ratio by which the signal is reduced. 1:2 means that if the level
413 rose 4dB above the threshold, it will be only 2dB above after the reduction.
414 Default is 2. Range is between 1 and 20.
417 Amount of milliseconds the signal has to rise above the threshold before gain
418 reduction starts. Default is 20. Range is between 0.01 and 2000.
421 Amount of milliseconds the signal has to fall below the threshold before
422 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
425 Set the amount by how much signal will be amplified after processing.
426 Default is 1. Range is from 1 to 64.
429 Curve the sharp knee around the threshold to enter gain reduction more softly.
430 Default is 2.82843. Range is between 1 and 8.
433 Choose if the @code{average} level between all channels of input stream
434 or the louder(@code{maximum}) channel of input stream affects the
435 reduction. Default is @code{average}.
438 Should the exact signal be taken in case of @code{peak} or an RMS one in case
439 of @code{rms}. Default is @code{rms} which is mostly smoother.
442 How much to use compressed signal in output. Default is 1.
443 Range is between 0 and 1.
448 This filter supports the all above options as @ref{commands}.
451 Simple audio dynamic range compression/expansion filter.
453 The filter accepts the following options:
457 Set contrast. Default is 33. Allowed range is between 0 and 100.
462 Copy the input audio source unchanged to the output. This is mainly useful for
467 Apply cross fade from one input audio stream to another input audio stream.
468 The cross fade is applied for specified duration near the end of first stream.
470 The filter accepts the following options:
474 Specify the number of samples for which the cross fade effect has to last.
475 At the end of the cross fade effect the first input audio will be completely
476 silent. Default is 44100.
479 Specify the duration of the cross fade effect. See
480 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
481 for the accepted syntax.
482 By default the duration is determined by @var{nb_samples}.
483 If set this option is used instead of @var{nb_samples}.
486 Should first stream end overlap with second stream start. Default is enabled.
489 Set curve for cross fade transition for first stream.
492 Set curve for cross fade transition for second stream.
494 For description of available curve types see @ref{afade} filter description.
501 Cross fade from one input to another:
503 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
507 Cross fade from one input to another but without overlapping:
509 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
514 Split audio stream into several bands.
516 This filter splits audio stream into two or more frequency ranges.
517 Summing all streams back will give flat output.
519 The filter accepts the following options:
523 Set split frequencies. Those must be positive and increasing.
526 Set filter order for each band split. This controls filter roll-off or steepness
527 of filter transfer function.
528 Available values are:
553 Default is @var{4th}.
556 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
559 Set output gain for each band. Default value is 1 for all bands.
566 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
567 each band will be in separate stream:
569 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
573 Same as above, but with higher filter order:
575 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
579 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
581 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=8th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
587 Reduce audio bit resolution.
589 This filter is bit crusher with enhanced functionality. A bit crusher
590 is used to audibly reduce number of bits an audio signal is sampled
591 with. This doesn't change the bit depth at all, it just produces the
592 effect. Material reduced in bit depth sounds more harsh and "digital".
593 This filter is able to even round to continuous values instead of discrete
595 Additionally it has a D/C offset which results in different crushing of
596 the lower and the upper half of the signal.
597 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
599 Another feature of this filter is the logarithmic mode.
600 This setting switches from linear distances between bits to logarithmic ones.
601 The result is a much more "natural" sounding crusher which doesn't gate low
602 signals for example. The human ear has a logarithmic perception,
603 so this kind of crushing is much more pleasant.
604 Logarithmic crushing is also able to get anti-aliased.
606 The filter accepts the following options:
622 Can be linear: @code{lin} or logarithmic: @code{log}.
631 Set sample reduction.
634 Enable LFO. By default disabled.
645 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
649 Remove impulsive noise from input audio.
651 Samples detected as impulsive noise are replaced by interpolated samples using
652 autoregressive modelling.
656 Set window size, in milliseconds. Allowed range is from @code{10} to
657 @code{100}. Default value is @code{55} milliseconds.
658 This sets size of window which will be processed at once.
661 Set window overlap, in percentage of window size. Allowed range is from
662 @code{50} to @code{95}. Default value is @code{75} percent.
663 Setting this to a very high value increases impulsive noise removal but makes
664 whole process much slower.
667 Set autoregression order, in percentage of window size. Allowed range is from
668 @code{0} to @code{25}. Default value is @code{2} percent. This option also
669 controls quality of interpolated samples using neighbour good samples.
672 Set threshold value. Allowed range is from @code{1} to @code{100}.
673 Default value is @code{2}.
674 This controls the strength of impulsive noise which is going to be removed.
675 The lower value, the more samples will be detected as impulsive noise.
678 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
679 @code{10}. Default value is @code{2}.
680 If any two samples detected as noise are spaced less than this value then any
681 sample between those two samples will be also detected as noise.
686 It accepts the following values:
689 Select overlap-add method. Even not interpolated samples are slightly
690 changed with this method.
693 Select overlap-save method. Not interpolated samples remain unchanged.
696 Default value is @code{a}.
700 Remove clipped samples from input audio.
702 Samples detected as clipped are replaced by interpolated samples using
703 autoregressive modelling.
707 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
708 Default value is @code{55} milliseconds.
709 This sets size of window which will be processed at once.
712 Set window overlap, in percentage of window size. Allowed range is from @code{50}
713 to @code{95}. Default value is @code{75} percent.
716 Set autoregression order, in percentage of window size. Allowed range is from
717 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
718 quality of interpolated samples using neighbour good samples.
721 Set threshold value. Allowed range is from @code{1} to @code{100}.
722 Default value is @code{10}. Higher values make clip detection less aggressive.
725 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
726 Default value is @code{1000}. Higher values make clip detection less aggressive.
731 It accepts the following values:
734 Select overlap-add method. Even not interpolated samples are slightly changed
738 Select overlap-save method. Not interpolated samples remain unchanged.
741 Default value is @code{a}.
746 Delay one or more audio channels.
748 Samples in delayed channel are filled with silence.
750 The filter accepts the following option:
754 Set list of delays in milliseconds for each channel separated by '|'.
755 Unused delays will be silently ignored. If number of given delays is
756 smaller than number of channels all remaining channels will not be delayed.
757 If you want to delay exact number of samples, append 'S' to number.
758 If you want instead to delay in seconds, append 's' to number.
761 Use last set delay for all remaining channels. By default is disabled.
762 This option if enabled changes how option @code{delays} is interpreted.
769 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
770 the second channel (and any other channels that may be present) unchanged.
776 Delay second channel by 500 samples, the third channel by 700 samples and leave
777 the first channel (and any other channels that may be present) unchanged.
783 Delay all channels by same number of samples:
785 adelay=delays=64S:all=1
790 Remedy denormals in audio by adding extremely low-level noise.
792 This filter shall be placed before any filter that can produce denormals.
794 A description of the accepted parameters follows.
798 Set level of added noise in dB. Default is @code{-351}.
799 Allowed range is from -451 to -90.
802 Set type of added noise.
815 Default is @code{dc}.
820 This filter supports the all above options as @ref{commands}.
822 @section aderivative, aintegral
824 Compute derivative/integral of audio stream.
826 Applying both filters one after another produces original audio.
830 Apply echoing to the input audio.
832 Echoes are reflected sound and can occur naturally amongst mountains
833 (and sometimes large buildings) when talking or shouting; digital echo
834 effects emulate this behaviour and are often used to help fill out the
835 sound of a single instrument or vocal. The time difference between the
836 original signal and the reflection is the @code{delay}, and the
837 loudness of the reflected signal is the @code{decay}.
838 Multiple echoes can have different delays and decays.
840 A description of the accepted parameters follows.
844 Set input gain of reflected signal. Default is @code{0.6}.
847 Set output gain of reflected signal. Default is @code{0.3}.
850 Set list of time intervals in milliseconds between original signal and reflections
851 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
852 Default is @code{1000}.
855 Set list of loudness of reflected signals separated by '|'.
856 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
857 Default is @code{0.5}.
864 Make it sound as if there are twice as many instruments as are actually playing:
866 aecho=0.8:0.88:60:0.4
870 If delay is very short, then it sounds like a (metallic) robot playing music:
876 A longer delay will sound like an open air concert in the mountains:
878 aecho=0.8:0.9:1000:0.3
882 Same as above but with one more mountain:
884 aecho=0.8:0.9:1000|1800:0.3|0.25
889 Audio emphasis filter creates or restores material directly taken from LPs or
890 emphased CDs with different filter curves. E.g. to store music on vinyl the
891 signal has to be altered by a filter first to even out the disadvantages of
892 this recording medium.
893 Once the material is played back the inverse filter has to be applied to
894 restore the distortion of the frequency response.
896 The filter accepts the following options:
906 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
907 use @code{production} mode. Default is @code{reproduction} mode.
910 Set filter type. Selects medium. Can be one of the following:
922 select Compact Disc (CD).
928 select 50µs (FM-KF).
930 select 75µs (FM-KF).
936 This filter supports the all above options as @ref{commands}.
940 Modify an audio signal according to the specified expressions.
942 This filter accepts one or more expressions (one for each channel),
943 which are evaluated and used to modify a corresponding audio signal.
945 It accepts the following parameters:
949 Set the '|'-separated expressions list for each separate channel. If
950 the number of input channels is greater than the number of
951 expressions, the last specified expression is used for the remaining
954 @item channel_layout, c
955 Set output channel layout. If not specified, the channel layout is
956 specified by the number of expressions. If set to @samp{same}, it will
957 use by default the same input channel layout.
960 Each expression in @var{exprs} can contain the following constants and functions:
964 channel number of the current expression
967 number of the evaluated sample, starting from 0
973 time of the evaluated sample expressed in seconds
976 @item nb_out_channels
977 input and output number of channels
980 the value of input channel with number @var{CH}
983 Note: this filter is slow. For faster processing you should use a
992 aeval=val(ch)/2:c=same
996 Invert phase of the second channel:
1005 Apply fade-in/out effect to input audio.
1007 A description of the accepted parameters follows.
1011 Specify the effect type, can be either @code{in} for fade-in, or
1012 @code{out} for a fade-out effect. Default is @code{in}.
1014 @item start_sample, ss
1015 Specify the number of the start sample for starting to apply the fade
1016 effect. Default is 0.
1018 @item nb_samples, ns
1019 Specify the number of samples for which the fade effect has to last. At
1020 the end of the fade-in effect the output audio will have the same
1021 volume as the input audio, at the end of the fade-out transition
1022 the output audio will be silence. Default is 44100.
1024 @item start_time, st
1025 Specify the start time of the fade effect. Default is 0.
1026 The value must be specified as a time duration; see
1027 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1028 for the accepted syntax.
1029 If set this option is used instead of @var{start_sample}.
1032 Specify the duration of the fade effect. See
1033 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1034 for the accepted syntax.
1035 At the end of the fade-in effect the output audio will have the same
1036 volume as the input audio, at the end of the fade-out transition
1037 the output audio will be silence.
1038 By default the duration is determined by @var{nb_samples}.
1039 If set this option is used instead of @var{nb_samples}.
1042 Set curve for fade transition.
1044 It accepts the following values:
1047 select triangular, linear slope (default)
1049 select quarter of sine wave
1051 select half of sine wave
1053 select exponential sine wave
1057 select inverted parabola
1071 select inverted quarter of sine wave
1073 select inverted half of sine wave
1075 select double-exponential seat
1077 select double-exponential sigmoid
1079 select logistic sigmoid
1081 select sine cardinal function
1083 select inverted sine cardinal function
1089 @subsection Commands
1091 This filter supports the all above options as @ref{commands}.
1093 @subsection Examples
1097 Fade in first 15 seconds of audio:
1099 afade=t=in:ss=0:d=15
1103 Fade out last 25 seconds of a 900 seconds audio:
1105 afade=t=out:st=875:d=25
1110 Denoise audio samples with FFT.
1112 A description of the accepted parameters follows.
1116 Set the noise reduction in dB, allowed range is 0.01 to 97.
1117 Default value is 12 dB.
1120 Set the noise floor in dB, allowed range is -80 to -20.
1121 Default value is -50 dB.
1126 It accepts the following values:
1135 Select shellac noise.
1138 Select custom noise, defined in @code{bn} option.
1140 Default value is white noise.
1144 Set custom band noise for every one of 15 bands.
1145 Bands are separated by ' ' or '|'.
1148 Set the residual floor in dB, allowed range is -80 to -20.
1149 Default value is -38 dB.
1152 Enable noise tracking. By default is disabled.
1153 With this enabled, noise floor is automatically adjusted.
1156 Enable residual tracking. By default is disabled.
1159 Set the output mode.
1161 It accepts the following values:
1164 Pass input unchanged.
1167 Pass noise filtered out.
1172 Default value is @var{o}.
1176 @subsection Commands
1178 This filter supports the following commands:
1180 @item sample_noise, sn
1181 Start or stop measuring noise profile.
1182 Syntax for the command is : "start" or "stop" string.
1183 After measuring noise profile is stopped it will be
1184 automatically applied in filtering.
1186 @item noise_reduction, nr
1187 Change noise reduction. Argument is single float number.
1188 Syntax for the command is : "@var{noise_reduction}"
1190 @item noise_floor, nf
1191 Change noise floor. Argument is single float number.
1192 Syntax for the command is : "@var{noise_floor}"
1194 @item output_mode, om
1195 Change output mode operation.
1196 Syntax for the command is : "i", "o" or "n" string.
1200 Apply arbitrary expressions to samples in frequency domain.
1204 Set frequency domain real expression for each separate channel separated
1205 by '|'. Default is "re".
1206 If the number of input channels is greater than the number of
1207 expressions, the last specified expression is used for the remaining
1211 Set frequency domain imaginary expression for each separate channel
1212 separated by '|'. Default is "im".
1214 Each expression in @var{real} and @var{imag} can contain the following
1215 constants and functions:
1222 current frequency bin number
1225 number of available bins
1228 channel number of the current expression
1237 current real part of frequency bin of current channel
1240 current imaginary part of frequency bin of current channel
1243 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1246 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1250 Set window size. Allowed range is from 16 to 131072.
1251 Default is @code{4096}
1254 Set window function. Default is @code{hann}.
1257 Set window overlap. If set to 1, the recommended overlap for selected
1258 window function will be picked. Default is @code{0.75}.
1261 @subsection Examples
1265 Leave almost only low frequencies in audio:
1267 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1271 Apply robotize effect:
1273 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1277 Apply whisper effect:
1279 afftfilt="real='hypot(re,im)*cos((random(0)*2-1)*2*3.14)':imag='hypot(re,im)*sin((random(1)*2-1)*2*3.14)':win_size=128:overlap=0.8"
1286 Apply an arbitrary Finite Impulse Response filter.
1288 This filter is designed for applying long FIR filters,
1289 up to 60 seconds long.
1291 It can be used as component for digital crossover filters,
1292 room equalization, cross talk cancellation, wavefield synthesis,
1293 auralization, ambiophonics, ambisonics and spatialization.
1295 This filter uses the streams higher than first one as FIR coefficients.
1296 If the non-first stream holds a single channel, it will be used
1297 for all input channels in the first stream, otherwise
1298 the number of channels in the non-first stream must be same as
1299 the number of channels in the first stream.
1301 It accepts the following parameters:
1305 Set dry gain. This sets input gain.
1308 Set wet gain. This sets final output gain.
1311 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1314 Enable applying gain measured from power of IR.
1316 Set which approach to use for auto gain measurement.
1320 Do not apply any gain.
1323 select peak gain, very conservative approach. This is default value.
1326 select DC gain, limited application.
1329 select gain to noise approach, this is most popular one.
1333 Set gain to be applied to IR coefficients before filtering.
1334 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1337 Set format of IR stream. Can be @code{mono} or @code{input}.
1338 Default is @code{input}.
1341 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1342 Allowed range is 0.1 to 60 seconds.
1345 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1346 By default it is disabled.
1349 Set for which IR channel to display frequency response. By default is first channel
1350 displayed. This option is used only when @var{response} is enabled.
1353 Set video stream size. This option is used only when @var{response} is enabled.
1356 Set video stream frame rate. This option is used only when @var{response} is enabled.
1359 Set minimal partition size used for convolution. Default is @var{8192}.
1360 Allowed range is from @var{1} to @var{32768}.
1361 Lower values decreases latency at cost of higher CPU usage.
1364 Set maximal partition size used for convolution. Default is @var{8192}.
1365 Allowed range is from @var{8} to @var{32768}.
1366 Lower values may increase CPU usage.
1369 Set number of input impulse responses streams which will be switchable at runtime.
1370 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1373 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1374 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1375 This option can be changed at runtime via @ref{commands}.
1378 @subsection Examples
1382 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1384 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1391 Set output format constraints for the input audio. The framework will
1392 negotiate the most appropriate format to minimize conversions.
1394 It accepts the following parameters:
1397 @item sample_fmts, f
1398 A '|'-separated list of requested sample formats.
1400 @item sample_rates, r
1401 A '|'-separated list of requested sample rates.
1403 @item channel_layouts, cl
1404 A '|'-separated list of requested channel layouts.
1406 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1407 for the required syntax.
1410 If a parameter is omitted, all values are allowed.
1412 Force the output to either unsigned 8-bit or signed 16-bit stereo
1414 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1418 Apply frequency shift to input audio samples.
1420 The filter accepts the following options:
1424 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1425 Default value is 0.0.
1428 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
1429 Default value is 1.0.
1432 @subsection Commands
1434 This filter supports the all above options as @ref{commands}.
1438 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1439 processing reduces disturbing noise between useful signals.
1441 Gating is done by detecting the volume below a chosen level @var{threshold}
1442 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1443 floor is set via @var{range}. Because an exact manipulation of the signal
1444 would cause distortion of the waveform the reduction can be levelled over
1445 time. This is done by setting @var{attack} and @var{release}.
1447 @var{attack} determines how long the signal has to fall below the threshold
1448 before any reduction will occur and @var{release} sets the time the signal
1449 has to rise above the threshold to reduce the reduction again.
1450 Shorter signals than the chosen attack time will be left untouched.
1454 Set input level before filtering.
1455 Default is 1. Allowed range is from 0.015625 to 64.
1458 Set the mode of operation. Can be @code{upward} or @code{downward}.
1459 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1460 will be amplified, expanding dynamic range in upward direction.
1461 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1464 Set the level of gain reduction when the signal is below the threshold.
1465 Default is 0.06125. Allowed range is from 0 to 1.
1466 Setting this to 0 disables reduction and then filter behaves like expander.
1469 If a signal rises above this level the gain reduction is released.
1470 Default is 0.125. Allowed range is from 0 to 1.
1473 Set a ratio by which the signal is reduced.
1474 Default is 2. Allowed range is from 1 to 9000.
1477 Amount of milliseconds the signal has to rise above the threshold before gain
1479 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1482 Amount of milliseconds the signal has to fall below the threshold before the
1483 reduction is increased again. Default is 250 milliseconds.
1484 Allowed range is from 0.01 to 9000.
1487 Set amount of amplification of signal after processing.
1488 Default is 1. Allowed range is from 1 to 64.
1491 Curve the sharp knee around the threshold to enter gain reduction more softly.
1492 Default is 2.828427125. Allowed range is from 1 to 8.
1495 Choose if exact signal should be taken for detection or an RMS like one.
1496 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1499 Choose if the average level between all channels or the louder channel affects
1501 Default is @code{average}. Can be @code{average} or @code{maximum}.
1504 @subsection Commands
1506 This filter supports the all above options as @ref{commands}.
1510 Apply an arbitrary Infinite Impulse Response filter.
1512 It accepts the following parameters:
1516 Set B/numerator/zeros/reflection coefficients.
1519 Set A/denominator/poles/ladder coefficients.
1531 Set coefficients format.
1535 lattice-ladder function
1537 analog transfer function
1539 digital transfer function
1541 Z-plane zeros/poles, cartesian (default)
1543 Z-plane zeros/poles, polar radians
1545 Z-plane zeros/poles, polar degrees
1551 Set type of processing.
1563 Set filtering precision.
1567 double-precision floating-point (default)
1569 single-precision floating-point
1577 Normalize filter coefficients, by default is enabled.
1578 Enabling it will normalize magnitude response at DC to 0dB.
1581 How much to use filtered signal in output. Default is 1.
1582 Range is between 0 and 1.
1585 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1586 By default it is disabled.
1589 Set for which IR channel to display frequency response. By default is first channel
1590 displayed. This option is used only when @var{response} is enabled.
1593 Set video stream size. This option is used only when @var{response} is enabled.
1596 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1599 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1600 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1603 Different coefficients and gains can be provided for every channel, in such case
1604 use '|' to separate coefficients or gains. Last provided coefficients will be
1605 used for all remaining channels.
1607 @subsection Examples
1611 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1613 aiir=k=1:z=7.957584807809675810E-1 -2.575128568908332300 3.674839853930788710 -2.57512875289799137 7.957586296317130880E-1:p=1 -2.86950072432325953 3.63022088054647218 -2.28075678147272232 6.361362326477423500E-1:f=tf:r=d
1617 Same as above but in @code{zp} format:
1619 aiir=k=0.79575848078096756:z=0.80918701+0.58773007i 0.80918701-0.58773007i 0.80884700+0.58784055i 0.80884700-0.58784055i:p=0.63892345+0.59951235i 0.63892345-0.59951235i 0.79582691+0.44198673i 0.79582691-0.44198673i:f=zp:r=s
1623 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1625 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1631 The limiter prevents an input signal from rising over a desired threshold.
1632 This limiter uses lookahead technology to prevent your signal from distorting.
1633 It means that there is a small delay after the signal is processed. Keep in mind
1634 that the delay it produces is the attack time you set.
1636 The filter accepts the following options:
1640 Set input gain. Default is 1.
1643 Set output gain. Default is 1.
1646 Don't let signals above this level pass the limiter. Default is 1.
1649 The limiter will reach its attenuation level in this amount of time in
1650 milliseconds. Default is 5 milliseconds.
1653 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1654 Default is 50 milliseconds.
1657 When gain reduction is always needed ASC takes care of releasing to an
1658 average reduction level rather than reaching a reduction of 0 in the release
1662 Select how much the release time is affected by ASC, 0 means nearly no changes
1663 in release time while 1 produces higher release times.
1666 Auto level output signal. Default is enabled.
1667 This normalizes audio back to 0dB if enabled.
1670 Depending on picked setting it is recommended to upsample input 2x or 4x times
1671 with @ref{aresample} before applying this filter.
1675 Apply a two-pole all-pass filter with central frequency (in Hz)
1676 @var{frequency}, and filter-width @var{width}.
1677 An all-pass filter changes the audio's frequency to phase relationship
1678 without changing its frequency to amplitude relationship.
1680 The filter accepts the following options:
1684 Set frequency in Hz.
1687 Set method to specify band-width of filter.
1702 Specify the band-width of a filter in width_type units.
1705 How much to use filtered signal in output. Default is 1.
1706 Range is between 0 and 1.
1709 Specify which channels to filter, by default all available are filtered.
1712 Normalize biquad coefficients, by default is disabled.
1713 Enabling it will normalize magnitude response at DC to 0dB.
1716 Set the filter order, can be 1 or 2. Default is 2.
1719 Set transform type of IIR filter.
1728 Set precison of filtering.
1731 Pick automatic sample format depending on surround filters.
1733 Always use signed 16-bit.
1735 Always use signed 32-bit.
1737 Always use float 32-bit.
1739 Always use float 64-bit.
1743 @subsection Commands
1745 This filter supports the following commands:
1748 Change allpass frequency.
1749 Syntax for the command is : "@var{frequency}"
1752 Change allpass width_type.
1753 Syntax for the command is : "@var{width_type}"
1756 Change allpass width.
1757 Syntax for the command is : "@var{width}"
1761 Syntax for the command is : "@var{mix}"
1768 The filter accepts the following options:
1772 Set the number of loops. Setting this value to -1 will result in infinite loops.
1776 Set maximal number of samples. Default is 0.
1779 Set first sample of loop. Default is 0.
1785 Merge two or more audio streams into a single multi-channel stream.
1787 The filter accepts the following options:
1792 Set the number of inputs. Default is 2.
1796 If the channel layouts of the inputs are disjoint, and therefore compatible,
1797 the channel layout of the output will be set accordingly and the channels
1798 will be reordered as necessary. If the channel layouts of the inputs are not
1799 disjoint, the output will have all the channels of the first input then all
1800 the channels of the second input, in that order, and the channel layout of
1801 the output will be the default value corresponding to the total number of
1804 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1805 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1806 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1807 first input, b1 is the first channel of the second input).
1809 On the other hand, if both input are in stereo, the output channels will be
1810 in the default order: a1, a2, b1, b2, and the channel layout will be
1811 arbitrarily set to 4.0, which may or may not be the expected value.
1813 All inputs must have the same sample rate, and format.
1815 If inputs do not have the same duration, the output will stop with the
1818 @subsection Examples
1822 Merge two mono files into a stereo stream:
1824 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1828 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1830 ffmpeg -i input.mkv -filter_complex "[0:1][0:2][0:3][0:4][0:5][0:6] amerge=inputs=6" -c:a pcm_s16le output.mkv
1836 Mixes multiple audio inputs into a single output.
1838 Note that this filter only supports float samples (the @var{amerge}
1839 and @var{pan} audio filters support many formats). If the @var{amix}
1840 input has integer samples then @ref{aresample} will be automatically
1841 inserted to perform the conversion to float samples.
1845 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1847 will mix 3 input audio streams to a single output with the same duration as the
1848 first input and a dropout transition time of 3 seconds.
1850 It accepts the following parameters:
1854 The number of inputs. If unspecified, it defaults to 2.
1857 How to determine the end-of-stream.
1861 The duration of the longest input. (default)
1864 The duration of the shortest input.
1867 The duration of the first input.
1871 @item dropout_transition
1872 The transition time, in seconds, for volume renormalization when an input
1873 stream ends. The default value is 2 seconds.
1876 Specify weight of each input audio stream as sequence.
1877 Each weight is separated by space. By default all inputs have same weight.
1880 @subsection Commands
1882 This filter supports the following commands:
1885 Syntax is same as option with same name.
1890 Multiply first audio stream with second audio stream and store result
1891 in output audio stream. Multiplication is done by multiplying each
1892 sample from first stream with sample at same position from second stream.
1894 With this element-wise multiplication one can create amplitude fades and
1895 amplitude modulations.
1897 @section anequalizer
1899 High-order parametric multiband equalizer for each channel.
1901 It accepts the following parameters:
1905 This option string is in format:
1906 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1907 Each equalizer band is separated by '|'.
1911 Set channel number to which equalization will be applied.
1912 If input doesn't have that channel the entry is ignored.
1915 Set central frequency for band.
1916 If input doesn't have that frequency the entry is ignored.
1919 Set band width in Hertz.
1922 Set band gain in dB.
1925 Set filter type for band, optional, can be:
1929 Butterworth, this is default.
1940 With this option activated frequency response of anequalizer is displayed
1944 Set video stream size. Only useful if curves option is activated.
1947 Set max gain that will be displayed. Only useful if curves option is activated.
1948 Setting this to a reasonable value makes it possible to display gain which is derived from
1949 neighbour bands which are too close to each other and thus produce higher gain
1950 when both are activated.
1953 Set frequency scale used to draw frequency response in video output.
1954 Can be linear or logarithmic. Default is logarithmic.
1957 Set color for each channel curve which is going to be displayed in video stream.
1958 This is list of color names separated by space or by '|'.
1959 Unrecognised or missing colors will be replaced by white color.
1962 @subsection Examples
1966 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1967 for first 2 channels using Chebyshev type 1 filter:
1969 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1973 @subsection Commands
1975 This filter supports the following commands:
1978 Alter existing filter parameters.
1979 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1981 @var{fN} is existing filter number, starting from 0, if no such filter is available
1983 @var{freq} set new frequency parameter.
1984 @var{width} set new width parameter in Hertz.
1985 @var{gain} set new gain parameter in dB.
1987 Full filter invocation with asendcmd may look like this:
1988 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1993 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1995 Each sample is adjusted by looking for other samples with similar contexts. This
1996 context similarity is defined by comparing their surrounding patches of size
1997 @option{p}. Patches are searched in an area of @option{r} around the sample.
1999 The filter accepts the following options:
2003 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
2006 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
2007 Default value is 2 milliseconds.
2010 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
2011 Default value is 6 milliseconds.
2014 Set the output mode.
2016 It accepts the following values:
2019 Pass input unchanged.
2022 Pass noise filtered out.
2027 Default value is @var{o}.
2031 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
2034 @subsection Commands
2036 This filter supports the all above options as @ref{commands}.
2039 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
2041 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
2042 relate to producing the least mean square of the error signal (difference between the desired,
2043 2nd input audio stream and the actual signal, the 1st input audio stream).
2045 A description of the accepted options follows.
2058 Set the filter leakage.
2061 It accepts the following values:
2070 Pass filtered samples.
2073 Pass difference between desired and filtered samples.
2075 Default value is @var{o}.
2079 @subsection Examples
2083 One of many usages of this filter is noise reduction, input audio is filtered
2084 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2086 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2090 @subsection Commands
2092 This filter supports the same commands as options, excluding option @code{order}.
2096 Pass the audio source unchanged to the output.
2100 Pad the end of an audio stream with silence.
2102 This can be used together with @command{ffmpeg} @option{-shortest} to
2103 extend audio streams to the same length as the video stream.
2105 A description of the accepted options follows.
2109 Set silence packet size. Default value is 4096.
2112 Set the number of samples of silence to add to the end. After the
2113 value is reached, the stream is terminated. This option is mutually
2114 exclusive with @option{whole_len}.
2117 Set the minimum total number of samples in the output audio stream. If
2118 the value is longer than the input audio length, silence is added to
2119 the end, until the value is reached. This option is mutually exclusive
2120 with @option{pad_len}.
2123 Specify the duration of samples of silence to add. See
2124 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2125 for the accepted syntax. Used only if set to non-zero value.
2128 Specify the minimum total duration in the output audio stream. See
2129 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2130 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2131 the input audio length, silence is added to the end, until the value is reached.
2132 This option is mutually exclusive with @option{pad_dur}
2135 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2136 nor @option{whole_dur} option is set, the filter will add silence to the end of
2137 the input stream indefinitely.
2139 @subsection Examples
2143 Add 1024 samples of silence to the end of the input:
2149 Make sure the audio output will contain at least 10000 samples, pad
2150 the input with silence if required:
2152 apad=whole_len=10000
2156 Use @command{ffmpeg} to pad the audio input with silence, so that the
2157 video stream will always result the shortest and will be converted
2158 until the end in the output file when using the @option{shortest}
2161 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2166 Add a phasing effect to the input audio.
2168 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2169 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2171 A description of the accepted parameters follows.
2175 Set input gain. Default is 0.4.
2178 Set output gain. Default is 0.74
2181 Set delay in milliseconds. Default is 3.0.
2184 Set decay. Default is 0.4.
2187 Set modulation speed in Hz. Default is 0.5.
2190 Set modulation type. Default is triangular.
2192 It accepts the following values:
2199 @section aphaseshift
2200 Apply phase shift to input audio samples.
2202 The filter accepts the following options:
2206 Specify phase shift. Allowed range is from -1.0 to 1.0.
2207 Default value is 0.0.
2210 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
2211 Default value is 1.0.
2214 @subsection Commands
2216 This filter supports the all above options as @ref{commands}.
2220 Audio pulsator is something between an autopanner and a tremolo.
2221 But it can produce funny stereo effects as well. Pulsator changes the volume
2222 of the left and right channel based on a LFO (low frequency oscillator) with
2223 different waveforms and shifted phases.
2224 This filter have the ability to define an offset between left and right
2225 channel. An offset of 0 means that both LFO shapes match each other.
2226 The left and right channel are altered equally - a conventional tremolo.
2227 An offset of 50% means that the shape of the right channel is exactly shifted
2228 in phase (or moved backwards about half of the frequency) - pulsator acts as
2229 an autopanner. At 1 both curves match again. Every setting in between moves the
2230 phase shift gapless between all stages and produces some "bypassing" sounds with
2231 sine and triangle waveforms. The more you set the offset near 1 (starting from
2232 the 0.5) the faster the signal passes from the left to the right speaker.
2234 The filter accepts the following options:
2238 Set input gain. By default it is 1. Range is [0.015625 - 64].
2241 Set output gain. By default it is 1. Range is [0.015625 - 64].
2244 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2245 sawup or sawdown. Default is sine.
2248 Set modulation. Define how much of original signal is affected by the LFO.
2251 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2254 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2257 Set pulse width. Default is 1. Allowed range is [0 - 2].
2260 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2263 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2267 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2271 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2272 if timing is set to hz.
2278 Resample the input audio to the specified parameters, using the
2279 libswresample library. If none are specified then the filter will
2280 automatically convert between its input and output.
2282 This filter is also able to stretch/squeeze the audio data to make it match
2283 the timestamps or to inject silence / cut out audio to make it match the
2284 timestamps, do a combination of both or do neither.
2286 The filter accepts the syntax
2287 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2288 expresses a sample rate and @var{resampler_options} is a list of
2289 @var{key}=@var{value} pairs, separated by ":". See the
2290 @ref{Resampler Options,,"Resampler Options" section in the
2291 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2292 for the complete list of supported options.
2294 @subsection Examples
2298 Resample the input audio to 44100Hz:
2304 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2305 samples per second compensation:
2307 aresample=async=1000
2313 Reverse an audio clip.
2315 Warning: This filter requires memory to buffer the entire clip, so trimming
2318 @subsection Examples
2322 Take the first 5 seconds of a clip, and reverse it.
2324 atrim=end=5,areverse
2330 Reduce noise from speech using Recurrent Neural Networks.
2332 This filter accepts the following options:
2336 Set train model file to load. This option is always required.
2339 Set how much to mix filtered samples into final output.
2340 Allowed range is from -1 to 1. Default value is 1.
2341 Negative values are special, they set how much to keep filtered noise
2342 in the final filter output. Set this option to -1 to hear actual
2343 noise removed from input signal.
2346 @section asetnsamples
2348 Set the number of samples per each output audio frame.
2350 The last output packet may contain a different number of samples, as
2351 the filter will flush all the remaining samples when the input audio
2354 The filter accepts the following options:
2358 @item nb_out_samples, n
2359 Set the number of frames per each output audio frame. The number is
2360 intended as the number of samples @emph{per each channel}.
2361 Default value is 1024.
2364 If set to 1, the filter will pad the last audio frame with zeroes, so
2365 that the last frame will contain the same number of samples as the
2366 previous ones. Default value is 1.
2369 For example, to set the number of per-frame samples to 1234 and
2370 disable padding for the last frame, use:
2372 asetnsamples=n=1234:p=0
2377 Set the sample rate without altering the PCM data.
2378 This will result in a change of speed and pitch.
2380 The filter accepts the following options:
2383 @item sample_rate, r
2384 Set the output sample rate. Default is 44100 Hz.
2389 Show a line containing various information for each input audio frame.
2390 The input audio is not modified.
2392 The shown line contains a sequence of key/value pairs of the form
2393 @var{key}:@var{value}.
2395 The following values are shown in the output:
2399 The (sequential) number of the input frame, starting from 0.
2402 The presentation timestamp of the input frame, in time base units; the time base
2403 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2406 The presentation timestamp of the input frame in seconds.
2409 position of the frame in the input stream, -1 if this information in
2410 unavailable and/or meaningless (for example in case of synthetic audio)
2419 The sample rate for the audio frame.
2422 The number of samples (per channel) in the frame.
2425 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2426 audio, the data is treated as if all the planes were concatenated.
2428 @item plane_checksums
2429 A list of Adler-32 checksums for each data plane.
2433 Apply audio soft clipping.
2435 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2436 along a smooth curve, rather than the abrupt shape of hard-clipping.
2438 This filter accepts the following options:
2442 Set type of soft-clipping.
2444 It accepts the following values:
2458 Set threshold from where to start clipping. Default value is 0dB or 1.
2461 Set gain applied to output. Default value is 0dB or 1.
2464 Set additional parameter which controls sigmoid function.
2467 Set oversampling factor.
2470 @subsection Commands
2472 This filter supports the all above options as @ref{commands}.
2475 Automatic Speech Recognition
2477 This filter uses PocketSphinx for speech recognition. To enable
2478 compilation of this filter, you need to configure FFmpeg with
2479 @code{--enable-pocketsphinx}.
2481 It accepts the following options:
2485 Set sampling rate of input audio. Defaults is @code{16000}.
2486 This need to match speech models, otherwise one will get poor results.
2489 Set dictionary containing acoustic model files.
2492 Set pronunciation dictionary.
2495 Set language model file.
2498 Set language model set.
2501 Set which language model to use.
2504 Set output for log messages.
2507 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2512 Display time domain statistical information about the audio channels.
2513 Statistics are calculated and displayed for each audio channel and,
2514 where applicable, an overall figure is also given.
2516 It accepts the following option:
2519 Short window length in seconds, used for peak and trough RMS measurement.
2520 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2524 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2525 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2528 Available keys for each channel are:
2574 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2575 this @code{lavfi.astats.Overall.Peak_count}.
2577 For description what each key means read below.
2580 Set number of frame after which stats are going to be recalculated.
2581 Default is disabled.
2583 @item measure_perchannel
2584 Select the entries which need to be measured per channel. The metadata keys can
2585 be used as flags, default is @option{all} which measures everything.
2586 @option{none} disables all per channel measurement.
2588 @item measure_overall
2589 Select the entries which need to be measured overall. The metadata keys can
2590 be used as flags, default is @option{all} which measures everything.
2591 @option{none} disables all overall measurement.
2595 A description of each shown parameter follows:
2599 Mean amplitude displacement from zero.
2602 Minimal sample level.
2605 Maximal sample level.
2607 @item Min difference
2608 Minimal difference between two consecutive samples.
2610 @item Max difference
2611 Maximal difference between two consecutive samples.
2613 @item Mean difference
2614 Mean difference between two consecutive samples.
2615 The average of each difference between two consecutive samples.
2617 @item RMS difference
2618 Root Mean Square difference between two consecutive samples.
2622 Standard peak and RMS level measured in dBFS.
2626 Peak and trough values for RMS level measured over a short window.
2629 Standard ratio of peak to RMS level (note: not in dB).
2632 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2633 (i.e. either @var{Min level} or @var{Max level}).
2636 Number of occasions (not the number of samples) that the signal attained either
2637 @var{Min level} or @var{Max level}.
2639 @item Noise floor dB
2640 Minimum local peak measured in dBFS over a short window.
2642 @item Noise floor count
2643 Number of occasions (not the number of samples) that the signal attained
2647 Overall bit depth of audio. Number of bits used for each sample.
2650 Measured dynamic range of audio in dB.
2652 @item Zero crossings
2653 Number of points where the waveform crosses the zero level axis.
2655 @item Zero crossings rate
2656 Rate of Zero crossings and number of audio samples.
2660 Boost subwoofer frequencies.
2662 The filter accepts the following options:
2666 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2667 Default value is 0.7.
2670 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2671 Default value is 0.7.
2674 Set delay line decay gain value. Allowed range is from 0 to 1.
2675 Default value is 0.7.
2678 Set delay line feedback gain value. Allowed range is from 0 to 1.
2679 Default value is 0.9.
2682 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2683 Default value is 100.
2686 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2687 Default value is 0.5.
2690 Set delay. Allowed range is from 1 to 100.
2691 Default value is 20.
2694 @subsection Commands
2696 This filter supports the all above options as @ref{commands}.
2699 Cut subwoofer frequencies.
2701 This filter allows to set custom, steeper
2702 roll off than highpass filter, and thus is able to more attenuate
2703 frequency content in stop-band.
2705 The filter accepts the following options:
2709 Set cutoff frequency in Hertz. Allowed range is 2 to 200.
2710 Default value is 20.
2713 Set filter order. Available values are from 3 to 20.
2714 Default value is 10.
2717 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2720 @subsection Commands
2722 This filter supports the all above options as @ref{commands}.
2725 Cut super frequencies.
2727 The filter accepts the following options:
2731 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2732 Default value is 20000.
2735 Set filter order. Available values are from 3 to 20.
2736 Default value is 10.
2739 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2742 @subsection Commands
2744 This filter supports the all above options as @ref{commands}.
2747 Apply high order Butterworth band-pass filter.
2749 The filter accepts the following options:
2753 Set center frequency in Hertz. Allowed range is 2 to 999999.
2754 Default value is 1000.
2757 Set filter order. Available values are from 4 to 20.
2761 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2764 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2767 @subsection Commands
2769 This filter supports the all above options as @ref{commands}.
2772 Apply high order Butterworth band-stop filter.
2774 The filter accepts the following options:
2778 Set center frequency in Hertz. Allowed range is 2 to 999999.
2779 Default value is 1000.
2782 Set filter order. Available values are from 4 to 20.
2786 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2789 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2792 @subsection Commands
2794 This filter supports the all above options as @ref{commands}.
2800 The filter accepts exactly one parameter, the audio tempo. If not
2801 specified then the filter will assume nominal 1.0 tempo. Tempo must
2802 be in the [0.5, 100.0] range.
2804 Note that tempo greater than 2 will skip some samples rather than
2805 blend them in. If for any reason this is a concern it is always
2806 possible to daisy-chain several instances of atempo to achieve the
2807 desired product tempo.
2809 @subsection Examples
2813 Slow down audio to 80% tempo:
2819 To speed up audio to 300% tempo:
2825 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2827 atempo=sqrt(3),atempo=sqrt(3)
2831 @subsection Commands
2833 This filter supports the following commands:
2836 Change filter tempo scale factor.
2837 Syntax for the command is : "@var{tempo}"
2842 Trim the input so that the output contains one continuous subpart of the input.
2844 It accepts the following parameters:
2847 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2848 sample with the timestamp @var{start} will be the first sample in the output.
2851 Specify time of the first audio sample that will be dropped, i.e. the
2852 audio sample immediately preceding the one with the timestamp @var{end} will be
2853 the last sample in the output.
2856 Same as @var{start}, except this option sets the start timestamp in samples
2860 Same as @var{end}, except this option sets the end timestamp in samples instead
2864 The maximum duration of the output in seconds.
2867 The number of the first sample that should be output.
2870 The number of the first sample that should be dropped.
2873 @option{start}, @option{end}, and @option{duration} are expressed as time
2874 duration specifications; see
2875 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2877 Note that the first two sets of the start/end options and the @option{duration}
2878 option look at the frame timestamp, while the _sample options simply count the
2879 samples that pass through the filter. So start/end_pts and start/end_sample will
2880 give different results when the timestamps are wrong, inexact or do not start at
2881 zero. Also note that this filter does not modify the timestamps. If you wish
2882 to have the output timestamps start at zero, insert the asetpts filter after the
2885 If multiple start or end options are set, this filter tries to be greedy and
2886 keep all samples that match at least one of the specified constraints. To keep
2887 only the part that matches all the constraints at once, chain multiple atrim
2890 The defaults are such that all the input is kept. So it is possible to set e.g.
2891 just the end values to keep everything before the specified time.
2896 Drop everything except the second minute of input:
2898 ffmpeg -i INPUT -af atrim=60:120
2902 Keep only the first 1000 samples:
2904 ffmpeg -i INPUT -af atrim=end_sample=1000
2909 @section axcorrelate
2910 Calculate normalized cross-correlation between two input audio streams.
2912 Resulted samples are always between -1 and 1 inclusive.
2913 If result is 1 it means two input samples are highly correlated in that selected segment.
2914 Result 0 means they are not correlated at all.
2915 If result is -1 it means two input samples are out of phase, which means they cancel each
2918 The filter accepts the following options:
2922 Set size of segment over which cross-correlation is calculated.
2923 Default is 256. Allowed range is from 2 to 131072.
2926 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2927 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2928 are always zero and thus need much less calculations to make.
2929 This is generally not true, but is valid for typical audio streams.
2932 @subsection Examples
2936 Calculate correlation between channels in stereo audio stream:
2938 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2944 Apply a two-pole Butterworth band-pass filter with central
2945 frequency @var{frequency}, and (3dB-point) band-width width.
2946 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2947 instead of the default: constant 0dB peak gain.
2948 The filter roll off at 6dB per octave (20dB per decade).
2950 The filter accepts the following options:
2954 Set the filter's central frequency. Default is @code{3000}.
2957 Constant skirt gain if set to 1. Defaults to 0.
2960 Set method to specify band-width of filter.
2975 Specify the band-width of a filter in width_type units.
2978 How much to use filtered signal in output. Default is 1.
2979 Range is between 0 and 1.
2982 Specify which channels to filter, by default all available are filtered.
2985 Normalize biquad coefficients, by default is disabled.
2986 Enabling it will normalize magnitude response at DC to 0dB.
2989 Set transform type of IIR filter.
2998 Set precison of filtering.
3001 Pick automatic sample format depending on surround filters.
3003 Always use signed 16-bit.
3005 Always use signed 32-bit.
3007 Always use float 32-bit.
3009 Always use float 64-bit.
3013 @subsection Commands
3015 This filter supports the following commands:
3018 Change bandpass frequency.
3019 Syntax for the command is : "@var{frequency}"
3022 Change bandpass width_type.
3023 Syntax for the command is : "@var{width_type}"
3026 Change bandpass width.
3027 Syntax for the command is : "@var{width}"
3030 Change bandpass mix.
3031 Syntax for the command is : "@var{mix}"
3036 Apply a two-pole Butterworth band-reject filter with central
3037 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
3038 The filter roll off at 6dB per octave (20dB per decade).
3040 The filter accepts the following options:
3044 Set the filter's central frequency. Default is @code{3000}.
3047 Set method to specify band-width of filter.
3062 Specify the band-width of a filter in width_type units.
3065 How much to use filtered signal in output. Default is 1.
3066 Range is between 0 and 1.
3069 Specify which channels to filter, by default all available are filtered.
3072 Normalize biquad coefficients, by default is disabled.
3073 Enabling it will normalize magnitude response at DC to 0dB.
3076 Set transform type of IIR filter.
3085 Set precison of filtering.
3088 Pick automatic sample format depending on surround filters.
3090 Always use signed 16-bit.
3092 Always use signed 32-bit.
3094 Always use float 32-bit.
3096 Always use float 64-bit.
3100 @subsection Commands
3102 This filter supports the following commands:
3105 Change bandreject frequency.
3106 Syntax for the command is : "@var{frequency}"
3109 Change bandreject width_type.
3110 Syntax for the command is : "@var{width_type}"
3113 Change bandreject width.
3114 Syntax for the command is : "@var{width}"
3117 Change bandreject mix.
3118 Syntax for the command is : "@var{mix}"
3121 @section bass, lowshelf
3123 Boost or cut the bass (lower) frequencies of the audio using a two-pole
3124 shelving filter with a response similar to that of a standard
3125 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
3127 The filter accepts the following options:
3131 Give the gain at 0 Hz. Its useful range is about -20
3132 (for a large cut) to +20 (for a large boost).
3133 Beware of clipping when using a positive gain.
3136 Set the filter's central frequency and so can be used
3137 to extend or reduce the frequency range to be boosted or cut.
3138 The default value is @code{100} Hz.
3141 Set method to specify band-width of filter.
3156 Determine how steep is the filter's shelf transition.
3159 Set number of poles. Default is 2.
3162 How much to use filtered signal in output. Default is 1.
3163 Range is between 0 and 1.
3166 Specify which channels to filter, by default all available are filtered.
3169 Normalize biquad coefficients, by default is disabled.
3170 Enabling it will normalize magnitude response at DC to 0dB.
3173 Set transform type of IIR filter.
3182 Set precison of filtering.
3185 Pick automatic sample format depending on surround filters.
3187 Always use signed 16-bit.
3189 Always use signed 32-bit.
3191 Always use float 32-bit.
3193 Always use float 64-bit.
3197 @subsection Commands
3199 This filter supports the following commands:
3202 Change bass frequency.
3203 Syntax for the command is : "@var{frequency}"
3206 Change bass width_type.
3207 Syntax for the command is : "@var{width_type}"
3211 Syntax for the command is : "@var{width}"
3215 Syntax for the command is : "@var{gain}"
3219 Syntax for the command is : "@var{mix}"
3224 Apply a biquad IIR filter with the given coefficients.
3225 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3226 are the numerator and denominator coefficients respectively.
3227 and @var{channels}, @var{c} specify which channels to filter, by default all
3228 available are filtered.
3230 @subsection Commands
3232 This filter supports the following commands:
3240 Change biquad parameter.
3241 Syntax for the command is : "@var{value}"
3244 How much to use filtered signal in output. Default is 1.
3245 Range is between 0 and 1.
3248 Specify which channels to filter, by default all available are filtered.
3251 Normalize biquad coefficients, by default is disabled.
3252 Enabling it will normalize magnitude response at DC to 0dB.
3255 Set transform type of IIR filter.
3264 Set precison of filtering.
3267 Pick automatic sample format depending on surround filters.
3269 Always use signed 16-bit.
3271 Always use signed 32-bit.
3273 Always use float 32-bit.
3275 Always use float 64-bit.
3280 Bauer stereo to binaural transformation, which improves headphone listening of
3281 stereo audio records.
3283 To enable compilation of this filter you need to configure FFmpeg with
3284 @code{--enable-libbs2b}.
3286 It accepts the following parameters:
3290 Pre-defined crossfeed level.
3294 Default level (fcut=700, feed=50).
3297 Chu Moy circuit (fcut=700, feed=60).
3300 Jan Meier circuit (fcut=650, feed=95).
3305 Cut frequency (in Hz).
3314 Remap input channels to new locations.
3316 It accepts the following parameters:
3319 Map channels from input to output. The argument is a '|'-separated list of
3320 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3321 @var{in_channel} form. @var{in_channel} can be either the name of the input
3322 channel (e.g. FL for front left) or its index in the input channel layout.
3323 @var{out_channel} is the name of the output channel or its index in the output
3324 channel layout. If @var{out_channel} is not given then it is implicitly an
3325 index, starting with zero and increasing by one for each mapping.
3327 @item channel_layout
3328 The channel layout of the output stream.
3331 If no mapping is present, the filter will implicitly map input channels to
3332 output channels, preserving indices.
3334 @subsection Examples
3338 For example, assuming a 5.1+downmix input MOV file,
3340 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3342 will create an output WAV file tagged as stereo from the downmix channels of
3346 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3348 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3352 @section channelsplit
3354 Split each channel from an input audio stream into a separate output stream.
3356 It accepts the following parameters:
3358 @item channel_layout
3359 The channel layout of the input stream. The default is "stereo".
3361 A channel layout describing the channels to be extracted as separate output streams
3362 or "all" to extract each input channel as a separate stream. The default is "all".
3364 Choosing channels not present in channel layout in the input will result in an error.
3367 @subsection Examples
3371 For example, assuming a stereo input MP3 file,
3373 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3375 will create an output Matroska file with two audio streams, one containing only
3376 the left channel and the other the right channel.
3379 Split a 5.1 WAV file into per-channel files:
3381 ffmpeg -i in.wav -filter_complex
3382 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3383 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3384 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3389 Extract only LFE from a 5.1 WAV file:
3391 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3392 -map '[LFE]' lfe.wav
3397 Add a chorus effect to the audio.
3399 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3401 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3402 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3403 The modulation depth defines the range the modulated delay is played before or after
3404 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3405 sound tuned around the original one, like in a chorus where some vocals are slightly
3408 It accepts the following parameters:
3411 Set input gain. Default is 0.4.
3414 Set output gain. Default is 0.4.
3417 Set delays. A typical delay is around 40ms to 60ms.
3429 @subsection Examples
3435 chorus=0.7:0.9:55:0.4:0.25:2
3441 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3445 Fuller sounding chorus with three delays:
3447 chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3
3452 Compress or expand the audio's dynamic range.
3454 It accepts the following parameters:
3460 A list of times in seconds for each channel over which the instantaneous level
3461 of the input signal is averaged to determine its volume. @var{attacks} refers to
3462 increase of volume and @var{decays} refers to decrease of volume. For most
3463 situations, the attack time (response to the audio getting louder) should be
3464 shorter than the decay time, because the human ear is more sensitive to sudden
3465 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3466 a typical value for decay is 0.8 seconds.
3467 If specified number of attacks & decays is lower than number of channels, the last
3468 set attack/decay will be used for all remaining channels.
3471 A list of points for the transfer function, specified in dB relative to the
3472 maximum possible signal amplitude. Each key points list must be defined using
3473 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3474 @code{x0/y0 x1/y1 x2/y2 ....}
3476 The input values must be in strictly increasing order but the transfer function
3477 does not have to be monotonically rising. The point @code{0/0} is assumed but
3478 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3479 function are @code{-70/-70|-60/-20|1/0}.
3482 Set the curve radius in dB for all joints. It defaults to 0.01.
3485 Set the additional gain in dB to be applied at all points on the transfer
3486 function. This allows for easy adjustment of the overall gain.
3490 Set an initial volume, in dB, to be assumed for each channel when filtering
3491 starts. This permits the user to supply a nominal level initially, so that, for
3492 example, a very large gain is not applied to initial signal levels before the
3493 companding has begun to operate. A typical value for audio which is initially
3494 quiet is -90 dB. It defaults to 0.
3497 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3498 delayed before being fed to the volume adjuster. Specifying a delay
3499 approximately equal to the attack/decay times allows the filter to effectively
3500 operate in predictive rather than reactive mode. It defaults to 0.
3504 @subsection Examples
3508 Make music with both quiet and loud passages suitable for listening to in a
3511 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3514 Another example for audio with whisper and explosion parts:
3516 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3520 A noise gate for when the noise is at a lower level than the signal:
3522 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3526 Here is another noise gate, this time for when the noise is at a higher level
3527 than the signal (making it, in some ways, similar to squelch):
3529 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3533 2:1 compression starting at -6dB:
3535 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3539 2:1 compression starting at -9dB:
3541 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3545 2:1 compression starting at -12dB:
3547 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3551 2:1 compression starting at -18dB:
3553 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3557 3:1 compression starting at -15dB:
3559 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3565 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3571 compand=attacks=0:points=-80/-169|-54/-80|-49.5/-64.6|-41.1/-41.1|-25.8/-15|-10.8/-4.5|0/0|20/8.3
3575 Hard limiter at -6dB:
3577 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3581 Hard limiter at -12dB:
3583 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3587 Hard noise gate at -35 dB:
3589 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3595 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3599 @section compensationdelay
3601 Compensation Delay Line is a metric based delay to compensate differing
3602 positions of microphones or speakers.
3604 For example, you have recorded guitar with two microphones placed in
3605 different locations. Because the front of sound wave has fixed speed in
3606 normal conditions, the phasing of microphones can vary and depends on
3607 their location and interposition. The best sound mix can be achieved when
3608 these microphones are in phase (synchronized). Note that a distance of
3609 ~30 cm between microphones makes one microphone capture the signal in
3610 antiphase to the other microphone. That makes the final mix sound moody.
3611 This filter helps to solve phasing problems by adding different delays
3612 to each microphone track and make them synchronized.
3614 The best result can be reached when you take one track as base and
3615 synchronize other tracks one by one with it.
3616 Remember that synchronization/delay tolerance depends on sample rate, too.
3617 Higher sample rates will give more tolerance.
3619 The filter accepts the following parameters:
3623 Set millimeters distance. This is compensation distance for fine tuning.
3627 Set cm distance. This is compensation distance for tightening distance setup.
3631 Set meters distance. This is compensation distance for hard distance setup.
3635 Set dry amount. Amount of unprocessed (dry) signal.
3639 Set wet amount. Amount of processed (wet) signal.
3643 Set temperature in degrees Celsius. This is the temperature of the environment.
3648 Apply headphone crossfeed filter.
3650 Crossfeed is the process of blending the left and right channels of stereo
3652 It is mainly used to reduce extreme stereo separation of low frequencies.
3654 The intent is to produce more speaker like sound to the listener.
3656 The filter accepts the following options:
3660 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3661 This sets gain of low shelf filter for side part of stereo image.
3662 Default is -6dB. Max allowed is -30db when strength is set to 1.
3665 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3666 This sets cut off frequency of low shelf filter. Default is cut off near
3667 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3670 Set curve slope of low shelf filter. Default is 0.5.
3671 Allowed range is from 0.01 to 1.
3674 Set input gain. Default is 0.9.
3677 Set output gain. Default is 1.
3680 @subsection Commands
3682 This filter supports the all above options as @ref{commands}.
3684 @section crystalizer
3685 Simple algorithm for audio noise sharpening.
3687 This filter linearly increases differences betweeen each audio sample.
3689 The filter accepts the following options:
3693 Sets the intensity of effect (default: 2.0). Must be in range between -10.0 to 0
3694 (unchanged sound) to 10.0 (maximum effect).
3695 To inverse filtering use negative value.
3698 Enable clipping. By default is enabled.
3701 @subsection Commands
3703 This filter supports the all above options as @ref{commands}.
3706 Apply a DC shift to the audio.
3708 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3709 in the recording chain) from the audio. The effect of a DC offset is reduced
3710 headroom and hence volume. The @ref{astats} filter can be used to determine if
3711 a signal has a DC offset.
3715 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3719 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3720 used to prevent clipping.
3725 Apply de-essing to the audio samples.
3729 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3733 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3737 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3741 Set the output mode.
3743 It accepts the following values:
3746 Pass input unchanged.
3749 Pass ess filtered out.
3754 Default value is @var{o}.
3760 Measure audio dynamic range.
3762 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3763 is found in transition material. And anything less that 8 have very poor dynamics
3764 and is very compressed.
3766 The filter accepts the following options:
3770 Set window length in seconds used to split audio into segments of equal length.
3771 Default is 3 seconds.
3775 Dynamic Audio Normalizer.
3777 This filter applies a certain amount of gain to the input audio in order
3778 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3779 contrast to more "simple" normalization algorithms, the Dynamic Audio
3780 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3781 This allows for applying extra gain to the "quiet" sections of the audio
3782 while avoiding distortions or clipping the "loud" sections. In other words:
3783 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3784 sections, in the sense that the volume of each section is brought to the
3785 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3786 this goal *without* applying "dynamic range compressing". It will retain 100%
3787 of the dynamic range *within* each section of the audio file.
3791 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3792 Default is 500 milliseconds.
3793 The Dynamic Audio Normalizer processes the input audio in small chunks,
3794 referred to as frames. This is required, because a peak magnitude has no
3795 meaning for just a single sample value. Instead, we need to determine the
3796 peak magnitude for a contiguous sequence of sample values. While a "standard"
3797 normalizer would simply use the peak magnitude of the complete file, the
3798 Dynamic Audio Normalizer determines the peak magnitude individually for each
3799 frame. The length of a frame is specified in milliseconds. By default, the
3800 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3801 been found to give good results with most files.
3802 Note that the exact frame length, in number of samples, will be determined
3803 automatically, based on the sampling rate of the individual input audio file.
3806 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3807 number. Default is 31.
3808 Probably the most important parameter of the Dynamic Audio Normalizer is the
3809 @code{window size} of the Gaussian smoothing filter. The filter's window size
3810 is specified in frames, centered around the current frame. For the sake of
3811 simplicity, this must be an odd number. Consequently, the default value of 31
3812 takes into account the current frame, as well as the 15 preceding frames and
3813 the 15 subsequent frames. Using a larger window results in a stronger
3814 smoothing effect and thus in less gain variation, i.e. slower gain
3815 adaptation. Conversely, using a smaller window results in a weaker smoothing
3816 effect and thus in more gain variation, i.e. faster gain adaptation.
3817 In other words, the more you increase this value, the more the Dynamic Audio
3818 Normalizer will behave like a "traditional" normalization filter. On the
3819 contrary, the more you decrease this value, the more the Dynamic Audio
3820 Normalizer will behave like a dynamic range compressor.
3823 Set the target peak value. This specifies the highest permissible magnitude
3824 level for the normalized audio input. This filter will try to approach the
3825 target peak magnitude as closely as possible, but at the same time it also
3826 makes sure that the normalized signal will never exceed the peak magnitude.
3827 A frame's maximum local gain factor is imposed directly by the target peak
3828 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3829 It is not recommended to go above this value.
3832 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3833 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3834 factor for each input frame, i.e. the maximum gain factor that does not
3835 result in clipping or distortion. The maximum gain factor is determined by
3836 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3837 additionally bounds the frame's maximum gain factor by a predetermined
3838 (global) maximum gain factor. This is done in order to avoid excessive gain
3839 factors in "silent" or almost silent frames. By default, the maximum gain
3840 factor is 10.0, For most inputs the default value should be sufficient and
3841 it usually is not recommended to increase this value. Though, for input
3842 with an extremely low overall volume level, it may be necessary to allow even
3843 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3844 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3845 Instead, a "sigmoid" threshold function will be applied. This way, the
3846 gain factors will smoothly approach the threshold value, but never exceed that
3850 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3851 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3852 This means that the maximum local gain factor for each frame is defined
3853 (only) by the frame's highest magnitude sample. This way, the samples can
3854 be amplified as much as possible without exceeding the maximum signal
3855 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3856 Normalizer can also take into account the frame's root mean square,
3857 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3858 determine the power of a time-varying signal. It is therefore considered
3859 that the RMS is a better approximation of the "perceived loudness" than
3860 just looking at the signal's peak magnitude. Consequently, by adjusting all
3861 frames to a constant RMS value, a uniform "perceived loudness" can be
3862 established. If a target RMS value has been specified, a frame's local gain
3863 factor is defined as the factor that would result in exactly that RMS value.
3864 Note, however, that the maximum local gain factor is still restricted by the
3865 frame's highest magnitude sample, in order to prevent clipping.
3868 Enable channels coupling. By default is enabled.
3869 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3870 amount. This means the same gain factor will be applied to all channels, i.e.
3871 the maximum possible gain factor is determined by the "loudest" channel.
3872 However, in some recordings, it may happen that the volume of the different
3873 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3874 In this case, this option can be used to disable the channel coupling. This way,
3875 the gain factor will be determined independently for each channel, depending
3876 only on the individual channel's highest magnitude sample. This allows for
3877 harmonizing the volume of the different channels.
3880 Enable DC bias correction. By default is disabled.
3881 An audio signal (in the time domain) is a sequence of sample values.
3882 In the Dynamic Audio Normalizer these sample values are represented in the
3883 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3884 audio signal, or "waveform", should be centered around the zero point.
3885 That means if we calculate the mean value of all samples in a file, or in a
3886 single frame, then the result should be 0.0 or at least very close to that
3887 value. If, however, there is a significant deviation of the mean value from
3888 0.0, in either positive or negative direction, this is referred to as a
3889 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3890 Audio Normalizer provides optional DC bias correction.
3891 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3892 the mean value, or "DC correction" offset, of each input frame and subtract
3893 that value from all of the frame's sample values which ensures those samples
3894 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3895 boundaries, the DC correction offset values will be interpolated smoothly
3896 between neighbouring frames.
3898 @item altboundary, b
3899 Enable alternative boundary mode. By default is disabled.
3900 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3901 around each frame. This includes the preceding frames as well as the
3902 subsequent frames. However, for the "boundary" frames, located at the very
3903 beginning and at the very end of the audio file, not all neighbouring
3904 frames are available. In particular, for the first few frames in the audio
3905 file, the preceding frames are not known. And, similarly, for the last few
3906 frames in the audio file, the subsequent frames are not known. Thus, the
3907 question arises which gain factors should be assumed for the missing frames
3908 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3909 to deal with this situation. The default boundary mode assumes a gain factor
3910 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3911 "fade out" at the beginning and at the end of the input, respectively.
3914 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3915 By default, the Dynamic Audio Normalizer does not apply "traditional"
3916 compression. This means that signal peaks will not be pruned and thus the
3917 full dynamic range will be retained within each local neighbourhood. However,
3918 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3919 normalization algorithm with a more "traditional" compression.
3920 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3921 (thresholding) function. If (and only if) the compression feature is enabled,
3922 all input frames will be processed by a soft knee thresholding function prior
3923 to the actual normalization process. Put simply, the thresholding function is
3924 going to prune all samples whose magnitude exceeds a certain threshold value.
3925 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3926 value. Instead, the threshold value will be adjusted for each individual
3928 In general, smaller parameters result in stronger compression, and vice versa.
3929 Values below 3.0 are not recommended, because audible distortion may appear.
3932 Set the target threshold value. This specifies the lowest permissible
3933 magnitude level for the audio input which will be normalized.
3934 If input frame volume is above this value frame will be normalized.
3935 Otherwise frame may not be normalized at all. The default value is set
3936 to 0, which means all input frames will be normalized.
3937 This option is mostly useful if digital noise is not wanted to be amplified.
3940 @subsection Commands
3942 This filter supports the all above options as @ref{commands}.
3946 Make audio easier to listen to on headphones.
3948 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3949 so that when listened to on headphones the stereo image is moved from
3950 inside your head (standard for headphones) to outside and in front of
3951 the listener (standard for speakers).
3957 Apply a two-pole peaking equalisation (EQ) filter. With this
3958 filter, the signal-level at and around a selected frequency can
3959 be increased or decreased, whilst (unlike bandpass and bandreject
3960 filters) that at all other frequencies is unchanged.
3962 In order to produce complex equalisation curves, this filter can
3963 be given several times, each with a different central frequency.
3965 The filter accepts the following options:
3969 Set the filter's central frequency in Hz.
3972 Set method to specify band-width of filter.
3987 Specify the band-width of a filter in width_type units.
3990 Set the required gain or attenuation in dB.
3991 Beware of clipping when using a positive gain.
3994 How much to use filtered signal in output. Default is 1.
3995 Range is between 0 and 1.
3998 Specify which channels to filter, by default all available are filtered.
4001 Normalize biquad coefficients, by default is disabled.
4002 Enabling it will normalize magnitude response at DC to 0dB.
4005 Set transform type of IIR filter.
4014 Set precison of filtering.
4017 Pick automatic sample format depending on surround filters.
4019 Always use signed 16-bit.
4021 Always use signed 32-bit.
4023 Always use float 32-bit.
4025 Always use float 64-bit.
4029 @subsection Examples
4032 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
4034 equalizer=f=1000:t=h:width=200:g=-10
4038 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
4040 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
4044 @subsection Commands
4046 This filter supports the following commands:
4049 Change equalizer frequency.
4050 Syntax for the command is : "@var{frequency}"
4053 Change equalizer width_type.
4054 Syntax for the command is : "@var{width_type}"
4057 Change equalizer width.
4058 Syntax for the command is : "@var{width}"
4061 Change equalizer gain.
4062 Syntax for the command is : "@var{gain}"
4065 Change equalizer mix.
4066 Syntax for the command is : "@var{mix}"
4069 @section extrastereo
4071 Linearly increases the difference between left and right channels which
4072 adds some sort of "live" effect to playback.
4074 The filter accepts the following options:
4078 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
4079 (average of both channels), with 1.0 sound will be unchanged, with
4080 -1.0 left and right channels will be swapped.
4083 Enable clipping. By default is enabled.
4086 @subsection Commands
4088 This filter supports the all above options as @ref{commands}.
4090 @section firequalizer
4091 Apply FIR Equalization using arbitrary frequency response.
4093 The filter accepts the following option:
4097 Set gain curve equation (in dB). The expression can contain variables:
4100 the evaluated frequency
4104 channel number, set to 0 when multichannels evaluation is disabled
4106 channel id, see libavutil/channel_layout.h, set to the first channel id when
4107 multichannels evaluation is disabled
4111 channel_layout, see libavutil/channel_layout.h
4116 @item gain_interpolate(f)
4117 interpolate gain on frequency f based on gain_entry
4118 @item cubic_interpolate(f)
4119 same as gain_interpolate, but smoother
4121 This option is also available as command. Default is @code{gain_interpolate(f)}.
4124 Set gain entry for gain_interpolate function. The expression can
4128 store gain entry at frequency f with value g
4130 This option is also available as command.
4133 Set filter delay in seconds. Higher value means more accurate.
4134 Default is @code{0.01}.
4137 Set filter accuracy in Hz. Lower value means more accurate.
4138 Default is @code{5}.
4141 Set window function. Acceptable values are:
4144 rectangular window, useful when gain curve is already smooth
4146 hann window (default)
4152 3-terms continuous 1st derivative nuttall window
4154 minimum 3-terms discontinuous nuttall window
4156 4-terms continuous 1st derivative nuttall window
4158 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4160 blackman-harris window
4166 If enabled, use fixed number of audio samples. This improves speed when
4167 filtering with large delay. Default is disabled.
4170 Enable multichannels evaluation on gain. Default is disabled.
4173 Enable zero phase mode by subtracting timestamp to compensate delay.
4174 Default is disabled.
4177 Set scale used by gain. Acceptable values are:
4180 linear frequency, linear gain
4182 linear frequency, logarithmic (in dB) gain (default)
4184 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
4186 logarithmic frequency, logarithmic gain
4190 Set file for dumping, suitable for gnuplot.
4193 Set scale for dumpfile. Acceptable values are same with scale option.
4197 Enable 2-channel convolution using complex FFT. This improves speed significantly.
4198 Default is disabled.
4201 Enable minimum phase impulse response. Default is disabled.
4204 @subsection Examples
4209 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4212 lowpass at 1000 Hz with gain_entry:
4214 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4217 custom equalization:
4219 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4222 higher delay with zero phase to compensate delay:
4224 firequalizer=delay=0.1:fixed=on:zero_phase=on
4227 lowpass on left channel, highpass on right channel:
4229 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4230 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4235 Apply a flanging effect to the audio.
4237 The filter accepts the following options:
4241 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4244 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4247 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4251 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4252 Default value is 71.
4255 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4258 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4259 Default value is @var{sinusoidal}.
4262 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4263 Default value is 25.
4266 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4267 Default is @var{linear}.
4271 Apply Haas effect to audio.
4273 Note that this makes most sense to apply on mono signals.
4274 With this filter applied to mono signals it give some directionality and
4275 stretches its stereo image.
4277 The filter accepts the following options:
4281 Set input level. By default is @var{1}, or 0dB
4284 Set output level. By default is @var{1}, or 0dB.
4287 Set gain applied to side part of signal. By default is @var{1}.
4290 Set kind of middle source. Can be one of the following:
4300 Pick middle part signal of stereo image.
4303 Pick side part signal of stereo image.
4307 Change middle phase. By default is disabled.
4310 Set left channel delay. By default is @var{2.05} milliseconds.
4313 Set left channel balance. By default is @var{-1}.
4316 Set left channel gain. By default is @var{1}.
4319 Change left phase. By default is disabled.
4322 Set right channel delay. By defaults is @var{2.12} milliseconds.
4325 Set right channel balance. By default is @var{1}.
4328 Set right channel gain. By default is @var{1}.
4331 Change right phase. By default is enabled.
4336 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4337 embedded HDCD codes is expanded into a 20-bit PCM stream.
4339 The filter supports the Peak Extend and Low-level Gain Adjustment features
4340 of HDCD, and detects the Transient Filter flag.
4343 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4346 When using the filter with wav, note the default encoding for wav is 16-bit,
4347 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4348 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4350 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4351 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4354 The filter accepts the following options:
4357 @item disable_autoconvert
4358 Disable any automatic format conversion or resampling in the filter graph.
4360 @item process_stereo
4361 Process the stereo channels together. If target_gain does not match between
4362 channels, consider it invalid and use the last valid target_gain.
4365 Set the code detect timer period in ms.
4368 Always extend peaks above -3dBFS even if PE isn't signaled.
4371 Replace audio with a solid tone and adjust the amplitude to signal some
4372 specific aspect of the decoding process. The output file can be loaded in
4373 an audio editor alongside the original to aid analysis.
4375 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4382 Gain adjustment level at each sample
4384 Samples where peak extend occurs
4386 Samples where the code detect timer is active
4388 Samples where the target gain does not match between channels
4394 Apply head-related transfer functions (HRTFs) to create virtual
4395 loudspeakers around the user for binaural listening via headphones.
4396 The HRIRs are provided via additional streams, for each channel
4397 one stereo input stream is needed.
4399 The filter accepts the following options:
4403 Set mapping of input streams for convolution.
4404 The argument is a '|'-separated list of channel names in order as they
4405 are given as additional stream inputs for filter.
4406 This also specify number of input streams. Number of input streams
4407 must be not less than number of channels in first stream plus one.
4410 Set gain applied to audio. Value is in dB. Default is 0.
4413 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4414 processing audio in time domain which is slow.
4415 @var{freq} is processing audio in frequency domain which is fast.
4416 Default is @var{freq}.
4419 Set custom gain for LFE channels. Value is in dB. Default is 0.
4422 Set size of frame in number of samples which will be processed at once.
4423 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4426 Set format of hrir stream.
4427 Default value is @var{stereo}. Alternative value is @var{multich}.
4428 If value is set to @var{stereo}, number of additional streams should
4429 be greater or equal to number of input channels in first input stream.
4430 Also each additional stream should have stereo number of channels.
4431 If value is set to @var{multich}, number of additional streams should
4432 be exactly one. Also number of input channels of additional stream
4433 should be equal or greater than twice number of channels of first input
4437 @subsection Examples
4441 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4442 each amovie filter use stereo file with IR coefficients as input.
4443 The files give coefficients for each position of virtual loudspeaker:
4446 -filter_complex "amovie=azi_270_ele_0_DFC.wav[sr];amovie=azi_90_ele_0_DFC.wav[sl];amovie=azi_225_ele_0_DFC.wav[br];amovie=azi_135_ele_0_DFC.wav[bl];amovie=azi_0_ele_0_DFC.wav,asplit[fc][lfe];amovie=azi_35_ele_0_DFC.wav[fl];amovie=azi_325_ele_0_DFC.wav[fr];[0:a][fl][fr][fc][lfe][bl][br][sl][sr]headphone=FL|FR|FC|LFE|BL|BR|SL|SR"
4451 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4452 but now in @var{multich} @var{hrir} format.
4454 ffmpeg -i input.wav -filter_complex "amovie=minp.wav[hrirs];[0:a][hrirs]headphone=map=FL|FR|FC|LFE|BL|BR|SL|SR:hrir=multich"
4461 Apply a high-pass filter with 3dB point frequency.
4462 The filter can be either single-pole, or double-pole (the default).
4463 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4465 The filter accepts the following options:
4469 Set frequency in Hz. Default is 3000.
4472 Set number of poles. Default is 2.
4475 Set method to specify band-width of filter.
4490 Specify the band-width of a filter in width_type units.
4491 Applies only to double-pole filter.
4492 The default is 0.707q and gives a Butterworth response.
4495 How much to use filtered signal in output. Default is 1.
4496 Range is between 0 and 1.
4499 Specify which channels to filter, by default all available are filtered.
4502 Normalize biquad coefficients, by default is disabled.
4503 Enabling it will normalize magnitude response at DC to 0dB.
4506 Set transform type of IIR filter.
4515 Set precison of filtering.
4518 Pick automatic sample format depending on surround filters.
4520 Always use signed 16-bit.
4522 Always use signed 32-bit.
4524 Always use float 32-bit.
4526 Always use float 64-bit.
4530 @subsection Commands
4532 This filter supports the following commands:
4535 Change highpass frequency.
4536 Syntax for the command is : "@var{frequency}"
4539 Change highpass width_type.
4540 Syntax for the command is : "@var{width_type}"
4543 Change highpass width.
4544 Syntax for the command is : "@var{width}"
4547 Change highpass mix.
4548 Syntax for the command is : "@var{mix}"
4553 Join multiple input streams into one multi-channel stream.
4555 It accepts the following parameters:
4559 The number of input streams. It defaults to 2.
4561 @item channel_layout
4562 The desired output channel layout. It defaults to stereo.
4565 Map channels from inputs to output. The argument is a '|'-separated list of
4566 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4567 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4568 can be either the name of the input channel (e.g. FL for front left) or its
4569 index in the specified input stream. @var{out_channel} is the name of the output
4573 The filter will attempt to guess the mappings when they are not specified
4574 explicitly. It does so by first trying to find an unused matching input channel
4575 and if that fails it picks the first unused input channel.
4577 Join 3 inputs (with properly set channel layouts):
4579 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4582 Build a 5.1 output from 6 single-channel streams:
4584 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4585 'join=inputs=6:channel_layout=5.1:map=0.0-FL|1.0-FR|2.0-FC|3.0-SL|4.0-SR|5.0-LFE'
4591 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4593 To enable compilation of this filter you need to configure FFmpeg with
4594 @code{--enable-ladspa}.
4598 Specifies the name of LADSPA plugin library to load. If the environment
4599 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4600 each one of the directories specified by the colon separated list in
4601 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4602 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4603 @file{/usr/lib/ladspa/}.
4606 Specifies the plugin within the library. Some libraries contain only
4607 one plugin, but others contain many of them. If this is not set filter
4608 will list all available plugins within the specified library.
4611 Set the '|' separated list of controls which are zero or more floating point
4612 values that determine the behavior of the loaded plugin (for example delay,
4614 Controls need to be defined using the following syntax:
4615 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4616 @var{valuei} is the value set on the @var{i}-th control.
4617 Alternatively they can be also defined using the following syntax:
4618 @var{value0}|@var{value1}|@var{value2}|..., where
4619 @var{valuei} is the value set on the @var{i}-th control.
4620 If @option{controls} is set to @code{help}, all available controls and
4621 their valid ranges are printed.
4623 @item sample_rate, s
4624 Specify the sample rate, default to 44100. Only used if plugin have
4628 Set the number of samples per channel per each output frame, default
4629 is 1024. Only used if plugin have zero inputs.
4632 Set the minimum duration of the sourced audio. See
4633 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4634 for the accepted syntax.
4635 Note that the resulting duration may be greater than the specified duration,
4636 as the generated audio is always cut at the end of a complete frame.
4637 If not specified, or the expressed duration is negative, the audio is
4638 supposed to be generated forever.
4639 Only used if plugin have zero inputs.
4642 Enable latency compensation, by default is disabled.
4643 Only used if plugin have inputs.
4646 @subsection Examples
4650 List all available plugins within amp (LADSPA example plugin) library:
4656 List all available controls and their valid ranges for @code{vcf_notch}
4657 plugin from @code{VCF} library:
4659 ladspa=f=vcf:p=vcf_notch:c=help
4663 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4666 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4670 Add reverberation to the audio using TAP-plugins
4671 (Tom's Audio Processing plugins):
4673 ladspa=file=tap_reverb:tap_reverb
4677 Generate white noise, with 0.2 amplitude:
4679 ladspa=file=cmt:noise_source_white:c=c0=.2
4683 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4684 @code{C* Audio Plugin Suite} (CAPS) library:
4686 ladspa=file=caps:Click:c=c1=20'
4690 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4692 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4696 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4697 @code{SWH Plugins} collection:
4699 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4703 Attenuate low frequencies using Multiband EQ from Steve Harris
4704 @code{SWH Plugins} collection:
4706 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4710 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4713 ladspa=caps:Narrower
4717 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4719 ladspa=caps:White:.2
4723 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4725 ladspa=caps:Fractal:c=c1=1
4729 Dynamic volume normalization using @code{VLevel} plugin:
4731 ladspa=vlevel-ladspa:vlevel_mono
4735 @subsection Commands
4737 This filter supports the following commands:
4740 Modify the @var{N}-th control value.
4742 If the specified value is not valid, it is ignored and prior one is kept.
4747 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4748 Support for both single pass (livestreams, files) and double pass (files) modes.
4749 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4750 detect true peaks, the audio stream will be upsampled to 192 kHz.
4751 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4753 The filter accepts the following options:
4757 Set integrated loudness target.
4758 Range is -70.0 - -5.0. Default value is -24.0.
4761 Set loudness range target.
4762 Range is 1.0 - 20.0. Default value is 7.0.
4765 Set maximum true peak.
4766 Range is -9.0 - +0.0. Default value is -2.0.
4768 @item measured_I, measured_i
4769 Measured IL of input file.
4770 Range is -99.0 - +0.0.
4772 @item measured_LRA, measured_lra
4773 Measured LRA of input file.
4774 Range is 0.0 - 99.0.
4776 @item measured_TP, measured_tp
4777 Measured true peak of input file.
4778 Range is -99.0 - +99.0.
4780 @item measured_thresh
4781 Measured threshold of input file.
4782 Range is -99.0 - +0.0.
4785 Set offset gain. Gain is applied before the true-peak limiter.
4786 Range is -99.0 - +99.0. Default is +0.0.
4789 Normalize by linearly scaling the source audio.
4790 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4791 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4792 be lower than source LRA and the change in integrated loudness shouldn't
4793 result in a true peak which exceeds the target TP. If any of these
4794 conditions aren't met, normalization mode will revert to @var{dynamic}.
4795 Options are @code{true} or @code{false}. Default is @code{true}.
4798 Treat mono input files as "dual-mono". If a mono file is intended for playback
4799 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4800 If set to @code{true}, this option will compensate for this effect.
4801 Multi-channel input files are not affected by this option.
4802 Options are true or false. Default is false.
4805 Set print format for stats. Options are summary, json, or none.
4806 Default value is none.
4811 Apply a low-pass filter with 3dB point frequency.
4812 The filter can be either single-pole or double-pole (the default).
4813 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4815 The filter accepts the following options:
4819 Set frequency in Hz. Default is 500.
4822 Set number of poles. Default is 2.
4825 Set method to specify band-width of filter.
4840 Specify the band-width of a filter in width_type units.
4841 Applies only to double-pole filter.
4842 The default is 0.707q and gives a Butterworth response.
4845 How much to use filtered signal in output. Default is 1.
4846 Range is between 0 and 1.
4849 Specify which channels to filter, by default all available are filtered.
4852 Normalize biquad coefficients, by default is disabled.
4853 Enabling it will normalize magnitude response at DC to 0dB.
4856 Set transform type of IIR filter.
4865 Set precison of filtering.
4868 Pick automatic sample format depending on surround filters.
4870 Always use signed 16-bit.
4872 Always use signed 32-bit.
4874 Always use float 32-bit.
4876 Always use float 64-bit.
4880 @subsection Examples
4883 Lowpass only LFE channel, it LFE is not present it does nothing:
4889 @subsection Commands
4891 This filter supports the following commands:
4894 Change lowpass frequency.
4895 Syntax for the command is : "@var{frequency}"
4898 Change lowpass width_type.
4899 Syntax for the command is : "@var{width_type}"
4902 Change lowpass width.
4903 Syntax for the command is : "@var{width}"
4907 Syntax for the command is : "@var{mix}"
4912 Load a LV2 (LADSPA Version 2) plugin.
4914 To enable compilation of this filter you need to configure FFmpeg with
4915 @code{--enable-lv2}.
4919 Specifies the plugin URI. You may need to escape ':'.
4922 Set the '|' separated list of controls which are zero or more floating point
4923 values that determine the behavior of the loaded plugin (for example delay,
4925 If @option{controls} is set to @code{help}, all available controls and
4926 their valid ranges are printed.
4928 @item sample_rate, s
4929 Specify the sample rate, default to 44100. Only used if plugin have
4933 Set the number of samples per channel per each output frame, default
4934 is 1024. Only used if plugin have zero inputs.
4937 Set the minimum duration of the sourced audio. See
4938 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4939 for the accepted syntax.
4940 Note that the resulting duration may be greater than the specified duration,
4941 as the generated audio is always cut at the end of a complete frame.
4942 If not specified, or the expressed duration is negative, the audio is
4943 supposed to be generated forever.
4944 Only used if plugin have zero inputs.
4947 @subsection Examples
4951 Apply bass enhancer plugin from Calf:
4953 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4957 Apply vinyl plugin from Calf:
4959 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4963 Apply bit crusher plugin from ArtyFX:
4965 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4970 Multiband Compress or expand the audio's dynamic range.
4972 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4973 This is akin to the crossover of a loudspeaker, and results in flat frequency
4974 response when absent compander action.
4976 It accepts the following parameters:
4980 This option syntax is:
4981 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4982 For explanation of each item refer to compand filter documentation.
4988 Mix channels with specific gain levels. The filter accepts the output
4989 channel layout followed by a set of channels definitions.
4991 This filter is also designed to efficiently remap the channels of an audio
4994 The filter accepts parameters of the form:
4995 "@var{l}|@var{outdef}|@var{outdef}|..."
4999 output channel layout or number of channels
5002 output channel specification, of the form:
5003 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
5006 output channel to define, either a channel name (FL, FR, etc.) or a channel
5007 number (c0, c1, etc.)
5010 multiplicative coefficient for the channel, 1 leaving the volume unchanged
5013 input channel to use, see out_name for details; it is not possible to mix
5014 named and numbered input channels
5017 If the `=' in a channel specification is replaced by `<', then the gains for
5018 that specification will be renormalized so that the total is 1, thus
5019 avoiding clipping noise.
5021 @subsection Mixing examples
5023 For example, if you want to down-mix from stereo to mono, but with a bigger
5024 factor for the left channel:
5026 pan=1c|c0=0.9*c0+0.1*c1
5029 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
5030 7-channels surround:
5032 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
5035 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
5036 that should be preferred (see "-ac" option) unless you have very specific
5039 @subsection Remapping examples
5041 The channel remapping will be effective if, and only if:
5044 @item gain coefficients are zeroes or ones,
5045 @item only one input per channel output,
5048 If all these conditions are satisfied, the filter will notify the user ("Pure
5049 channel mapping detected"), and use an optimized and lossless method to do the
5052 For example, if you have a 5.1 source and want a stereo audio stream by
5053 dropping the extra channels:
5055 pan="stereo| c0=FL | c1=FR"
5058 Given the same source, you can also switch front left and front right channels
5059 and keep the input channel layout:
5061 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
5064 If the input is a stereo audio stream, you can mute the front left channel (and
5065 still keep the stereo channel layout) with:
5070 Still with a stereo audio stream input, you can copy the right channel in both
5071 front left and right:
5073 pan="stereo| c0=FR | c1=FR"
5078 ReplayGain scanner filter. This filter takes an audio stream as an input and
5079 outputs it unchanged.
5080 At end of filtering it displays @code{track_gain} and @code{track_peak}.
5084 Convert the audio sample format, sample rate and channel layout. It is
5085 not meant to be used directly.
5088 Apply time-stretching and pitch-shifting with librubberband.
5090 To enable compilation of this filter, you need to configure FFmpeg with
5091 @code{--enable-librubberband}.
5093 The filter accepts the following options:
5097 Set tempo scale factor.
5100 Set pitch scale factor.
5103 Set transients detector.
5104 Possible values are:
5113 Possible values are:
5122 Possible values are:
5129 Set processing window size.
5130 Possible values are:
5139 Possible values are:
5146 Enable formant preservation when shift pitching.
5147 Possible values are:
5155 Possible values are:
5164 Possible values are:
5171 @subsection Commands
5173 This filter supports the following commands:
5176 Change filter tempo scale factor.
5177 Syntax for the command is : "@var{tempo}"
5180 Change filter pitch scale factor.
5181 Syntax for the command is : "@var{pitch}"
5184 @section sidechaincompress
5186 This filter acts like normal compressor but has the ability to compress
5187 detected signal using second input signal.
5188 It needs two input streams and returns one output stream.
5189 First input stream will be processed depending on second stream signal.
5190 The filtered signal then can be filtered with other filters in later stages of
5191 processing. See @ref{pan} and @ref{amerge} filter.
5193 The filter accepts the following options:
5197 Set input gain. Default is 1. Range is between 0.015625 and 64.
5200 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
5201 Default is @code{downward}.
5204 If a signal of second stream raises above this level it will affect the gain
5205 reduction of first stream.
5206 By default is 0.125. Range is between 0.00097563 and 1.
5209 Set a ratio about which the signal is reduced. 1:2 means that if the level
5210 raised 4dB above the threshold, it will be only 2dB above after the reduction.
5211 Default is 2. Range is between 1 and 20.
5214 Amount of milliseconds the signal has to rise above the threshold before gain
5215 reduction starts. Default is 20. Range is between 0.01 and 2000.
5218 Amount of milliseconds the signal has to fall below the threshold before
5219 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
5222 Set the amount by how much signal will be amplified after processing.
5223 Default is 1. Range is from 1 to 64.
5226 Curve the sharp knee around the threshold to enter gain reduction more softly.
5227 Default is 2.82843. Range is between 1 and 8.
5230 Choose if the @code{average} level between all channels of side-chain stream
5231 or the louder(@code{maximum}) channel of side-chain stream affects the
5232 reduction. Default is @code{average}.
5235 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5236 of @code{rms}. Default is @code{rms} which is mainly smoother.
5239 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5242 How much to use compressed signal in output. Default is 1.
5243 Range is between 0 and 1.
5246 @subsection Commands
5248 This filter supports the all above options as @ref{commands}.
5250 @subsection Examples
5254 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5255 depending on the signal of 2nd input and later compressed signal to be
5256 merged with 2nd input:
5258 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5262 @section sidechaingate
5264 A sidechain gate acts like a normal (wideband) gate but has the ability to
5265 filter the detected signal before sending it to the gain reduction stage.
5266 Normally a gate uses the full range signal to detect a level above the
5268 For example: If you cut all lower frequencies from your sidechain signal
5269 the gate will decrease the volume of your track only if not enough highs
5270 appear. With this technique you are able to reduce the resonation of a
5271 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5273 It needs two input streams and returns one output stream.
5274 First input stream will be processed depending on second stream signal.
5276 The filter accepts the following options:
5280 Set input level before filtering.
5281 Default is 1. Allowed range is from 0.015625 to 64.
5284 Set the mode of operation. Can be @code{upward} or @code{downward}.
5285 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5286 will be amplified, expanding dynamic range in upward direction.
5287 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5290 Set the level of gain reduction when the signal is below the threshold.
5291 Default is 0.06125. Allowed range is from 0 to 1.
5292 Setting this to 0 disables reduction and then filter behaves like expander.
5295 If a signal rises above this level the gain reduction is released.
5296 Default is 0.125. Allowed range is from 0 to 1.
5299 Set a ratio about which the signal is reduced.
5300 Default is 2. Allowed range is from 1 to 9000.
5303 Amount of milliseconds the signal has to rise above the threshold before gain
5305 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5308 Amount of milliseconds the signal has to fall below the threshold before the
5309 reduction is increased again. Default is 250 milliseconds.
5310 Allowed range is from 0.01 to 9000.
5313 Set amount of amplification of signal after processing.
5314 Default is 1. Allowed range is from 1 to 64.
5317 Curve the sharp knee around the threshold to enter gain reduction more softly.
5318 Default is 2.828427125. Allowed range is from 1 to 8.
5321 Choose if exact signal should be taken for detection or an RMS like one.
5322 Default is rms. Can be peak or rms.
5325 Choose if the average level between all channels or the louder channel affects
5327 Default is average. Can be average or maximum.
5330 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5333 @subsection Commands
5335 This filter supports the all above options as @ref{commands}.
5337 @section silencedetect
5339 Detect silence in an audio stream.
5341 This filter logs a message when it detects that the input audio volume is less
5342 or equal to a noise tolerance value for a duration greater or equal to the
5343 minimum detected noise duration.
5345 The printed times and duration are expressed in seconds. The
5346 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5347 is set on the first frame whose timestamp equals or exceeds the detection
5348 duration and it contains the timestamp of the first frame of the silence.
5350 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5351 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5352 keys are set on the first frame after the silence. If @option{mono} is
5353 enabled, and each channel is evaluated separately, the @code{.X}
5354 suffixed keys are used, and @code{X} corresponds to the channel number.
5356 The filter accepts the following options:
5360 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5361 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5364 Set silence duration until notification (default is 2 seconds). See
5365 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5366 for the accepted syntax.
5369 Process each channel separately, instead of combined. By default is disabled.
5372 @subsection Examples
5376 Detect 5 seconds of silence with -50dB noise tolerance:
5378 silencedetect=n=-50dB:d=5
5382 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5383 tolerance in @file{silence.mp3}:
5385 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5389 @section silenceremove
5391 Remove silence from the beginning, middle or end of the audio.
5393 The filter accepts the following options:
5397 This value is used to indicate if audio should be trimmed at beginning of
5398 the audio. A value of zero indicates no silence should be trimmed from the
5399 beginning. When specifying a non-zero value, it trims audio up until it
5400 finds non-silence. Normally, when trimming silence from beginning of audio
5401 the @var{start_periods} will be @code{1} but it can be increased to higher
5402 values to trim all audio up to specific count of non-silence periods.
5403 Default value is @code{0}.
5405 @item start_duration
5406 Specify the amount of time that non-silence must be detected before it stops
5407 trimming audio. By increasing the duration, bursts of noises can be treated
5408 as silence and trimmed off. Default value is @code{0}.
5410 @item start_threshold
5411 This indicates what sample value should be treated as silence. For digital
5412 audio, a value of @code{0} may be fine but for audio recorded from analog,
5413 you may wish to increase the value to account for background noise.
5414 Can be specified in dB (in case "dB" is appended to the specified value)
5415 or amplitude ratio. Default value is @code{0}.
5418 Specify max duration of silence at beginning that will be kept after
5419 trimming. Default is 0, which is equal to trimming all samples detected
5423 Specify mode of detection of silence end in start of multi-channel audio.
5424 Can be @var{any} or @var{all}. Default is @var{any}.
5425 With @var{any}, any sample that is detected as non-silence will cause
5426 stopped trimming of silence.
5427 With @var{all}, only if all channels are detected as non-silence will cause
5428 stopped trimming of silence.
5431 Set the count for trimming silence from the end of audio.
5432 To remove silence from the middle of a file, specify a @var{stop_periods}
5433 that is negative. This value is then treated as a positive value and is
5434 used to indicate the effect should restart processing as specified by
5435 @var{start_periods}, making it suitable for removing periods of silence
5436 in the middle of the audio.
5437 Default value is @code{0}.
5440 Specify a duration of silence that must exist before audio is not copied any
5441 more. By specifying a higher duration, silence that is wanted can be left in
5443 Default value is @code{0}.
5445 @item stop_threshold
5446 This is the same as @option{start_threshold} but for trimming silence from
5448 Can be specified in dB (in case "dB" is appended to the specified value)
5449 or amplitude ratio. Default value is @code{0}.
5452 Specify max duration of silence at end that will be kept after
5453 trimming. Default is 0, which is equal to trimming all samples detected
5457 Specify mode of detection of silence start in end of multi-channel audio.
5458 Can be @var{any} or @var{all}. Default is @var{any}.
5459 With @var{any}, any sample that is detected as non-silence will cause
5460 stopped trimming of silence.
5461 With @var{all}, only if all channels are detected as non-silence will cause
5462 stopped trimming of silence.
5465 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5466 and works better with digital silence which is exactly 0.
5467 Default value is @code{rms}.
5470 Set duration in number of seconds used to calculate size of window in number
5471 of samples for detecting silence.
5472 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5475 @subsection Examples
5479 The following example shows how this filter can be used to start a recording
5480 that does not contain the delay at the start which usually occurs between
5481 pressing the record button and the start of the performance:
5483 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5487 Trim all silence encountered from beginning to end where there is more than 1
5488 second of silence in audio:
5490 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5494 Trim all digital silence samples, using peak detection, from beginning to end
5495 where there is more than 0 samples of digital silence in audio and digital
5496 silence is detected in all channels at same positions in stream:
5498 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5504 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5505 loudspeakers around the user for binaural listening via headphones (audio
5506 formats up to 9 channels supported).
5507 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5508 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5509 Austrian Academy of Sciences.
5511 To enable compilation of this filter you need to configure FFmpeg with
5512 @code{--enable-libmysofa}.
5514 The filter accepts the following options:
5518 Set the SOFA file used for rendering.
5521 Set gain applied to audio. Value is in dB. Default is 0.
5524 Set rotation of virtual loudspeakers in deg. Default is 0.
5527 Set elevation of virtual speakers in deg. Default is 0.
5530 Set distance in meters between loudspeakers and the listener with near-field
5531 HRTFs. Default is 1.
5534 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5535 processing audio in time domain which is slow.
5536 @var{freq} is processing audio in frequency domain which is fast.
5537 Default is @var{freq}.
5540 Set custom positions of virtual loudspeakers. Syntax for this option is:
5541 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5542 Each virtual loudspeaker is described with short channel name following with
5543 azimuth and elevation in degrees.
5544 Each virtual loudspeaker description is separated by '|'.
5545 For example to override front left and front right channel positions use:
5546 'speakers=FL 45 15|FR 345 15'.
5547 Descriptions with unrecognised channel names are ignored.
5550 Set custom gain for LFE channels. Value is in dB. Default is 0.
5553 Set custom frame size in number of samples. Default is 1024.
5554 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5555 is set to @var{freq}.
5558 Should all IRs be normalized upon importing SOFA file.
5559 By default is enabled.
5562 Should nearest IRs be interpolated with neighbor IRs if exact position
5563 does not match. By default is disabled.
5566 Minphase all IRs upon loading of SOFA file. By default is disabled.
5569 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5572 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5575 @subsection Examples
5579 Using ClubFritz6 sofa file:
5581 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5585 Using ClubFritz12 sofa file and bigger radius with small rotation:
5587 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5591 Similar as above but with custom speaker positions for front left, front right, back left and back right
5592 and also with custom gain:
5594 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5601 This filter expands or compresses each half-cycle of audio samples
5602 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5603 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5605 The filter accepts the following options:
5609 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5610 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5613 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5614 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5615 would be such that local peak value reaches target peak value but never to surpass it and that
5616 ratio between new and previous peak value does not surpass this option value.
5618 @item compression, c
5619 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5620 This option controls maximum local half-cycle of samples compression. This option is used
5621 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5622 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5623 that peak's half-cycle will be compressed by current compression factor.
5626 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5627 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5628 Any half-cycle samples with their local peak value below or same as this option value will be
5629 compressed by current compression factor, otherwise, if greater than threshold value they will be
5630 expanded with expansion factor so that it could reach peak target value but never surpass it.
5633 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5634 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5635 each new half-cycle until it reaches @option{expansion} value.
5636 Setting this options too high may lead to distortions.
5639 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5640 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5641 each new half-cycle until it reaches @option{compression} value.
5644 Specify which channels to filter, by default all available channels are filtered.
5647 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5648 option. When enabled any half-cycle of samples with their local peak value below or same as
5649 @option{threshold} option will be expanded otherwise it will be compressed.
5652 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5653 When disabled each filtered channel gain calculation is independent, otherwise when this option
5654 is enabled the minimum of all possible gains for each filtered channel is used.
5657 @subsection Commands
5659 This filter supports the all above options as @ref{commands}.
5661 @section stereotools
5663 This filter has some handy utilities to manage stereo signals, for converting
5664 M/S stereo recordings to L/R signal while having control over the parameters
5665 or spreading the stereo image of master track.
5667 The filter accepts the following options:
5671 Set input level before filtering for both channels. Defaults is 1.
5672 Allowed range is from 0.015625 to 64.
5675 Set output level after filtering for both channels. Defaults is 1.
5676 Allowed range is from 0.015625 to 64.
5679 Set input balance between both channels. Default is 0.
5680 Allowed range is from -1 to 1.
5683 Set output balance between both channels. Default is 0.
5684 Allowed range is from -1 to 1.
5687 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5688 clipping. Disabled by default.
5691 Mute the left channel. Disabled by default.
5694 Mute the right channel. Disabled by default.
5697 Change the phase of the left channel. Disabled by default.
5700 Change the phase of the right channel. Disabled by default.
5703 Set stereo mode. Available values are:
5707 Left/Right to Left/Right, this is default.
5710 Left/Right to Mid/Side.
5713 Mid/Side to Left/Right.
5716 Left/Right to Left/Left.
5719 Left/Right to Right/Right.
5722 Left/Right to Left + Right.
5725 Left/Right to Right/Left.
5728 Mid/Side to Left/Left.
5731 Mid/Side to Right/Right.
5734 Mid/Side to Right/Left.
5737 Left/Right to Left - Right.
5741 Set level of side signal. Default is 1.
5742 Allowed range is from 0.015625 to 64.
5745 Set balance of side signal. Default is 0.
5746 Allowed range is from -1 to 1.
5749 Set level of the middle signal. Default is 1.
5750 Allowed range is from 0.015625 to 64.
5753 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5756 Set stereo base between mono and inversed channels. Default is 0.
5757 Allowed range is from -1 to 1.
5760 Set delay in milliseconds how much to delay left from right channel and
5761 vice versa. Default is 0. Allowed range is from -20 to 20.
5764 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5767 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5769 @item bmode_in, bmode_out
5770 Set balance mode for balance_in/balance_out option.
5772 Can be one of the following:
5776 Classic balance mode. Attenuate one channel at time.
5777 Gain is raised up to 1.
5780 Similar as classic mode above but gain is raised up to 2.
5783 Equal power distribution, from -6dB to +6dB range.
5787 @subsection Commands
5789 This filter supports the all above options as @ref{commands}.
5791 @subsection Examples
5795 Apply karaoke like effect:
5797 stereotools=mlev=0.015625
5801 Convert M/S signal to L/R:
5803 "stereotools=mode=ms>lr"
5807 @section stereowiden
5809 This filter enhance the stereo effect by suppressing signal common to both
5810 channels and by delaying the signal of left into right and vice versa,
5811 thereby widening the stereo effect.
5813 The filter accepts the following options:
5817 Time in milliseconds of the delay of left signal into right and vice versa.
5818 Default is 20 milliseconds.
5821 Amount of gain in delayed signal into right and vice versa. Gives a delay
5822 effect of left signal in right output and vice versa which gives widening
5823 effect. Default is 0.3.
5826 Cross feed of left into right with inverted phase. This helps in suppressing
5827 the mono. If the value is 1 it will cancel all the signal common to both
5828 channels. Default is 0.3.
5831 Set level of input signal of original channel. Default is 0.8.
5834 @subsection Commands
5836 This filter supports the all above options except @code{delay} as @ref{commands}.
5838 @section superequalizer
5839 Apply 18 band equalizer.
5841 The filter accepts the following options:
5848 Set 131Hz band gain.
5850 Set 185Hz band gain.
5852 Set 262Hz band gain.
5854 Set 370Hz band gain.
5856 Set 523Hz band gain.
5858 Set 740Hz band gain.
5860 Set 1047Hz band gain.
5862 Set 1480Hz band gain.
5864 Set 2093Hz band gain.
5866 Set 2960Hz band gain.
5868 Set 4186Hz band gain.
5870 Set 5920Hz band gain.
5872 Set 8372Hz band gain.
5874 Set 11840Hz band gain.
5876 Set 16744Hz band gain.
5878 Set 20000Hz band gain.
5882 Apply audio surround upmix filter.
5884 This filter allows to produce multichannel output from audio stream.
5886 The filter accepts the following options:
5890 Set output channel layout. By default, this is @var{5.1}.
5892 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5893 for the required syntax.
5896 Set input channel layout. By default, this is @var{stereo}.
5898 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5899 for the required syntax.
5902 Set input volume level. By default, this is @var{1}.
5905 Set output volume level. By default, this is @var{1}.
5908 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5911 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5914 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5917 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5918 In @var{add} mode, LFE channel is created from input audio and added to output.
5919 In @var{sub} mode, LFE channel is created from input audio and added to output but
5920 also all non-LFE output channels are subtracted with output LFE channel.
5923 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5924 Default is @var{90}.
5927 Set front center input volume. By default, this is @var{1}.
5930 Set front center output volume. By default, this is @var{1}.
5933 Set front left input volume. By default, this is @var{1}.
5936 Set front left output volume. By default, this is @var{1}.
5939 Set front right input volume. By default, this is @var{1}.
5942 Set front right output volume. By default, this is @var{1}.
5945 Set side left input volume. By default, this is @var{1}.
5948 Set side left output volume. By default, this is @var{1}.
5951 Set side right input volume. By default, this is @var{1}.
5954 Set side right output volume. By default, this is @var{1}.
5957 Set back left input volume. By default, this is @var{1}.
5960 Set back left output volume. By default, this is @var{1}.
5963 Set back right input volume. By default, this is @var{1}.
5966 Set back right output volume. By default, this is @var{1}.
5969 Set back center input volume. By default, this is @var{1}.
5972 Set back center output volume. By default, this is @var{1}.
5975 Set LFE input volume. By default, this is @var{1}.
5978 Set LFE output volume. By default, this is @var{1}.
5981 Set spread usage of stereo image across X axis for all channels.
5984 Set spread usage of stereo image across Y axis for all channels.
5986 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5987 Set spread usage of stereo image across X axis for each channel.
5989 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5990 Set spread usage of stereo image across Y axis for each channel.
5993 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5996 Set window function.
5998 It accepts the following values:
6021 Default is @code{hann}.
6024 Set window overlap. If set to 1, the recommended overlap for selected
6025 window function will be picked. Default is @code{0.5}.
6028 @section treble, highshelf
6030 Boost or cut treble (upper) frequencies of the audio using a two-pole
6031 shelving filter with a response similar to that of a standard
6032 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
6034 The filter accepts the following options:
6038 Give the gain at whichever is the lower of ~22 kHz and the
6039 Nyquist frequency. Its useful range is about -20 (for a large cut)
6040 to +20 (for a large boost). Beware of clipping when using a positive gain.
6043 Set the filter's central frequency and so can be used
6044 to extend or reduce the frequency range to be boosted or cut.
6045 The default value is @code{3000} Hz.
6048 Set method to specify band-width of filter.
6063 Determine how steep is the filter's shelf transition.
6066 Set number of poles. Default is 2.
6069 How much to use filtered signal in output. Default is 1.
6070 Range is between 0 and 1.
6073 Specify which channels to filter, by default all available are filtered.
6076 Normalize biquad coefficients, by default is disabled.
6077 Enabling it will normalize magnitude response at DC to 0dB.
6080 Set transform type of IIR filter.
6089 Set precison of filtering.
6092 Pick automatic sample format depending on surround filters.
6094 Always use signed 16-bit.
6096 Always use signed 32-bit.
6098 Always use float 32-bit.
6100 Always use float 64-bit.
6104 @subsection Commands
6106 This filter supports the following commands:
6109 Change treble frequency.
6110 Syntax for the command is : "@var{frequency}"
6113 Change treble width_type.
6114 Syntax for the command is : "@var{width_type}"
6117 Change treble width.
6118 Syntax for the command is : "@var{width}"
6122 Syntax for the command is : "@var{gain}"
6126 Syntax for the command is : "@var{mix}"
6131 Sinusoidal amplitude modulation.
6133 The filter accepts the following options:
6137 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
6138 (20 Hz or lower) will result in a tremolo effect.
6139 This filter may also be used as a ring modulator by specifying
6140 a modulation frequency higher than 20 Hz.
6141 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6144 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6145 Default value is 0.5.
6150 Sinusoidal phase modulation.
6152 The filter accepts the following options:
6156 Modulation frequency in Hertz.
6157 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6160 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6161 Default value is 0.5.
6166 Adjust the input audio volume.
6168 It accepts the following parameters:
6172 Set audio volume expression.
6174 Output values are clipped to the maximum value.
6176 The output audio volume is given by the relation:
6178 @var{output_volume} = @var{volume} * @var{input_volume}
6181 The default value for @var{volume} is "1.0".
6184 This parameter represents the mathematical precision.
6186 It determines which input sample formats will be allowed, which affects the
6187 precision of the volume scaling.
6191 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
6193 32-bit floating-point; this limits input sample format to FLT. (default)
6195 64-bit floating-point; this limits input sample format to DBL.
6199 Choose the behaviour on encountering ReplayGain side data in input frames.
6203 Remove ReplayGain side data, ignoring its contents (the default).
6206 Ignore ReplayGain side data, but leave it in the frame.
6209 Prefer the track gain, if present.
6212 Prefer the album gain, if present.
6215 @item replaygain_preamp
6216 Pre-amplification gain in dB to apply to the selected replaygain gain.
6218 Default value for @var{replaygain_preamp} is 0.0.
6220 @item replaygain_noclip
6221 Prevent clipping by limiting the gain applied.
6223 Default value for @var{replaygain_noclip} is 1.
6226 Set when the volume expression is evaluated.
6228 It accepts the following values:
6231 only evaluate expression once during the filter initialization, or
6232 when the @samp{volume} command is sent
6235 evaluate expression for each incoming frame
6238 Default value is @samp{once}.
6241 The volume expression can contain the following parameters.
6245 frame number (starting at zero)
6248 @item nb_consumed_samples
6249 number of samples consumed by the filter
6251 number of samples in the current frame
6253 original frame position in the file
6259 PTS at start of stream
6261 time at start of stream
6267 last set volume value
6270 Note that when @option{eval} is set to @samp{once} only the
6271 @var{sample_rate} and @var{tb} variables are available, all other
6272 variables will evaluate to NAN.
6274 @subsection Commands
6276 This filter supports the following commands:
6279 Modify the volume expression.
6280 The command accepts the same syntax of the corresponding option.
6282 If the specified expression is not valid, it is kept at its current
6286 @subsection Examples
6290 Halve the input audio volume:
6294 volume=volume=-6.0206dB
6297 In all the above example the named key for @option{volume} can be
6298 omitted, for example like in:
6304 Increase input audio power by 6 decibels using fixed-point precision:
6306 volume=volume=6dB:precision=fixed
6310 Fade volume after time 10 with an annihilation period of 5 seconds:
6312 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6316 @section volumedetect
6318 Detect the volume of the input video.
6320 The filter has no parameters. The input is not modified. Statistics about
6321 the volume will be printed in the log when the input stream end is reached.
6323 In particular it will show the mean volume (root mean square), maximum
6324 volume (on a per-sample basis), and the beginning of a histogram of the
6325 registered volume values (from the maximum value to a cumulated 1/1000 of
6328 All volumes are in decibels relative to the maximum PCM value.
6330 @subsection Examples
6332 Here is an excerpt of the output:
6334 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6335 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6336 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6337 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6338 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6339 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6340 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6341 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6342 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6348 The mean square energy is approximately -27 dB, or 10^-2.7.
6350 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6352 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6355 In other words, raising the volume by +4 dB does not cause any clipping,
6356 raising it by +5 dB causes clipping for 6 samples, etc.
6358 @c man end AUDIO FILTERS
6360 @chapter Audio Sources
6361 @c man begin AUDIO SOURCES
6363 Below is a description of the currently available audio sources.
6367 Buffer audio frames, and make them available to the filter chain.
6369 This source is mainly intended for a programmatic use, in particular
6370 through the interface defined in @file{libavfilter/buffersrc.h}.
6372 It accepts the following parameters:
6376 The timebase which will be used for timestamps of submitted frames. It must be
6377 either a floating-point number or in @var{numerator}/@var{denominator} form.
6380 The sample rate of the incoming audio buffers.
6383 The sample format of the incoming audio buffers.
6384 Either a sample format name or its corresponding integer representation from
6385 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6387 @item channel_layout
6388 The channel layout of the incoming audio buffers.
6389 Either a channel layout name from channel_layout_map in
6390 @file{libavutil/channel_layout.c} or its corresponding integer representation
6391 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6394 The number of channels of the incoming audio buffers.
6395 If both @var{channels} and @var{channel_layout} are specified, then they
6400 @subsection Examples
6403 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6406 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6407 Since the sample format with name "s16p" corresponds to the number
6408 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6411 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6416 Generate an audio signal specified by an expression.
6418 This source accepts in input one or more expressions (one for each
6419 channel), which are evaluated and used to generate a corresponding
6422 This source accepts the following options:
6426 Set the '|'-separated expressions list for each separate channel. In case the
6427 @option{channel_layout} option is not specified, the selected channel layout
6428 depends on the number of provided expressions. Otherwise the last
6429 specified expression is applied to the remaining output channels.
6431 @item channel_layout, c
6432 Set the channel layout. The number of channels in the specified layout
6433 must be equal to the number of specified expressions.
6436 Set the minimum duration of the sourced audio. See
6437 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6438 for the accepted syntax.
6439 Note that the resulting duration may be greater than the specified
6440 duration, as the generated audio is always cut at the end of a
6443 If not specified, or the expressed duration is negative, the audio is
6444 supposed to be generated forever.
6447 Set the number of samples per channel per each output frame,
6450 @item sample_rate, s
6451 Specify the sample rate, default to 44100.
6454 Each expression in @var{exprs} can contain the following constants:
6458 number of the evaluated sample, starting from 0
6461 time of the evaluated sample expressed in seconds, starting from 0
6468 @subsection Examples
6478 Generate a sin signal with frequency of 440 Hz, set sample rate to
6481 aevalsrc="sin(440*2*PI*t):s=8000"
6485 Generate a two channels signal, specify the channel layout (Front
6486 Center + Back Center) explicitly:
6488 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6492 Generate white noise:
6494 aevalsrc="-2+random(0)"
6498 Generate an amplitude modulated signal:
6500 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6504 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6506 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6513 Generate a FIR coefficients using frequency sampling method.
6515 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6517 The filter accepts the following options:
6521 Set number of filter coefficents in output audio stream.
6522 Default value is 1025.
6525 Set frequency points from where magnitude and phase are set.
6526 This must be in non decreasing order, and first element must be 0, while last element
6527 must be 1. Elements are separated by white spaces.
6530 Set magnitude value for every frequency point set by @option{frequency}.
6531 Number of values must be same as number of frequency points.
6532 Values are separated by white spaces.
6535 Set phase value for every frequency point set by @option{frequency}.
6536 Number of values must be same as number of frequency points.
6537 Values are separated by white spaces.
6539 @item sample_rate, r
6540 Set sample rate, default is 44100.
6543 Set number of samples per each frame. Default is 1024.
6546 Set window function. Default is blackman.
6551 The null audio source, return unprocessed audio frames. It is mainly useful
6552 as a template and to be employed in analysis / debugging tools, or as
6553 the source for filters which ignore the input data (for example the sox
6556 This source accepts the following options:
6560 @item channel_layout, cl
6562 Specifies the channel layout, and can be either an integer or a string
6563 representing a channel layout. The default value of @var{channel_layout}
6566 Check the channel_layout_map definition in
6567 @file{libavutil/channel_layout.c} for the mapping between strings and
6568 channel layout values.
6570 @item sample_rate, r
6571 Specifies the sample rate, and defaults to 44100.
6574 Set the number of samples per requested frames.
6577 Set the duration of the sourced audio. See
6578 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6579 for the accepted syntax.
6581 If not specified, or the expressed duration is negative, the audio is
6582 supposed to be generated forever.
6585 @subsection Examples
6589 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6591 anullsrc=r=48000:cl=4
6595 Do the same operation with a more obvious syntax:
6597 anullsrc=r=48000:cl=mono
6601 All the parameters need to be explicitly defined.
6605 Synthesize a voice utterance using the libflite library.
6607 To enable compilation of this filter you need to configure FFmpeg with
6608 @code{--enable-libflite}.
6610 Note that versions of the flite library prior to 2.0 are not thread-safe.
6612 The filter accepts the following options:
6617 If set to 1, list the names of the available voices and exit
6618 immediately. Default value is 0.
6621 Set the maximum number of samples per frame. Default value is 512.
6624 Set the filename containing the text to speak.
6627 Set the text to speak.
6630 Set the voice to use for the speech synthesis. Default value is
6631 @code{kal}. See also the @var{list_voices} option.
6634 @subsection Examples
6638 Read from file @file{speech.txt}, and synthesize the text using the
6639 standard flite voice:
6641 flite=textfile=speech.txt
6645 Read the specified text selecting the @code{slt} voice:
6647 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6651 Input text to ffmpeg:
6653 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6657 Make @file{ffplay} speak the specified text, using @code{flite} and
6658 the @code{lavfi} device:
6660 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6664 For more information about libflite, check:
6665 @url{http://www.festvox.org/flite/}
6669 Generate a noise audio signal.
6671 The filter accepts the following options:
6674 @item sample_rate, r
6675 Specify the sample rate. Default value is 48000 Hz.
6678 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6682 Specify the duration of the generated audio stream. Not specifying this option
6683 results in noise with an infinite length.
6685 @item color, colour, c
6686 Specify the color of noise. Available noise colors are white, pink, brown,
6687 blue, violet and velvet. Default color is white.
6690 Specify a value used to seed the PRNG.
6693 Set the number of samples per each output frame, default is 1024.
6696 @subsection Examples
6701 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6703 anoisesrc=d=60:c=pink:r=44100:a=0.5
6709 Generate odd-tap Hilbert transform FIR coefficients.
6711 The resulting stream can be used with @ref{afir} filter for phase-shifting
6712 the signal by 90 degrees.
6714 This is used in many matrix coding schemes and for analytic signal generation.
6715 The process is often written as a multiplication by i (or j), the imaginary unit.
6717 The filter accepts the following options:
6721 @item sample_rate, s
6722 Set sample rate, default is 44100.
6725 Set length of FIR filter, default is 22051.
6728 Set number of samples per each frame.
6731 Set window function to be used when generating FIR coefficients.
6736 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6738 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6740 The filter accepts the following options:
6743 @item sample_rate, r
6744 Set sample rate, default is 44100.
6747 Set number of samples per each frame. Default is 1024.
6750 Set high-pass frequency. Default is 0.
6753 Set low-pass frequency. Default is 0.
6754 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6755 is higher than 0 then filter will create band-pass filter coefficients,
6756 otherwise band-reject filter coefficients.
6759 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6762 Set Kaiser window beta.
6765 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6768 Enable rounding, by default is disabled.
6771 Set number of taps for high-pass filter.
6774 Set number of taps for low-pass filter.
6779 Generate an audio signal made of a sine wave with amplitude 1/8.
6781 The audio signal is bit-exact.
6783 The filter accepts the following options:
6788 Set the carrier frequency. Default is 440 Hz.
6790 @item beep_factor, b
6791 Enable a periodic beep every second with frequency @var{beep_factor} times
6792 the carrier frequency. Default is 0, meaning the beep is disabled.
6794 @item sample_rate, r
6795 Specify the sample rate, default is 44100.
6798 Specify the duration of the generated audio stream.
6800 @item samples_per_frame
6801 Set the number of samples per output frame.
6803 The expression can contain the following constants:
6807 The (sequential) number of the output audio frame, starting from 0.
6810 The PTS (Presentation TimeStamp) of the output audio frame,
6811 expressed in @var{TB} units.
6814 The PTS of the output audio frame, expressed in seconds.
6817 The timebase of the output audio frames.
6820 Default is @code{1024}.
6823 @subsection Examples
6828 Generate a simple 440 Hz sine wave:
6834 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6838 sine=frequency=220:beep_factor=4:duration=5
6842 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6845 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6849 @c man end AUDIO SOURCES
6851 @chapter Audio Sinks
6852 @c man begin AUDIO SINKS
6854 Below is a description of the currently available audio sinks.
6856 @section abuffersink
6858 Buffer audio frames, and make them available to the end of filter chain.
6860 This sink is mainly intended for programmatic use, in particular
6861 through the interface defined in @file{libavfilter/buffersink.h}
6862 or the options system.
6864 It accepts a pointer to an AVABufferSinkContext structure, which
6865 defines the incoming buffers' formats, to be passed as the opaque
6866 parameter to @code{avfilter_init_filter} for initialization.
6869 Null audio sink; do absolutely nothing with the input audio. It is
6870 mainly useful as a template and for use in analysis / debugging
6873 @c man end AUDIO SINKS
6875 @chapter Video Filters
6876 @c man begin VIDEO FILTERS
6878 When you configure your FFmpeg build, you can disable any of the
6879 existing filters using @code{--disable-filters}.
6880 The configure output will show the video filters included in your
6883 Below is a description of the currently available video filters.
6887 Mark a region of interest in a video frame.
6889 The frame data is passed through unchanged, but metadata is attached
6890 to the frame indicating regions of interest which can affect the
6891 behaviour of later encoding. Multiple regions can be marked by
6892 applying the filter multiple times.
6896 Region distance in pixels from the left edge of the frame.
6898 Region distance in pixels from the top edge of the frame.
6900 Region width in pixels.
6902 Region height in pixels.
6904 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6905 and may contain the following variables:
6908 Width of the input frame.
6910 Height of the input frame.
6914 Quantisation offset to apply within the region.
6916 This must be a real value in the range -1 to +1. A value of zero
6917 indicates no quality change. A negative value asks for better quality
6918 (less quantisation), while a positive value asks for worse quality
6919 (greater quantisation).
6921 The range is calibrated so that the extreme values indicate the
6922 largest possible offset - if the rest of the frame is encoded with the
6923 worst possible quality, an offset of -1 indicates that this region
6924 should be encoded with the best possible quality anyway. Intermediate
6925 values are then interpolated in some codec-dependent way.
6927 For example, in 10-bit H.264 the quantisation parameter varies between
6928 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6929 this region should be encoded with a QP around one-tenth of the full
6930 range better than the rest of the frame. So, if most of the frame
6931 were to be encoded with a QP of around 30, this region would get a QP
6932 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6933 An extreme value of -1 would indicate that this region should be
6934 encoded with the best possible quality regardless of the treatment of
6935 the rest of the frame - that is, should be encoded at a QP of -12.
6937 If set to true, remove any existing regions of interest marked on the
6938 frame before adding the new one.
6941 @subsection Examples
6945 Mark the centre quarter of the frame as interesting.
6947 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6950 Mark the 100-pixel-wide region on the left edge of the frame as very
6951 uninteresting (to be encoded at much lower quality than the rest of
6954 addroi=0:0:100:ih:+1/5
6958 @section alphaextract
6960 Extract the alpha component from the input as a grayscale video. This
6961 is especially useful with the @var{alphamerge} filter.
6965 Add or replace the alpha component of the primary input with the
6966 grayscale value of a second input. This is intended for use with
6967 @var{alphaextract} to allow the transmission or storage of frame
6968 sequences that have alpha in a format that doesn't support an alpha
6971 For example, to reconstruct full frames from a normal YUV-encoded video
6972 and a separate video created with @var{alphaextract}, you might use:
6974 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6979 Amplify differences between current pixel and pixels of adjacent frames in
6980 same pixel location.
6982 This filter accepts the following options:
6986 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6987 For example radius of 3 will instruct filter to calculate average of 7 frames.
6990 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6993 Set threshold for difference amplification. Any difference greater or equal to
6994 this value will not alter source pixel. Default is 10.
6995 Allowed range is from 0 to 65535.
6998 Set tolerance for difference amplification. Any difference lower to
6999 this value will not alter source pixel. Default is 0.
7000 Allowed range is from 0 to 65535.
7003 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7004 This option controls maximum possible value that will decrease source pixel value.
7007 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7008 This option controls maximum possible value that will increase source pixel value.
7011 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
7014 @subsection Commands
7016 This filter supports the following @ref{commands} that corresponds to option of same name:
7028 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
7029 and libavformat to work. On the other hand, it is limited to ASS (Advanced
7030 Substation Alpha) subtitles files.
7032 This filter accepts the following option in addition to the common options from
7033 the @ref{subtitles} filter:
7037 Set the shaping engine
7039 Available values are:
7042 The default libass shaping engine, which is the best available.
7044 Fast, font-agnostic shaper that can do only substitutions
7046 Slower shaper using OpenType for substitutions and positioning
7049 The default is @code{auto}.
7053 Apply an Adaptive Temporal Averaging Denoiser to the video input.
7055 The filter accepts the following options:
7059 Set threshold A for 1st plane. Default is 0.02.
7060 Valid range is 0 to 0.3.
7063 Set threshold B for 1st plane. Default is 0.04.
7064 Valid range is 0 to 5.
7067 Set threshold A for 2nd plane. Default is 0.02.
7068 Valid range is 0 to 0.3.
7071 Set threshold B for 2nd plane. Default is 0.04.
7072 Valid range is 0 to 5.
7075 Set threshold A for 3rd plane. Default is 0.02.
7076 Valid range is 0 to 0.3.
7079 Set threshold B for 3rd plane. Default is 0.04.
7080 Valid range is 0 to 5.
7082 Threshold A is designed to react on abrupt changes in the input signal and
7083 threshold B is designed to react on continuous changes in the input signal.
7086 Set number of frames filter will use for averaging. Default is 9. Must be odd
7087 number in range [5, 129].
7090 Set what planes of frame filter will use for averaging. Default is all.
7093 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
7094 Alternatively can be set to @code{s} serial.
7096 Parallel can be faster then serial, while other way around is never true.
7097 Parallel will abort early on first change being greater then thresholds, while serial
7098 will continue processing other side of frames if they are equal or below thresholds.
7101 @subsection Commands
7102 This filter supports same @ref{commands} as options except option @code{s}.
7103 The command accepts the same syntax of the corresponding option.
7107 Apply average blur filter.
7109 The filter accepts the following options:
7113 Set horizontal radius size.
7116 Set which planes to filter. By default all planes are filtered.
7119 Set vertical radius size, if zero it will be same as @code{sizeX}.
7120 Default is @code{0}.
7123 @subsection Commands
7124 This filter supports same commands as options.
7125 The command accepts the same syntax of the corresponding option.
7127 If the specified expression is not valid, it is kept at its current
7132 Compute the bounding box for the non-black pixels in the input frame
7135 This filter computes the bounding box containing all the pixels with a
7136 luminance value greater than the minimum allowed value.
7137 The parameters describing the bounding box are printed on the filter
7140 The filter accepts the following option:
7144 Set the minimal luminance value. Default is @code{16}.
7147 @subsection Commands
7149 This filter supports the all above options as @ref{commands}.
7152 Apply bilateral filter, spatial smoothing while preserving edges.
7154 The filter accepts the following options:
7157 Set sigma of gaussian function to calculate spatial weight.
7158 Allowed range is 0 to 512. Default is 0.1.
7161 Set sigma of gaussian function to calculate range weight.
7162 Allowed range is 0 to 1. Default is 0.1.
7165 Set planes to filter. Default is first only.
7168 @subsection Commands
7170 This filter supports the all above options as @ref{commands}.
7172 @section bitplanenoise
7174 Show and measure bit plane noise.
7176 The filter accepts the following options:
7180 Set which plane to analyze. Default is @code{1}.
7183 Filter out noisy pixels from @code{bitplane} set above.
7184 Default is disabled.
7187 @section blackdetect
7189 Detect video intervals that are (almost) completely black. Can be
7190 useful to detect chapter transitions, commercials, or invalid
7193 The filter outputs its detection analysis to both the log as well as
7194 frame metadata. If a black segment of at least the specified minimum
7195 duration is found, a line with the start and end timestamps as well
7196 as duration is printed to the log with level @code{info}. In addition,
7197 a log line with level @code{debug} is printed per frame showing the
7198 black amount detected for that frame.
7200 The filter also attaches metadata to the first frame of a black
7201 segment with key @code{lavfi.black_start} and to the first frame
7202 after the black segment ends with key @code{lavfi.black_end}. The
7203 value is the frame's timestamp. This metadata is added regardless
7204 of the minimum duration specified.
7206 The filter accepts the following options:
7209 @item black_min_duration, d
7210 Set the minimum detected black duration expressed in seconds. It must
7211 be a non-negative floating point number.
7213 Default value is 2.0.
7215 @item picture_black_ratio_th, pic_th
7216 Set the threshold for considering a picture "black".
7217 Express the minimum value for the ratio:
7219 @var{nb_black_pixels} / @var{nb_pixels}
7222 for which a picture is considered black.
7223 Default value is 0.98.
7225 @item pixel_black_th, pix_th
7226 Set the threshold for considering a pixel "black".
7228 The threshold expresses the maximum pixel luminance value for which a
7229 pixel is considered "black". The provided value is scaled according to
7230 the following equation:
7232 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
7235 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
7236 the input video format, the range is [0-255] for YUV full-range
7237 formats and [16-235] for YUV non full-range formats.
7239 Default value is 0.10.
7242 The following example sets the maximum pixel threshold to the minimum
7243 value, and detects only black intervals of 2 or more seconds:
7245 blackdetect=d=2:pix_th=0.00
7250 Detect frames that are (almost) completely black. Can be useful to
7251 detect chapter transitions or commercials. Output lines consist of
7252 the frame number of the detected frame, the percentage of blackness,
7253 the position in the file if known or -1 and the timestamp in seconds.
7255 In order to display the output lines, you need to set the loglevel at
7256 least to the AV_LOG_INFO value.
7258 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
7259 The value represents the percentage of pixels in the picture that
7260 are below the threshold value.
7262 It accepts the following parameters:
7267 The percentage of the pixels that have to be below the threshold; it defaults to
7270 @item threshold, thresh
7271 The threshold below which a pixel value is considered black; it defaults to
7279 Blend two video frames into each other.
7281 The @code{blend} filter takes two input streams and outputs one
7282 stream, the first input is the "top" layer and second input is
7283 "bottom" layer. By default, the output terminates when the longest input terminates.
7285 The @code{tblend} (time blend) filter takes two consecutive frames
7286 from one single stream, and outputs the result obtained by blending
7287 the new frame on top of the old frame.
7289 A description of the accepted options follows.
7297 Set blend mode for specific pixel component or all pixel components in case
7298 of @var{all_mode}. Default value is @code{normal}.
7300 Available values for component modes are:
7342 Set blend opacity for specific pixel component or all pixel components in case
7343 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7350 Set blend expression for specific pixel component or all pixel components in case
7351 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7353 The expressions can use the following variables:
7357 The sequential number of the filtered frame, starting from @code{0}.
7361 the coordinates of the current sample
7365 the width and height of currently filtered plane
7369 Width and height scale for the plane being filtered. It is the
7370 ratio between the dimensions of the current plane to the luma plane,
7371 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7372 the luma plane and @code{0.5,0.5} for the chroma planes.
7375 Time of the current frame, expressed in seconds.
7378 Value of pixel component at current location for first video frame (top layer).
7381 Value of pixel component at current location for second video frame (bottom layer).
7385 The @code{blend} filter also supports the @ref{framesync} options.
7387 @subsection Examples
7391 Apply transition from bottom layer to top layer in first 10 seconds:
7393 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7397 Apply linear horizontal transition from top layer to bottom layer:
7399 blend=all_expr='A*(X/W)+B*(1-X/W)'
7403 Apply 1x1 checkerboard effect:
7405 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7409 Apply uncover left effect:
7411 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7415 Apply uncover down effect:
7417 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7421 Apply uncover up-left effect:
7423 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7427 Split diagonally video and shows top and bottom layer on each side:
7429 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7433 Display differences between the current and the previous frame:
7435 tblend=all_mode=grainextract
7441 Denoise frames using Block-Matching 3D algorithm.
7443 The filter accepts the following options.
7447 Set denoising strength. Default value is 1.
7448 Allowed range is from 0 to 999.9.
7449 The denoising algorithm is very sensitive to sigma, so adjust it
7450 according to the source.
7453 Set local patch size. This sets dimensions in 2D.
7456 Set sliding step for processing blocks. Default value is 4.
7457 Allowed range is from 1 to 64.
7458 Smaller values allows processing more reference blocks and is slower.
7461 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7462 When set to 1, no block matching is done. Larger values allows more blocks
7464 Allowed range is from 1 to 256.
7467 Set radius for search block matching. Default is 9.
7468 Allowed range is from 1 to INT32_MAX.
7471 Set step between two search locations for block matching. Default is 1.
7472 Allowed range is from 1 to 64. Smaller is slower.
7475 Set threshold of mean square error for block matching. Valid range is 0 to
7479 Set thresholding parameter for hard thresholding in 3D transformed domain.
7480 Larger values results in stronger hard-thresholding filtering in frequency
7484 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7485 Default is @code{basic}.
7488 If enabled, filter will use 2nd stream for block matching.
7489 Default is disabled for @code{basic} value of @var{estim} option,
7490 and always enabled if value of @var{estim} is @code{final}.
7493 Set planes to filter. Default is all available except alpha.
7496 @subsection Examples
7500 Basic filtering with bm3d:
7502 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7506 Same as above, but filtering only luma:
7508 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7512 Same as above, but with both estimation modes:
7514 split[a][b],[a]bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7518 Same as above, but prefilter with @ref{nlmeans} filter instead:
7520 split[a][b],[a]nlmeans=s=3:r=7:p=3[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7526 Apply a boxblur algorithm to the input video.
7528 It accepts the following parameters:
7532 @item luma_radius, lr
7533 @item luma_power, lp
7534 @item chroma_radius, cr
7535 @item chroma_power, cp
7536 @item alpha_radius, ar
7537 @item alpha_power, ap
7541 A description of the accepted options follows.
7544 @item luma_radius, lr
7545 @item chroma_radius, cr
7546 @item alpha_radius, ar
7547 Set an expression for the box radius in pixels used for blurring the
7548 corresponding input plane.
7550 The radius value must be a non-negative number, and must not be
7551 greater than the value of the expression @code{min(w,h)/2} for the
7552 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7555 Default value for @option{luma_radius} is "2". If not specified,
7556 @option{chroma_radius} and @option{alpha_radius} default to the
7557 corresponding value set for @option{luma_radius}.
7559 The expressions can contain the following constants:
7563 The input width and height in pixels.
7567 The input chroma image width and height in pixels.
7571 The horizontal and vertical chroma subsample values. For example, for the
7572 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7575 @item luma_power, lp
7576 @item chroma_power, cp
7577 @item alpha_power, ap
7578 Specify how many times the boxblur filter is applied to the
7579 corresponding plane.
7581 Default value for @option{luma_power} is 2. If not specified,
7582 @option{chroma_power} and @option{alpha_power} default to the
7583 corresponding value set for @option{luma_power}.
7585 A value of 0 will disable the effect.
7588 @subsection Examples
7592 Apply a boxblur filter with the luma, chroma, and alpha radii
7595 boxblur=luma_radius=2:luma_power=1
7600 Set the luma radius to 2, and alpha and chroma radius to 0:
7602 boxblur=2:1:cr=0:ar=0
7606 Set the luma and chroma radii to a fraction of the video dimension:
7608 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7614 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7615 Deinterlacing Filter").
7617 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7618 interpolation algorithms.
7619 It accepts the following parameters:
7623 The interlacing mode to adopt. It accepts one of the following values:
7627 Output one frame for each frame.
7629 Output one frame for each field.
7632 The default value is @code{send_field}.
7635 The picture field parity assumed for the input interlaced video. It accepts one
7636 of the following values:
7640 Assume the top field is first.
7642 Assume the bottom field is first.
7644 Enable automatic detection of field parity.
7647 The default value is @code{auto}.
7648 If the interlacing is unknown or the decoder does not export this information,
7649 top field first will be assumed.
7652 Specify which frames to deinterlace. Accepts one of the following
7657 Deinterlace all frames.
7659 Only deinterlace frames marked as interlaced.
7662 The default value is @code{all}.
7667 Apply Contrast Adaptive Sharpen filter to video stream.
7669 The filter accepts the following options:
7673 Set the sharpening strength. Default value is 0.
7676 Set planes to filter. Default value is to filter all
7677 planes except alpha plane.
7680 @subsection Commands
7681 This filter supports same @ref{commands} as options.
7684 Remove all color information for all colors except for certain one.
7686 The filter accepts the following options:
7690 The color which will not be replaced with neutral chroma.
7693 Similarity percentage with the above color.
7694 0.01 matches only the exact key color, while 1.0 matches everything.
7698 0.0 makes pixels either fully gray, or not gray at all.
7699 Higher values result in more preserved color.
7702 Signals that the color passed is already in YUV instead of RGB.
7704 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7705 This can be used to pass exact YUV values as hexadecimal numbers.
7708 @subsection Commands
7709 This filter supports same @ref{commands} as options.
7710 The command accepts the same syntax of the corresponding option.
7712 If the specified expression is not valid, it is kept at its current
7716 YUV colorspace color/chroma keying.
7718 The filter accepts the following options:
7722 The color which will be replaced with transparency.
7725 Similarity percentage with the key color.
7727 0.01 matches only the exact key color, while 1.0 matches everything.
7732 0.0 makes pixels either fully transparent, or not transparent at all.
7734 Higher values result in semi-transparent pixels, with a higher transparency
7735 the more similar the pixels color is to the key color.
7738 Signals that the color passed is already in YUV instead of RGB.
7740 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7741 This can be used to pass exact YUV values as hexadecimal numbers.
7744 @subsection Commands
7745 This filter supports same @ref{commands} as options.
7746 The command accepts the same syntax of the corresponding option.
7748 If the specified expression is not valid, it is kept at its current
7751 @subsection Examples
7755 Make every green pixel in the input image transparent:
7757 ffmpeg -i input.png -vf chromakey=green out.png
7761 Overlay a greenscreen-video on top of a static black background.
7763 ffmpeg -f lavfi -i color=c=black:s=1280x720 -i video.mp4 -shortest -filter_complex "[1:v]chromakey=0x70de77:0.1:0.2[ckout];[0:v][ckout]overlay[out]" -map "[out]" output.mkv
7768 Reduce chrominance noise.
7770 The filter accepts the following options:
7774 Set threshold for averaging chrominance values.
7775 Sum of absolute difference of Y, U and V pixel components of current
7776 pixel and neighbour pixels lower than this threshold will be used in
7777 averaging. Luma component is left unchanged and is copied to output.
7778 Default value is 30. Allowed range is from 1 to 200.
7781 Set horizontal radius of rectangle used for averaging.
7782 Allowed range is from 1 to 100. Default value is 5.
7785 Set vertical radius of rectangle used for averaging.
7786 Allowed range is from 1 to 100. Default value is 5.
7789 Set horizontal step when averaging. Default value is 1.
7790 Allowed range is from 1 to 50.
7791 Mostly useful to speed-up filtering.
7794 Set vertical step when averaging. Default value is 1.
7795 Allowed range is from 1 to 50.
7796 Mostly useful to speed-up filtering.
7799 Set Y threshold for averaging chrominance values.
7800 Set finer control for max allowed difference between Y components
7801 of current pixel and neigbour pixels.
7802 Default value is 200. Allowed range is from 1 to 200.
7805 Set U threshold for averaging chrominance values.
7806 Set finer control for max allowed difference between U components
7807 of current pixel and neigbour pixels.
7808 Default value is 200. Allowed range is from 1 to 200.
7811 Set V threshold for averaging chrominance values.
7812 Set finer control for max allowed difference between V components
7813 of current pixel and neigbour pixels.
7814 Default value is 200. Allowed range is from 1 to 200.
7817 @subsection Commands
7818 This filter supports same @ref{commands} as options.
7819 The command accepts the same syntax of the corresponding option.
7821 @section chromashift
7822 Shift chroma pixels horizontally and/or vertically.
7824 The filter accepts the following options:
7827 Set amount to shift chroma-blue horizontally.
7829 Set amount to shift chroma-blue vertically.
7831 Set amount to shift chroma-red horizontally.
7833 Set amount to shift chroma-red vertically.
7835 Set edge mode, can be @var{smear}, default, or @var{warp}.
7838 @subsection Commands
7840 This filter supports the all above options as @ref{commands}.
7844 Display CIE color diagram with pixels overlaid onto it.
7846 The filter accepts the following options:
7861 @item uhdtv, rec2020
7875 Set what gamuts to draw.
7877 See @code{system} option for available values.
7880 Set ciescope size, by default set to 512.
7883 Set intensity used to map input pixel values to CIE diagram.
7886 Set contrast used to draw tongue colors that are out of active color system gamut.
7889 Correct gamma displayed on scope, by default enabled.
7892 Show white point on CIE diagram, by default disabled.
7895 Set input gamma. Used only with XYZ input color space.
7900 Visualize information exported by some codecs.
7902 Some codecs can export information through frames using side-data or other
7903 means. For example, some MPEG based codecs export motion vectors through the
7904 @var{export_mvs} flag in the codec @option{flags2} option.
7906 The filter accepts the following option:
7910 Set motion vectors to visualize.
7912 Available flags for @var{mv} are:
7916 forward predicted MVs of P-frames
7918 forward predicted MVs of B-frames
7920 backward predicted MVs of B-frames
7924 Display quantization parameters using the chroma planes.
7927 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7929 Available flags for @var{mv_type} are:
7933 forward predicted MVs
7935 backward predicted MVs
7938 @item frame_type, ft
7939 Set frame type to visualize motion vectors of.
7941 Available flags for @var{frame_type} are:
7945 intra-coded frames (I-frames)
7947 predicted frames (P-frames)
7949 bi-directionally predicted frames (B-frames)
7953 @subsection Examples
7957 Visualize forward predicted MVs of all frames using @command{ffplay}:
7959 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7963 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7965 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7969 @section colorbalance
7970 Modify intensity of primary colors (red, green and blue) of input frames.
7972 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7973 regions for the red-cyan, green-magenta or blue-yellow balance.
7975 A positive adjustment value shifts the balance towards the primary color, a negative
7976 value towards the complementary color.
7978 The filter accepts the following options:
7984 Adjust red, green and blue shadows (darkest pixels).
7989 Adjust red, green and blue midtones (medium pixels).
7994 Adjust red, green and blue highlights (brightest pixels).
7996 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7999 Preserve lightness when changing color balance. Default is disabled.
8002 @subsection Examples
8006 Add red color cast to shadows:
8012 @subsection Commands
8014 This filter supports the all above options as @ref{commands}.
8016 @section colorchannelmixer
8018 Adjust video input frames by re-mixing color channels.
8020 This filter modifies a color channel by adding the values associated to
8021 the other channels of the same pixels. For example if the value to
8022 modify is red, the output value will be:
8024 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
8027 The filter accepts the following options:
8034 Adjust contribution of input red, green, blue and alpha channels for output red channel.
8035 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
8041 Adjust contribution of input red, green, blue and alpha channels for output green channel.
8042 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
8048 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
8049 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
8055 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
8056 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
8058 Allowed ranges for options are @code{[-2.0, 2.0]}.
8061 @subsection Examples
8065 Convert source to grayscale:
8067 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
8070 Simulate sepia tones:
8072 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
8076 @subsection Commands
8078 This filter supports the all above options as @ref{commands}.
8081 RGB colorspace color keying.
8083 The filter accepts the following options:
8087 The color which will be replaced with transparency.
8090 Similarity percentage with the key color.
8092 0.01 matches only the exact key color, while 1.0 matches everything.
8097 0.0 makes pixels either fully transparent, or not transparent at all.
8099 Higher values result in semi-transparent pixels, with a higher transparency
8100 the more similar the pixels color is to the key color.
8103 @subsection Examples
8107 Make every green pixel in the input image transparent:
8109 ffmpeg -i input.png -vf colorkey=green out.png
8113 Overlay a greenscreen-video on top of a static background image.
8115 ffmpeg -i background.png -i video.mp4 -filter_complex "[1:v]colorkey=0x3BBD1E:0.3:0.2[ckout];[0:v][ckout]overlay[out]" -map "[out]" output.flv
8119 @subsection Commands
8120 This filter supports same @ref{commands} as options.
8121 The command accepts the same syntax of the corresponding option.
8123 If the specified expression is not valid, it is kept at its current
8127 Remove all color information for all RGB colors except for certain one.
8129 The filter accepts the following options:
8133 The color which will not be replaced with neutral gray.
8136 Similarity percentage with the above color.
8137 0.01 matches only the exact key color, while 1.0 matches everything.
8140 Blend percentage. 0.0 makes pixels fully gray.
8141 Higher values result in more preserved color.
8144 @subsection Commands
8145 This filter supports same @ref{commands} as options.
8146 The command accepts the same syntax of the corresponding option.
8148 If the specified expression is not valid, it is kept at its current
8151 @section colorlevels
8153 Adjust video input frames using levels.
8155 The filter accepts the following options:
8162 Adjust red, green, blue and alpha input black point.
8163 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
8169 Adjust red, green, blue and alpha input white point.
8170 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
8172 Input levels are used to lighten highlights (bright tones), darken shadows
8173 (dark tones), change the balance of bright and dark tones.
8179 Adjust red, green, blue and alpha output black point.
8180 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
8186 Adjust red, green, blue and alpha output white point.
8187 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
8189 Output levels allows manual selection of a constrained output level range.
8192 @subsection Examples
8196 Make video output darker:
8198 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
8204 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
8208 Make video output lighter:
8210 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
8214 Increase brightness:
8216 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
8220 @subsection Commands
8222 This filter supports the all above options as @ref{commands}.
8224 @section colormatrix
8226 Convert color matrix.
8228 The filter accepts the following options:
8233 Specify the source and destination color matrix. Both values must be
8236 The accepted values are:
8264 For example to convert from BT.601 to SMPTE-240M, use the command:
8266 colormatrix=bt601:smpte240m
8271 Convert colorspace, transfer characteristics or color primaries.
8272 Input video needs to have an even size.
8274 The filter accepts the following options:
8279 Specify all color properties at once.
8281 The accepted values are:
8311 Specify output colorspace.
8313 The accepted values are:
8322 BT.470BG or BT.601-6 625
8325 SMPTE-170M or BT.601-6 525
8334 BT.2020 with non-constant luminance
8340 Specify output transfer characteristics.
8342 The accepted values are:
8354 Constant gamma of 2.2
8357 Constant gamma of 2.8
8360 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8378 BT.2020 for 10-bits content
8381 BT.2020 for 12-bits content
8387 Specify output color primaries.
8389 The accepted values are:
8398 BT.470BG or BT.601-6 625
8401 SMPTE-170M or BT.601-6 525
8425 Specify output color range.
8427 The accepted values are:
8430 TV (restricted) range
8433 MPEG (restricted) range
8444 Specify output color format.
8446 The accepted values are:
8449 YUV 4:2:0 planar 8-bits
8452 YUV 4:2:0 planar 10-bits
8455 YUV 4:2:0 planar 12-bits
8458 YUV 4:2:2 planar 8-bits
8461 YUV 4:2:2 planar 10-bits
8464 YUV 4:2:2 planar 12-bits
8467 YUV 4:4:4 planar 8-bits
8470 YUV 4:4:4 planar 10-bits
8473 YUV 4:4:4 planar 12-bits
8478 Do a fast conversion, which skips gamma/primary correction. This will take
8479 significantly less CPU, but will be mathematically incorrect. To get output
8480 compatible with that produced by the colormatrix filter, use fast=1.
8483 Specify dithering mode.
8485 The accepted values are:
8491 Floyd-Steinberg dithering
8495 Whitepoint adaptation mode.
8497 The accepted values are:
8500 Bradford whitepoint adaptation
8503 von Kries whitepoint adaptation
8506 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8510 Override all input properties at once. Same accepted values as @ref{all}.
8513 Override input colorspace. Same accepted values as @ref{space}.
8516 Override input color primaries. Same accepted values as @ref{primaries}.
8519 Override input transfer characteristics. Same accepted values as @ref{trc}.
8522 Override input color range. Same accepted values as @ref{range}.
8526 The filter converts the transfer characteristics, color space and color
8527 primaries to the specified user values. The output value, if not specified,
8528 is set to a default value based on the "all" property. If that property is
8529 also not specified, the filter will log an error. The output color range and
8530 format default to the same value as the input color range and format. The
8531 input transfer characteristics, color space, color primaries and color range
8532 should be set on the input data. If any of these are missing, the filter will
8533 log an error and no conversion will take place.
8535 For example to convert the input to SMPTE-240M, use the command:
8537 colorspace=smpte240m
8540 @section convolution
8542 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8544 The filter accepts the following options:
8551 Set matrix for each plane.
8552 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8553 and from 1 to 49 odd number of signed integers in @var{row} mode.
8559 Set multiplier for calculated value for each plane.
8560 If unset or 0, it will be sum of all matrix elements.
8566 Set bias for each plane. This value is added to the result of the multiplication.
8567 Useful for making the overall image brighter or darker. Default is 0.0.
8573 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8574 Default is @var{square}.
8577 @subsection Commands
8579 This filter supports the all above options as @ref{commands}.
8581 @subsection Examples
8587 convolution="0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0"
8593 convolution="1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1/9:1/9:1/9:1/9"
8599 convolution="0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:5:1:1:1:0:128:128:128"
8605 convolution="0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:5:5:5:1:0:128:128:128"
8609 Apply laplacian edge detector which includes diagonals:
8611 convolution="1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:5:5:5:1:0:128:128:0"
8617 convolution="-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2"
8623 Apply 2D convolution of video stream in frequency domain using second stream
8626 The filter accepts the following options:
8630 Set which planes to process.
8633 Set which impulse video frames will be processed, can be @var{first}
8634 or @var{all}. Default is @var{all}.
8637 The @code{convolve} filter also supports the @ref{framesync} options.
8641 Copy the input video source unchanged to the output. This is mainly useful for
8646 Video filtering on GPU using Apple's CoreImage API on OSX.
8648 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8649 processed by video hardware. However, software-based OpenGL implementations
8650 exist which means there is no guarantee for hardware processing. It depends on
8653 There are many filters and image generators provided by Apple that come with a
8654 large variety of options. The filter has to be referenced by its name along
8657 The coreimage filter accepts the following options:
8660 List all available filters and generators along with all their respective
8661 options as well as possible minimum and maximum values along with the default
8668 Specify all filters by their respective name and options.
8669 Use @var{list_filters} to determine all valid filter names and options.
8670 Numerical options are specified by a float value and are automatically clamped
8671 to their respective value range. Vector and color options have to be specified
8672 by a list of space separated float values. Character escaping has to be done.
8673 A special option name @code{default} is available to use default options for a
8676 It is required to specify either @code{default} or at least one of the filter options.
8677 All omitted options are used with their default values.
8678 The syntax of the filter string is as follows:
8680 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8684 Specify a rectangle where the output of the filter chain is copied into the
8685 input image. It is given by a list of space separated float values:
8687 output_rect=x\ y\ width\ height
8689 If not given, the output rectangle equals the dimensions of the input image.
8690 The output rectangle is automatically cropped at the borders of the input
8691 image. Negative values are valid for each component.
8693 output_rect=25\ 25\ 100\ 100
8697 Several filters can be chained for successive processing without GPU-HOST
8698 transfers allowing for fast processing of complex filter chains.
8699 Currently, only filters with zero (generators) or exactly one (filters) input
8700 image and one output image are supported. Also, transition filters are not yet
8703 Some filters generate output images with additional padding depending on the
8704 respective filter kernel. The padding is automatically removed to ensure the
8705 filter output has the same size as the input image.
8707 For image generators, the size of the output image is determined by the
8708 previous output image of the filter chain or the input image of the whole
8709 filterchain, respectively. The generators do not use the pixel information of
8710 this image to generate their output. However, the generated output is
8711 blended onto this image, resulting in partial or complete coverage of the
8714 The @ref{coreimagesrc} video source can be used for generating input images
8715 which are directly fed into the filter chain. By using it, providing input
8716 images by another video source or an input video is not required.
8718 @subsection Examples
8723 List all filters available:
8725 coreimage=list_filters=true
8729 Use the CIBoxBlur filter with default options to blur an image:
8731 coreimage=filter=CIBoxBlur@@default
8735 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8736 its center at 100x100 and a radius of 50 pixels:
8738 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8742 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8743 given as complete and escaped command-line for Apple's standard bash shell:
8745 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8751 Cover a rectangular object
8753 It accepts the following options:
8757 Filepath of the optional cover image, needs to be in yuv420.
8762 It accepts the following values:
8765 cover it by the supplied image
8767 cover it by interpolating the surrounding pixels
8770 Default value is @var{blur}.
8773 @subsection Examples
8777 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8779 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8785 Crop the input video to given dimensions.
8787 It accepts the following parameters:
8791 The width of the output video. It defaults to @code{iw}.
8792 This expression is evaluated only once during the filter
8793 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8796 The height of the output video. It defaults to @code{ih}.
8797 This expression is evaluated only once during the filter
8798 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8801 The horizontal position, in the input video, of the left edge of the output
8802 video. It defaults to @code{(in_w-out_w)/2}.
8803 This expression is evaluated per-frame.
8806 The vertical position, in the input video, of the top edge of the output video.
8807 It defaults to @code{(in_h-out_h)/2}.
8808 This expression is evaluated per-frame.
8811 If set to 1 will force the output display aspect ratio
8812 to be the same of the input, by changing the output sample aspect
8813 ratio. It defaults to 0.
8816 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8817 width/height/x/y as specified and will not be rounded to nearest smaller value.
8821 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8822 expressions containing the following constants:
8827 The computed values for @var{x} and @var{y}. They are evaluated for
8832 The input width and height.
8836 These are the same as @var{in_w} and @var{in_h}.
8840 The output (cropped) width and height.
8844 These are the same as @var{out_w} and @var{out_h}.
8847 same as @var{iw} / @var{ih}
8850 input sample aspect ratio
8853 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8857 horizontal and vertical chroma subsample values. For example for the
8858 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8861 The number of the input frame, starting from 0.
8864 the position in the file of the input frame, NAN if unknown
8867 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8871 The expression for @var{out_w} may depend on the value of @var{out_h},
8872 and the expression for @var{out_h} may depend on @var{out_w}, but they
8873 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8874 evaluated after @var{out_w} and @var{out_h}.
8876 The @var{x} and @var{y} parameters specify the expressions for the
8877 position of the top-left corner of the output (non-cropped) area. They
8878 are evaluated for each frame. If the evaluated value is not valid, it
8879 is approximated to the nearest valid value.
8881 The expression for @var{x} may depend on @var{y}, and the expression
8882 for @var{y} may depend on @var{x}.
8884 @subsection Examples
8888 Crop area with size 100x100 at position (12,34).
8893 Using named options, the example above becomes:
8895 crop=w=100:h=100:x=12:y=34
8899 Crop the central input area with size 100x100:
8905 Crop the central input area with size 2/3 of the input video:
8907 crop=2/3*in_w:2/3*in_h
8911 Crop the input video central square:
8918 Delimit the rectangle with the top-left corner placed at position
8919 100:100 and the right-bottom corner corresponding to the right-bottom
8920 corner of the input image.
8922 crop=in_w-100:in_h-100:100:100
8926 Crop 10 pixels from the left and right borders, and 20 pixels from
8927 the top and bottom borders
8929 crop=in_w-2*10:in_h-2*20
8933 Keep only the bottom right quarter of the input image:
8935 crop=in_w/2:in_h/2:in_w/2:in_h/2
8939 Crop height for getting Greek harmony:
8941 crop=in_w:1/PHI*in_w
8945 Apply trembling effect:
8947 crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(n/10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(n/7)
8951 Apply erratic camera effect depending on timestamp:
8953 crop=in_w/2:in_h/2:(in_w-out_w)/2+((in_w-out_w)/2)*sin(t*10):(in_h-out_h)/2 +((in_h-out_h)/2)*sin(t*13)"
8957 Set x depending on the value of y:
8959 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8963 @subsection Commands
8965 This filter supports the following commands:
8971 Set width/height of the output video and the horizontal/vertical position
8973 The command accepts the same syntax of the corresponding option.
8975 If the specified expression is not valid, it is kept at its current
8981 Auto-detect the crop size.
8983 It calculates the necessary cropping parameters and prints the
8984 recommended parameters via the logging system. The detected dimensions
8985 correspond to the non-black area of the input video.
8987 It accepts the following parameters:
8992 Set higher black value threshold, which can be optionally specified
8993 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8994 value greater to the set value is considered non-black. It defaults to 24.
8995 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8996 on the bitdepth of the pixel format.
8999 The value which the width/height should be divisible by. It defaults to
9000 16. The offset is automatically adjusted to center the video. Use 2 to
9001 get only even dimensions (needed for 4:2:2 video). 16 is best when
9002 encoding to most video codecs.
9005 Set the number of initial frames for which evaluation is skipped.
9006 Default is 2. Range is 0 to INT_MAX.
9008 @item reset_count, reset
9009 Set the counter that determines after how many frames cropdetect will
9010 reset the previously detected largest video area and start over to
9011 detect the current optimal crop area. Default value is 0.
9013 This can be useful when channel logos distort the video area. 0
9014 indicates 'never reset', and returns the largest area encountered during
9021 Delay video filtering until a given wallclock timestamp. The filter first
9022 passes on @option{preroll} amount of frames, then it buffers at most
9023 @option{buffer} amount of frames and waits for the cue. After reaching the cue
9024 it forwards the buffered frames and also any subsequent frames coming in its
9027 The filter can be used synchronize the output of multiple ffmpeg processes for
9028 realtime output devices like decklink. By putting the delay in the filtering
9029 chain and pre-buffering frames the process can pass on data to output almost
9030 immediately after the target wallclock timestamp is reached.
9032 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
9038 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
9041 The duration of content to pass on as preroll expressed in seconds. Default is 0.
9044 The maximum duration of content to buffer before waiting for the cue expressed
9045 in seconds. Default is 0.
9052 Apply color adjustments using curves.
9054 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
9055 component (red, green and blue) has its values defined by @var{N} key points
9056 tied from each other using a smooth curve. The x-axis represents the pixel
9057 values from the input frame, and the y-axis the new pixel values to be set for
9060 By default, a component curve is defined by the two points @var{(0;0)} and
9061 @var{(1;1)}. This creates a straight line where each original pixel value is
9062 "adjusted" to its own value, which means no change to the image.
9064 The filter allows you to redefine these two points and add some more. A new
9065 curve (using a natural cubic spline interpolation) will be define to pass
9066 smoothly through all these new coordinates. The new defined points needs to be
9067 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
9068 be in the @var{[0;1]} interval. If the computed curves happened to go outside
9069 the vector spaces, the values will be clipped accordingly.
9071 The filter accepts the following options:
9075 Select one of the available color presets. This option can be used in addition
9076 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
9077 options takes priority on the preset values.
9078 Available presets are:
9081 @item color_negative
9084 @item increase_contrast
9086 @item linear_contrast
9087 @item medium_contrast
9089 @item strong_contrast
9092 Default is @code{none}.
9094 Set the master key points. These points will define a second pass mapping. It
9095 is sometimes called a "luminance" or "value" mapping. It can be used with
9096 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
9097 post-processing LUT.
9099 Set the key points for the red component.
9101 Set the key points for the green component.
9103 Set the key points for the blue component.
9105 Set the key points for all components (not including master).
9106 Can be used in addition to the other key points component
9107 options. In this case, the unset component(s) will fallback on this
9108 @option{all} setting.
9110 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
9112 Save Gnuplot script of the curves in specified file.
9115 To avoid some filtergraph syntax conflicts, each key points list need to be
9116 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
9118 @subsection Examples
9122 Increase slightly the middle level of blue:
9124 curves=blue='0/0 0.5/0.58 1/1'
9130 curves=r='0/0.11 .42/.51 1/0.95':g='0/0 0.50/0.48 1/1':b='0/0.22 .49/.44 1/0.8'
9132 Here we obtain the following coordinates for each components:
9135 @code{(0;0.11) (0.42;0.51) (1;0.95)}
9137 @code{(0;0) (0.50;0.48) (1;1)}
9139 @code{(0;0.22) (0.49;0.44) (1;0.80)}
9143 The previous example can also be achieved with the associated built-in preset:
9145 curves=preset=vintage
9155 Use a Photoshop preset and redefine the points of the green component:
9157 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
9161 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
9162 and @command{gnuplot}:
9164 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
9165 gnuplot -p /tmp/curves.plt
9171 Video data analysis filter.
9173 This filter shows hexadecimal pixel values of part of video.
9175 The filter accepts the following options:
9179 Set output video size.
9182 Set x offset from where to pick pixels.
9185 Set y offset from where to pick pixels.
9188 Set scope mode, can be one of the following:
9191 Draw hexadecimal pixel values with white color on black background.
9194 Draw hexadecimal pixel values with input video pixel color on black
9198 Draw hexadecimal pixel values on color background picked from input video,
9199 the text color is picked in such way so its always visible.
9203 Draw rows and columns numbers on left and top of video.
9206 Set background opacity.
9209 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
9212 Set pixel components to display. By default all pixel components are displayed.
9216 Apply Directional blur filter.
9218 The filter accepts the following options:
9222 Set angle of directional blur. Default is @code{45}.
9225 Set radius of directional blur. Default is @code{5}.
9228 Set which planes to filter. By default all planes are filtered.
9231 @subsection Commands
9232 This filter supports same @ref{commands} as options.
9233 The command accepts the same syntax of the corresponding option.
9235 If the specified expression is not valid, it is kept at its current
9240 Denoise frames using 2D DCT (frequency domain filtering).
9242 This filter is not designed for real time.
9244 The filter accepts the following options:
9248 Set the noise sigma constant.
9250 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
9251 coefficient (absolute value) below this threshold with be dropped.
9253 If you need a more advanced filtering, see @option{expr}.
9255 Default is @code{0}.
9258 Set number overlapping pixels for each block. Since the filter can be slow, you
9259 may want to reduce this value, at the cost of a less effective filter and the
9260 risk of various artefacts.
9262 If the overlapping value doesn't permit processing the whole input width or
9263 height, a warning will be displayed and according borders won't be denoised.
9265 Default value is @var{blocksize}-1, which is the best possible setting.
9268 Set the coefficient factor expression.
9270 For each coefficient of a DCT block, this expression will be evaluated as a
9271 multiplier value for the coefficient.
9273 If this is option is set, the @option{sigma} option will be ignored.
9275 The absolute value of the coefficient can be accessed through the @var{c}
9279 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
9280 @var{blocksize}, which is the width and height of the processed blocks.
9282 The default value is @var{3} (8x8) and can be raised to @var{4} for a
9283 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
9284 on the speed processing. Also, a larger block size does not necessarily means a
9288 @subsection Examples
9290 Apply a denoise with a @option{sigma} of @code{4.5}:
9295 The same operation can be achieved using the expression system:
9297 dctdnoiz=e='gte(c, 4.5*3)'
9300 Violent denoise using a block size of @code{16x16}:
9307 Remove banding artifacts from input video.
9308 It works by replacing banded pixels with average value of referenced pixels.
9310 The filter accepts the following options:
9317 Set banding detection threshold for each plane. Default is 0.02.
9318 Valid range is 0.00003 to 0.5.
9319 If difference between current pixel and reference pixel is less than threshold,
9320 it will be considered as banded.
9323 Banding detection range in pixels. Default is 16. If positive, random number
9324 in range 0 to set value will be used. If negative, exact absolute value
9326 The range defines square of four pixels around current pixel.
9329 Set direction in radians from which four pixel will be compared. If positive,
9330 random direction from 0 to set direction will be picked. If negative, exact of
9331 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9332 will pick only pixels on same row and -PI/2 will pick only pixels on same
9336 If enabled, current pixel is compared with average value of all four
9337 surrounding pixels. The default is enabled. If disabled current pixel is
9338 compared with all four surrounding pixels. The pixel is considered banded
9339 if only all four differences with surrounding pixels are less than threshold.
9342 If enabled, current pixel is changed if and only if all pixel components are banded,
9343 e.g. banding detection threshold is triggered for all color components.
9344 The default is disabled.
9349 Remove blocking artifacts from input video.
9351 The filter accepts the following options:
9355 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9356 This controls what kind of deblocking is applied.
9359 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9365 Set blocking detection thresholds. Allowed range is 0 to 1.
9366 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9367 Using higher threshold gives more deblocking strength.
9368 Setting @var{alpha} controls threshold detection at exact edge of block.
9369 Remaining options controls threshold detection near the edge. Each one for
9370 below/above or left/right. Setting any of those to @var{0} disables
9374 Set planes to filter. Default is to filter all available planes.
9377 @subsection Examples
9381 Deblock using weak filter and block size of 4 pixels.
9383 deblock=filter=weak:block=4
9387 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9388 deblocking more edges.
9390 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9394 Similar as above, but filter only first plane.
9396 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9400 Similar as above, but filter only second and third plane.
9402 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9409 Drop duplicated frames at regular intervals.
9411 The filter accepts the following options:
9415 Set the number of frames from which one will be dropped. Setting this to
9416 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9417 Default is @code{5}.
9420 Set the threshold for duplicate detection. If the difference metric for a frame
9421 is less than or equal to this value, then it is declared as duplicate. Default
9425 Set scene change threshold. Default is @code{15}.
9429 Set the size of the x and y-axis blocks used during metric calculations.
9430 Larger blocks give better noise suppression, but also give worse detection of
9431 small movements. Must be a power of two. Default is @code{32}.
9434 Mark main input as a pre-processed input and activate clean source input
9435 stream. This allows the input to be pre-processed with various filters to help
9436 the metrics calculation while keeping the frame selection lossless. When set to
9437 @code{1}, the first stream is for the pre-processed input, and the second
9438 stream is the clean source from where the kept frames are chosen. Default is
9442 Set whether or not chroma is considered in the metric calculations. Default is
9448 Apply 2D deconvolution of video stream in frequency domain using second stream
9451 The filter accepts the following options:
9455 Set which planes to process.
9458 Set which impulse video frames will be processed, can be @var{first}
9459 or @var{all}. Default is @var{all}.
9462 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9463 and height are not same and not power of 2 or if stream prior to convolving
9467 The @code{deconvolve} filter also supports the @ref{framesync} options.
9471 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9473 It accepts the following options:
9477 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9478 @var{rainbows} for cross-color reduction.
9481 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9484 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9487 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9490 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9495 Apply deflate effect to the video.
9497 This filter replaces the pixel by the local(3x3) average by taking into account
9498 only values lower than the pixel.
9500 It accepts the following options:
9507 Limit the maximum change for each plane, default is 65535.
9508 If 0, plane will remain unchanged.
9511 @subsection Commands
9513 This filter supports the all above options as @ref{commands}.
9517 Remove temporal frame luminance variations.
9519 It accepts the following options:
9523 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9526 Set averaging mode to smooth temporal luminance variations.
9528 Available values are:
9553 Do not actually modify frame. Useful when one only wants metadata.
9558 Remove judder produced by partially interlaced telecined content.
9560 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9561 source was partially telecined content then the output of @code{pullup,dejudder}
9562 will have a variable frame rate. May change the recorded frame rate of the
9563 container. Aside from that change, this filter will not affect constant frame
9566 The option available in this filter is:
9570 Specify the length of the window over which the judder repeats.
9572 Accepts any integer greater than 1. Useful values are:
9576 If the original was telecined from 24 to 30 fps (Film to NTSC).
9579 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9582 If a mixture of the two.
9585 The default is @samp{4}.
9590 Suppress a TV station logo by a simple interpolation of the surrounding
9591 pixels. Just set a rectangle covering the logo and watch it disappear
9592 (and sometimes something even uglier appear - your mileage may vary).
9594 It accepts the following parameters:
9599 Specify the top left corner coordinates of the logo. They must be
9604 Specify the width and height of the logo to clear. They must be
9608 Specify the thickness of the fuzzy edge of the rectangle (added to
9609 @var{w} and @var{h}). The default value is 1. This option is
9610 deprecated, setting higher values should no longer be necessary and
9614 When set to 1, a green rectangle is drawn on the screen to simplify
9615 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9616 The default value is 0.
9618 The rectangle is drawn on the outermost pixels which will be (partly)
9619 replaced with interpolated values. The values of the next pixels
9620 immediately outside this rectangle in each direction will be used to
9621 compute the interpolated pixel values inside the rectangle.
9625 @subsection Examples
9629 Set a rectangle covering the area with top left corner coordinates 0,0
9630 and size 100x77, and a band of size 10:
9632 delogo=x=0:y=0:w=100:h=77:band=10
9640 Remove the rain in the input image/video by applying the derain methods based on
9641 convolutional neural networks. Supported models:
9645 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9646 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9649 Training as well as model generation scripts are provided in
9650 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9652 Native model files (.model) can be generated from TensorFlow model
9653 files (.pb) by using tools/python/convert.py
9655 The filter accepts the following options:
9659 Specify which filter to use. This option accepts the following values:
9663 Derain filter. To conduct derain filter, you need to use a derain model.
9666 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9668 Default value is @samp{derain}.
9671 Specify which DNN backend to use for model loading and execution. This option accepts
9672 the following values:
9676 Native implementation of DNN loading and execution.
9679 TensorFlow backend. To enable this backend you
9680 need to install the TensorFlow for C library (see
9681 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9682 @code{--enable-libtensorflow}
9684 Default value is @samp{native}.
9687 Set path to model file specifying network architecture and its parameters.
9688 Note that different backends use different file formats. TensorFlow and native
9689 backend can load files for only its format.
9692 It can also be finished with @ref{dnn_processing} filter.
9696 Attempt to fix small changes in horizontal and/or vertical shift. This
9697 filter helps remove camera shake from hand-holding a camera, bumping a
9698 tripod, moving on a vehicle, etc.
9700 The filter accepts the following options:
9708 Specify a rectangular area where to limit the search for motion
9710 If desired the search for motion vectors can be limited to a
9711 rectangular area of the frame defined by its top left corner, width
9712 and height. These parameters have the same meaning as the drawbox
9713 filter which can be used to visualise the position of the bounding
9716 This is useful when simultaneous movement of subjects within the frame
9717 might be confused for camera motion by the motion vector search.
9719 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9720 then the full frame is used. This allows later options to be set
9721 without specifying the bounding box for the motion vector search.
9723 Default - search the whole frame.
9727 Specify the maximum extent of movement in x and y directions in the
9728 range 0-64 pixels. Default 16.
9731 Specify how to generate pixels to fill blanks at the edge of the
9732 frame. Available values are:
9735 Fill zeroes at blank locations
9737 Original image at blank locations
9739 Extruded edge value at blank locations
9741 Mirrored edge at blank locations
9743 Default value is @samp{mirror}.
9746 Specify the blocksize to use for motion search. Range 4-128 pixels,
9750 Specify the contrast threshold for blocks. Only blocks with more than
9751 the specified contrast (difference between darkest and lightest
9752 pixels) will be considered. Range 1-255, default 125.
9755 Specify the search strategy. Available values are:
9758 Set exhaustive search
9760 Set less exhaustive search.
9762 Default value is @samp{exhaustive}.
9765 If set then a detailed log of the motion search is written to the
9772 Remove unwanted contamination of foreground colors, caused by reflected color of
9773 greenscreen or bluescreen.
9775 This filter accepts the following options:
9779 Set what type of despill to use.
9782 Set how spillmap will be generated.
9785 Set how much to get rid of still remaining spill.
9788 Controls amount of red in spill area.
9791 Controls amount of green in spill area.
9792 Should be -1 for greenscreen.
9795 Controls amount of blue in spill area.
9796 Should be -1 for bluescreen.
9799 Controls brightness of spill area, preserving colors.
9802 Modify alpha from generated spillmap.
9805 @subsection Commands
9807 This filter supports the all above options as @ref{commands}.
9811 Apply an exact inverse of the telecine operation. It requires a predefined
9812 pattern specified using the pattern option which must be the same as that passed
9813 to the telecine filter.
9815 This filter accepts the following options:
9824 The default value is @code{top}.
9828 A string of numbers representing the pulldown pattern you wish to apply.
9829 The default value is @code{23}.
9832 A number representing position of the first frame with respect to the telecine
9833 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9838 Apply dilation effect to the video.
9840 This filter replaces the pixel by the local(3x3) maximum.
9842 It accepts the following options:
9849 Limit the maximum change for each plane, default is 65535.
9850 If 0, plane will remain unchanged.
9853 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9856 Flags to local 3x3 coordinates maps like this:
9863 @subsection Commands
9865 This filter supports the all above options as @ref{commands}.
9869 Displace pixels as indicated by second and third input stream.
9871 It takes three input streams and outputs one stream, the first input is the
9872 source, and second and third input are displacement maps.
9874 The second input specifies how much to displace pixels along the
9875 x-axis, while the third input specifies how much to displace pixels
9877 If one of displacement map streams terminates, last frame from that
9878 displacement map will be used.
9880 Note that once generated, displacements maps can be reused over and over again.
9882 A description of the accepted options follows.
9886 Set displace behavior for pixels that are out of range.
9888 Available values are:
9891 Missing pixels are replaced by black pixels.
9894 Adjacent pixels will spread out to replace missing pixels.
9897 Out of range pixels are wrapped so they point to pixels of other side.
9900 Out of range pixels will be replaced with mirrored pixels.
9902 Default is @samp{smear}.
9906 @subsection Examples
9910 Add ripple effect to rgb input of video size hd720:
9912 ffmpeg -i INPUT -f lavfi -i nullsrc=s=hd720,lutrgb=128:128:128 -f lavfi -i nullsrc=s=hd720,geq='r=128+30*sin(2*PI*X/400+T):g=128+30*sin(2*PI*X/400+T):b=128+30*sin(2*PI*X/400+T)' -lavfi '[0][1][2]displace' OUTPUT
9916 Add wave effect to rgb input of video size hd720:
9918 ffmpeg -i INPUT -f lavfi -i nullsrc=hd720,geq='r=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):g=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T)):b=128+80*(sin(sqrt((X-W/2)*(X-W/2)+(Y-H/2)*(Y-H/2))/220*2*PI+T))' -lavfi '[1]split[x][y],[0][x][y]displace' OUTPUT
9922 @anchor{dnn_processing}
9923 @section dnn_processing
9925 Do image processing with deep neural networks. It works together with another filter
9926 which converts the pixel format of the Frame to what the dnn network requires.
9928 The filter accepts the following options:
9932 Specify which DNN backend to use for model loading and execution. This option accepts
9933 the following values:
9937 Native implementation of DNN loading and execution.
9940 TensorFlow backend. To enable this backend you
9941 need to install the TensorFlow for C library (see
9942 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9943 @code{--enable-libtensorflow}
9946 OpenVINO backend. To enable this backend you
9947 need to build and install the OpenVINO for C library (see
9948 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9949 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9950 be needed if the header files and libraries are not installed into system path)
9954 Default value is @samp{native}.
9957 Set path to model file specifying network architecture and its parameters.
9958 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9959 backend can load files for only its format.
9961 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9964 Set the input name of the dnn network.
9967 Set the output name of the dnn network.
9970 use DNN async execution if set (default: set),
9971 roll back to sync execution if the backend does not support async.
9975 @subsection Examples
9979 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9981 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9985 Halve the pixel value of the frame with format gray32f:
9987 ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
9991 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9993 ./ffmpeg -i 480p.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.jpg
9997 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9999 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
10006 Draw a colored box on the input image.
10008 It accepts the following parameters:
10013 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
10017 The expressions which specify the width and height of the box; if 0 they are interpreted as
10018 the input width and height. It defaults to 0.
10021 Specify the color of the box to write. For the general syntax of this option,
10022 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10023 value @code{invert} is used, the box edge color is the same as the
10024 video with inverted luma.
10027 The expression which sets the thickness of the box edge.
10028 A value of @code{fill} will create a filled box. Default value is @code{3}.
10030 See below for the list of accepted constants.
10033 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
10034 will overwrite the video's color and alpha pixels.
10035 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
10038 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10039 following constants:
10043 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10047 horizontal and vertical chroma subsample values. For example for the
10048 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10052 The input width and height.
10055 The input sample aspect ratio.
10059 The x and y offset coordinates where the box is drawn.
10063 The width and height of the drawn box.
10066 The thickness of the drawn box.
10068 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10069 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10073 @subsection Examples
10077 Draw a black box around the edge of the input image:
10083 Draw a box with color red and an opacity of 50%:
10085 drawbox=10:20:200:60:red@@0.5
10088 The previous example can be specified as:
10090 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
10094 Fill the box with pink color:
10096 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
10100 Draw a 2-pixel red 2.40:1 mask:
10102 drawbox=x=-t:y=0.5*(ih-iw/2.4)-t:w=iw+t*2:h=iw/2.4+t*2:t=2:c=red
10106 @subsection Commands
10107 This filter supports same commands as options.
10108 The command accepts the same syntax of the corresponding option.
10110 If the specified expression is not valid, it is kept at its current
10115 Draw a graph using input video metadata.
10117 It accepts the following parameters:
10121 Set 1st frame metadata key from which metadata values will be used to draw a graph.
10124 Set 1st foreground color expression.
10127 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
10130 Set 2nd foreground color expression.
10133 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
10136 Set 3rd foreground color expression.
10139 Set 4th frame metadata key from which metadata values will be used to draw a graph.
10142 Set 4th foreground color expression.
10145 Set minimal value of metadata value.
10148 Set maximal value of metadata value.
10151 Set graph background color. Default is white.
10156 Available values for mode is:
10163 Default is @code{line}.
10168 Available values for slide is:
10171 Draw new frame when right border is reached.
10174 Replace old columns with new ones.
10177 Scroll from right to left.
10180 Scroll from left to right.
10183 Draw single picture.
10186 Default is @code{frame}.
10189 Set size of graph video. For the syntax of this option, check the
10190 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
10191 The default value is @code{900x256}.
10194 Set the output frame rate. Default value is @code{25}.
10196 The foreground color expressions can use the following variables:
10199 Minimal value of metadata value.
10202 Maximal value of metadata value.
10205 Current metadata key value.
10208 The color is defined as 0xAABBGGRR.
10211 Example using metadata from @ref{signalstats} filter:
10213 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
10216 Example using metadata from @ref{ebur128} filter:
10218 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
10223 Draw a grid on the input image.
10225 It accepts the following parameters:
10230 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
10234 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
10235 input width and height, respectively, minus @code{thickness}, so image gets
10236 framed. Default to 0.
10239 Specify the color of the grid. For the general syntax of this option,
10240 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10241 value @code{invert} is used, the grid color is the same as the
10242 video with inverted luma.
10245 The expression which sets the thickness of the grid line. Default value is @code{1}.
10247 See below for the list of accepted constants.
10250 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
10251 will overwrite the video's color and alpha pixels.
10252 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
10255 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10256 following constants:
10260 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10264 horizontal and vertical chroma subsample values. For example for the
10265 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10269 The input grid cell width and height.
10272 The input sample aspect ratio.
10276 The x and y coordinates of some point of grid intersection (meant to configure offset).
10280 The width and height of the drawn cell.
10283 The thickness of the drawn cell.
10285 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10286 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10290 @subsection Examples
10294 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
10296 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
10300 Draw a white 3x3 grid with an opacity of 50%:
10302 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10306 @subsection Commands
10307 This filter supports same commands as options.
10308 The command accepts the same syntax of the corresponding option.
10310 If the specified expression is not valid, it is kept at its current
10316 Draw a text string or text from a specified file on top of a video, using the
10317 libfreetype library.
10319 To enable compilation of this filter, you need to configure FFmpeg with
10320 @code{--enable-libfreetype}.
10321 To enable default font fallback and the @var{font} option you need to
10322 configure FFmpeg with @code{--enable-libfontconfig}.
10323 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10324 @code{--enable-libfribidi}.
10328 It accepts the following parameters:
10333 Used to draw a box around text using the background color.
10334 The value must be either 1 (enable) or 0 (disable).
10335 The default value of @var{box} is 0.
10338 Set the width of the border to be drawn around the box using @var{boxcolor}.
10339 The default value of @var{boxborderw} is 0.
10342 The color to be used for drawing box around text. For the syntax of this
10343 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10345 The default value of @var{boxcolor} is "white".
10348 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10349 The default value of @var{line_spacing} is 0.
10352 Set the width of the border to be drawn around the text using @var{bordercolor}.
10353 The default value of @var{borderw} is 0.
10356 Set the color to be used for drawing border around text. For the syntax of this
10357 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10359 The default value of @var{bordercolor} is "black".
10362 Select how the @var{text} is expanded. Can be either @code{none},
10363 @code{strftime} (deprecated) or
10364 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10368 Set a start time for the count. Value is in microseconds. Only applied
10369 in the deprecated strftime expansion mode. To emulate in normal expansion
10370 mode use the @code{pts} function, supplying the start time (in seconds)
10371 as the second argument.
10374 If true, check and fix text coords to avoid clipping.
10377 The color to be used for drawing fonts. For the syntax of this option, check
10378 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10380 The default value of @var{fontcolor} is "black".
10382 @item fontcolor_expr
10383 String which is expanded the same way as @var{text} to obtain dynamic
10384 @var{fontcolor} value. By default this option has empty value and is not
10385 processed. When this option is set, it overrides @var{fontcolor} option.
10388 The font family to be used for drawing text. By default Sans.
10391 The font file to be used for drawing text. The path must be included.
10392 This parameter is mandatory if the fontconfig support is disabled.
10395 Draw the text applying alpha blending. The value can
10396 be a number between 0.0 and 1.0.
10397 The expression accepts the same variables @var{x, y} as well.
10398 The default value is 1.
10399 Please see @var{fontcolor_expr}.
10402 The font size to be used for drawing text.
10403 The default value of @var{fontsize} is 16.
10406 If set to 1, attempt to shape the text (for example, reverse the order of
10407 right-to-left text and join Arabic characters) before drawing it.
10408 Otherwise, just draw the text exactly as given.
10409 By default 1 (if supported).
10411 @item ft_load_flags
10412 The flags to be used for loading the fonts.
10414 The flags map the corresponding flags supported by libfreetype, and are
10415 a combination of the following values:
10422 @item vertical_layout
10423 @item force_autohint
10426 @item ignore_global_advance_width
10428 @item ignore_transform
10430 @item linear_design
10434 Default value is "default".
10436 For more information consult the documentation for the FT_LOAD_*
10440 The color to be used for drawing a shadow behind the drawn text. For the
10441 syntax of this option, check the @ref{color syntax,,"Color" section in the
10442 ffmpeg-utils manual,ffmpeg-utils}.
10444 The default value of @var{shadowcolor} is "black".
10448 The x and y offsets for the text shadow position with respect to the
10449 position of the text. They can be either positive or negative
10450 values. The default value for both is "0".
10453 The starting frame number for the n/frame_num variable. The default value
10457 The size in number of spaces to use for rendering the tab.
10458 Default value is 4.
10461 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10462 format. It can be used with or without text parameter. @var{timecode_rate}
10463 option must be specified.
10465 @item timecode_rate, rate, r
10466 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10467 integer. Minimum value is "1".
10468 Drop-frame timecode is supported for frame rates 30 & 60.
10471 If set to 1, the output of the timecode option will wrap around at 24 hours.
10472 Default is 0 (disabled).
10475 The text string to be drawn. The text must be a sequence of UTF-8
10476 encoded characters.
10477 This parameter is mandatory if no file is specified with the parameter
10481 A text file containing text to be drawn. The text must be a sequence
10482 of UTF-8 encoded characters.
10484 This parameter is mandatory if no text string is specified with the
10485 parameter @var{text}.
10487 If both @var{text} and @var{textfile} are specified, an error is thrown.
10490 If set to 1, the @var{textfile} will be reloaded before each frame.
10491 Be sure to update it atomically, or it may be read partially, or even fail.
10495 The expressions which specify the offsets where text will be drawn
10496 within the video frame. They are relative to the top/left border of the
10499 The default value of @var{x} and @var{y} is "0".
10501 See below for the list of accepted constants and functions.
10504 The parameters for @var{x} and @var{y} are expressions containing the
10505 following constants and functions:
10509 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10513 horizontal and vertical chroma subsample values. For example for the
10514 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10517 the height of each text line
10525 @item max_glyph_a, ascent
10526 the maximum distance from the baseline to the highest/upper grid
10527 coordinate used to place a glyph outline point, for all the rendered
10529 It is a positive value, due to the grid's orientation with the Y axis
10532 @item max_glyph_d, descent
10533 the maximum distance from the baseline to the lowest grid coordinate
10534 used to place a glyph outline point, for all the rendered glyphs.
10535 This is a negative value, due to the grid's orientation, with the Y axis
10539 maximum glyph height, that is the maximum height for all the glyphs
10540 contained in the rendered text, it is equivalent to @var{ascent} -
10544 maximum glyph width, that is the maximum width for all the glyphs
10545 contained in the rendered text
10548 the number of input frame, starting from 0
10550 @item rand(min, max)
10551 return a random number included between @var{min} and @var{max}
10554 The input sample aspect ratio.
10557 timestamp expressed in seconds, NAN if the input timestamp is unknown
10560 the height of the rendered text
10563 the width of the rendered text
10567 the x and y offset coordinates where the text is drawn.
10569 These parameters allow the @var{x} and @var{y} expressions to refer
10570 to each other, so you can for example specify @code{y=x/dar}.
10573 A one character description of the current frame's picture type.
10576 The current packet's position in the input file or stream
10577 (in bytes, from the start of the input). A value of -1 indicates
10578 this info is not available.
10581 The current packet's duration, in seconds.
10584 The current packet's size (in bytes).
10587 @anchor{drawtext_expansion}
10588 @subsection Text expansion
10590 If @option{expansion} is set to @code{strftime},
10591 the filter recognizes strftime() sequences in the provided text and
10592 expands them accordingly. Check the documentation of strftime(). This
10593 feature is deprecated.
10595 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10597 If @option{expansion} is set to @code{normal} (which is the default),
10598 the following expansion mechanism is used.
10600 The backslash character @samp{\}, followed by any character, always expands to
10601 the second character.
10603 Sequences of the form @code{%@{...@}} are expanded. The text between the
10604 braces is a function name, possibly followed by arguments separated by ':'.
10605 If the arguments contain special characters or delimiters (':' or '@}'),
10606 they should be escaped.
10608 Note that they probably must also be escaped as the value for the
10609 @option{text} option in the filter argument string and as the filter
10610 argument in the filtergraph description, and possibly also for the shell,
10611 that makes up to four levels of escaping; using a text file avoids these
10614 The following functions are available:
10619 The expression evaluation result.
10621 It must take one argument specifying the expression to be evaluated,
10622 which accepts the same constants and functions as the @var{x} and
10623 @var{y} values. Note that not all constants should be used, for
10624 example the text size is not known when evaluating the expression, so
10625 the constants @var{text_w} and @var{text_h} will have an undefined
10628 @item expr_int_format, eif
10629 Evaluate the expression's value and output as formatted integer.
10631 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10632 The second argument specifies the output format. Allowed values are @samp{x},
10633 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10634 @code{printf} function.
10635 The third parameter is optional and sets the number of positions taken by the output.
10636 It can be used to add padding with zeros from the left.
10639 The time at which the filter is running, expressed in UTC.
10640 It can accept an argument: a strftime() format string.
10643 The time at which the filter is running, expressed in the local time zone.
10644 It can accept an argument: a strftime() format string.
10647 Frame metadata. Takes one or two arguments.
10649 The first argument is mandatory and specifies the metadata key.
10651 The second argument is optional and specifies a default value, used when the
10652 metadata key is not found or empty.
10654 Available metadata can be identified by inspecting entries
10655 starting with TAG included within each frame section
10656 printed by running @code{ffprobe -show_frames}.
10658 String metadata generated in filters leading to
10659 the drawtext filter are also available.
10662 The frame number, starting from 0.
10665 A one character description of the current picture type.
10668 The timestamp of the current frame.
10669 It can take up to three arguments.
10671 The first argument is the format of the timestamp; it defaults to @code{flt}
10672 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10673 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10674 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10675 @code{localtime} stands for the timestamp of the frame formatted as
10676 local time zone time.
10678 The second argument is an offset added to the timestamp.
10680 If the format is set to @code{hms}, a third argument @code{24HH} may be
10681 supplied to present the hour part of the formatted timestamp in 24h format
10684 If the format is set to @code{localtime} or @code{gmtime},
10685 a third argument may be supplied: a strftime() format string.
10686 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10689 @subsection Commands
10691 This filter supports altering parameters via commands:
10694 Alter existing filter parameters.
10696 Syntax for the argument is the same as for filter invocation, e.g.
10699 fontsize=56:fontcolor=green:text='Hello World'
10702 Full filter invocation with sendcmd would look like this:
10705 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10709 If the entire argument can't be parsed or applied as valid values then the filter will
10710 continue with its existing parameters.
10712 @subsection Examples
10716 Draw "Test Text" with font FreeSerif, using the default values for the
10717 optional parameters.
10720 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10724 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10725 and y=50 (counting from the top-left corner of the screen), text is
10726 yellow with a red box around it. Both the text and the box have an
10730 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10731 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10734 Note that the double quotes are not necessary if spaces are not used
10735 within the parameter list.
10738 Show the text at the center of the video frame:
10740 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10744 Show the text at a random position, switching to a new position every 30 seconds:
10746 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=if(eq(mod(t\,30)\,0)\,rand(0\,(w-text_w))\,x):y=if(eq(mod(t\,30)\,0)\,rand(0\,(h-text_h))\,y)"
10750 Show a text line sliding from right to left in the last row of the video
10751 frame. The file @file{LONG_LINE} is assumed to contain a single line
10754 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10758 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10760 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10764 Draw a single green letter "g", at the center of the input video.
10765 The glyph baseline is placed at half screen height.
10767 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10771 Show text for 1 second every 3 seconds:
10773 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10777 Use fontconfig to set the font. Note that the colons need to be escaped.
10779 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10783 Draw "Test Text" with font size dependent on height of the video.
10785 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10789 Print the date of a real-time encoding (see strftime(3)):
10791 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10795 Show text fading in and out (appearing/disappearing):
10798 DS=1.0 # display start
10799 DE=10.0 # display end
10800 FID=1.5 # fade in duration
10801 FOD=5 # fade out duration
10802 ffplay -f lavfi "color,drawtext=text=TEST:fontsize=50:fontfile=FreeSerif.ttf:fontcolor_expr=ff0000%@{eif\\\\: clip(255*(1*between(t\\, $DS + $FID\\, $DE - $FOD) + ((t - $DS)/$FID)*between(t\\, $DS\\, $DS + $FID) + (-(t - $DE)/$FOD)*between(t\\, $DE - $FOD\\, $DE) )\\, 0\\, 255) \\\\: x\\\\: 2 @}"
10806 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10807 and the @option{fontsize} value are included in the @option{y} offset.
10809 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10810 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10814 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10815 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10816 must have option @option{-export_path_metadata 1} for the special metadata fields
10817 to be available for filters.
10819 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10824 For more information about libfreetype, check:
10825 @url{http://www.freetype.org/}.
10827 For more information about fontconfig, check:
10828 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10830 For more information about libfribidi, check:
10831 @url{http://fribidi.org/}.
10833 @section edgedetect
10835 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10837 The filter accepts the following options:
10842 Set low and high threshold values used by the Canny thresholding
10845 The high threshold selects the "strong" edge pixels, which are then
10846 connected through 8-connectivity with the "weak" edge pixels selected
10847 by the low threshold.
10849 @var{low} and @var{high} threshold values must be chosen in the range
10850 [0,1], and @var{low} should be lesser or equal to @var{high}.
10852 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10856 Define the drawing mode.
10860 Draw white/gray wires on black background.
10863 Mix the colors to create a paint/cartoon effect.
10866 Apply Canny edge detector on all selected planes.
10868 Default value is @var{wires}.
10871 Select planes for filtering. By default all available planes are filtered.
10874 @subsection Examples
10878 Standard edge detection with custom values for the hysteresis thresholding:
10880 edgedetect=low=0.1:high=0.4
10884 Painting effect without thresholding:
10886 edgedetect=mode=colormix:high=0
10892 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10894 For each input image, the filter will compute the optimal mapping from
10895 the input to the output given the codebook length, that is the number
10896 of distinct output colors.
10898 This filter accepts the following options.
10901 @item codebook_length, l
10902 Set codebook length. The value must be a positive integer, and
10903 represents the number of distinct output colors. Default value is 256.
10906 Set the maximum number of iterations to apply for computing the optimal
10907 mapping. The higher the value the better the result and the higher the
10908 computation time. Default value is 1.
10911 Set a random seed, must be an integer included between 0 and
10912 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10913 will try to use a good random seed on a best effort basis.
10916 Set pal8 output pixel format. This option does not work with codebook
10917 length greater than 256.
10922 Measure graylevel entropy in histogram of color channels of video frames.
10924 It accepts the following parameters:
10928 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10930 @var{diff} mode measures entropy of histogram delta values, absolute differences
10931 between neighbour histogram values.
10935 Set brightness, contrast, saturation and approximate gamma adjustment.
10937 The filter accepts the following options:
10941 Set the contrast expression. The value must be a float value in range
10942 @code{-1000.0} to @code{1000.0}. The default value is "1".
10945 Set the brightness expression. The value must be a float value in
10946 range @code{-1.0} to @code{1.0}. The default value is "0".
10949 Set the saturation expression. The value must be a float in
10950 range @code{0.0} to @code{3.0}. The default value is "1".
10953 Set the gamma expression. The value must be a float in range
10954 @code{0.1} to @code{10.0}. The default value is "1".
10957 Set the gamma expression for red. The value must be a float in
10958 range @code{0.1} to @code{10.0}. The default value is "1".
10961 Set the gamma expression for green. The value must be a float in range
10962 @code{0.1} to @code{10.0}. The default value is "1".
10965 Set the gamma expression for blue. The value must be a float in range
10966 @code{0.1} to @code{10.0}. The default value is "1".
10969 Set the gamma weight expression. It can be used to reduce the effect
10970 of a high gamma value on bright image areas, e.g. keep them from
10971 getting overamplified and just plain white. The value must be a float
10972 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10973 gamma correction all the way down while @code{1.0} leaves it at its
10974 full strength. Default is "1".
10977 Set when the expressions for brightness, contrast, saturation and
10978 gamma expressions are evaluated.
10980 It accepts the following values:
10983 only evaluate expressions once during the filter initialization or
10984 when a command is processed
10987 evaluate expressions for each incoming frame
10990 Default value is @samp{init}.
10993 The expressions accept the following parameters:
10996 frame count of the input frame starting from 0
10999 byte position of the corresponding packet in the input file, NAN if
11003 frame rate of the input video, NAN if the input frame rate is unknown
11006 timestamp expressed in seconds, NAN if the input timestamp is unknown
11009 @subsection Commands
11010 The filter supports the following commands:
11014 Set the contrast expression.
11017 Set the brightness expression.
11020 Set the saturation expression.
11023 Set the gamma expression.
11026 Set the gamma_r expression.
11029 Set gamma_g expression.
11032 Set gamma_b expression.
11035 Set gamma_weight expression.
11037 The command accepts the same syntax of the corresponding option.
11039 If the specified expression is not valid, it is kept at its current
11046 Apply erosion effect to the video.
11048 This filter replaces the pixel by the local(3x3) minimum.
11050 It accepts the following options:
11057 Limit the maximum change for each plane, default is 65535.
11058 If 0, plane will remain unchanged.
11061 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
11064 Flags to local 3x3 coordinates maps like this:
11071 @subsection Commands
11073 This filter supports the all above options as @ref{commands}.
11077 Deinterlace the input video ("estdif" stands for "Edge Slope
11078 Tracing Deinterlacing Filter").
11080 Spatial only filter that uses edge slope tracing algorithm
11081 to interpolate missing lines.
11082 It accepts the following parameters:
11086 The interlacing mode to adopt. It accepts one of the following values:
11090 Output one frame for each frame.
11092 Output one frame for each field.
11095 The default value is @code{field}.
11098 The picture field parity assumed for the input interlaced video. It accepts one
11099 of the following values:
11103 Assume the top field is first.
11105 Assume the bottom field is first.
11107 Enable automatic detection of field parity.
11110 The default value is @code{auto}.
11111 If the interlacing is unknown or the decoder does not export this information,
11112 top field first will be assumed.
11115 Specify which frames to deinterlace. Accepts one of the following
11120 Deinterlace all frames.
11122 Only deinterlace frames marked as interlaced.
11125 The default value is @code{all}.
11128 Specify the search radius for edge slope tracing. Default value is 1.
11129 Allowed range is from 1 to 15.
11132 Specify the search radius for best edge matching. Default value is 2.
11133 Allowed range is from 0 to 15.
11136 @subsection Commands
11137 This filter supports same @ref{commands} as options.
11139 @section extractplanes
11141 Extract color channel components from input video stream into
11142 separate grayscale video streams.
11144 The filter accepts the following option:
11148 Set plane(s) to extract.
11150 Available values for planes are:
11161 Choosing planes not available in the input will result in an error.
11162 That means you cannot select @code{r}, @code{g}, @code{b} planes
11163 with @code{y}, @code{u}, @code{v} planes at same time.
11166 @subsection Examples
11170 Extract luma, u and v color channel component from input video frame
11171 into 3 grayscale outputs:
11173 ffmpeg -i video.avi -filter_complex 'extractplanes=y+u+v[y][u][v]' -map '[y]' y.avi -map '[u]' u.avi -map '[v]' v.avi
11179 Apply a fade-in/out effect to the input video.
11181 It accepts the following parameters:
11185 The effect type can be either "in" for a fade-in, or "out" for a fade-out
11187 Default is @code{in}.
11189 @item start_frame, s
11190 Specify the number of the frame to start applying the fade
11191 effect at. Default is 0.
11194 The number of frames that the fade effect lasts. At the end of the
11195 fade-in effect, the output video will have the same intensity as the input video.
11196 At the end of the fade-out transition, the output video will be filled with the
11197 selected @option{color}.
11201 If set to 1, fade only alpha channel, if one exists on the input.
11202 Default value is 0.
11204 @item start_time, st
11205 Specify the timestamp (in seconds) of the frame to start to apply the fade
11206 effect. If both start_frame and start_time are specified, the fade will start at
11207 whichever comes last. Default is 0.
11210 The number of seconds for which the fade effect has to last. At the end of the
11211 fade-in effect the output video will have the same intensity as the input video,
11212 at the end of the fade-out transition the output video will be filled with the
11213 selected @option{color}.
11214 If both duration and nb_frames are specified, duration is used. Default is 0
11215 (nb_frames is used by default).
11218 Specify the color of the fade. Default is "black".
11221 @subsection Examples
11225 Fade in the first 30 frames of video:
11230 The command above is equivalent to:
11236 Fade out the last 45 frames of a 200-frame video:
11239 fade=type=out:start_frame=155:nb_frames=45
11243 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
11245 fade=in:0:25, fade=out:975:25
11249 Make the first 5 frames yellow, then fade in from frame 5-24:
11251 fade=in:5:20:color=yellow
11255 Fade in alpha over first 25 frames of video:
11257 fade=in:0:25:alpha=1
11261 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
11263 fade=t=in:st=5.5:d=0.5
11269 Denoise frames using 3D FFT (frequency domain filtering).
11271 The filter accepts the following options:
11275 Set the noise sigma constant. This sets denoising strength.
11276 Default value is 1. Allowed range is from 0 to 30.
11277 Using very high sigma with low overlap may give blocking artifacts.
11280 Set amount of denoising. By default all detected noise is reduced.
11281 Default value is 1. Allowed range is from 0 to 1.
11284 Set size of block, Default is 4, can be 3, 4, 5 or 6.
11285 Actual size of block in pixels is 2 to power of @var{block}, so by default
11286 block size in pixels is 2^4 which is 16.
11289 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
11292 Set number of previous frames to use for denoising. By default is set to 0.
11295 Set number of next frames to to use for denoising. By default is set to 0.
11298 Set planes which will be filtered, by default are all available filtered
11303 Apply arbitrary expressions to samples in frequency domain
11307 Adjust the dc value (gain) of the luma plane of the image. The filter
11308 accepts an integer value in range @code{0} to @code{1000}. The default
11309 value is set to @code{0}.
11312 Adjust the dc value (gain) of the 1st chroma plane of the image. The
11313 filter accepts an integer value in range @code{0} to @code{1000}. The
11314 default value is set to @code{0}.
11317 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
11318 filter accepts an integer value in range @code{0} to @code{1000}. The
11319 default value is set to @code{0}.
11322 Set the frequency domain weight expression for the luma plane.
11325 Set the frequency domain weight expression for the 1st chroma plane.
11328 Set the frequency domain weight expression for the 2nd chroma plane.
11331 Set when the expressions are evaluated.
11333 It accepts the following values:
11336 Only evaluate expressions once during the filter initialization.
11339 Evaluate expressions for each incoming frame.
11342 Default value is @samp{init}.
11344 The filter accepts the following variables:
11347 The coordinates of the current sample.
11351 The width and height of the image.
11354 The number of input frame, starting from 0.
11357 @subsection Examples
11363 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11369 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11375 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11381 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11388 Extract a single field from an interlaced image using stride
11389 arithmetic to avoid wasting CPU time. The output frames are marked as
11392 The filter accepts the following options:
11396 Specify whether to extract the top (if the value is @code{0} or
11397 @code{top}) or the bottom field (if the value is @code{1} or
11403 Create new frames by copying the top and bottom fields from surrounding frames
11404 supplied as numbers by the hint file.
11408 Set file containing hints: absolute/relative frame numbers.
11410 There must be one line for each frame in a clip. Each line must contain two
11411 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11412 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11413 is current frame number for @code{absolute} mode or out of [-1, 1] range
11414 for @code{relative} mode. First number tells from which frame to pick up top
11415 field and second number tells from which frame to pick up bottom field.
11417 If optionally followed by @code{+} output frame will be marked as interlaced,
11418 else if followed by @code{-} output frame will be marked as progressive, else
11419 it will be marked same as input frame.
11420 If optionally followed by @code{t} output frame will use only top field, or in
11421 case of @code{b} it will use only bottom field.
11422 If line starts with @code{#} or @code{;} that line is skipped.
11425 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11428 Example of first several lines of @code{hint} file for @code{relative} mode:
11430 0,0 - # first frame
11431 1,0 - # second frame, use third's frame top field and second's frame bottom field
11432 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11447 @section fieldmatch
11449 Field matching filter for inverse telecine. It is meant to reconstruct the
11450 progressive frames from a telecined stream. The filter does not drop duplicated
11451 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11452 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11454 The separation of the field matching and the decimation is notably motivated by
11455 the possibility of inserting a de-interlacing filter fallback between the two.
11456 If the source has mixed telecined and real interlaced content,
11457 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11458 But these remaining combed frames will be marked as interlaced, and thus can be
11459 de-interlaced by a later filter such as @ref{yadif} before decimation.
11461 In addition to the various configuration options, @code{fieldmatch} can take an
11462 optional second stream, activated through the @option{ppsrc} option. If
11463 enabled, the frames reconstruction will be based on the fields and frames from
11464 this second stream. This allows the first input to be pre-processed in order to
11465 help the various algorithms of the filter, while keeping the output lossless
11466 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11467 or brightness/contrast adjustments can help.
11469 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11470 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11471 which @code{fieldmatch} is based on. While the semantic and usage are very
11472 close, some behaviour and options names can differ.
11474 The @ref{decimate} filter currently only works for constant frame rate input.
11475 If your input has mixed telecined (30fps) and progressive content with a lower
11476 framerate like 24fps use the following filterchain to produce the necessary cfr
11477 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11479 The filter accepts the following options:
11483 Specify the assumed field order of the input stream. Available values are:
11487 Auto detect parity (use FFmpeg's internal parity value).
11489 Assume bottom field first.
11491 Assume top field first.
11494 Note that it is sometimes recommended not to trust the parity announced by the
11497 Default value is @var{auto}.
11500 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11501 sense that it won't risk creating jerkiness due to duplicate frames when
11502 possible, but if there are bad edits or blended fields it will end up
11503 outputting combed frames when a good match might actually exist. On the other
11504 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11505 but will almost always find a good frame if there is one. The other values are
11506 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11507 jerkiness and creating duplicate frames versus finding good matches in sections
11508 with bad edits, orphaned fields, blended fields, etc.
11510 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11512 Available values are:
11516 2-way matching (p/c)
11518 2-way matching, and trying 3rd match if still combed (p/c + n)
11520 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11522 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11523 still combed (p/c + n + u/b)
11525 3-way matching (p/c/n)
11527 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11528 detected as combed (p/c/n + u/b)
11531 The parenthesis at the end indicate the matches that would be used for that
11532 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11535 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11538 Default value is @var{pc_n}.
11541 Mark the main input stream as a pre-processed input, and enable the secondary
11542 input stream as the clean source to pick the fields from. See the filter
11543 introduction for more details. It is similar to the @option{clip2} feature from
11546 Default value is @code{0} (disabled).
11549 Set the field to match from. It is recommended to set this to the same value as
11550 @option{order} unless you experience matching failures with that setting. In
11551 certain circumstances changing the field that is used to match from can have a
11552 large impact on matching performance. Available values are:
11556 Automatic (same value as @option{order}).
11558 Match from the bottom field.
11560 Match from the top field.
11563 Default value is @var{auto}.
11566 Set whether or not chroma is included during the match comparisons. In most
11567 cases it is recommended to leave this enabled. You should set this to @code{0}
11568 only if your clip has bad chroma problems such as heavy rainbowing or other
11569 artifacts. Setting this to @code{0} could also be used to speed things up at
11570 the cost of some accuracy.
11572 Default value is @code{1}.
11576 These define an exclusion band which excludes the lines between @option{y0} and
11577 @option{y1} from being included in the field matching decision. An exclusion
11578 band can be used to ignore subtitles, a logo, or other things that may
11579 interfere with the matching. @option{y0} sets the starting scan line and
11580 @option{y1} sets the ending line; all lines in between @option{y0} and
11581 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11582 @option{y0} and @option{y1} to the same value will disable the feature.
11583 @option{y0} and @option{y1} defaults to @code{0}.
11586 Set the scene change detection threshold as a percentage of maximum change on
11587 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11588 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11589 @option{scthresh} is @code{[0.0, 100.0]}.
11591 Default value is @code{12.0}.
11594 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11595 account the combed scores of matches when deciding what match to use as the
11596 final match. Available values are:
11600 No final matching based on combed scores.
11602 Combed scores are only used when a scene change is detected.
11604 Use combed scores all the time.
11607 Default is @var{sc}.
11610 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11611 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11612 Available values are:
11616 No forced calculation.
11618 Force p/c/n calculations.
11620 Force p/c/n/u/b calculations.
11623 Default value is @var{none}.
11626 This is the area combing threshold used for combed frame detection. This
11627 essentially controls how "strong" or "visible" combing must be to be detected.
11628 Larger values mean combing must be more visible and smaller values mean combing
11629 can be less visible or strong and still be detected. Valid settings are from
11630 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11631 be detected as combed). This is basically a pixel difference value. A good
11632 range is @code{[8, 12]}.
11634 Default value is @code{9}.
11637 Sets whether or not chroma is considered in the combed frame decision. Only
11638 disable this if your source has chroma problems (rainbowing, etc.) that are
11639 causing problems for the combed frame detection with chroma enabled. Actually,
11640 using @option{chroma}=@var{0} is usually more reliable, except for the case
11641 where there is chroma only combing in the source.
11643 Default value is @code{0}.
11647 Respectively set the x-axis and y-axis size of the window used during combed
11648 frame detection. This has to do with the size of the area in which
11649 @option{combpel} pixels are required to be detected as combed for a frame to be
11650 declared combed. See the @option{combpel} parameter description for more info.
11651 Possible values are any number that is a power of 2 starting at 4 and going up
11654 Default value is @code{16}.
11657 The number of combed pixels inside any of the @option{blocky} by
11658 @option{blockx} size blocks on the frame for the frame to be detected as
11659 combed. While @option{cthresh} controls how "visible" the combing must be, this
11660 setting controls "how much" combing there must be in any localized area (a
11661 window defined by the @option{blockx} and @option{blocky} settings) on the
11662 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11663 which point no frames will ever be detected as combed). This setting is known
11664 as @option{MI} in TFM/VFM vocabulary.
11666 Default value is @code{80}.
11669 @anchor{p/c/n/u/b meaning}
11670 @subsection p/c/n/u/b meaning
11672 @subsubsection p/c/n
11674 We assume the following telecined stream:
11677 Top fields: 1 2 2 3 4
11678 Bottom fields: 1 2 3 4 4
11681 The numbers correspond to the progressive frame the fields relate to. Here, the
11682 first two frames are progressive, the 3rd and 4th are combed, and so on.
11684 When @code{fieldmatch} is configured to run a matching from bottom
11685 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11690 B 1 2 3 4 4 <-- matching reference
11699 As a result of the field matching, we can see that some frames get duplicated.
11700 To perform a complete inverse telecine, you need to rely on a decimation filter
11701 after this operation. See for instance the @ref{decimate} filter.
11703 The same operation now matching from top fields (@option{field}=@var{top})
11708 T 1 2 2 3 4 <-- matching reference
11718 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11719 basically, they refer to the frame and field of the opposite parity:
11722 @item @var{p} matches the field of the opposite parity in the previous frame
11723 @item @var{c} matches the field of the opposite parity in the current frame
11724 @item @var{n} matches the field of the opposite parity in the next frame
11729 The @var{u} and @var{b} matching are a bit special in the sense that they match
11730 from the opposite parity flag. In the following examples, we assume that we are
11731 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11732 'x' is placed above and below each matched fields.
11734 With bottom matching (@option{field}=@var{bottom}):
11739 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11740 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11748 With top matching (@option{field}=@var{top}):
11753 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11754 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11762 @subsection Examples
11764 Simple IVTC of a top field first telecined stream:
11766 fieldmatch=order=tff:combmatch=none, decimate
11769 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11771 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11774 @section fieldorder
11776 Transform the field order of the input video.
11778 It accepts the following parameters:
11783 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11784 for bottom field first.
11787 The default value is @samp{tff}.
11789 The transformation is done by shifting the picture content up or down
11790 by one line, and filling the remaining line with appropriate picture content.
11791 This method is consistent with most broadcast field order converters.
11793 If the input video is not flagged as being interlaced, or it is already
11794 flagged as being of the required output field order, then this filter does
11795 not alter the incoming video.
11797 It is very useful when converting to or from PAL DV material,
11798 which is bottom field first.
11802 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11805 @section fifo, afifo
11807 Buffer input images and send them when they are requested.
11809 It is mainly useful when auto-inserted by the libavfilter
11812 It does not take parameters.
11814 @section fillborders
11816 Fill borders of the input video, without changing video stream dimensions.
11817 Sometimes video can have garbage at the four edges and you may not want to
11818 crop video input to keep size multiple of some number.
11820 This filter accepts the following options:
11824 Number of pixels to fill from left border.
11827 Number of pixels to fill from right border.
11830 Number of pixels to fill from top border.
11833 Number of pixels to fill from bottom border.
11838 It accepts the following values:
11841 fill pixels using outermost pixels
11844 fill pixels using mirroring (half sample symmetric)
11847 fill pixels with constant value
11850 fill pixels using reflecting (whole sample symmetric)
11853 fill pixels using wrapping
11856 fade pixels to constant value
11859 Default is @var{smear}.
11862 Set color for pixels in fixed or fade mode. Default is @var{black}.
11865 @subsection Commands
11866 This filter supports same @ref{commands} as options.
11867 The command accepts the same syntax of the corresponding option.
11869 If the specified expression is not valid, it is kept at its current
11874 Find a rectangular object
11876 It accepts the following options:
11880 Filepath of the object image, needs to be in gray8.
11883 Detection threshold, default is 0.5.
11886 Number of mipmaps, default is 3.
11888 @item xmin, ymin, xmax, ymax
11889 Specifies the rectangle in which to search.
11892 @subsection Examples
11896 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11898 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11904 Flood area with values of same pixel components with another values.
11906 It accepts the following options:
11909 Set pixel x coordinate.
11912 Set pixel y coordinate.
11915 Set source #0 component value.
11918 Set source #1 component value.
11921 Set source #2 component value.
11924 Set source #3 component value.
11927 Set destination #0 component value.
11930 Set destination #1 component value.
11933 Set destination #2 component value.
11936 Set destination #3 component value.
11942 Convert the input video to one of the specified pixel formats.
11943 Libavfilter will try to pick one that is suitable as input to
11946 It accepts the following parameters:
11950 A '|'-separated list of pixel format names, such as
11951 "pix_fmts=yuv420p|monow|rgb24".
11955 @subsection Examples
11959 Convert the input video to the @var{yuv420p} format
11961 format=pix_fmts=yuv420p
11964 Convert the input video to any of the formats in the list
11966 format=pix_fmts=yuv420p|yuv444p|yuv410p
11973 Convert the video to specified constant frame rate by duplicating or dropping
11974 frames as necessary.
11976 It accepts the following parameters:
11980 The desired output frame rate. The default is @code{25}.
11983 Assume the first PTS should be the given value, in seconds. This allows for
11984 padding/trimming at the start of stream. By default, no assumption is made
11985 about the first frame's expected PTS, so no padding or trimming is done.
11986 For example, this could be set to 0 to pad the beginning with duplicates of
11987 the first frame if a video stream starts after the audio stream or to trim any
11988 frames with a negative PTS.
11991 Timestamp (PTS) rounding method.
11993 Possible values are:
12000 round towards -infinity
12002 round towards +infinity
12006 The default is @code{near}.
12009 Action performed when reading the last frame.
12011 Possible values are:
12014 Use same timestamp rounding method as used for other frames.
12016 Pass through last frame if input duration has not been reached yet.
12018 The default is @code{round}.
12022 Alternatively, the options can be specified as a flat string:
12023 @var{fps}[:@var{start_time}[:@var{round}]].
12025 See also the @ref{setpts} filter.
12027 @subsection Examples
12031 A typical usage in order to set the fps to 25:
12037 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
12039 fps=fps=film:round=near
12045 Pack two different video streams into a stereoscopic video, setting proper
12046 metadata on supported codecs. The two views should have the same size and
12047 framerate and processing will stop when the shorter video ends. Please note
12048 that you may conveniently adjust view properties with the @ref{scale} and
12051 It accepts the following parameters:
12055 The desired packing format. Supported values are:
12060 The views are next to each other (default).
12063 The views are on top of each other.
12066 The views are packed by line.
12069 The views are packed by column.
12072 The views are temporally interleaved.
12081 # Convert left and right views into a frame-sequential video
12082 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
12084 # Convert views into a side-by-side video with the same output resolution as the input
12085 ffmpeg -i LEFT -i RIGHT -filter_complex [0:v]scale=w=iw/2[left],[1:v]scale=w=iw/2[right],[left][right]framepack=sbs OUTPUT
12090 Change the frame rate by interpolating new video output frames from the source
12093 This filter is not designed to function correctly with interlaced media. If
12094 you wish to change the frame rate of interlaced media then you are required
12095 to deinterlace before this filter and re-interlace after this filter.
12097 A description of the accepted options follows.
12101 Specify the output frames per second. This option can also be specified
12102 as a value alone. The default is @code{50}.
12105 Specify the start of a range where the output frame will be created as a
12106 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12107 the default is @code{15}.
12110 Specify the end of a range where the output frame will be created as a
12111 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12112 the default is @code{240}.
12115 Specify the level at which a scene change is detected as a value between
12116 0 and 100 to indicate a new scene; a low value reflects a low
12117 probability for the current frame to introduce a new scene, while a higher
12118 value means the current frame is more likely to be one.
12119 The default is @code{8.2}.
12122 Specify flags influencing the filter process.
12124 Available value for @var{flags} is:
12127 @item scene_change_detect, scd
12128 Enable scene change detection using the value of the option @var{scene}.
12129 This flag is enabled by default.
12135 Select one frame every N-th frame.
12137 This filter accepts the following option:
12140 Select frame after every @code{step} frames.
12141 Allowed values are positive integers higher than 0. Default value is @code{1}.
12144 @section freezedetect
12146 Detect frozen video.
12148 This filter logs a message and sets frame metadata when it detects that the
12149 input video has no significant change in content during a specified duration.
12150 Video freeze detection calculates the mean average absolute difference of all
12151 the components of video frames and compares it to a noise floor.
12153 The printed times and duration are expressed in seconds. The
12154 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
12155 whose timestamp equals or exceeds the detection duration and it contains the
12156 timestamp of the first frame of the freeze. The
12157 @code{lavfi.freezedetect.freeze_duration} and
12158 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
12161 The filter accepts the following options:
12165 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
12166 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
12170 Set freeze duration until notification (default is 2 seconds).
12173 @section freezeframes
12175 Freeze video frames.
12177 This filter freezes video frames using frame from 2nd input.
12179 The filter accepts the following options:
12183 Set number of first frame from which to start freeze.
12186 Set number of last frame from which to end freeze.
12189 Set number of frame from 2nd input which will be used instead of replaced frames.
12195 Apply a frei0r effect to the input video.
12197 To enable the compilation of this filter, you need to install the frei0r
12198 header and configure FFmpeg with @code{--enable-frei0r}.
12200 It accepts the following parameters:
12205 The name of the frei0r effect to load. If the environment variable
12206 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
12207 directories specified by the colon-separated list in @env{FREI0R_PATH}.
12208 Otherwise, the standard frei0r paths are searched, in this order:
12209 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
12210 @file{/usr/lib/frei0r-1/}.
12212 @item filter_params
12213 A '|'-separated list of parameters to pass to the frei0r effect.
12217 A frei0r effect parameter can be a boolean (its value is either
12218 "y" or "n"), a double, a color (specified as
12219 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
12220 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
12221 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
12222 a position (specified as @var{X}/@var{Y}, where
12223 @var{X} and @var{Y} are floating point numbers) and/or a string.
12225 The number and types of parameters depend on the loaded effect. If an
12226 effect parameter is not specified, the default value is set.
12228 @subsection Examples
12232 Apply the distort0r effect, setting the first two double parameters:
12234 frei0r=filter_name=distort0r:filter_params=0.5|0.01
12238 Apply the colordistance effect, taking a color as the first parameter:
12240 frei0r=colordistance:0.2/0.3/0.4
12241 frei0r=colordistance:violet
12242 frei0r=colordistance:0x112233
12246 Apply the perspective effect, specifying the top left and top right image
12249 frei0r=perspective:0.2/0.2|0.8/0.2
12253 For more information, see
12254 @url{http://frei0r.dyne.org}
12256 @subsection Commands
12258 This filter supports the @option{filter_params} option as @ref{commands}.
12262 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
12264 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
12265 processing filter, one of them is performed once per block, not per pixel.
12266 This allows for much higher speed.
12268 The filter accepts the following options:
12272 Set quality. This option defines the number of levels for averaging. It accepts
12273 an integer in the range 4-5. Default value is @code{4}.
12276 Force a constant quantization parameter. It accepts an integer in range 0-63.
12277 If not set, the filter will use the QP from the video stream (if available).
12280 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
12281 more details but also more artifacts, while higher values make the image smoother
12282 but also blurrier. Default value is @code{0} − PSNR optimal.
12284 @item use_bframe_qp
12285 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
12286 option may cause flicker since the B-Frames have often larger QP. Default is
12287 @code{0} (not enabled).
12293 Apply Gaussian blur filter.
12295 The filter accepts the following options:
12299 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
12302 Set number of steps for Gaussian approximation. Default is @code{1}.
12305 Set which planes to filter. By default all planes are filtered.
12308 Set vertical sigma, if negative it will be same as @code{sigma}.
12309 Default is @code{-1}.
12312 @subsection Commands
12313 This filter supports same commands as options.
12314 The command accepts the same syntax of the corresponding option.
12316 If the specified expression is not valid, it is kept at its current
12321 Apply generic equation to each pixel.
12323 The filter accepts the following options:
12326 @item lum_expr, lum
12327 Set the luminance expression.
12329 Set the chrominance blue expression.
12331 Set the chrominance red expression.
12332 @item alpha_expr, a
12333 Set the alpha expression.
12335 Set the red expression.
12336 @item green_expr, g
12337 Set the green expression.
12339 Set the blue expression.
12342 The colorspace is selected according to the specified options. If one
12343 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
12344 options is specified, the filter will automatically select a YCbCr
12345 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
12346 @option{blue_expr} options is specified, it will select an RGB
12349 If one of the chrominance expression is not defined, it falls back on the other
12350 one. If no alpha expression is specified it will evaluate to opaque value.
12351 If none of chrominance expressions are specified, they will evaluate
12352 to the luminance expression.
12354 The expressions can use the following variables and functions:
12358 The sequential number of the filtered frame, starting from @code{0}.
12362 The coordinates of the current sample.
12366 The width and height of the image.
12370 Width and height scale depending on the currently filtered plane. It is the
12371 ratio between the corresponding luma plane number of pixels and the current
12372 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
12373 @code{0.5,0.5} for chroma planes.
12376 Time of the current frame, expressed in seconds.
12379 Return the value of the pixel at location (@var{x},@var{y}) of the current
12383 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12387 Return the value of the pixel at location (@var{x},@var{y}) of the
12388 blue-difference chroma plane. Return 0 if there is no such plane.
12391 Return the value of the pixel at location (@var{x},@var{y}) of the
12392 red-difference chroma plane. Return 0 if there is no such plane.
12397 Return the value of the pixel at location (@var{x},@var{y}) of the
12398 red/green/blue component. Return 0 if there is no such component.
12401 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12402 plane. Return 0 if there is no such plane.
12404 @item psum(x,y), lumsum(x, y), cbsum(x,y), crsum(x,y), rsum(x,y), gsum(x,y), bsum(x,y), alphasum(x,y)
12405 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12406 sums of samples within a rectangle. See the functions without the sum postfix.
12408 @item interpolation
12409 Set one of interpolation methods:
12414 Default is bilinear.
12417 For functions, if @var{x} and @var{y} are outside the area, the value will be
12418 automatically clipped to the closer edge.
12420 Please note that this filter can use multiple threads in which case each slice
12421 will have its own expression state. If you want to use only a single expression
12422 state because your expressions depend on previous state then you should limit
12423 the number of filter threads to 1.
12425 @subsection Examples
12429 Flip the image horizontally:
12435 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12436 wavelength of 100 pixels:
12438 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12442 Generate a fancy enigmatic moving light:
12444 nullsrc=s=256x256,geq=random(1)/hypot(X-cos(N*0.07)*W/2-W/2\,Y-sin(N*0.09)*H/2-H/2)^2*1000000*sin(N*0.02):128:128
12448 Generate a quick emboss effect:
12450 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12454 Modify RGB components depending on pixel position:
12456 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12460 Create a radial gradient that is the same size as the input (also see
12461 the @ref{vignette} filter):
12463 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12469 Fix the banding artifacts that are sometimes introduced into nearly flat
12470 regions by truncation to 8-bit color depth.
12471 Interpolate the gradients that should go where the bands are, and
12474 It is designed for playback only. Do not use it prior to
12475 lossy compression, because compression tends to lose the dither and
12476 bring back the bands.
12478 It accepts the following parameters:
12483 The maximum amount by which the filter will change any one pixel. This is also
12484 the threshold for detecting nearly flat regions. Acceptable values range from
12485 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12489 The neighborhood to fit the gradient to. A larger radius makes for smoother
12490 gradients, but also prevents the filter from modifying the pixels near detailed
12491 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12492 values will be clipped to the valid range.
12496 Alternatively, the options can be specified as a flat string:
12497 @var{strength}[:@var{radius}]
12499 @subsection Examples
12503 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12509 Specify radius, omitting the strength (which will fall-back to the default
12517 @anchor{graphmonitor}
12518 @section graphmonitor
12519 Show various filtergraph stats.
12521 With this filter one can debug complete filtergraph.
12522 Especially issues with links filling with queued frames.
12524 The filter accepts the following options:
12528 Set video output size. Default is @var{hd720}.
12531 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12534 Set output mode, can be @var{fulll} or @var{compact}.
12535 In @var{compact} mode only filters with some queued frames have displayed stats.
12538 Set flags which enable which stats are shown in video.
12540 Available values for flags are:
12543 Display number of queued frames in each link.
12545 @item frame_count_in
12546 Display number of frames taken from filter.
12548 @item frame_count_out
12549 Display number of frames given out from filter.
12552 Display current filtered frame pts.
12555 Display current filtered frame time.
12558 Display time base for filter link.
12561 Display used format for filter link.
12564 Display video size or number of audio channels in case of audio used by filter link.
12567 Display video frame rate or sample rate in case of audio used by filter link.
12570 Display link output status.
12574 Set upper limit for video rate of output stream, Default value is @var{25}.
12575 This guarantee that output video frame rate will not be higher than this value.
12579 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12580 and corrects the scene colors accordingly.
12582 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12584 The filter accepts the following options:
12588 The order of differentiation to be applied on the scene. Must be chosen in the range
12589 [0,2] and default value is 1.
12592 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12593 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12594 max value instead of calculating Minkowski distance.
12597 The standard deviation of Gaussian blur to be applied on the scene. Must be
12598 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12599 can't be equal to 0 if @var{difford} is greater than 0.
12602 @subsection Examples
12608 greyedge=difford=1:minknorm=5:sigma=2
12614 greyedge=difford=1:minknorm=0:sigma=2
12622 Apply a Hald CLUT to a video stream.
12624 First input is the video stream to process, and second one is the Hald CLUT.
12625 The Hald CLUT input can be a simple picture or a complete video stream.
12627 The filter accepts the following options:
12631 Force termination when the shortest input terminates. Default is @code{0}.
12633 Continue applying the last CLUT after the end of the stream. A value of
12634 @code{0} disable the filter after the last frame of the CLUT is reached.
12635 Default is @code{1}.
12638 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12639 filters share the same internals).
12641 This filter also supports the @ref{framesync} options.
12643 More information about the Hald CLUT can be found on Eskil Steenberg's website
12644 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12646 @subsection Workflow examples
12648 @subsubsection Hald CLUT video stream
12650 Generate an identity Hald CLUT stream altered with various effects:
12652 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "hue=H=2*PI*t:s=sin(2*PI*t)+1, curves=cross_process" -t 10 -c:v ffv1 clut.nut
12655 Note: make sure you use a lossless codec.
12657 Then use it with @code{haldclut} to apply it on some random stream:
12659 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12662 The Hald CLUT will be applied to the 10 first seconds (duration of
12663 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12664 to the remaining frames of the @code{mandelbrot} stream.
12666 @subsubsection Hald CLUT with preview
12668 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12669 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12670 biggest possible square starting at the top left of the picture. The remaining
12671 padding pixels (bottom or right) will be ignored. This area can be used to add
12672 a preview of the Hald CLUT.
12674 Typically, the following generated Hald CLUT will be supported by the
12675 @code{haldclut} filter:
12678 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12679 pad=iw+320 [padded_clut];
12680 smptebars=s=320x256, split [a][b];
12681 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12682 [main][b] overlay=W-320" -frames:v 1 clut.png
12685 It contains the original and a preview of the effect of the CLUT: SMPTE color
12686 bars are displayed on the right-top, and below the same color bars processed by
12689 Then, the effect of this Hald CLUT can be visualized with:
12691 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12696 Flip the input video horizontally.
12698 For example, to horizontally flip the input video with @command{ffmpeg}:
12700 ffmpeg -i in.avi -vf "hflip" out.avi
12704 This filter applies a global color histogram equalization on a
12707 It can be used to correct video that has a compressed range of pixel
12708 intensities. The filter redistributes the pixel intensities to
12709 equalize their distribution across the intensity range. It may be
12710 viewed as an "automatically adjusting contrast filter". This filter is
12711 useful only for correcting degraded or poorly captured source
12714 The filter accepts the following options:
12718 Determine the amount of equalization to be applied. As the strength
12719 is reduced, the distribution of pixel intensities more-and-more
12720 approaches that of the input frame. The value must be a float number
12721 in the range [0,1] and defaults to 0.200.
12724 Set the maximum intensity that can generated and scale the output
12725 values appropriately. The strength should be set as desired and then
12726 the intensity can be limited if needed to avoid washing-out. The value
12727 must be a float number in the range [0,1] and defaults to 0.210.
12730 Set the antibanding level. If enabled the filter will randomly vary
12731 the luminance of output pixels by a small amount to avoid banding of
12732 the histogram. Possible values are @code{none}, @code{weak} or
12733 @code{strong}. It defaults to @code{none}.
12739 Compute and draw a color distribution histogram for the input video.
12741 The computed histogram is a representation of the color component
12742 distribution in an image.
12744 Standard histogram displays the color components distribution in an image.
12745 Displays color graph for each color component. Shows distribution of
12746 the Y, U, V, A or R, G, B components, depending on input format, in the
12747 current frame. Below each graph a color component scale meter is shown.
12749 The filter accepts the following options:
12753 Set height of level. Default value is @code{200}.
12754 Allowed range is [50, 2048].
12757 Set height of color scale. Default value is @code{12}.
12758 Allowed range is [0, 40].
12762 It accepts the following values:
12765 Per color component graphs are placed below each other.
12768 Per color component graphs are placed side by side.
12771 Presents information identical to that in the @code{parade}, except
12772 that the graphs representing color components are superimposed directly
12775 Default is @code{stack}.
12778 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12779 Default is @code{linear}.
12782 Set what color components to display.
12783 Default is @code{7}.
12786 Set foreground opacity. Default is @code{0.7}.
12789 Set background opacity. Default is @code{0.5}.
12792 @subsection Examples
12797 Calculate and draw histogram:
12799 ffplay -i input -vf histogram
12807 This is a high precision/quality 3d denoise filter. It aims to reduce
12808 image noise, producing smooth images and making still images really
12809 still. It should enhance compressibility.
12811 It accepts the following optional parameters:
12815 A non-negative floating point number which specifies spatial luma strength.
12816 It defaults to 4.0.
12818 @item chroma_spatial
12819 A non-negative floating point number which specifies spatial chroma strength.
12820 It defaults to 3.0*@var{luma_spatial}/4.0.
12823 A floating point number which specifies luma temporal strength. It defaults to
12824 6.0*@var{luma_spatial}/4.0.
12827 A floating point number which specifies chroma temporal strength. It defaults to
12828 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12831 @subsection Commands
12832 This filter supports same @ref{commands} as options.
12833 The command accepts the same syntax of the corresponding option.
12835 If the specified expression is not valid, it is kept at its current
12838 @anchor{hwdownload}
12839 @section hwdownload
12841 Download hardware frames to system memory.
12843 The input must be in hardware frames, and the output a non-hardware format.
12844 Not all formats will be supported on the output - it may be necessary to insert
12845 an additional @option{format} filter immediately following in the graph to get
12846 the output in a supported format.
12850 Map hardware frames to system memory or to another device.
12852 This filter has several different modes of operation; which one is used depends
12853 on the input and output formats:
12856 Hardware frame input, normal frame output
12858 Map the input frames to system memory and pass them to the output. If the
12859 original hardware frame is later required (for example, after overlaying
12860 something else on part of it), the @option{hwmap} filter can be used again
12861 in the next mode to retrieve it.
12863 Normal frame input, hardware frame output
12865 If the input is actually a software-mapped hardware frame, then unmap it -
12866 that is, return the original hardware frame.
12868 Otherwise, a device must be provided. Create new hardware surfaces on that
12869 device for the output, then map them back to the software format at the input
12870 and give those frames to the preceding filter. This will then act like the
12871 @option{hwupload} filter, but may be able to avoid an additional copy when
12872 the input is already in a compatible format.
12874 Hardware frame input and output
12876 A device must be supplied for the output, either directly or with the
12877 @option{derive_device} option. The input and output devices must be of
12878 different types and compatible - the exact meaning of this is
12879 system-dependent, but typically it means that they must refer to the same
12880 underlying hardware context (for example, refer to the same graphics card).
12882 If the input frames were originally created on the output device, then unmap
12883 to retrieve the original frames.
12885 Otherwise, map the frames to the output device - create new hardware frames
12886 on the output corresponding to the frames on the input.
12889 The following additional parameters are accepted:
12893 Set the frame mapping mode. Some combination of:
12896 The mapped frame should be readable.
12898 The mapped frame should be writeable.
12900 The mapping will always overwrite the entire frame.
12902 This may improve performance in some cases, as the original contents of the
12903 frame need not be loaded.
12905 The mapping must not involve any copying.
12907 Indirect mappings to copies of frames are created in some cases where either
12908 direct mapping is not possible or it would have unexpected properties.
12909 Setting this flag ensures that the mapping is direct and will fail if that is
12912 Defaults to @var{read+write} if not specified.
12914 @item derive_device @var{type}
12915 Rather than using the device supplied at initialisation, instead derive a new
12916 device of type @var{type} from the device the input frames exist on.
12919 In a hardware to hardware mapping, map in reverse - create frames in the sink
12920 and map them back to the source. This may be necessary in some cases where
12921 a mapping in one direction is required but only the opposite direction is
12922 supported by the devices being used.
12924 This option is dangerous - it may break the preceding filter in undefined
12925 ways if there are any additional constraints on that filter's output.
12926 Do not use it without fully understanding the implications of its use.
12932 Upload system memory frames to hardware surfaces.
12934 The device to upload to must be supplied when the filter is initialised. If
12935 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12936 option or with the @option{derive_device} option. The input and output devices
12937 must be of different types and compatible - the exact meaning of this is
12938 system-dependent, but typically it means that they must refer to the same
12939 underlying hardware context (for example, refer to the same graphics card).
12941 The following additional parameters are accepted:
12944 @item derive_device @var{type}
12945 Rather than using the device supplied at initialisation, instead derive a new
12946 device of type @var{type} from the device the input frames exist on.
12949 @anchor{hwupload_cuda}
12950 @section hwupload_cuda
12952 Upload system memory frames to a CUDA device.
12954 It accepts the following optional parameters:
12958 The number of the CUDA device to use
12963 Apply a high-quality magnification filter designed for pixel art. This filter
12964 was originally created by Maxim Stepin.
12966 It accepts the following option:
12970 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12971 @code{hq3x} and @code{4} for @code{hq4x}.
12972 Default is @code{3}.
12976 Stack input videos horizontally.
12978 All streams must be of same pixel format and of same height.
12980 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12981 to create same output.
12983 The filter accepts the following option:
12987 Set number of input streams. Default is 2.
12990 If set to 1, force the output to terminate when the shortest input
12991 terminates. Default value is 0.
12996 Modify the hue and/or the saturation of the input.
12998 It accepts the following parameters:
13002 Specify the hue angle as a number of degrees. It accepts an expression,
13003 and defaults to "0".
13006 Specify the saturation in the [-10,10] range. It accepts an expression and
13010 Specify the hue angle as a number of radians. It accepts an
13011 expression, and defaults to "0".
13014 Specify the brightness in the [-10,10] range. It accepts an expression and
13018 @option{h} and @option{H} are mutually exclusive, and can't be
13019 specified at the same time.
13021 The @option{b}, @option{h}, @option{H} and @option{s} option values are
13022 expressions containing the following constants:
13026 frame count of the input frame starting from 0
13029 presentation timestamp of the input frame expressed in time base units
13032 frame rate of the input video, NAN if the input frame rate is unknown
13035 timestamp expressed in seconds, NAN if the input timestamp is unknown
13038 time base of the input video
13041 @subsection Examples
13045 Set the hue to 90 degrees and the saturation to 1.0:
13051 Same command but expressing the hue in radians:
13057 Rotate hue and make the saturation swing between 0
13058 and 2 over a period of 1 second:
13060 hue="H=2*PI*t: s=sin(2*PI*t)+1"
13064 Apply a 3 seconds saturation fade-in effect starting at 0:
13066 hue="s=min(t/3\,1)"
13069 The general fade-in expression can be written as:
13071 hue="s=min(0\, max((t-START)/DURATION\, 1))"
13075 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
13077 hue="s=max(0\, min(1\, (8-t)/3))"
13080 The general fade-out expression can be written as:
13082 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
13087 @subsection Commands
13089 This filter supports the following commands:
13095 Modify the hue and/or the saturation and/or brightness of the input video.
13096 The command accepts the same syntax of the corresponding option.
13098 If the specified expression is not valid, it is kept at its current
13102 @section hysteresis
13104 Grow first stream into second stream by connecting components.
13105 This makes it possible to build more robust edge masks.
13107 This filter accepts the following options:
13111 Set which planes will be processed as bitmap, unprocessed planes will be
13112 copied from first stream.
13113 By default value 0xf, all planes will be processed.
13116 Set threshold which is used in filtering. If pixel component value is higher than
13117 this value filter algorithm for connecting components is activated.
13118 By default value is 0.
13121 The @code{hysteresis} filter also supports the @ref{framesync} options.
13125 Detect video interlacing type.
13127 This filter tries to detect if the input frames are interlaced, progressive,
13128 top or bottom field first. It will also try to detect fields that are
13129 repeated between adjacent frames (a sign of telecine).
13131 Single frame detection considers only immediately adjacent frames when classifying each frame.
13132 Multiple frame detection incorporates the classification history of previous frames.
13134 The filter will log these metadata values:
13137 @item single.current_frame
13138 Detected type of current frame using single-frame detection. One of:
13139 ``tff'' (top field first), ``bff'' (bottom field first),
13140 ``progressive'', or ``undetermined''
13143 Cumulative number of frames detected as top field first using single-frame detection.
13146 Cumulative number of frames detected as top field first using multiple-frame detection.
13149 Cumulative number of frames detected as bottom field first using single-frame detection.
13151 @item multiple.current_frame
13152 Detected type of current frame using multiple-frame detection. One of:
13153 ``tff'' (top field first), ``bff'' (bottom field first),
13154 ``progressive'', or ``undetermined''
13157 Cumulative number of frames detected as bottom field first using multiple-frame detection.
13159 @item single.progressive
13160 Cumulative number of frames detected as progressive using single-frame detection.
13162 @item multiple.progressive
13163 Cumulative number of frames detected as progressive using multiple-frame detection.
13165 @item single.undetermined
13166 Cumulative number of frames that could not be classified using single-frame detection.
13168 @item multiple.undetermined
13169 Cumulative number of frames that could not be classified using multiple-frame detection.
13171 @item repeated.current_frame
13172 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
13174 @item repeated.neither
13175 Cumulative number of frames with no repeated field.
13178 Cumulative number of frames with the top field repeated from the previous frame's top field.
13180 @item repeated.bottom
13181 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
13184 The filter accepts the following options:
13188 Set interlacing threshold.
13190 Set progressive threshold.
13192 Threshold for repeated field detection.
13194 Number of frames after which a given frame's contribution to the
13195 statistics is halved (i.e., it contributes only 0.5 to its
13196 classification). The default of 0 means that all frames seen are given
13197 full weight of 1.0 forever.
13198 @item analyze_interlaced_flag
13199 When this is not 0 then idet will use the specified number of frames to determine
13200 if the interlaced flag is accurate, it will not count undetermined frames.
13201 If the flag is found to be accurate it will be used without any further
13202 computations, if it is found to be inaccurate it will be cleared without any
13203 further computations. This allows inserting the idet filter as a low computational
13204 method to clean up the interlaced flag
13209 Deinterleave or interleave fields.
13211 This filter allows one to process interlaced images fields without
13212 deinterlacing them. Deinterleaving splits the input frame into 2
13213 fields (so called half pictures). Odd lines are moved to the top
13214 half of the output image, even lines to the bottom half.
13215 You can process (filter) them independently and then re-interleave them.
13217 The filter accepts the following options:
13221 @item chroma_mode, c
13222 @item alpha_mode, a
13223 Available values for @var{luma_mode}, @var{chroma_mode} and
13224 @var{alpha_mode} are:
13230 @item deinterleave, d
13231 Deinterleave fields, placing one above the other.
13233 @item interleave, i
13234 Interleave fields. Reverse the effect of deinterleaving.
13236 Default value is @code{none}.
13238 @item luma_swap, ls
13239 @item chroma_swap, cs
13240 @item alpha_swap, as
13241 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
13244 @subsection Commands
13246 This filter supports the all above options as @ref{commands}.
13250 Apply inflate effect to the video.
13252 This filter replaces the pixel by the local(3x3) average by taking into account
13253 only values higher than the pixel.
13255 It accepts the following options:
13262 Limit the maximum change for each plane, default is 65535.
13263 If 0, plane will remain unchanged.
13266 @subsection Commands
13268 This filter supports the all above options as @ref{commands}.
13272 Simple interlacing filter from progressive contents. This interleaves upper (or
13273 lower) lines from odd frames with lower (or upper) lines from even frames,
13274 halving the frame rate and preserving image height.
13277 Original Original New Frame
13278 Frame 'j' Frame 'j+1' (tff)
13279 ========== =========== ==================
13280 Line 0 --------------------> Frame 'j' Line 0
13281 Line 1 Line 1 ----> Frame 'j+1' Line 1
13282 Line 2 ---------------------> Frame 'j' Line 2
13283 Line 3 Line 3 ----> Frame 'j+1' Line 3
13285 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
13288 It accepts the following optional parameters:
13292 This determines whether the interlaced frame is taken from the even
13293 (tff - default) or odd (bff) lines of the progressive frame.
13296 Vertical lowpass filter to avoid twitter interlacing and
13297 reduce moire patterns.
13301 Disable vertical lowpass filter
13304 Enable linear filter (default)
13307 Enable complex filter. This will slightly less reduce twitter and moire
13308 but better retain detail and subjective sharpness impression.
13315 Deinterlace input video by applying Donald Graft's adaptive kernel
13316 deinterling. Work on interlaced parts of a video to produce
13317 progressive frames.
13319 The description of the accepted parameters follows.
13323 Set the threshold which affects the filter's tolerance when
13324 determining if a pixel line must be processed. It must be an integer
13325 in the range [0,255] and defaults to 10. A value of 0 will result in
13326 applying the process on every pixels.
13329 Paint pixels exceeding the threshold value to white if set to 1.
13333 Set the fields order. Swap fields if set to 1, leave fields alone if
13337 Enable additional sharpening if set to 1. Default is 0.
13340 Enable twoway sharpening if set to 1. Default is 0.
13343 @subsection Examples
13347 Apply default values:
13349 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
13353 Enable additional sharpening:
13359 Paint processed pixels in white:
13367 Slowly update darker pixels.
13369 This filter makes short flashes of light appear longer.
13370 This filter accepts the following options:
13374 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13377 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13380 @section lenscorrection
13382 Correct radial lens distortion
13384 This filter can be used to correct for radial distortion as can result from the use
13385 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13386 one can use tools available for example as part of opencv or simply trial-and-error.
13387 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13388 and extract the k1 and k2 coefficients from the resulting matrix.
13390 Note that effectively the same filter is available in the open-source tools Krita and
13391 Digikam from the KDE project.
13393 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13394 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13395 brightness distribution, so you may want to use both filters together in certain
13396 cases, though you will have to take care of ordering, i.e. whether vignetting should
13397 be applied before or after lens correction.
13399 @subsection Options
13401 The filter accepts the following options:
13405 Relative x-coordinate of the focal point of the image, and thereby the center of the
13406 distortion. This value has a range [0,1] and is expressed as fractions of the image
13407 width. Default is 0.5.
13409 Relative y-coordinate of the focal point of the image, and thereby the center of the
13410 distortion. This value has a range [0,1] and is expressed as fractions of the image
13411 height. Default is 0.5.
13413 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13414 no correction. Default is 0.
13416 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13417 0 means no correction. Default is 0.
13420 The formula that generates the correction is:
13422 @var{r_src} = @var{r_tgt} * (1 + @var{k1} * (@var{r_tgt} / @var{r_0})^2 + @var{k2} * (@var{r_tgt} / @var{r_0})^4)
13424 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13425 distances from the focal point in the source and target images, respectively.
13429 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13431 The @code{lensfun} filter requires the camera make, camera model, and lens model
13432 to apply the lens correction. The filter will load the lensfun database and
13433 query it to find the corresponding camera and lens entries in the database. As
13434 long as these entries can be found with the given options, the filter can
13435 perform corrections on frames. Note that incomplete strings will result in the
13436 filter choosing the best match with the given options, and the filter will
13437 output the chosen camera and lens models (logged with level "info"). You must
13438 provide the make, camera model, and lens model as they are required.
13440 The filter accepts the following options:
13444 The make of the camera (for example, "Canon"). This option is required.
13447 The model of the camera (for example, "Canon EOS 100D"). This option is
13451 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13452 option is required.
13455 The type of correction to apply. The following values are valid options:
13459 Enables fixing lens vignetting.
13462 Enables fixing lens geometry. This is the default.
13465 Enables fixing chromatic aberrations.
13468 Enables fixing lens vignetting and lens geometry.
13471 Enables fixing lens vignetting and chromatic aberrations.
13474 Enables fixing both lens geometry and chromatic aberrations.
13477 Enables all possible corrections.
13481 The focal length of the image/video (zoom; expected constant for video). For
13482 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13483 range should be chosen when using that lens. Default 18.
13486 The aperture of the image/video (expected constant for video). Note that
13487 aperture is only used for vignetting correction. Default 3.5.
13489 @item focus_distance
13490 The focus distance of the image/video (expected constant for video). Note that
13491 focus distance is only used for vignetting and only slightly affects the
13492 vignetting correction process. If unknown, leave it at the default value (which
13496 The scale factor which is applied after transformation. After correction the
13497 video is no longer necessarily rectangular. This parameter controls how much of
13498 the resulting image is visible. The value 0 means that a value will be chosen
13499 automatically such that there is little or no unmapped area in the output
13500 image. 1.0 means that no additional scaling is done. Lower values may result
13501 in more of the corrected image being visible, while higher values may avoid
13502 unmapped areas in the output.
13504 @item target_geometry
13505 The target geometry of the output image/video. The following values are valid
13509 @item rectilinear (default)
13512 @item equirectangular
13513 @item fisheye_orthographic
13514 @item fisheye_stereographic
13515 @item fisheye_equisolid
13516 @item fisheye_thoby
13519 Apply the reverse of image correction (instead of correcting distortion, apply
13522 @item interpolation
13523 The type of interpolation used when correcting distortion. The following values
13528 @item linear (default)
13533 @subsection Examples
13537 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13538 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13542 ffmpeg -i input.mov -vf lensfun=make=Canon:model="Canon EOS 100D":lens_model="Canon EF-S 18-55mm f/3.5-5.6 IS STM":focal_length=18:aperture=8 -c:v h264 -b:v 8000k output.mov
13546 Apply the same as before, but only for the first 5 seconds of video.
13549 ffmpeg -i input.mov -vf lensfun=make=Canon:model="Canon EOS 100D":lens_model="Canon EF-S 18-55mm f/3.5-5.6 IS STM":focal_length=18:aperture=8:enable='lte(t\,5)' -c:v h264 -b:v 8000k output.mov
13556 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13557 score between two input videos.
13559 The obtained VMAF score is printed through the logging system.
13561 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13562 After installing the library it can be enabled using:
13563 @code{./configure --enable-libvmaf}.
13564 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13566 The filter has following options:
13570 Set the model path which is to be used for SVM.
13571 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13574 Set the file path to be used to store logs.
13577 Set the format of the log file (csv, json or xml).
13579 @item enable_transform
13580 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13581 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13582 Default value: @code{false}
13585 Invokes the phone model which will generate VMAF scores higher than in the
13586 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13587 Default value: @code{false}
13590 Enables computing psnr along with vmaf.
13591 Default value: @code{false}
13594 Enables computing ssim along with vmaf.
13595 Default value: @code{false}
13598 Enables computing ms_ssim along with vmaf.
13599 Default value: @code{false}
13602 Set the pool method to be used for computing vmaf.
13603 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13606 Set number of threads to be used when computing vmaf.
13607 Default value: @code{0}, which makes use of all available logical processors.
13610 Set interval for frame subsampling used when computing vmaf.
13611 Default value: @code{1}
13613 @item enable_conf_interval
13614 Enables confidence interval.
13615 Default value: @code{false}
13618 This filter also supports the @ref{framesync} options.
13620 @subsection Examples
13623 On the below examples the input file @file{main.mpg} being processed is
13624 compared with the reference file @file{ref.mpg}.
13627 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13631 Example with options:
13633 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13637 Example with options and different containers:
13639 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]libvmaf=psnr=1:log_fmt=json" -f null -
13645 Limits the pixel components values to the specified range [min, max].
13647 The filter accepts the following options:
13651 Lower bound. Defaults to the lowest allowed value for the input.
13654 Upper bound. Defaults to the highest allowed value for the input.
13657 Specify which planes will be processed. Defaults to all available.
13660 @subsection Commands
13662 This filter supports the all above options as @ref{commands}.
13668 The filter accepts the following options:
13672 Set the number of loops. Setting this value to -1 will result in infinite loops.
13676 Set maximal size in number of frames. Default is 0.
13679 Set first frame of loop. Default is 0.
13682 @subsection Examples
13686 Loop single first frame infinitely:
13688 loop=loop=-1:size=1:start=0
13692 Loop single first frame 10 times:
13694 loop=loop=10:size=1:start=0
13698 Loop 10 first frames 5 times:
13700 loop=loop=5:size=10:start=0
13706 Apply a 1D LUT to an input video.
13708 The filter accepts the following options:
13712 Set the 1D LUT file name.
13714 Currently supported formats:
13723 Select interpolation mode.
13725 Available values are:
13729 Use values from the nearest defined point.
13731 Interpolate values using the linear interpolation.
13733 Interpolate values using the cosine interpolation.
13735 Interpolate values using the cubic interpolation.
13737 Interpolate values using the spline interpolation.
13744 Apply a 3D LUT to an input video.
13746 The filter accepts the following options:
13750 Set the 3D LUT file name.
13752 Currently supported formats:
13766 Select interpolation mode.
13768 Available values are:
13772 Use values from the nearest defined point.
13774 Interpolate values using the 8 points defining a cube.
13776 Interpolate values using a tetrahedron.
13782 Turn certain luma values into transparency.
13784 The filter accepts the following options:
13788 Set the luma which will be used as base for transparency.
13789 Default value is @code{0}.
13792 Set the range of luma values to be keyed out.
13793 Default value is @code{0.01}.
13796 Set the range of softness. Default value is @code{0}.
13797 Use this to control gradual transition from zero to full transparency.
13800 @subsection Commands
13801 This filter supports same @ref{commands} as options.
13802 The command accepts the same syntax of the corresponding option.
13804 If the specified expression is not valid, it is kept at its current
13807 @section lut, lutrgb, lutyuv
13809 Compute a look-up table for binding each pixel component input value
13810 to an output value, and apply it to the input video.
13812 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13813 to an RGB input video.
13815 These filters accept the following parameters:
13818 set first pixel component expression
13820 set second pixel component expression
13822 set third pixel component expression
13824 set fourth pixel component expression, corresponds to the alpha component
13827 set red component expression
13829 set green component expression
13831 set blue component expression
13833 alpha component expression
13836 set Y/luminance component expression
13838 set U/Cb component expression
13840 set V/Cr component expression
13843 Each of them specifies the expression to use for computing the lookup table for
13844 the corresponding pixel component values.
13846 The exact component associated to each of the @var{c*} options depends on the
13849 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13850 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13852 The expressions can contain the following constants and functions:
13857 The input width and height.
13860 The input value for the pixel component.
13863 The input value, clipped to the @var{minval}-@var{maxval} range.
13866 The maximum value for the pixel component.
13869 The minimum value for the pixel component.
13872 The negated value for the pixel component value, clipped to the
13873 @var{minval}-@var{maxval} range; it corresponds to the expression
13874 "maxval-clipval+minval".
13877 The computed value in @var{val}, clipped to the
13878 @var{minval}-@var{maxval} range.
13880 @item gammaval(gamma)
13881 The computed gamma correction value of the pixel component value,
13882 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13884 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13888 All expressions default to "val".
13890 @subsection Examples
13894 Negate input video:
13896 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13897 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13900 The above is the same as:
13902 lutrgb="r=negval:g=negval:b=negval"
13903 lutyuv="y=negval:u=negval:v=negval"
13913 Remove chroma components, turning the video into a graytone image:
13915 lutyuv="u=128:v=128"
13919 Apply a luma burning effect:
13925 Remove green and blue components:
13931 Set a constant alpha channel value on input:
13933 format=rgba,lutrgb=a="maxval-minval/2"
13937 Correct luminance gamma by a factor of 0.5:
13939 lutyuv=y=gammaval(0.5)
13943 Discard least significant bits of luma:
13945 lutyuv=y='bitand(val, 128+64+32)'
13949 Technicolor like effect:
13951 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13955 @section lut2, tlut2
13957 The @code{lut2} filter takes two input streams and outputs one
13960 The @code{tlut2} (time lut2) filter takes two consecutive frames
13961 from one single stream.
13963 This filter accepts the following parameters:
13966 set first pixel component expression
13968 set second pixel component expression
13970 set third pixel component expression
13972 set fourth pixel component expression, corresponds to the alpha component
13975 set output bit depth, only available for @code{lut2} filter. By default is 0,
13976 which means bit depth is automatically picked from first input format.
13979 The @code{lut2} filter also supports the @ref{framesync} options.
13981 Each of them specifies the expression to use for computing the lookup table for
13982 the corresponding pixel component values.
13984 The exact component associated to each of the @var{c*} options depends on the
13987 The expressions can contain the following constants:
13992 The input width and height.
13995 The first input value for the pixel component.
13998 The second input value for the pixel component.
14001 The first input video bit depth.
14004 The second input video bit depth.
14007 All expressions default to "x".
14009 @subsection Examples
14013 Highlight differences between two RGB video streams:
14015 lut2='ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,0,pow(2,bdx)-1)'
14019 Highlight differences between two YUV video streams:
14021 lut2='ifnot(x-y,0,pow(2,bdx)-1):ifnot(x-y,pow(2,bdx-1),pow(2,bdx)-1):ifnot(x-y,pow(2,bdx-1),pow(2,bdx)-1)'
14025 Show max difference between two video streams:
14027 lut2='if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1))):if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1))):if(lt(x,y),0,if(gt(x,y),pow(2,bdx)-1,pow(2,bdx-1)))'
14031 @section maskedclamp
14033 Clamp the first input stream with the second input and third input stream.
14035 Returns the value of first stream to be between second input
14036 stream - @code{undershoot} and third input stream + @code{overshoot}.
14038 This filter accepts the following options:
14041 Default value is @code{0}.
14044 Default value is @code{0}.
14047 Set which planes will be processed as bitmap, unprocessed planes will be
14048 copied from first stream.
14049 By default value 0xf, all planes will be processed.
14052 @subsection Commands
14054 This filter supports the all above options as @ref{commands}.
14058 Merge the second and third input stream into output stream using absolute differences
14059 between second input stream and first input stream and absolute difference between
14060 third input stream and first input stream. The picked value will be from second input
14061 stream if second absolute difference is greater than first one or from third input stream
14064 This filter accepts the following options:
14067 Set which planes will be processed as bitmap, unprocessed planes will be
14068 copied from first stream.
14069 By default value 0xf, all planes will be processed.
14072 @subsection Commands
14074 This filter supports the all above options as @ref{commands}.
14076 @section maskedmerge
14078 Merge the first input stream with the second input stream using per pixel
14079 weights in the third input stream.
14081 A value of 0 in the third stream pixel component means that pixel component
14082 from first stream is returned unchanged, while maximum value (eg. 255 for
14083 8-bit videos) means that pixel component from second stream is returned
14084 unchanged. Intermediate values define the amount of merging between both
14085 input stream's pixel components.
14087 This filter accepts the following options:
14090 Set which planes will be processed as bitmap, unprocessed planes will be
14091 copied from first stream.
14092 By default value 0xf, all planes will be processed.
14095 @subsection Commands
14097 This filter supports the all above options as @ref{commands}.
14101 Merge the second and third input stream into output stream using absolute differences
14102 between second input stream and first input stream and absolute difference between
14103 third input stream and first input stream. The picked value will be from second input
14104 stream if second absolute difference is less than first one or from third input stream
14107 This filter accepts the following options:
14110 Set which planes will be processed as bitmap, unprocessed planes will be
14111 copied from first stream.
14112 By default value 0xf, all planes will be processed.
14115 @subsection Commands
14117 This filter supports the all above options as @ref{commands}.
14119 @section maskedthreshold
14120 Pick pixels comparing absolute difference of two video streams with fixed
14123 If absolute difference between pixel component of first and second video
14124 stream is equal or lower than user supplied threshold than pixel component
14125 from first video stream is picked, otherwise pixel component from second
14126 video stream is picked.
14128 This filter accepts the following options:
14131 Set threshold used when picking pixels from absolute difference from two input
14135 Set which planes will be processed as bitmap, unprocessed planes will be
14136 copied from second stream.
14137 By default value 0xf, all planes will be processed.
14140 @subsection Commands
14142 This filter supports the all above options as @ref{commands}.
14145 Create mask from input video.
14147 For example it is useful to create motion masks after @code{tblend} filter.
14149 This filter accepts the following options:
14153 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
14156 Set high threshold. Any pixel component higher than this value will be set to max value
14157 allowed for current pixel format.
14160 Set planes to filter, by default all available planes are filtered.
14163 Fill all frame pixels with this value.
14166 Set max average pixel value for frame. If sum of all pixel components is higher that this
14167 average, output frame will be completely filled with value set by @var{fill} option.
14168 Typically useful for scene changes when used in combination with @code{tblend} filter.
14173 Apply motion-compensation deinterlacing.
14175 It needs one field per frame as input and must thus be used together
14176 with yadif=1/3 or equivalent.
14178 This filter accepts the following options:
14181 Set the deinterlacing mode.
14183 It accepts one of the following values:
14188 use iterative motion estimation
14190 like @samp{slow}, but use multiple reference frames.
14192 Default value is @samp{fast}.
14195 Set the picture field parity assumed for the input video. It must be
14196 one of the following values:
14200 assume top field first
14202 assume bottom field first
14205 Default value is @samp{bff}.
14208 Set per-block quantization parameter (QP) used by the internal
14211 Higher values should result in a smoother motion vector field but less
14212 optimal individual vectors. Default value is 1.
14217 Pick median pixel from certain rectangle defined by radius.
14219 This filter accepts the following options:
14223 Set horizontal radius size. Default value is @code{1}.
14224 Allowed range is integer from 1 to 127.
14227 Set which planes to process. Default is @code{15}, which is all available planes.
14230 Set vertical radius size. Default value is @code{0}.
14231 Allowed range is integer from 0 to 127.
14232 If it is 0, value will be picked from horizontal @code{radius} option.
14235 Set median percentile. Default value is @code{0.5}.
14236 Default value of @code{0.5} will pick always median values, while @code{0} will pick
14237 minimum values, and @code{1} maximum values.
14240 @subsection Commands
14241 This filter supports same @ref{commands} as options.
14242 The command accepts the same syntax of the corresponding option.
14244 If the specified expression is not valid, it is kept at its current
14247 @section mergeplanes
14249 Merge color channel components from several video streams.
14251 The filter accepts up to 4 input streams, and merge selected input
14252 planes to the output video.
14254 This filter accepts the following options:
14257 Set input to output plane mapping. Default is @code{0}.
14259 The mappings is specified as a bitmap. It should be specified as a
14260 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
14261 mapping for the first plane of the output stream. 'A' sets the number of
14262 the input stream to use (from 0 to 3), and 'a' the plane number of the
14263 corresponding input to use (from 0 to 3). The rest of the mappings is
14264 similar, 'Bb' describes the mapping for the output stream second
14265 plane, 'Cc' describes the mapping for the output stream third plane and
14266 'Dd' describes the mapping for the output stream fourth plane.
14269 Set output pixel format. Default is @code{yuva444p}.
14272 @subsection Examples
14276 Merge three gray video streams of same width and height into single video stream:
14278 [a0][a1][a2]mergeplanes=0x001020:yuv444p
14282 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
14284 [a0][a1]mergeplanes=0x00010210:yuva444p
14288 Swap Y and A plane in yuva444p stream:
14290 format=yuva444p,mergeplanes=0x03010200:yuva444p
14294 Swap U and V plane in yuv420p stream:
14296 format=yuv420p,mergeplanes=0x000201:yuv420p
14300 Cast a rgb24 clip to yuv444p:
14302 format=rgb24,mergeplanes=0x000102:yuv444p
14308 Estimate and export motion vectors using block matching algorithms.
14309 Motion vectors are stored in frame side data to be used by other filters.
14311 This filter accepts the following options:
14314 Specify the motion estimation method. Accepts one of the following values:
14318 Exhaustive search algorithm.
14320 Three step search algorithm.
14322 Two dimensional logarithmic search algorithm.
14324 New three step search algorithm.
14326 Four step search algorithm.
14328 Diamond search algorithm.
14330 Hexagon-based search algorithm.
14332 Enhanced predictive zonal search algorithm.
14334 Uneven multi-hexagon search algorithm.
14336 Default value is @samp{esa}.
14339 Macroblock size. Default @code{16}.
14342 Search parameter. Default @code{7}.
14345 @section midequalizer
14347 Apply Midway Image Equalization effect using two video streams.
14349 Midway Image Equalization adjusts a pair of images to have the same
14350 histogram, while maintaining their dynamics as much as possible. It's
14351 useful for e.g. matching exposures from a pair of stereo cameras.
14353 This filter has two inputs and one output, which must be of same pixel format, but
14354 may be of different sizes. The output of filter is first input adjusted with
14355 midway histogram of both inputs.
14357 This filter accepts the following option:
14361 Set which planes to process. Default is @code{15}, which is all available planes.
14364 @section minterpolate
14366 Convert the video to specified frame rate using motion interpolation.
14368 This filter accepts the following options:
14371 Specify the output frame rate. This can be rational e.g. @code{60000/1001}. Frames are dropped if @var{fps} is lower than source fps. Default @code{60}.
14374 Motion interpolation mode. Following values are accepted:
14377 Duplicate previous or next frame for interpolating new ones.
14379 Blend source frames. Interpolated frame is mean of previous and next frames.
14381 Motion compensated interpolation. Following options are effective when this mode is selected:
14385 Motion compensation mode. Following values are accepted:
14388 Overlapped block motion compensation.
14390 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
14392 Default mode is @samp{obmc}.
14395 Motion estimation mode. Following values are accepted:
14398 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14400 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14402 Default mode is @samp{bilat}.
14405 The algorithm to be used for motion estimation. Following values are accepted:
14408 Exhaustive search algorithm.
14410 Three step search algorithm.
14412 Two dimensional logarithmic search algorithm.
14414 New three step search algorithm.
14416 Four step search algorithm.
14418 Diamond search algorithm.
14420 Hexagon-based search algorithm.
14422 Enhanced predictive zonal search algorithm.
14424 Uneven multi-hexagon search algorithm.
14426 Default algorithm is @samp{epzs}.
14429 Macroblock size. Default @code{16}.
14432 Motion estimation search parameter. Default @code{32}.
14435 Enable variable-size block motion compensation. Motion estimation is applied with smaller block sizes at object boundaries in order to make the them less blur. Default is @code{0} (disabled).
14440 Scene change detection method. Scene change leads motion vectors to be in random direction. Scene change detection replace interpolated frames by duplicate ones. May not be needed for other modes. Following values are accepted:
14443 Disable scene change detection.
14445 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14447 Default method is @samp{fdiff}.
14449 @item scd_threshold
14450 Scene change detection threshold. Default is @code{10.}.
14455 Mix several video input streams into one video stream.
14457 A description of the accepted options follows.
14461 The number of inputs. If unspecified, it defaults to 2.
14464 Specify weight of each input video stream as sequence.
14465 Each weight is separated by space. If number of weights
14466 is smaller than number of @var{frames} last specified
14467 weight will be used for all remaining unset weights.
14470 Specify scale, if it is set it will be multiplied with sum
14471 of each weight multiplied with pixel values to give final destination
14472 pixel value. By default @var{scale} is auto scaled to sum of weights.
14475 Specify how end of stream is determined.
14478 The duration of the longest input. (default)
14481 The duration of the shortest input.
14484 The duration of the first input.
14488 @section mpdecimate
14490 Drop frames that do not differ greatly from the previous frame in
14491 order to reduce frame rate.
14493 The main use of this filter is for very-low-bitrate encoding
14494 (e.g. streaming over dialup modem), but it could in theory be used for
14495 fixing movies that were inverse-telecined incorrectly.
14497 A description of the accepted options follows.
14501 Set the maximum number of consecutive frames which can be dropped (if
14502 positive), or the minimum interval between dropped frames (if
14503 negative). If the value is 0, the frame is dropped disregarding the
14504 number of previous sequentially dropped frames.
14506 Default value is 0.
14511 Set the dropping threshold values.
14513 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14514 represent actual pixel value differences, so a threshold of 64
14515 corresponds to 1 unit of difference for each pixel, or the same spread
14516 out differently over the block.
14518 A frame is a candidate for dropping if no 8x8 blocks differ by more
14519 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14520 meaning the whole image) differ by more than a threshold of @option{lo}.
14522 Default value for @option{hi} is 64*12, default value for @option{lo} is
14523 64*5, and default value for @option{frac} is 0.33.
14529 Negate (invert) the input video.
14531 It accepts the following option:
14536 With value 1, it negates the alpha component, if present. Default value is 0.
14542 Denoise frames using Non-Local Means algorithm.
14544 Each pixel is adjusted by looking for other pixels with similar contexts. This
14545 context similarity is defined by comparing their surrounding patches of size
14546 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14549 Note that the research area defines centers for patches, which means some
14550 patches will be made of pixels outside that research area.
14552 The filter accepts the following options.
14556 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14559 Set patch size. Default is 7. Must be odd number in range [0, 99].
14562 Same as @option{p} but for chroma planes.
14564 The default value is @var{0} and means automatic.
14567 Set research size. Default is 15. Must be odd number in range [0, 99].
14570 Same as @option{r} but for chroma planes.
14572 The default value is @var{0} and means automatic.
14577 Deinterlace video using neural network edge directed interpolation.
14579 This filter accepts the following options:
14583 Mandatory option, without binary file filter can not work.
14584 Currently file can be found here:
14585 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14588 Set which frames to deinterlace, by default it is @code{all}.
14589 Can be @code{all} or @code{interlaced}.
14592 Set mode of operation.
14594 Can be one of the following:
14598 Use frame flags, both fields.
14600 Use frame flags, single field.
14602 Use top field only.
14604 Use bottom field only.
14606 Use both fields, top first.
14608 Use both fields, bottom first.
14612 Set which planes to process, by default filter process all frames.
14615 Set size of local neighborhood around each pixel, used by the predictor neural
14618 Can be one of the following:
14631 Set the number of neurons in predictor neural network.
14632 Can be one of the following:
14643 Controls the number of different neural network predictions that are blended
14644 together to compute the final output value. Can be @code{fast}, default or
14648 Set which set of weights to use in the predictor.
14649 Can be one of the following:
14653 weights trained to minimize absolute error
14655 weights trained to minimize squared error
14659 Controls whether or not the prescreener neural network is used to decide
14660 which pixels should be processed by the predictor neural network and which
14661 can be handled by simple cubic interpolation.
14662 The prescreener is trained to know whether cubic interpolation will be
14663 sufficient for a pixel or whether it should be predicted by the predictor nn.
14664 The computational complexity of the prescreener nn is much less than that of
14665 the predictor nn. Since most pixels can be handled by cubic interpolation,
14666 using the prescreener generally results in much faster processing.
14667 The prescreener is pretty accurate, so the difference between using it and not
14668 using it is almost always unnoticeable.
14670 Can be one of the following:
14680 Default is @code{new}.
14683 @subsection Commands
14684 This filter supports same @ref{commands} as options, excluding @var{weights} option.
14688 Force libavfilter not to use any of the specified pixel formats for the
14689 input to the next filter.
14691 It accepts the following parameters:
14695 A '|'-separated list of pixel format names, such as
14696 pix_fmts=yuv420p|monow|rgb24".
14700 @subsection Examples
14704 Force libavfilter to use a format different from @var{yuv420p} for the
14705 input to the vflip filter:
14707 noformat=pix_fmts=yuv420p,vflip
14711 Convert the input video to any of the formats not contained in the list:
14713 noformat=yuv420p|yuv444p|yuv410p
14719 Add noise on video input frame.
14721 The filter accepts the following options:
14729 Set noise seed for specific pixel component or all pixel components in case
14730 of @var{all_seed}. Default value is @code{123457}.
14732 @item all_strength, alls
14733 @item c0_strength, c0s
14734 @item c1_strength, c1s
14735 @item c2_strength, c2s
14736 @item c3_strength, c3s
14737 Set noise strength for specific pixel component or all pixel components in case
14738 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14740 @item all_flags, allf
14741 @item c0_flags, c0f
14742 @item c1_flags, c1f
14743 @item c2_flags, c2f
14744 @item c3_flags, c3f
14745 Set pixel component flags or set flags for all components if @var{all_flags}.
14746 Available values for component flags are:
14749 averaged temporal noise (smoother)
14751 mix random noise with a (semi)regular pattern
14753 temporal noise (noise pattern changes between frames)
14755 uniform noise (gaussian otherwise)
14759 @subsection Examples
14761 Add temporal and uniform noise to input video:
14763 noise=alls=20:allf=t+u
14768 Normalize RGB video (aka histogram stretching, contrast stretching).
14769 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14771 For each channel of each frame, the filter computes the input range and maps
14772 it linearly to the user-specified output range. The output range defaults
14773 to the full dynamic range from pure black to pure white.
14775 Temporal smoothing can be used on the input range to reduce flickering (rapid
14776 changes in brightness) caused when small dark or bright objects enter or leave
14777 the scene. This is similar to the auto-exposure (automatic gain control) on a
14778 video camera, and, like a video camera, it may cause a period of over- or
14779 under-exposure of the video.
14781 The R,G,B channels can be normalized independently, which may cause some
14782 color shifting, or linked together as a single channel, which prevents
14783 color shifting. Linked normalization preserves hue. Independent normalization
14784 does not, so it can be used to remove some color casts. Independent and linked
14785 normalization can be combined in any ratio.
14787 The normalize filter accepts the following options:
14792 Colors which define the output range. The minimum input value is mapped to
14793 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14794 The defaults are black and white respectively. Specifying white for
14795 @var{blackpt} and black for @var{whitept} will give color-inverted,
14796 normalized video. Shades of grey can be used to reduce the dynamic range
14797 (contrast). Specifying saturated colors here can create some interesting
14801 The number of previous frames to use for temporal smoothing. The input range
14802 of each channel is smoothed using a rolling average over the current frame
14803 and the @var{smoothing} previous frames. The default is 0 (no temporal
14807 Controls the ratio of independent (color shifting) channel normalization to
14808 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14809 independent. Defaults to 1.0 (fully independent).
14812 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14813 expensive no-op. Defaults to 1.0 (full strength).
14817 @subsection Commands
14818 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14819 The command accepts the same syntax of the corresponding option.
14821 If the specified expression is not valid, it is kept at its current
14824 @subsection Examples
14826 Stretch video contrast to use the full dynamic range, with no temporal
14827 smoothing; may flicker depending on the source content:
14829 normalize=blackpt=black:whitept=white:smoothing=0
14832 As above, but with 50 frames of temporal smoothing; flicker should be
14833 reduced, depending on the source content:
14835 normalize=blackpt=black:whitept=white:smoothing=50
14838 As above, but with hue-preserving linked channel normalization:
14840 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14843 As above, but with half strength:
14845 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14848 Map the darkest input color to red, the brightest input color to cyan:
14850 normalize=blackpt=red:whitept=cyan
14855 Pass the video source unchanged to the output.
14858 Optical Character Recognition
14860 This filter uses Tesseract for optical character recognition. To enable
14861 compilation of this filter, you need to configure FFmpeg with
14862 @code{--enable-libtesseract}.
14864 It accepts the following options:
14868 Set datapath to tesseract data. Default is to use whatever was
14869 set at installation.
14872 Set language, default is "eng".
14875 Set character whitelist.
14878 Set character blacklist.
14881 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14882 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14886 Apply a video transform using libopencv.
14888 To enable this filter, install the libopencv library and headers and
14889 configure FFmpeg with @code{--enable-libopencv}.
14891 It accepts the following parameters:
14896 The name of the libopencv filter to apply.
14898 @item filter_params
14899 The parameters to pass to the libopencv filter. If not specified, the default
14900 values are assumed.
14904 Refer to the official libopencv documentation for more precise
14906 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14908 Several libopencv filters are supported; see the following subsections.
14913 Dilate an image by using a specific structuring element.
14914 It corresponds to the libopencv function @code{cvDilate}.
14916 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14918 @var{struct_el} represents a structuring element, and has the syntax:
14919 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14921 @var{cols} and @var{rows} represent the number of columns and rows of
14922 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14923 point, and @var{shape} the shape for the structuring element. @var{shape}
14924 must be "rect", "cross", "ellipse", or "custom".
14926 If the value for @var{shape} is "custom", it must be followed by a
14927 string of the form "=@var{filename}". The file with name
14928 @var{filename} is assumed to represent a binary image, with each
14929 printable character corresponding to a bright pixel. When a custom
14930 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14931 or columns and rows of the read file are assumed instead.
14933 The default value for @var{struct_el} is "3x3+0x0/rect".
14935 @var{nb_iterations} specifies the number of times the transform is
14936 applied to the image, and defaults to 1.
14940 # Use the default values
14943 # Dilate using a structuring element with a 5x5 cross, iterating two times
14944 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14946 # Read the shape from the file diamond.shape, iterating two times.
14947 # The file diamond.shape may contain a pattern of characters like this
14953 # The specified columns and rows are ignored
14954 # but the anchor point coordinates are not
14955 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14960 Erode an image by using a specific structuring element.
14961 It corresponds to the libopencv function @code{cvErode}.
14963 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14964 with the same syntax and semantics as the @ref{dilate} filter.
14968 Smooth the input video.
14970 The filter takes the following parameters:
14971 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14973 @var{type} is the type of smooth filter to apply, and must be one of
14974 the following values: "blur", "blur_no_scale", "median", "gaussian",
14975 or "bilateral". The default value is "gaussian".
14977 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14978 depends on the smooth type. @var{param1} and
14979 @var{param2} accept integer positive values or 0. @var{param3} and
14980 @var{param4} accept floating point values.
14982 The default value for @var{param1} is 3. The default value for the
14983 other parameters is 0.
14985 These parameters correspond to the parameters assigned to the
14986 libopencv function @code{cvSmooth}.
14988 @section oscilloscope
14990 2D Video Oscilloscope.
14992 Useful to measure spatial impulse, step responses, chroma delays, etc.
14994 It accepts the following parameters:
14998 Set scope center x position.
15001 Set scope center y position.
15004 Set scope size, relative to frame diagonal.
15007 Set scope tilt/rotation.
15013 Set trace center x position.
15016 Set trace center y position.
15019 Set trace width, relative to width of frame.
15022 Set trace height, relative to height of frame.
15025 Set which components to trace. By default it traces first three components.
15028 Draw trace grid. By default is enabled.
15031 Draw some statistics. By default is enabled.
15034 Draw scope. By default is enabled.
15037 @subsection Commands
15038 This filter supports same @ref{commands} as options.
15039 The command accepts the same syntax of the corresponding option.
15041 If the specified expression is not valid, it is kept at its current
15044 @subsection Examples
15048 Inspect full first row of video frame.
15050 oscilloscope=x=0.5:y=0:s=1
15054 Inspect full last row of video frame.
15056 oscilloscope=x=0.5:y=1:s=1
15060 Inspect full 5th line of video frame of height 1080.
15062 oscilloscope=x=0.5:y=5/1080:s=1
15066 Inspect full last column of video frame.
15068 oscilloscope=x=1:y=0.5:s=1:t=1
15076 Overlay one video on top of another.
15078 It takes two inputs and has one output. The first input is the "main"
15079 video on which the second input is overlaid.
15081 It accepts the following parameters:
15083 A description of the accepted options follows.
15088 Set the expression for the x and y coordinates of the overlaid video
15089 on the main video. Default value is "0" for both expressions. In case
15090 the expression is invalid, it is set to a huge value (meaning that the
15091 overlay will not be displayed within the output visible area).
15094 See @ref{framesync}.
15097 Set when the expressions for @option{x}, and @option{y} are evaluated.
15099 It accepts the following values:
15102 only evaluate expressions once during the filter initialization or
15103 when a command is processed
15106 evaluate expressions for each incoming frame
15109 Default value is @samp{frame}.
15112 See @ref{framesync}.
15115 Set the format for the output video.
15117 It accepts the following values:
15120 force YUV420 output
15123 force YUV420p10 output
15126 force YUV422 output
15129 force YUV422p10 output
15132 force YUV444 output
15135 force packed RGB output
15138 force planar RGB output
15141 automatically pick format
15144 Default value is @samp{yuv420}.
15147 See @ref{framesync}.
15150 Set format of alpha of the overlaid video, it can be @var{straight} or
15151 @var{premultiplied}. Default is @var{straight}.
15154 The @option{x}, and @option{y} expressions can contain the following
15160 The main input width and height.
15164 The overlay input width and height.
15168 The computed values for @var{x} and @var{y}. They are evaluated for
15173 horizontal and vertical chroma subsample values of the output
15174 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
15178 the number of input frame, starting from 0
15181 the position in the file of the input frame, NAN if unknown
15184 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
15188 This filter also supports the @ref{framesync} options.
15190 Note that the @var{n}, @var{pos}, @var{t} variables are available only
15191 when evaluation is done @emph{per frame}, and will evaluate to NAN
15192 when @option{eval} is set to @samp{init}.
15194 Be aware that frames are taken from each input video in timestamp
15195 order, hence, if their initial timestamps differ, it is a good idea
15196 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
15197 have them begin in the same zero timestamp, as the example for
15198 the @var{movie} filter does.
15200 You can chain together more overlays but you should test the
15201 efficiency of such approach.
15203 @subsection Commands
15205 This filter supports the following commands:
15209 Modify the x and y of the overlay input.
15210 The command accepts the same syntax of the corresponding option.
15212 If the specified expression is not valid, it is kept at its current
15216 @subsection Examples
15220 Draw the overlay at 10 pixels from the bottom right corner of the main
15223 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15226 Using named options the example above becomes:
15228 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15232 Insert a transparent PNG logo in the bottom left corner of the input,
15233 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
15235 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15239 Insert 2 different transparent PNG logos (second logo on bottom
15240 right corner) using the @command{ffmpeg} tool:
15242 ffmpeg -i input -i logo1 -i logo2 -filter_complex 'overlay=x=10:y=H-h-10,overlay=x=W-w-10:y=H-h-10' output
15246 Add a transparent color layer on top of the main video; @code{WxH}
15247 must specify the size of the main input to the overlay filter:
15249 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
15253 Play an original video and a filtered version (here with the deshake
15254 filter) side by side using the @command{ffplay} tool:
15256 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15259 The above command is the same as:
15261 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15265 Make a sliding overlay appearing from the left to the right top part of the
15266 screen starting since time 2:
15268 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15272 Compose output by putting two input videos side to side:
15274 ffmpeg -i left.avi -i right.avi -filter_complex "
15275 nullsrc=size=200x100 [background];
15276 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15277 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15278 [background][left] overlay=shortest=1 [background+left];
15279 [background+left][right] overlay=shortest=1:x=100 [left+right]
15284 Mask 10-20 seconds of a video by applying the delogo filter to a section
15286 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15287 -vf '[in]split[split_main][split_delogo];[split_delogo]trim=start=360:end=371,delogo=0:0:640:480[delogoed];[split_main][delogoed]overlay=eof_action=pass[out]'
15292 Chain several overlays in cascade:
15294 nullsrc=s=200x200 [bg];
15295 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15296 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15297 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15298 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15299 [in3] null, [mid2] overlay=100:100 [out0]
15304 @anchor{overlay_cuda}
15305 @section overlay_cuda
15307 Overlay one video on top of another.
15309 This is the CUDA variant of the @ref{overlay} filter.
15310 It only accepts CUDA frames. The underlying input pixel formats have to match.
15312 It takes two inputs and has one output. The first input is the "main"
15313 video on which the second input is overlaid.
15315 It accepts the following parameters:
15320 Set the x and y coordinates of the overlaid video on the main video.
15321 Default value is "0" for both expressions.
15324 See @ref{framesync}.
15327 See @ref{framesync}.
15330 See @ref{framesync}.
15334 This filter also supports the @ref{framesync} options.
15338 Apply Overcomplete Wavelet denoiser.
15340 The filter accepts the following options:
15346 Larger depth values will denoise lower frequency components more, but
15347 slow down filtering.
15349 Must be an int in the range 8-16, default is @code{8}.
15351 @item luma_strength, ls
15354 Must be a double value in the range 0-1000, default is @code{1.0}.
15356 @item chroma_strength, cs
15357 Set chroma strength.
15359 Must be a double value in the range 0-1000, default is @code{1.0}.
15365 Add paddings to the input image, and place the original input at the
15366 provided @var{x}, @var{y} coordinates.
15368 It accepts the following parameters:
15373 Specify an expression for the size of the output image with the
15374 paddings added. If the value for @var{width} or @var{height} is 0, the
15375 corresponding input size is used for the output.
15377 The @var{width} expression can reference the value set by the
15378 @var{height} expression, and vice versa.
15380 The default value of @var{width} and @var{height} is 0.
15384 Specify the offsets to place the input image at within the padded area,
15385 with respect to the top/left border of the output image.
15387 The @var{x} expression can reference the value set by the @var{y}
15388 expression, and vice versa.
15390 The default value of @var{x} and @var{y} is 0.
15392 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15393 so the input image is centered on the padded area.
15396 Specify the color of the padded area. For the syntax of this option,
15397 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15398 manual,ffmpeg-utils}.
15400 The default value of @var{color} is "black".
15403 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15405 It accepts the following values:
15409 Only evaluate expressions once during the filter initialization or when
15410 a command is processed.
15413 Evaluate expressions for each incoming frame.
15417 Default value is @samp{init}.
15420 Pad to aspect instead to a resolution.
15424 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15425 options are expressions containing the following constants:
15430 The input video width and height.
15434 These are the same as @var{in_w} and @var{in_h}.
15438 The output width and height (the size of the padded area), as
15439 specified by the @var{width} and @var{height} expressions.
15443 These are the same as @var{out_w} and @var{out_h}.
15447 The x and y offsets as specified by the @var{x} and @var{y}
15448 expressions, or NAN if not yet specified.
15451 same as @var{iw} / @var{ih}
15454 input sample aspect ratio
15457 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15461 The horizontal and vertical chroma subsample values. For example for the
15462 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15465 @subsection Examples
15469 Add paddings with the color "violet" to the input video. The output video
15470 size is 640x480, and the top-left corner of the input video is placed at
15473 pad=640:480:0:40:violet
15476 The example above is equivalent to the following command:
15478 pad=width=640:height=480:x=0:y=40:color=violet
15482 Pad the input to get an output with dimensions increased by 3/2,
15483 and put the input video at the center of the padded area:
15485 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15489 Pad the input to get a squared output with size equal to the maximum
15490 value between the input width and height, and put the input video at
15491 the center of the padded area:
15493 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15497 Pad the input to get a final w/h ratio of 16:9:
15499 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15503 In case of anamorphic video, in order to set the output display aspect
15504 correctly, it is necessary to use @var{sar} in the expression,
15505 according to the relation:
15507 (ih * X / ih) * sar = output_dar
15508 X = output_dar / sar
15511 Thus the previous example needs to be modified to:
15513 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15517 Double the output size and put the input video in the bottom-right
15518 corner of the output padded area:
15520 pad="2*iw:2*ih:ow-iw:oh-ih"
15524 @anchor{palettegen}
15525 @section palettegen
15527 Generate one palette for a whole video stream.
15529 It accepts the following options:
15533 Set the maximum number of colors to quantize in the palette.
15534 Note: the palette will still contain 256 colors; the unused palette entries
15537 @item reserve_transparent
15538 Create a palette of 255 colors maximum and reserve the last one for
15539 transparency. Reserving the transparency color is useful for GIF optimization.
15540 If not set, the maximum of colors in the palette will be 256. You probably want
15541 to disable this option for a standalone image.
15544 @item transparency_color
15545 Set the color that will be used as background for transparency.
15548 Set statistics mode.
15550 It accepts the following values:
15553 Compute full frame histograms.
15555 Compute histograms only for the part that differs from previous frame. This
15556 might be relevant to give more importance to the moving part of your input if
15557 the background is static.
15559 Compute new histogram for each frame.
15562 Default value is @var{full}.
15565 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15566 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15567 color quantization of the palette. This information is also visible at
15568 @var{info} logging level.
15570 @subsection Examples
15574 Generate a representative palette of a given video using @command{ffmpeg}:
15576 ffmpeg -i input.mkv -vf palettegen palette.png
15580 @section paletteuse
15582 Use a palette to downsample an input video stream.
15584 The filter takes two inputs: one video stream and a palette. The palette must
15585 be a 256 pixels image.
15587 It accepts the following options:
15591 Select dithering mode. Available algorithms are:
15594 Ordered 8x8 bayer dithering (deterministic)
15596 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15597 Note: this dithering is sometimes considered "wrong" and is included as a
15599 @item floyd_steinberg
15600 Floyd and Steingberg dithering (error diffusion)
15602 Frankie Sierra dithering v2 (error diffusion)
15604 Frankie Sierra dithering v2 "Lite" (error diffusion)
15607 Default is @var{sierra2_4a}.
15610 When @var{bayer} dithering is selected, this option defines the scale of the
15611 pattern (how much the crosshatch pattern is visible). A low value means more
15612 visible pattern for less banding, and higher value means less visible pattern
15613 at the cost of more banding.
15615 The option must be an integer value in the range [0,5]. Default is @var{2}.
15618 If set, define the zone to process
15622 Only the changing rectangle will be reprocessed. This is similar to GIF
15623 cropping/offsetting compression mechanism. This option can be useful for speed
15624 if only a part of the image is changing, and has use cases such as limiting the
15625 scope of the error diffusal @option{dither} to the rectangle that bounds the
15626 moving scene (it leads to more deterministic output if the scene doesn't change
15627 much, and as a result less moving noise and better GIF compression).
15630 Default is @var{none}.
15633 Take new palette for each output frame.
15635 @item alpha_threshold
15636 Sets the alpha threshold for transparency. Alpha values above this threshold
15637 will be treated as completely opaque, and values below this threshold will be
15638 treated as completely transparent.
15640 The option must be an integer value in the range [0,255]. Default is @var{128}.
15643 @subsection Examples
15647 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15648 using @command{ffmpeg}:
15650 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15654 @section perspective
15656 Correct perspective of video not recorded perpendicular to the screen.
15658 A description of the accepted parameters follows.
15669 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15670 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15671 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15672 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15673 then the corners of the source will be sent to the specified coordinates.
15675 The expressions can use the following variables:
15680 the width and height of video frame.
15684 Output frame count.
15687 @item interpolation
15688 Set interpolation for perspective correction.
15690 It accepts the following values:
15696 Default value is @samp{linear}.
15699 Set interpretation of coordinate options.
15701 It accepts the following values:
15705 Send point in the source specified by the given coordinates to
15706 the corners of the destination.
15708 @item 1, destination
15710 Send the corners of the source to the point in the destination specified
15711 by the given coordinates.
15713 Default value is @samp{source}.
15717 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15719 It accepts the following values:
15722 only evaluate expressions once during the filter initialization or
15723 when a command is processed
15726 evaluate expressions for each incoming frame
15729 Default value is @samp{init}.
15734 Delay interlaced video by one field time so that the field order changes.
15736 The intended use is to fix PAL movies that have been captured with the
15737 opposite field order to the film-to-video transfer.
15739 A description of the accepted parameters follows.
15745 It accepts the following values:
15748 Capture field order top-first, transfer bottom-first.
15749 Filter will delay the bottom field.
15752 Capture field order bottom-first, transfer top-first.
15753 Filter will delay the top field.
15756 Capture and transfer with the same field order. This mode only exists
15757 for the documentation of the other options to refer to, but if you
15758 actually select it, the filter will faithfully do nothing.
15761 Capture field order determined automatically by field flags, transfer
15763 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15764 basis using field flags. If no field information is available,
15765 then this works just like @samp{u}.
15768 Capture unknown or varying, transfer opposite.
15769 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15770 analyzing the images and selecting the alternative that produces best
15771 match between the fields.
15774 Capture top-first, transfer unknown or varying.
15775 Filter selects among @samp{t} and @samp{p} using image analysis.
15778 Capture bottom-first, transfer unknown or varying.
15779 Filter selects among @samp{b} and @samp{p} using image analysis.
15782 Capture determined by field flags, transfer unknown or varying.
15783 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15784 image analysis. If no field information is available, then this works just
15785 like @samp{U}. This is the default mode.
15788 Both capture and transfer unknown or varying.
15789 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15793 @subsection Commands
15795 This filter supports the all above options as @ref{commands}.
15797 @section photosensitivity
15798 Reduce various flashes in video, so to help users with epilepsy.
15800 It accepts the following options:
15803 Set how many frames to use when filtering. Default is 30.
15806 Set detection threshold factor. Default is 1.
15810 Set how many pixels to skip when sampling frames. Default is 1.
15811 Allowed range is from 1 to 1024.
15814 Leave frames unchanged. Default is disabled.
15817 @section pixdesctest
15819 Pixel format descriptor test filter, mainly useful for internal
15820 testing. The output video should be equal to the input video.
15824 format=monow, pixdesctest
15827 can be used to test the monowhite pixel format descriptor definition.
15831 Display sample values of color channels. Mainly useful for checking color
15832 and levels. Minimum supported resolution is 640x480.
15834 The filters accept the following options:
15838 Set scope X position, relative offset on X axis.
15841 Set scope Y position, relative offset on Y axis.
15850 Set window opacity. This window also holds statistics about pixel area.
15853 Set window X position, relative offset on X axis.
15856 Set window Y position, relative offset on Y axis.
15861 Enable the specified chain of postprocessing subfilters using libpostproc. This
15862 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15863 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15864 Each subfilter and some options have a short and a long name that can be used
15865 interchangeably, i.e. dr/dering are the same.
15867 The filters accept the following options:
15871 Set postprocessing subfilters string.
15874 All subfilters share common options to determine their scope:
15878 Honor the quality commands for this subfilter.
15881 Do chrominance filtering, too (default).
15884 Do luminance filtering only (no chrominance).
15887 Do chrominance filtering only (no luminance).
15890 These options can be appended after the subfilter name, separated by a '|'.
15892 Available subfilters are:
15895 @item hb/hdeblock[|difference[|flatness]]
15896 Horizontal deblocking filter
15899 Difference factor where higher values mean more deblocking (default: @code{32}).
15901 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15904 @item vb/vdeblock[|difference[|flatness]]
15905 Vertical deblocking filter
15908 Difference factor where higher values mean more deblocking (default: @code{32}).
15910 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15913 @item ha/hadeblock[|difference[|flatness]]
15914 Accurate horizontal deblocking filter
15917 Difference factor where higher values mean more deblocking (default: @code{32}).
15919 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15922 @item va/vadeblock[|difference[|flatness]]
15923 Accurate vertical deblocking filter
15926 Difference factor where higher values mean more deblocking (default: @code{32}).
15928 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15932 The horizontal and vertical deblocking filters share the difference and
15933 flatness values so you cannot set different horizontal and vertical
15937 @item h1/x1hdeblock
15938 Experimental horizontal deblocking filter
15940 @item v1/x1vdeblock
15941 Experimental vertical deblocking filter
15946 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15949 larger -> stronger filtering
15951 larger -> stronger filtering
15953 larger -> stronger filtering
15956 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15959 Stretch luminance to @code{0-255}.
15962 @item lb/linblenddeint
15963 Linear blend deinterlacing filter that deinterlaces the given block by
15964 filtering all lines with a @code{(1 2 1)} filter.
15966 @item li/linipoldeint
15967 Linear interpolating deinterlacing filter that deinterlaces the given block by
15968 linearly interpolating every second line.
15970 @item ci/cubicipoldeint
15971 Cubic interpolating deinterlacing filter deinterlaces the given block by
15972 cubically interpolating every second line.
15974 @item md/mediandeint
15975 Median deinterlacing filter that deinterlaces the given block by applying a
15976 median filter to every second line.
15978 @item fd/ffmpegdeint
15979 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15980 second line with a @code{(-1 4 2 4 -1)} filter.
15983 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15984 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15986 @item fq/forceQuant[|quantizer]
15987 Overrides the quantizer table from the input with the constant quantizer you
15995 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15998 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
16001 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
16004 @subsection Examples
16008 Apply horizontal and vertical deblocking, deringing and automatic
16009 brightness/contrast:
16015 Apply default filters without brightness/contrast correction:
16021 Apply default filters and temporal denoiser:
16023 pp=default/tmpnoise|1|2|3
16027 Apply deblocking on luminance only, and switch vertical deblocking on or off
16028 automatically depending on available CPU time:
16035 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
16036 similar to spp = 6 with 7 point DCT, where only the center sample is
16039 The filter accepts the following options:
16043 Force a constant quantization parameter. It accepts an integer in range
16044 0 to 63. If not set, the filter will use the QP from the video stream
16048 Set thresholding mode. Available modes are:
16052 Set hard thresholding.
16054 Set soft thresholding (better de-ringing effect, but likely blurrier).
16056 Set medium thresholding (good results, default).
16060 @section premultiply
16061 Apply alpha premultiply effect to input video stream using first plane
16062 of second stream as alpha.
16064 Both streams must have same dimensions and same pixel format.
16066 The filter accepts the following option:
16070 Set which planes will be processed, unprocessed planes will be copied.
16071 By default value 0xf, all planes will be processed.
16074 Do not require 2nd input for processing, instead use alpha plane from input stream.
16078 Apply prewitt operator to input video stream.
16080 The filter accepts the following option:
16084 Set which planes will be processed, unprocessed planes will be copied.
16085 By default value 0xf, all planes will be processed.
16088 Set value which will be multiplied with filtered result.
16091 Set value which will be added to filtered result.
16094 @subsection Commands
16096 This filter supports the all above options as @ref{commands}.
16098 @section pseudocolor
16100 Alter frame colors in video with pseudocolors.
16102 This filter accepts the following options:
16106 set pixel first component expression
16109 set pixel second component expression
16112 set pixel third component expression
16115 set pixel fourth component expression, corresponds to the alpha component
16118 set component to use as base for altering colors
16121 Each of them specifies the expression to use for computing the lookup table for
16122 the corresponding pixel component values.
16124 The expressions can contain the following constants and functions:
16129 The input width and height.
16132 The input value for the pixel component.
16134 @item ymin, umin, vmin, amin
16135 The minimum allowed component value.
16137 @item ymax, umax, vmax, amax
16138 The maximum allowed component value.
16141 All expressions default to "val".
16143 @subsection Examples
16147 Change too high luma values to gradient:
16149 pseudocolor="'if(between(val,ymax,amax),lerp(ymin,ymax,(val-ymax)/(amax-ymax)),-1):if(between(val,ymax,amax),lerp(umax,umin,(val-ymax)/(amax-ymax)),-1):if(between(val,ymax,amax),lerp(vmin,vmax,(val-ymax)/(amax-ymax)),-1):-1'"
16155 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16156 Ratio) between two input videos.
16158 This filter takes in input two input videos, the first input is
16159 considered the "main" source and is passed unchanged to the
16160 output. The second input is used as a "reference" video for computing
16163 Both video inputs must have the same resolution and pixel format for
16164 this filter to work correctly. Also it assumes that both inputs
16165 have the same number of frames, which are compared one by one.
16167 The obtained average PSNR is printed through the logging system.
16169 The filter stores the accumulated MSE (mean squared error) of each
16170 frame, and at the end of the processing it is averaged across all frames
16171 equally, and the following formula is applied to obtain the PSNR:
16174 PSNR = 10*log10(MAX^2/MSE)
16177 Where MAX is the average of the maximum values of each component of the
16180 The description of the accepted parameters follows.
16183 @item stats_file, f
16184 If specified the filter will use the named file to save the PSNR of
16185 each individual frame. When filename equals "-" the data is sent to
16188 @item stats_version
16189 Specifies which version of the stats file format to use. Details of
16190 each format are written below.
16191 Default value is 1.
16193 @item stats_add_max
16194 Determines whether the max value is output to the stats log.
16195 Default value is 0.
16196 Requires stats_version >= 2. If this is set and stats_version < 2,
16197 the filter will return an error.
16200 This filter also supports the @ref{framesync} options.
16202 The file printed if @var{stats_file} is selected, contains a sequence of
16203 key/value pairs of the form @var{key}:@var{value} for each compared
16206 If a @var{stats_version} greater than 1 is specified, a header line precedes
16207 the list of per-frame-pair stats, with key value pairs following the frame
16208 format with the following parameters:
16211 @item psnr_log_version
16212 The version of the log file format. Will match @var{stats_version}.
16215 A comma separated list of the per-frame-pair parameters included in
16219 A description of each shown per-frame-pair parameter follows:
16223 sequential number of the input frame, starting from 1
16226 Mean Square Error pixel-by-pixel average difference of the compared
16227 frames, averaged over all the image components.
16229 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16230 Mean Square Error pixel-by-pixel average difference of the compared
16231 frames for the component specified by the suffix.
16233 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16234 Peak Signal to Noise ratio of the compared frames for the component
16235 specified by the suffix.
16237 @item max_avg, max_y, max_u, max_v
16238 Maximum allowed value for each channel, and average over all
16242 @subsection Examples
16247 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16248 [main][ref] psnr="stats_file=stats.log" [out]
16251 On this example the input file being processed is compared with the
16252 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
16253 is stored in @file{stats.log}.
16256 Another example with different containers:
16258 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]psnr" -f null -
16265 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16266 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
16269 The pullup filter is designed to take advantage of future context in making
16270 its decisions. This filter is stateless in the sense that it does not lock
16271 onto a pattern to follow, but it instead looks forward to the following
16272 fields in order to identify matches and rebuild progressive frames.
16274 To produce content with an even framerate, insert the fps filter after
16275 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
16276 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
16278 The filter accepts the following options:
16285 These options set the amount of "junk" to ignore at the left, right, top, and
16286 bottom of the image, respectively. Left and right are in units of 8 pixels,
16287 while top and bottom are in units of 2 lines.
16288 The default is 8 pixels on each side.
16291 Set the strict breaks. Setting this option to 1 will reduce the chances of
16292 filter generating an occasional mismatched frame, but it may also cause an
16293 excessive number of frames to be dropped during high motion sequences.
16294 Conversely, setting it to -1 will make filter match fields more easily.
16295 This may help processing of video where there is slight blurring between
16296 the fields, but may also cause there to be interlaced frames in the output.
16297 Default value is @code{0}.
16300 Set the metric plane to use. It accepts the following values:
16306 Use chroma blue plane.
16309 Use chroma red plane.
16312 This option may be set to use chroma plane instead of the default luma plane
16313 for doing filter's computations. This may improve accuracy on very clean
16314 source material, but more likely will decrease accuracy, especially if there
16315 is chroma noise (rainbow effect) or any grayscale video.
16316 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
16317 load and make pullup usable in realtime on slow machines.
16320 For best results (without duplicated frames in the output file) it is
16321 necessary to change the output frame rate. For example, to inverse
16322 telecine NTSC input:
16324 ffmpeg -i input -vf pullup -r 24000/1001 ...
16329 Change video quantization parameters (QP).
16331 The filter accepts the following option:
16335 Set expression for quantization parameter.
16338 The expression is evaluated through the eval API and can contain, among others,
16339 the following constants:
16343 1 if index is not 129, 0 otherwise.
16346 Sequential index starting from -129 to 128.
16349 @subsection Examples
16353 Some equation like:
16361 Flush video frames from internal cache of frames into a random order.
16362 No frame is discarded.
16363 Inspired by @ref{frei0r} nervous filter.
16367 Set size in number of frames of internal cache, in range from @code{2} to
16368 @code{512}. Default is @code{30}.
16371 Set seed for random number generator, must be an integer included between
16372 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
16373 less than @code{0}, the filter will try to use a good random seed on a
16377 @section readeia608
16379 Read closed captioning (EIA-608) information from the top lines of a video frame.
16381 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
16382 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
16383 with EIA-608 data (starting from 0). A description of each metadata value follows:
16386 @item lavfi.readeia608.X.cc
16387 The two bytes stored as EIA-608 data (printed in hexadecimal).
16389 @item lavfi.readeia608.X.line
16390 The number of the line on which the EIA-608 data was identified and read.
16393 This filter accepts the following options:
16397 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16400 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16403 Set the ratio of width reserved for sync code detection.
16404 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16407 Enable checking the parity bit. In the event of a parity error, the filter will output
16408 @code{0x00} for that character. Default is false.
16411 Lowpass lines prior to further processing. Default is enabled.
16414 @subsection Commands
16416 This filter supports the all above options as @ref{commands}.
16418 @subsection Examples
16422 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16424 ffprobe -f lavfi -i movie=captioned_video.mov,readeia608 -show_entries frame=pkt_pts_time:frame_tags=lavfi.readeia608.0.cc,lavfi.readeia608.1.cc -of csv
16430 Read vertical interval timecode (VITC) information from the top lines of a
16433 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16434 timecode value, if a valid timecode has been detected. Further metadata key
16435 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16436 timecode data has been found or not.
16438 This filter accepts the following options:
16442 Set the maximum number of lines to scan for VITC data. If the value is set to
16443 @code{-1} the full video frame is scanned. Default is @code{45}.
16446 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16447 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16450 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16451 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16454 @subsection Examples
16458 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16459 draw @code{--:--:--:--} as a placeholder:
16461 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16467 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16469 Destination pixel at position (X, Y) will be picked from source (x, y) position
16470 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16471 value for pixel will be used for destination pixel.
16473 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16474 will have Xmap/Ymap video stream dimensions.
16475 Xmap and Ymap input video streams are 16bit depth, single channel.
16479 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16480 Default is @code{color}.
16483 Specify the color of the unmapped pixels. For the syntax of this option,
16484 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16485 manual,ffmpeg-utils}. Default color is @code{black}.
16488 @section removegrain
16490 The removegrain filter is a spatial denoiser for progressive video.
16494 Set mode for the first plane.
16497 Set mode for the second plane.
16500 Set mode for the third plane.
16503 Set mode for the fourth plane.
16506 Range of mode is from 0 to 24. Description of each mode follows:
16510 Leave input plane unchanged. Default.
16513 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16516 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16519 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16522 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16523 This is equivalent to a median filter.
16526 Line-sensitive clipping giving the minimal change.
16529 Line-sensitive clipping, intermediate.
16532 Line-sensitive clipping, intermediate.
16535 Line-sensitive clipping, intermediate.
16538 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16541 Replaces the target pixel with the closest neighbour.
16544 [1 2 1] horizontal and vertical kernel blur.
16550 Bob mode, interpolates top field from the line where the neighbours
16551 pixels are the closest.
16554 Bob mode, interpolates bottom field from the line where the neighbours
16555 pixels are the closest.
16558 Bob mode, interpolates top field. Same as 13 but with a more complicated
16559 interpolation formula.
16562 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16563 interpolation formula.
16566 Clips the pixel with the minimum and maximum of respectively the maximum and
16567 minimum of each pair of opposite neighbour pixels.
16570 Line-sensitive clipping using opposite neighbours whose greatest distance from
16571 the current pixel is minimal.
16574 Replaces the pixel with the average of its 8 neighbours.
16577 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16580 Clips pixels using the averages of opposite neighbour.
16583 Same as mode 21 but simpler and faster.
16586 Small edge and halo removal, but reputed useless.
16592 @section removelogo
16594 Suppress a TV station logo, using an image file to determine which
16595 pixels comprise the logo. It works by filling in the pixels that
16596 comprise the logo with neighboring pixels.
16598 The filter accepts the following options:
16602 Set the filter bitmap file, which can be any image format supported by
16603 libavformat. The width and height of the image file must match those of the
16604 video stream being processed.
16607 Pixels in the provided bitmap image with a value of zero are not
16608 considered part of the logo, non-zero pixels are considered part of
16609 the logo. If you use white (255) for the logo and black (0) for the
16610 rest, you will be safe. For making the filter bitmap, it is
16611 recommended to take a screen capture of a black frame with the logo
16612 visible, and then using a threshold filter followed by the erode
16613 filter once or twice.
16615 If needed, little splotches can be fixed manually. Remember that if
16616 logo pixels are not covered, the filter quality will be much
16617 reduced. Marking too many pixels as part of the logo does not hurt as
16618 much, but it will increase the amount of blurring needed to cover over
16619 the image and will destroy more information than necessary, and extra
16620 pixels will slow things down on a large logo.
16622 @section repeatfields
16624 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16625 fields based on its value.
16629 Reverse a video clip.
16631 Warning: This filter requires memory to buffer the entire clip, so trimming
16634 @subsection Examples
16638 Take the first 5 seconds of a clip, and reverse it.
16645 Shift R/G/B/A pixels horizontally and/or vertically.
16647 The filter accepts the following options:
16650 Set amount to shift red horizontally.
16652 Set amount to shift red vertically.
16654 Set amount to shift green horizontally.
16656 Set amount to shift green vertically.
16658 Set amount to shift blue horizontally.
16660 Set amount to shift blue vertically.
16662 Set amount to shift alpha horizontally.
16664 Set amount to shift alpha vertically.
16666 Set edge mode, can be @var{smear}, default, or @var{warp}.
16669 @subsection Commands
16671 This filter supports the all above options as @ref{commands}.
16674 Apply roberts cross operator to input video stream.
16676 The filter accepts the following option:
16680 Set which planes will be processed, unprocessed planes will be copied.
16681 By default value 0xf, all planes will be processed.
16684 Set value which will be multiplied with filtered result.
16687 Set value which will be added to filtered result.
16690 @subsection Commands
16692 This filter supports the all above options as @ref{commands}.
16696 Rotate video by an arbitrary angle expressed in radians.
16698 The filter accepts the following options:
16700 A description of the optional parameters follows.
16703 Set an expression for the angle by which to rotate the input video
16704 clockwise, expressed as a number of radians. A negative value will
16705 result in a counter-clockwise rotation. By default it is set to "0".
16707 This expression is evaluated for each frame.
16710 Set the output width expression, default value is "iw".
16711 This expression is evaluated just once during configuration.
16714 Set the output height expression, default value is "ih".
16715 This expression is evaluated just once during configuration.
16718 Enable bilinear interpolation if set to 1, a value of 0 disables
16719 it. Default value is 1.
16722 Set the color used to fill the output area not covered by the rotated
16723 image. For the general syntax of this option, check the
16724 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16725 If the special value "none" is selected then no
16726 background is printed (useful for example if the background is never shown).
16728 Default value is "black".
16731 The expressions for the angle and the output size can contain the
16732 following constants and functions:
16736 sequential number of the input frame, starting from 0. It is always NAN
16737 before the first frame is filtered.
16740 time in seconds of the input frame, it is set to 0 when the filter is
16741 configured. It is always NAN before the first frame is filtered.
16745 horizontal and vertical chroma subsample values. For example for the
16746 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16750 the input video width and height
16754 the output width and height, that is the size of the padded area as
16755 specified by the @var{width} and @var{height} expressions
16759 the minimal width/height required for completely containing the input
16760 video rotated by @var{a} radians.
16762 These are only available when computing the @option{out_w} and
16763 @option{out_h} expressions.
16766 @subsection Examples
16770 Rotate the input by PI/6 radians clockwise:
16776 Rotate the input by PI/6 radians counter-clockwise:
16782 Rotate the input by 45 degrees clockwise:
16788 Apply a constant rotation with period T, starting from an angle of PI/3:
16790 rotate=PI/3+2*PI*t/T
16794 Make the input video rotation oscillating with a period of T
16795 seconds and an amplitude of A radians:
16797 rotate=A*sin(2*PI/T*t)
16801 Rotate the video, output size is chosen so that the whole rotating
16802 input video is always completely contained in the output:
16804 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16808 Rotate the video, reduce the output size so that no background is ever
16811 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16815 @subsection Commands
16817 The filter supports the following commands:
16821 Set the angle expression.
16822 The command accepts the same syntax of the corresponding option.
16824 If the specified expression is not valid, it is kept at its current
16830 Apply Shape Adaptive Blur.
16832 The filter accepts the following options:
16835 @item luma_radius, lr
16836 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16837 value is 1.0. A greater value will result in a more blurred image, and
16838 in slower processing.
16840 @item luma_pre_filter_radius, lpfr
16841 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16844 @item luma_strength, ls
16845 Set luma maximum difference between pixels to still be considered, must
16846 be a value in the 0.1-100.0 range, default value is 1.0.
16848 @item chroma_radius, cr
16849 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16850 greater value will result in a more blurred image, and in slower
16853 @item chroma_pre_filter_radius, cpfr
16854 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16856 @item chroma_strength, cs
16857 Set chroma maximum difference between pixels to still be considered,
16858 must be a value in the -0.9-100.0 range.
16861 Each chroma option value, if not explicitly specified, is set to the
16862 corresponding luma option value.
16867 Scale (resize) the input video, using the libswscale library.
16869 The scale filter forces the output display aspect ratio to be the same
16870 of the input, by changing the output sample aspect ratio.
16872 If the input image format is different from the format requested by
16873 the next filter, the scale filter will convert the input to the
16876 @subsection Options
16877 The filter accepts the following options, or any of the options
16878 supported by the libswscale scaler.
16880 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16881 the complete list of scaler options.
16886 Set the output video dimension expression. Default value is the input
16889 If the @var{width} or @var{w} value is 0, the input width is used for
16890 the output. If the @var{height} or @var{h} value is 0, the input height
16891 is used for the output.
16893 If one and only one of the values is -n with n >= 1, the scale filter
16894 will use a value that maintains the aspect ratio of the input image,
16895 calculated from the other specified dimension. After that it will,
16896 however, make sure that the calculated dimension is divisible by n and
16897 adjust the value if necessary.
16899 If both values are -n with n >= 1, the behavior will be identical to
16900 both values being set to 0 as previously detailed.
16902 See below for the list of accepted constants for use in the dimension
16906 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16910 Only evaluate expressions once during the filter initialization or when a command is processed.
16913 Evaluate expressions for each incoming frame.
16917 Default value is @samp{init}.
16921 Set the interlacing mode. It accepts the following values:
16925 Force interlaced aware scaling.
16928 Do not apply interlaced scaling.
16931 Select interlaced aware scaling depending on whether the source frames
16932 are flagged as interlaced or not.
16935 Default value is @samp{0}.
16938 Set libswscale scaling flags. See
16939 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16940 complete list of values. If not explicitly specified the filter applies
16944 @item param0, param1
16945 Set libswscale input parameters for scaling algorithms that need them. See
16946 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16947 complete documentation. If not explicitly specified the filter applies
16953 Set the video size. For the syntax of this option, check the
16954 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16956 @item in_color_matrix
16957 @item out_color_matrix
16958 Set in/output YCbCr color space type.
16960 This allows the autodetected value to be overridden as well as allows forcing
16961 a specific value used for the output and encoder.
16963 If not specified, the color space type depends on the pixel format.
16969 Choose automatically.
16972 Format conforming to International Telecommunication Union (ITU)
16973 Recommendation BT.709.
16976 Set color space conforming to the United States Federal Communications
16977 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16982 Set color space conforming to:
16986 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16989 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16992 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16997 Set color space conforming to SMPTE ST 240:1999.
17000 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
17005 Set in/output YCbCr sample range.
17007 This allows the autodetected value to be overridden as well as allows forcing
17008 a specific value used for the output and encoder. If not specified, the
17009 range depends on the pixel format. Possible values:
17013 Choose automatically.
17016 Set full range (0-255 in case of 8-bit luma).
17018 @item mpeg/limited/tv
17019 Set "MPEG" range (16-235 in case of 8-bit luma).
17022 @item force_original_aspect_ratio
17023 Enable decreasing or increasing output video width or height if necessary to
17024 keep the original aspect ratio. Possible values:
17028 Scale the video as specified and disable this feature.
17031 The output video dimensions will automatically be decreased if needed.
17034 The output video dimensions will automatically be increased if needed.
17038 One useful instance of this option is that when you know a specific device's
17039 maximum allowed resolution, you can use this to limit the output video to
17040 that, while retaining the aspect ratio. For example, device A allows
17041 1280x720 playback, and your video is 1920x800. Using this option (set it to
17042 decrease) and specifying 1280x720 to the command line makes the output
17045 Please note that this is a different thing than specifying -1 for @option{w}
17046 or @option{h}, you still need to specify the output resolution for this option
17049 @item force_divisible_by
17050 Ensures that both the output dimensions, width and height, are divisible by the
17051 given integer when used together with @option{force_original_aspect_ratio}. This
17052 works similar to using @code{-n} in the @option{w} and @option{h} options.
17054 This option respects the value set for @option{force_original_aspect_ratio},
17055 increasing or decreasing the resolution accordingly. The video's aspect ratio
17056 may be slightly modified.
17058 This option can be handy if you need to have a video fit within or exceed
17059 a defined resolution using @option{force_original_aspect_ratio} but also have
17060 encoder restrictions on width or height divisibility.
17064 The values of the @option{w} and @option{h} options are expressions
17065 containing the following constants:
17070 The input width and height
17074 These are the same as @var{in_w} and @var{in_h}.
17078 The output (scaled) width and height
17082 These are the same as @var{out_w} and @var{out_h}
17085 The same as @var{iw} / @var{ih}
17088 input sample aspect ratio
17091 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
17095 horizontal and vertical input chroma subsample values. For example for the
17096 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17100 horizontal and vertical output chroma subsample values. For example for the
17101 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17104 The (sequential) number of the input frame, starting from 0.
17105 Only available with @code{eval=frame}.
17108 The presentation timestamp of the input frame, expressed as a number of
17109 seconds. Only available with @code{eval=frame}.
17112 The position (byte offset) of the frame in the input stream, or NaN if
17113 this information is unavailable and/or meaningless (for example in case of synthetic video).
17114 Only available with @code{eval=frame}.
17117 @subsection Examples
17121 Scale the input video to a size of 200x100
17126 This is equivalent to:
17137 Specify a size abbreviation for the output size:
17142 which can also be written as:
17148 Scale the input to 2x:
17150 scale=w=2*iw:h=2*ih
17154 The above is the same as:
17156 scale=2*in_w:2*in_h
17160 Scale the input to 2x with forced interlaced scaling:
17162 scale=2*iw:2*ih:interl=1
17166 Scale the input to half size:
17168 scale=w=iw/2:h=ih/2
17172 Increase the width, and set the height to the same size:
17178 Seek Greek harmony:
17185 Increase the height, and set the width to 3/2 of the height:
17187 scale=w=3/2*oh:h=3/5*ih
17191 Increase the size, making the size a multiple of the chroma
17194 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17198 Increase the width to a maximum of 500 pixels,
17199 keeping the same aspect ratio as the input:
17201 scale=w='min(500\, iw*3/2):h=-1'
17205 Make pixels square by combining scale and setsar:
17207 scale='trunc(ih*dar):ih',setsar=1/1
17211 Make pixels square by combining scale and setsar,
17212 making sure the resulting resolution is even (required by some codecs):
17214 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17218 @subsection Commands
17220 This filter supports the following commands:
17224 Set the output video dimension expression.
17225 The command accepts the same syntax of the corresponding option.
17227 If the specified expression is not valid, it is kept at its current
17233 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
17234 format conversion on CUDA video frames. Setting the output width and height
17235 works in the same way as for the @var{scale} filter.
17237 The following additional options are accepted:
17240 The pixel format of the output CUDA frames. If set to the string "same" (the
17241 default), the input format will be kept. Note that automatic format negotiation
17242 and conversion is not yet supported for hardware frames
17245 The interpolation algorithm used for resizing. One of the following:
17252 @item cubic2p_bspline
17253 2-parameter cubic (B=1, C=0)
17255 @item cubic2p_catmullrom
17256 2-parameter cubic (B=0, C=1/2)
17258 @item cubic2p_b05c03
17259 2-parameter cubic (B=1/2, C=3/10)
17267 @item force_original_aspect_ratio
17268 Enable decreasing or increasing output video width or height if necessary to
17269 keep the original aspect ratio. Possible values:
17273 Scale the video as specified and disable this feature.
17276 The output video dimensions will automatically be decreased if needed.
17279 The output video dimensions will automatically be increased if needed.
17283 One useful instance of this option is that when you know a specific device's
17284 maximum allowed resolution, you can use this to limit the output video to
17285 that, while retaining the aspect ratio. For example, device A allows
17286 1280x720 playback, and your video is 1920x800. Using this option (set it to
17287 decrease) and specifying 1280x720 to the command line makes the output
17290 Please note that this is a different thing than specifying -1 for @option{w}
17291 or @option{h}, you still need to specify the output resolution for this option
17294 @item force_divisible_by
17295 Ensures that both the output dimensions, width and height, are divisible by the
17296 given integer when used together with @option{force_original_aspect_ratio}. This
17297 works similar to using @code{-n} in the @option{w} and @option{h} options.
17299 This option respects the value set for @option{force_original_aspect_ratio},
17300 increasing or decreasing the resolution accordingly. The video's aspect ratio
17301 may be slightly modified.
17303 This option can be handy if you need to have a video fit within or exceed
17304 a defined resolution using @option{force_original_aspect_ratio} but also have
17305 encoder restrictions on width or height divisibility.
17311 Scale (resize) the input video, based on a reference video.
17313 See the scale filter for available options, scale2ref supports the same but
17314 uses the reference video instead of the main input as basis. scale2ref also
17315 supports the following additional constants for the @option{w} and
17316 @option{h} options:
17321 The main input video's width and height
17324 The same as @var{main_w} / @var{main_h}
17327 The main input video's sample aspect ratio
17329 @item main_dar, mdar
17330 The main input video's display aspect ratio. Calculated from
17331 @code{(main_w / main_h) * main_sar}.
17335 The main input video's horizontal and vertical chroma subsample values.
17336 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
17340 The (sequential) number of the main input frame, starting from 0.
17341 Only available with @code{eval=frame}.
17344 The presentation timestamp of the main input frame, expressed as a number of
17345 seconds. Only available with @code{eval=frame}.
17348 The position (byte offset) of the frame in the main input stream, or NaN if
17349 this information is unavailable and/or meaningless (for example in case of synthetic video).
17350 Only available with @code{eval=frame}.
17353 @subsection Examples
17357 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
17359 'scale2ref[b][a];[a][b]overlay'
17363 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
17365 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17369 @subsection Commands
17371 This filter supports the following commands:
17375 Set the output video dimension expression.
17376 The command accepts the same syntax of the corresponding option.
17378 If the specified expression is not valid, it is kept at its current
17383 Scroll input video horizontally and/or vertically by constant speed.
17385 The filter accepts the following options:
17387 @item horizontal, h
17388 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
17389 Negative values changes scrolling direction.
17392 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
17393 Negative values changes scrolling direction.
17396 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
17399 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
17402 @subsection Commands
17404 This filter supports the following @ref{commands}:
17406 @item horizontal, h
17407 Set the horizontal scrolling speed.
17409 Set the vertical scrolling speed.
17415 Detect video scene change.
17417 This filter sets frame metadata with mafd between frame, the scene score, and
17418 forward the frame to the next filter, so they can use these metadata to detect
17419 scene change or others.
17421 In addition, this filter logs a message and sets frame metadata when it detects
17422 a scene change by @option{threshold}.
17424 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17426 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17427 to detect scene change.
17429 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17430 detect scene change with @option{threshold}.
17432 The filter accepts the following options:
17436 Set the scene change detection threshold as a percentage of maximum change. Good
17437 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17440 Default value is @code{10.}.
17443 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17444 You can enable it if you want to get snapshot of scene change frames only.
17447 @anchor{selectivecolor}
17448 @section selectivecolor
17450 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17451 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17452 by the "purity" of the color (that is, how saturated it already is).
17454 This filter is similar to the Adobe Photoshop Selective Color tool.
17456 The filter accepts the following options:
17459 @item correction_method
17460 Select color correction method.
17462 Available values are:
17465 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17468 Specified adjustments are relative to the original component value.
17470 Default is @code{absolute}.
17472 Adjustments for red pixels (pixels where the red component is the maximum)
17474 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17476 Adjustments for green pixels (pixels where the green component is the maximum)
17478 Adjustments for cyan pixels (pixels where the red component is the minimum)
17480 Adjustments for blue pixels (pixels where the blue component is the maximum)
17482 Adjustments for magenta pixels (pixels where the green component is the minimum)
17484 Adjustments for white pixels (pixels where all components are greater than 128)
17486 Adjustments for all pixels except pure black and pure white
17488 Adjustments for black pixels (pixels where all components are lesser than 128)
17490 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17493 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17494 4 space separated floating point adjustment values in the [-1,1] range,
17495 respectively to adjust the amount of cyan, magenta, yellow and black for the
17496 pixels of its range.
17498 @subsection Examples
17502 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17503 increase magenta by 27% in blue areas:
17505 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17509 Use a Photoshop selective color preset:
17511 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17515 @anchor{separatefields}
17516 @section separatefields
17518 The @code{separatefields} takes a frame-based video input and splits
17519 each frame into its components fields, producing a new half height clip
17520 with twice the frame rate and twice the frame count.
17522 This filter use field-dominance information in frame to decide which
17523 of each pair of fields to place first in the output.
17524 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17526 @section setdar, setsar
17528 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17531 This is done by changing the specified Sample (aka Pixel) Aspect
17532 Ratio, according to the following equation:
17534 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17537 Keep in mind that the @code{setdar} filter does not modify the pixel
17538 dimensions of the video frame. Also, the display aspect ratio set by
17539 this filter may be changed by later filters in the filterchain,
17540 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17543 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17544 the filter output video.
17546 Note that as a consequence of the application of this filter, the
17547 output display aspect ratio will change according to the equation
17550 Keep in mind that the sample aspect ratio set by the @code{setsar}
17551 filter may be changed by later filters in the filterchain, e.g. if
17552 another "setsar" or a "setdar" filter is applied.
17554 It accepts the following parameters:
17557 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17558 Set the aspect ratio used by the filter.
17560 The parameter can be a floating point number string, an expression, or
17561 a string of the form @var{num}:@var{den}, where @var{num} and
17562 @var{den} are the numerator and denominator of the aspect ratio. If
17563 the parameter is not specified, it is assumed the value "0".
17564 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17568 Set the maximum integer value to use for expressing numerator and
17569 denominator when reducing the expressed aspect ratio to a rational.
17570 Default value is @code{100}.
17574 The parameter @var{sar} is an expression containing
17575 the following constants:
17579 These are approximated values for the mathematical constants e
17580 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17583 The input width and height.
17586 These are the same as @var{w} / @var{h}.
17589 The input sample aspect ratio.
17592 The input display aspect ratio. It is the same as
17593 (@var{w} / @var{h}) * @var{sar}.
17596 Horizontal and vertical chroma subsample values. For example, for the
17597 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17600 @subsection Examples
17605 To change the display aspect ratio to 16:9, specify one of the following:
17612 To change the sample aspect ratio to 10:11, specify:
17618 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17619 1000 in the aspect ratio reduction, use the command:
17621 setdar=ratio=16/9:max=1000
17629 Force field for the output video frame.
17631 The @code{setfield} filter marks the interlace type field for the
17632 output frames. It does not change the input frame, but only sets the
17633 corresponding property, which affects how the frame is treated by
17634 following filters (e.g. @code{fieldorder} or @code{yadif}).
17636 The filter accepts the following options:
17641 Available values are:
17645 Keep the same field property.
17648 Mark the frame as bottom-field-first.
17651 Mark the frame as top-field-first.
17654 Mark the frame as progressive.
17661 Force frame parameter for the output video frame.
17663 The @code{setparams} filter marks interlace and color range for the
17664 output frames. It does not change the input frame, but only sets the
17665 corresponding property, which affects how the frame is treated by
17670 Available values are:
17674 Keep the same field property (default).
17677 Mark the frame as bottom-field-first.
17680 Mark the frame as top-field-first.
17683 Mark the frame as progressive.
17687 Available values are:
17691 Keep the same color range property (default).
17693 @item unspecified, unknown
17694 Mark the frame as unspecified color range.
17696 @item limited, tv, mpeg
17697 Mark the frame as limited range.
17699 @item full, pc, jpeg
17700 Mark the frame as full range.
17703 @item color_primaries
17704 Set the color primaries.
17705 Available values are:
17709 Keep the same color primaries property (default).
17726 Set the color transfer.
17727 Available values are:
17731 Keep the same color trc property (default).
17753 Set the colorspace.
17754 Available values are:
17758 Keep the same colorspace property (default).
17771 @item chroma-derived-nc
17772 @item chroma-derived-c
17779 Show a line containing various information for each input video frame.
17780 The input video is not modified.
17782 This filter supports the following options:
17786 Calculate checksums of each plane. By default enabled.
17789 The shown line contains a sequence of key/value pairs of the form
17790 @var{key}:@var{value}.
17792 The following values are shown in the output:
17796 The (sequential) number of the input frame, starting from 0.
17799 The Presentation TimeStamp of the input frame, expressed as a number of
17800 time base units. The time base unit depends on the filter input pad.
17803 The Presentation TimeStamp of the input frame, expressed as a number of
17807 The position of the frame in the input stream, or -1 if this information is
17808 unavailable and/or meaningless (for example in case of synthetic video).
17811 The pixel format name.
17814 The sample aspect ratio of the input frame, expressed in the form
17815 @var{num}/@var{den}.
17818 The size of the input frame. For the syntax of this option, check the
17819 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17822 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17823 for bottom field first).
17826 This is 1 if the frame is a key frame, 0 otherwise.
17829 The picture type of the input frame ("I" for an I-frame, "P" for a
17830 P-frame, "B" for a B-frame, or "?" for an unknown type).
17831 Also refer to the documentation of the @code{AVPictureType} enum and of
17832 the @code{av_get_picture_type_char} function defined in
17833 @file{libavutil/avutil.h}.
17836 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17838 @item plane_checksum
17839 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17840 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17843 The mean value of pixels in each plane of the input frame, expressed in the form
17844 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17847 The standard deviation of pixel values in each plane of the input frame, expressed
17848 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17852 @section showpalette
17854 Displays the 256 colors palette of each frame. This filter is only relevant for
17855 @var{pal8} pixel format frames.
17857 It accepts the following option:
17861 Set the size of the box used to represent one palette color entry. Default is
17862 @code{30} (for a @code{30x30} pixel box).
17865 @section shuffleframes
17867 Reorder and/or duplicate and/or drop video frames.
17869 It accepts the following parameters:
17873 Set the destination indexes of input frames.
17874 This is space or '|' separated list of indexes that maps input frames to output
17875 frames. Number of indexes also sets maximal value that each index may have.
17876 '-1' index have special meaning and that is to drop frame.
17879 The first frame has the index 0. The default is to keep the input unchanged.
17881 @subsection Examples
17885 Swap second and third frame of every three frames of the input:
17887 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17891 Swap 10th and 1st frame of every ten frames of the input:
17893 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17897 @section shufflepixels
17899 Reorder pixels in video frames.
17901 This filter accepts the following options:
17905 Set shuffle direction. Can be forward or inverse direction.
17906 Default direction is forward.
17909 Set shuffle mode. Can be horizontal, vertical or block mode.
17913 Set shuffle block_size. In case of horizontal shuffle mode only width
17914 part of size is used, and in case of vertical shuffle mode only height
17915 part of size is used.
17918 Set random seed used with shuffling pixels. Mainly useful to set to be able
17919 to reverse filtering process to get original input.
17920 For example, to reverse forward shuffle you need to use same parameters
17921 and exact same seed and to set direction to inverse.
17924 @section shuffleplanes
17926 Reorder and/or duplicate video planes.
17928 It accepts the following parameters:
17933 The index of the input plane to be used as the first output plane.
17936 The index of the input plane to be used as the second output plane.
17939 The index of the input plane to be used as the third output plane.
17942 The index of the input plane to be used as the fourth output plane.
17946 The first plane has the index 0. The default is to keep the input unchanged.
17948 @subsection Examples
17952 Swap the second and third planes of the input:
17954 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17958 @anchor{signalstats}
17959 @section signalstats
17960 Evaluate various visual metrics that assist in determining issues associated
17961 with the digitization of analog video media.
17963 By default the filter will log these metadata values:
17967 Display the minimal Y value contained within the input frame. Expressed in
17971 Display the Y value at the 10% percentile within the input frame. Expressed in
17975 Display the average Y value within the input frame. Expressed in range of
17979 Display the Y value at the 90% percentile within the input frame. Expressed in
17983 Display the maximum Y value contained within the input frame. Expressed in
17987 Display the minimal U value contained within the input frame. Expressed in
17991 Display the U value at the 10% percentile within the input frame. Expressed in
17995 Display the average U value within the input frame. Expressed in range of
17999 Display the U value at the 90% percentile within the input frame. Expressed in
18003 Display the maximum U value contained within the input frame. Expressed in
18007 Display the minimal V value contained within the input frame. Expressed in
18011 Display the V value at the 10% percentile within the input frame. Expressed in
18015 Display the average V value within the input frame. Expressed in range of
18019 Display the V value at the 90% percentile within the input frame. Expressed in
18023 Display the maximum V value contained within the input frame. Expressed in
18027 Display the minimal saturation value contained within the input frame.
18028 Expressed in range of [0-~181.02].
18031 Display the saturation value at the 10% percentile within the input frame.
18032 Expressed in range of [0-~181.02].
18035 Display the average saturation value within the input frame. Expressed in range
18039 Display the saturation value at the 90% percentile within the input frame.
18040 Expressed in range of [0-~181.02].
18043 Display the maximum saturation value contained within the input frame.
18044 Expressed in range of [0-~181.02].
18047 Display the median value for hue within the input frame. Expressed in range of
18051 Display the average value for hue within the input frame. Expressed in range of
18055 Display the average of sample value difference between all values of the Y
18056 plane in the current frame and corresponding values of the previous input frame.
18057 Expressed in range of [0-255].
18060 Display the average of sample value difference between all values of the U
18061 plane in the current frame and corresponding values of the previous input frame.
18062 Expressed in range of [0-255].
18065 Display the average of sample value difference between all values of the V
18066 plane in the current frame and corresponding values of the previous input frame.
18067 Expressed in range of [0-255].
18070 Display bit depth of Y plane in current frame.
18071 Expressed in range of [0-16].
18074 Display bit depth of U plane in current frame.
18075 Expressed in range of [0-16].
18078 Display bit depth of V plane in current frame.
18079 Expressed in range of [0-16].
18082 The filter accepts the following options:
18088 @option{stat} specify an additional form of image analysis.
18089 @option{out} output video with the specified type of pixel highlighted.
18091 Both options accept the following values:
18095 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
18096 unlike the neighboring pixels of the same field. Examples of temporal outliers
18097 include the results of video dropouts, head clogs, or tape tracking issues.
18100 Identify @var{vertical line repetition}. Vertical line repetition includes
18101 similar rows of pixels within a frame. In born-digital video vertical line
18102 repetition is common, but this pattern is uncommon in video digitized from an
18103 analog source. When it occurs in video that results from the digitization of an
18104 analog source it can indicate concealment from a dropout compensator.
18107 Identify pixels that fall outside of legal broadcast range.
18111 Set the highlight color for the @option{out} option. The default color is
18115 @subsection Examples
18119 Output data of various video metrics:
18121 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
18125 Output specific data about the minimum and maximum values of the Y plane per frame:
18127 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
18131 Playback video while highlighting pixels that are outside of broadcast range in red.
18133 ffplay example.mov -vf signalstats="out=brng:color=red"
18137 Playback video with signalstats metadata drawn over the frame.
18139 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18142 The contents of signalstat_drawtext.txt used in the command are:
18145 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
18146 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
18147 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
18148 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
18156 Calculates the MPEG-7 Video Signature. The filter can handle more than one
18157 input. In this case the matching between the inputs can be calculated additionally.
18158 The filter always passes through the first input. The signature of each stream can
18159 be written into a file.
18161 It accepts the following options:
18165 Enable or disable the matching process.
18167 Available values are:
18171 Disable the calculation of a matching (default).
18173 Calculate the matching for the whole video and output whether the whole video
18174 matches or only parts.
18176 Calculate only until a matching is found or the video ends. Should be faster in
18181 Set the number of inputs. The option value must be a non negative integer.
18182 Default value is 1.
18185 Set the path to which the output is written. If there is more than one input,
18186 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
18187 integer), that will be replaced with the input number. If no filename is
18188 specified, no output will be written. This is the default.
18191 Choose the output format.
18193 Available values are:
18197 Use the specified binary representation (default).
18199 Use the specified xml representation.
18203 Set threshold to detect one word as similar. The option value must be an integer
18204 greater than zero. The default value is 9000.
18207 Set threshold to detect all words as similar. The option value must be an integer
18208 greater than zero. The default value is 60000.
18211 Set threshold to detect frames as similar. The option value must be an integer
18212 greater than zero. The default value is 116.
18215 Set the minimum length of a sequence in frames to recognize it as matching
18216 sequence. The option value must be a non negative integer value.
18217 The default value is 0.
18220 Set the minimum relation, that matching frames to all frames must have.
18221 The option value must be a double value between 0 and 1. The default value is 0.5.
18224 @subsection Examples
18228 To calculate the signature of an input video and store it in signature.bin:
18230 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18234 To detect whether two videos match and store the signatures in XML format in
18235 signature0.xml and signature1.xml:
18237 ffmpeg -i input1.mkv -i input2.mkv -filter_complex "[0:v][1:v] signature=nb_inputs=2:detectmode=full:format=xml:filename=signature%d.xml" -map :v -f null -
18245 Blur the input video without impacting the outlines.
18247 It accepts the following options:
18250 @item luma_radius, lr
18251 Set the luma radius. The option value must be a float number in
18252 the range [0.1,5.0] that specifies the variance of the gaussian filter
18253 used to blur the image (slower if larger). Default value is 1.0.
18255 @item luma_strength, ls
18256 Set the luma strength. The option value must be a float number
18257 in the range [-1.0,1.0] that configures the blurring. A value included
18258 in [0.0,1.0] will blur the image whereas a value included in
18259 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18261 @item luma_threshold, lt
18262 Set the luma threshold used as a coefficient to determine
18263 whether a pixel should be blurred or not. The option value must be an
18264 integer in the range [-30,30]. A value of 0 will filter all the image,
18265 a value included in [0,30] will filter flat areas and a value included
18266 in [-30,0] will filter edges. Default value is 0.
18268 @item chroma_radius, cr
18269 Set the chroma radius. The option value must be a float number in
18270 the range [0.1,5.0] that specifies the variance of the gaussian filter
18271 used to blur the image (slower if larger). Default value is @option{luma_radius}.
18273 @item chroma_strength, cs
18274 Set the chroma strength. The option value must be a float number
18275 in the range [-1.0,1.0] that configures the blurring. A value included
18276 in [0.0,1.0] will blur the image whereas a value included in
18277 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
18279 @item chroma_threshold, ct
18280 Set the chroma threshold used as a coefficient to determine
18281 whether a pixel should be blurred or not. The option value must be an
18282 integer in the range [-30,30]. A value of 0 will filter all the image,
18283 a value included in [0,30] will filter flat areas and a value included
18284 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
18287 If a chroma option is not explicitly set, the corresponding luma value
18291 Apply sobel operator to input video stream.
18293 The filter accepts the following option:
18297 Set which planes will be processed, unprocessed planes will be copied.
18298 By default value 0xf, all planes will be processed.
18301 Set value which will be multiplied with filtered result.
18304 Set value which will be added to filtered result.
18307 @subsection Commands
18309 This filter supports the all above options as @ref{commands}.
18314 Apply a simple postprocessing filter that compresses and decompresses the image
18315 at several (or - in the case of @option{quality} level @code{6} - all) shifts
18316 and average the results.
18318 The filter accepts the following options:
18322 Set quality. This option defines the number of levels for averaging. It accepts
18323 an integer in the range 0-6. If set to @code{0}, the filter will have no
18324 effect. A value of @code{6} means the higher quality. For each increment of
18325 that value the speed drops by a factor of approximately 2. Default value is
18329 Force a constant quantization parameter. If not set, the filter will use the QP
18330 from the video stream (if available).
18333 Set thresholding mode. Available modes are:
18337 Set hard thresholding (default).
18339 Set soft thresholding (better de-ringing effect, but likely blurrier).
18342 @item use_bframe_qp
18343 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
18344 option may cause flicker since the B-Frames have often larger QP. Default is
18345 @code{0} (not enabled).
18348 @subsection Commands
18350 This filter supports the following commands:
18352 @item quality, level
18353 Set quality level. The value @code{max} can be used to set the maximum level,
18354 currently @code{6}.
18360 Scale the input by applying one of the super-resolution methods based on
18361 convolutional neural networks. Supported models:
18365 Super-Resolution Convolutional Neural Network model (SRCNN).
18366 See @url{https://arxiv.org/abs/1501.00092}.
18369 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18370 See @url{https://arxiv.org/abs/1609.05158}.
18373 Training scripts as well as scripts for model file (.pb) saving can be found at
18374 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
18375 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
18377 Native model files (.model) can be generated from TensorFlow model
18378 files (.pb) by using tools/python/convert.py
18380 The filter accepts the following options:
18384 Specify which DNN backend to use for model loading and execution. This option accepts
18385 the following values:
18389 Native implementation of DNN loading and execution.
18392 TensorFlow backend. To enable this backend you
18393 need to install the TensorFlow for C library (see
18394 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
18395 @code{--enable-libtensorflow}
18398 Default value is @samp{native}.
18401 Set path to model file specifying network architecture and its parameters.
18402 Note that different backends use different file formats. TensorFlow backend
18403 can load files for both formats, while native backend can load files for only
18407 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
18408 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
18409 input upscaled using bicubic upscaling with proper scale factor.
18412 This feature can also be finished with @ref{dnn_processing} filter.
18416 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
18418 This filter takes in input two input videos, the first input is
18419 considered the "main" source and is passed unchanged to the
18420 output. The second input is used as a "reference" video for computing
18423 Both video inputs must have the same resolution and pixel format for
18424 this filter to work correctly. Also it assumes that both inputs
18425 have the same number of frames, which are compared one by one.
18427 The filter stores the calculated SSIM of each frame.
18429 The description of the accepted parameters follows.
18432 @item stats_file, f
18433 If specified the filter will use the named file to save the SSIM of
18434 each individual frame. When filename equals "-" the data is sent to
18438 The file printed if @var{stats_file} is selected, contains a sequence of
18439 key/value pairs of the form @var{key}:@var{value} for each compared
18442 A description of each shown parameter follows:
18446 sequential number of the input frame, starting from 1
18448 @item Y, U, V, R, G, B
18449 SSIM of the compared frames for the component specified by the suffix.
18452 SSIM of the compared frames for the whole frame.
18455 Same as above but in dB representation.
18458 This filter also supports the @ref{framesync} options.
18460 @subsection Examples
18465 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18466 [main][ref] ssim="stats_file=stats.log" [out]
18469 On this example the input file being processed is compared with the
18470 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18471 is stored in @file{stats.log}.
18474 Another example with both psnr and ssim at same time:
18476 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18480 Another example with different containers:
18482 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]ssim" -f null -
18488 Convert between different stereoscopic image formats.
18490 The filters accept the following options:
18494 Set stereoscopic image format of input.
18496 Available values for input image formats are:
18499 side by side parallel (left eye left, right eye right)
18502 side by side crosseye (right eye left, left eye right)
18505 side by side parallel with half width resolution
18506 (left eye left, right eye right)
18509 side by side crosseye with half width resolution
18510 (right eye left, left eye right)
18514 above-below (left eye above, right eye below)
18518 above-below (right eye above, left eye below)
18522 above-below with half height resolution
18523 (left eye above, right eye below)
18527 above-below with half height resolution
18528 (right eye above, left eye below)
18531 alternating frames (left eye first, right eye second)
18534 alternating frames (right eye first, left eye second)
18537 interleaved rows (left eye has top row, right eye starts on next row)
18540 interleaved rows (right eye has top row, left eye starts on next row)
18543 interleaved columns, left eye first
18546 interleaved columns, right eye first
18548 Default value is @samp{sbsl}.
18552 Set stereoscopic image format of output.
18556 side by side parallel (left eye left, right eye right)
18559 side by side crosseye (right eye left, left eye right)
18562 side by side parallel with half width resolution
18563 (left eye left, right eye right)
18566 side by side crosseye with half width resolution
18567 (right eye left, left eye right)
18571 above-below (left eye above, right eye below)
18575 above-below (right eye above, left eye below)
18579 above-below with half height resolution
18580 (left eye above, right eye below)
18584 above-below with half height resolution
18585 (right eye above, left eye below)
18588 alternating frames (left eye first, right eye second)
18591 alternating frames (right eye first, left eye second)
18594 interleaved rows (left eye has top row, right eye starts on next row)
18597 interleaved rows (right eye has top row, left eye starts on next row)
18600 anaglyph red/blue gray
18601 (red filter on left eye, blue filter on right eye)
18604 anaglyph red/green gray
18605 (red filter on left eye, green filter on right eye)
18608 anaglyph red/cyan gray
18609 (red filter on left eye, cyan filter on right eye)
18612 anaglyph red/cyan half colored
18613 (red filter on left eye, cyan filter on right eye)
18616 anaglyph red/cyan color
18617 (red filter on left eye, cyan filter on right eye)
18620 anaglyph red/cyan color optimized with the least squares projection of dubois
18621 (red filter on left eye, cyan filter on right eye)
18624 anaglyph green/magenta gray
18625 (green filter on left eye, magenta filter on right eye)
18628 anaglyph green/magenta half colored
18629 (green filter on left eye, magenta filter on right eye)
18632 anaglyph green/magenta colored
18633 (green filter on left eye, magenta filter on right eye)
18636 anaglyph green/magenta color optimized with the least squares projection of dubois
18637 (green filter on left eye, magenta filter on right eye)
18640 anaglyph yellow/blue gray
18641 (yellow filter on left eye, blue filter on right eye)
18644 anaglyph yellow/blue half colored
18645 (yellow filter on left eye, blue filter on right eye)
18648 anaglyph yellow/blue colored
18649 (yellow filter on left eye, blue filter on right eye)
18652 anaglyph yellow/blue color optimized with the least squares projection of dubois
18653 (yellow filter on left eye, blue filter on right eye)
18656 mono output (left eye only)
18659 mono output (right eye only)
18662 checkerboard, left eye first
18665 checkerboard, right eye first
18668 interleaved columns, left eye first
18671 interleaved columns, right eye first
18677 Default value is @samp{arcd}.
18680 @subsection Examples
18684 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18690 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18696 @section streamselect, astreamselect
18697 Select video or audio streams.
18699 The filter accepts the following options:
18703 Set number of inputs. Default is 2.
18706 Set input indexes to remap to outputs.
18709 @subsection Commands
18711 The @code{streamselect} and @code{astreamselect} filter supports the following
18716 Set input indexes to remap to outputs.
18719 @subsection Examples
18723 Select first 5 seconds 1st stream and rest of time 2nd stream:
18725 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18729 Same as above, but for audio:
18731 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18738 Draw subtitles on top of input video using the libass library.
18740 To enable compilation of this filter you need to configure FFmpeg with
18741 @code{--enable-libass}. This filter also requires a build with libavcodec and
18742 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18743 Alpha) subtitles format.
18745 The filter accepts the following options:
18749 Set the filename of the subtitle file to read. It must be specified.
18751 @item original_size
18752 Specify the size of the original video, the video for which the ASS file
18753 was composed. For the syntax of this option, check the
18754 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18755 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18756 correctly scale the fonts if the aspect ratio has been changed.
18759 Set a directory path containing fonts that can be used by the filter.
18760 These fonts will be used in addition to whatever the font provider uses.
18763 Process alpha channel, by default alpha channel is untouched.
18766 Set subtitles input character encoding. @code{subtitles} filter only. Only
18767 useful if not UTF-8.
18769 @item stream_index, si
18770 Set subtitles stream index. @code{subtitles} filter only.
18773 Override default style or script info parameters of the subtitles. It accepts a
18774 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18777 If the first key is not specified, it is assumed that the first value
18778 specifies the @option{filename}.
18780 For example, to render the file @file{sub.srt} on top of the input
18781 video, use the command:
18786 which is equivalent to:
18788 subtitles=filename=sub.srt
18791 To render the default subtitles stream from file @file{video.mkv}, use:
18793 subtitles=video.mkv
18796 To render the second subtitles stream from that file, use:
18798 subtitles=video.mkv:si=1
18801 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18802 @code{DejaVu Serif}, use:
18804 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18807 @section super2xsai
18809 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18810 Interpolate) pixel art scaling algorithm.
18812 Useful for enlarging pixel art images without reducing sharpness.
18816 Swap two rectangular objects in video.
18818 This filter accepts the following options:
18828 Set 1st rect x coordinate.
18831 Set 1st rect y coordinate.
18834 Set 2nd rect x coordinate.
18837 Set 2nd rect y coordinate.
18839 All expressions are evaluated once for each frame.
18842 The all options are expressions containing the following constants:
18847 The input width and height.
18850 same as @var{w} / @var{h}
18853 input sample aspect ratio
18856 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18859 The number of the input frame, starting from 0.
18862 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18865 the position in the file of the input frame, NAN if unknown
18872 Blend successive video frames.
18878 Apply telecine process to the video.
18880 This filter accepts the following options:
18889 The default value is @code{top}.
18893 A string of numbers representing the pulldown pattern you wish to apply.
18894 The default value is @code{23}.
18898 Some typical patterns:
18903 24p: 2332 (preferred)
18910 24p: 222222222223 ("Euro pulldown")
18915 @section thistogram
18917 Compute and draw a color distribution histogram for the input video across time.
18919 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18920 at certain time, this filter shows also past histograms of number of frames defined
18921 by @code{width} option.
18923 The computed histogram is a representation of the color component
18924 distribution in an image.
18926 The filter accepts the following options:
18930 Set width of single color component output. Default value is @code{0}.
18931 Value of @code{0} means width will be picked from input video.
18932 This also set number of passed histograms to keep.
18933 Allowed range is [0, 8192].
18935 @item display_mode, d
18937 It accepts the following values:
18940 Per color component graphs are placed below each other.
18943 Per color component graphs are placed side by side.
18946 Presents information identical to that in the @code{parade}, except
18947 that the graphs representing color components are superimposed directly
18950 Default is @code{stack}.
18952 @item levels_mode, m
18953 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18954 Default is @code{linear}.
18956 @item components, c
18957 Set what color components to display.
18958 Default is @code{7}.
18961 Set background opacity. Default is @code{0.9}.
18964 Show envelope. Default is disabled.
18967 Set envelope color. Default is @code{gold}.
18972 Available values for slide is:
18975 Draw new frame when right border is reached.
18978 Replace old columns with new ones.
18981 Scroll from right to left.
18984 Scroll from left to right.
18987 Draw single picture.
18990 Default is @code{replace}.
18995 Apply threshold effect to video stream.
18997 This filter needs four video streams to perform thresholding.
18998 First stream is stream we are filtering.
18999 Second stream is holding threshold values, third stream is holding min values,
19000 and last, fourth stream is holding max values.
19002 The filter accepts the following option:
19006 Set which planes will be processed, unprocessed planes will be copied.
19007 By default value 0xf, all planes will be processed.
19010 For example if first stream pixel's component value is less then threshold value
19011 of pixel component from 2nd threshold stream, third stream value will picked,
19012 otherwise fourth stream pixel component value will be picked.
19014 Using color source filter one can perform various types of thresholding:
19016 @subsection Examples
19020 Binary threshold, using gray color as threshold:
19022 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
19026 Inverted binary threshold, using gray color as threshold:
19028 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
19032 Truncate binary threshold, using gray color as threshold:
19034 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
19038 Threshold to zero, using gray color as threshold:
19040 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
19044 Inverted threshold to zero, using gray color as threshold:
19046 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
19051 Select the most representative frame in a given sequence of consecutive frames.
19053 The filter accepts the following options:
19057 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
19058 will pick one of them, and then handle the next batch of @var{n} frames until
19059 the end. Default is @code{100}.
19062 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
19063 value will result in a higher memory usage, so a high value is not recommended.
19065 @subsection Examples
19069 Extract one picture each 50 frames:
19075 Complete example of a thumbnail creation with @command{ffmpeg}:
19077 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
19084 Tile several successive frames together.
19086 The @ref{untile} filter can do the reverse.
19088 The filter accepts the following options:
19093 Set the grid size (i.e. the number of lines and columns). For the syntax of
19094 this option, check the
19095 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19098 Set the maximum number of frames to render in the given area. It must be less
19099 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
19100 the area will be used.
19103 Set the outer border margin in pixels.
19106 Set the inner border thickness (i.e. the number of pixels between frames). For
19107 more advanced padding options (such as having different values for the edges),
19108 refer to the pad video filter.
19111 Specify the color of the unused area. For the syntax of this option, check the
19112 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
19113 The default value of @var{color} is "black".
19116 Set the number of frames to overlap when tiling several successive frames together.
19117 The value must be between @code{0} and @var{nb_frames - 1}.
19120 Set the number of frames to initially be empty before displaying first output frame.
19121 This controls how soon will one get first output frame.
19122 The value must be between @code{0} and @var{nb_frames - 1}.
19125 @subsection Examples
19129 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
19131 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
19133 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
19134 duplicating each output frame to accommodate the originally detected frame
19138 Display @code{5} pictures in an area of @code{3x2} frames,
19139 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
19140 mixed flat and named options:
19142 tile=3x2:nb_frames=5:padding=7:margin=2
19146 @section tinterlace
19148 Perform various types of temporal field interlacing.
19150 Frames are counted starting from 1, so the first input frame is
19153 The filter accepts the following options:
19158 Specify the mode of the interlacing. This option can also be specified
19159 as a value alone. See below for a list of values for this option.
19161 Available values are:
19165 Move odd frames into the upper field, even into the lower field,
19166 generating a double height frame at half frame rate.
19170 Frame 1 Frame 2 Frame 3 Frame 4
19172 11111 22222 33333 44444
19173 11111 22222 33333 44444
19174 11111 22222 33333 44444
19175 11111 22222 33333 44444
19189 Only output odd frames, even frames are dropped, generating a frame with
19190 unchanged height at half frame rate.
19195 Frame 1 Frame 2 Frame 3 Frame 4
19197 11111 22222 33333 44444
19198 11111 22222 33333 44444
19199 11111 22222 33333 44444
19200 11111 22222 33333 44444
19210 Only output even frames, odd frames are dropped, generating a frame with
19211 unchanged height at half frame rate.
19216 Frame 1 Frame 2 Frame 3 Frame 4
19218 11111 22222 33333 44444
19219 11111 22222 33333 44444
19220 11111 22222 33333 44444
19221 11111 22222 33333 44444
19231 Expand each frame to full height, but pad alternate lines with black,
19232 generating a frame with double height at the same input frame rate.
19237 Frame 1 Frame 2 Frame 3 Frame 4
19239 11111 22222 33333 44444
19240 11111 22222 33333 44444
19241 11111 22222 33333 44444
19242 11111 22222 33333 44444
19245 11111 ..... 33333 .....
19246 ..... 22222 ..... 44444
19247 11111 ..... 33333 .....
19248 ..... 22222 ..... 44444
19249 11111 ..... 33333 .....
19250 ..... 22222 ..... 44444
19251 11111 ..... 33333 .....
19252 ..... 22222 ..... 44444
19256 @item interleave_top, 4
19257 Interleave the upper field from odd frames with the lower field from
19258 even frames, generating a frame with unchanged height at half frame rate.
19263 Frame 1 Frame 2 Frame 3 Frame 4
19265 11111<- 22222 33333<- 44444
19266 11111 22222<- 33333 44444<-
19267 11111<- 22222 33333<- 44444
19268 11111 22222<- 33333 44444<-
19278 @item interleave_bottom, 5
19279 Interleave the lower field from odd frames with the upper field from
19280 even frames, generating a frame with unchanged height at half frame rate.
19285 Frame 1 Frame 2 Frame 3 Frame 4
19287 11111 22222<- 33333 44444<-
19288 11111<- 22222 33333<- 44444
19289 11111 22222<- 33333 44444<-
19290 11111<- 22222 33333<- 44444
19300 @item interlacex2, 6
19301 Double frame rate with unchanged height. Frames are inserted each
19302 containing the second temporal field from the previous input frame and
19303 the first temporal field from the next input frame. This mode relies on
19304 the top_field_first flag. Useful for interlaced video displays with no
19305 field synchronisation.
19310 Frame 1 Frame 2 Frame 3 Frame 4
19312 11111 22222 33333 44444
19313 11111 22222 33333 44444
19314 11111 22222 33333 44444
19315 11111 22222 33333 44444
19318 11111 22222 22222 33333 33333 44444 44444
19319 11111 11111 22222 22222 33333 33333 44444
19320 11111 22222 22222 33333 33333 44444 44444
19321 11111 11111 22222 22222 33333 33333 44444
19326 Move odd frames into the upper field, even into the lower field,
19327 generating a double height frame at same frame rate.
19332 Frame 1 Frame 2 Frame 3 Frame 4
19334 11111 22222 33333 44444
19335 11111 22222 33333 44444
19336 11111 22222 33333 44444
19337 11111 22222 33333 44444
19340 11111 33333 33333 55555
19341 22222 22222 44444 44444
19342 11111 33333 33333 55555
19343 22222 22222 44444 44444
19344 11111 33333 33333 55555
19345 22222 22222 44444 44444
19346 11111 33333 33333 55555
19347 22222 22222 44444 44444
19352 Numeric values are deprecated but are accepted for backward
19353 compatibility reasons.
19355 Default mode is @code{merge}.
19358 Specify flags influencing the filter process.
19360 Available value for @var{flags} is:
19363 @item low_pass_filter, vlpf
19364 Enable linear vertical low-pass filtering in the filter.
19365 Vertical low-pass filtering is required when creating an interlaced
19366 destination from a progressive source which contains high-frequency
19367 vertical detail. Filtering will reduce interlace 'twitter' and Moire
19370 @item complex_filter, cvlpf
19371 Enable complex vertical low-pass filtering.
19372 This will slightly less reduce interlace 'twitter' and Moire
19373 patterning but better retain detail and subjective sharpness impression.
19376 Bypass already interlaced frames, only adjust the frame rate.
19379 Vertical low-pass filtering and bypassing already interlaced frames can only be
19380 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
19385 Pick median pixels from several successive input video frames.
19387 The filter accepts the following options:
19391 Set radius of median filter.
19392 Default is 1. Allowed range is from 1 to 127.
19395 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19398 Set median percentile. Default value is @code{0.5}.
19399 Default value of @code{0.5} will pick always median values, while @code{0} will pick
19400 minimum values, and @code{1} maximum values.
19403 @subsection Commands
19405 This filter supports all above options as @ref{commands}, excluding option @code{radius}.
19407 @section tmidequalizer
19409 Apply Temporal Midway Video Equalization effect.
19411 Midway Video Equalization adjusts a sequence of video frames to have the same
19412 histograms, while maintaining their dynamics as much as possible. It's
19413 useful for e.g. matching exposures from a video frames sequence.
19415 This filter accepts the following option:
19419 Set filtering radius. Default is @code{5}. Allowed range is from 1 to 127.
19422 Set filtering sigma. Default is @code{0.5}. This controls strength of filtering.
19423 Setting this option to 0 effectively does nothing.
19426 Set which planes to process. Default is @code{15}, which is all available planes.
19431 Mix successive video frames.
19433 A description of the accepted options follows.
19437 The number of successive frames to mix. If unspecified, it defaults to 3.
19440 Specify weight of each input video frame.
19441 Each weight is separated by space. If number of weights is smaller than
19442 number of @var{frames} last specified weight will be used for all remaining
19446 Specify scale, if it is set it will be multiplied with sum
19447 of each weight multiplied with pixel values to give final destination
19448 pixel value. By default @var{scale} is auto scaled to sum of weights.
19451 @subsection Examples
19455 Average 7 successive frames:
19457 tmix=frames=7:weights="1 1 1 1 1 1 1"
19461 Apply simple temporal convolution:
19463 tmix=frames=3:weights="-1 3 -1"
19467 Similar as above but only showing temporal differences:
19469 tmix=frames=3:weights="-1 2 -1":scale=1
19475 Tone map colors from different dynamic ranges.
19477 This filter expects data in single precision floating point, as it needs to
19478 operate on (and can output) out-of-range values. Another filter, such as
19479 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19481 The tonemapping algorithms implemented only work on linear light, so input
19482 data should be linearized beforehand (and possibly correctly tagged).
19485 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19488 @subsection Options
19489 The filter accepts the following options.
19493 Set the tone map algorithm to use.
19495 Possible values are:
19498 Do not apply any tone map, only desaturate overbright pixels.
19501 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19502 in-range values, while distorting out-of-range values.
19505 Stretch the entire reference gamut to a linear multiple of the display.
19508 Fit a logarithmic transfer between the tone curves.
19511 Preserve overall image brightness with a simple curve, using nonlinear
19512 contrast, which results in flattening details and degrading color accuracy.
19515 Preserve both dark and bright details better than @var{reinhard}, at the cost
19516 of slightly darkening everything. Use it when detail preservation is more
19517 important than color and brightness accuracy.
19520 Smoothly map out-of-range values, while retaining contrast and colors for
19521 in-range material as much as possible. Use it when color accuracy is more
19522 important than detail preservation.
19528 Tune the tone mapping algorithm.
19530 This affects the following algorithms:
19536 Specifies the scale factor to use while stretching.
19540 Specifies the exponent of the function.
19544 Specify an extra linear coefficient to multiply into the signal before clipping.
19548 Specify the local contrast coefficient at the display peak.
19549 Default to 0.5, which means that in-gamut values will be about half as bright
19556 Specify the transition point from linear to mobius transform. Every value
19557 below this point is guaranteed to be mapped 1:1. The higher the value, the
19558 more accurate the result will be, at the cost of losing bright details.
19559 Default to 0.3, which due to the steep initial slope still preserves in-range
19560 colors fairly accurately.
19564 Apply desaturation for highlights that exceed this level of brightness. The
19565 higher the parameter, the more color information will be preserved. This
19566 setting helps prevent unnaturally blown-out colors for super-highlights, by
19567 (smoothly) turning into white instead. This makes images feel more natural,
19568 at the cost of reducing information about out-of-range colors.
19570 The default of 2.0 is somewhat conservative and will mostly just apply to
19571 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19573 This option works only if the input frame has a supported color tag.
19576 Override signal/nominal/reference peak with this value. Useful when the
19577 embedded peak information in display metadata is not reliable or when tone
19578 mapping from a lower range to a higher range.
19583 Temporarily pad video frames.
19585 The filter accepts the following options:
19589 Specify number of delay frames before input video stream. Default is 0.
19592 Specify number of padding frames after input video stream.
19593 Set to -1 to pad indefinitely. Default is 0.
19596 Set kind of frames added to beginning of stream.
19597 Can be either @var{add} or @var{clone}.
19598 With @var{add} frames of solid-color are added.
19599 With @var{clone} frames are clones of first frame.
19600 Default is @var{add}.
19603 Set kind of frames added to end of stream.
19604 Can be either @var{add} or @var{clone}.
19605 With @var{add} frames of solid-color are added.
19606 With @var{clone} frames are clones of last frame.
19607 Default is @var{add}.
19609 @item start_duration, stop_duration
19610 Specify the duration of the start/stop delay. See
19611 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19612 for the accepted syntax.
19613 These options override @var{start} and @var{stop}. Default is 0.
19616 Specify the color of the padded area. For the syntax of this option,
19617 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19618 manual,ffmpeg-utils}.
19620 The default value of @var{color} is "black".
19626 Transpose rows with columns in the input video and optionally flip it.
19628 It accepts the following parameters:
19633 Specify the transposition direction.
19635 Can assume the following values:
19637 @item 0, 4, cclock_flip
19638 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19646 Rotate by 90 degrees clockwise, that is:
19654 Rotate by 90 degrees counterclockwise, that is:
19661 @item 3, 7, clock_flip
19662 Rotate by 90 degrees clockwise and vertically flip, that is:
19670 For values between 4-7, the transposition is only done if the input
19671 video geometry is portrait and not landscape. These values are
19672 deprecated, the @code{passthrough} option should be used instead.
19674 Numerical values are deprecated, and should be dropped in favor of
19675 symbolic constants.
19678 Do not apply the transposition if the input geometry matches the one
19679 specified by the specified value. It accepts the following values:
19682 Always apply transposition.
19684 Preserve portrait geometry (when @var{height} >= @var{width}).
19686 Preserve landscape geometry (when @var{width} >= @var{height}).
19689 Default value is @code{none}.
19692 For example to rotate by 90 degrees clockwise and preserve portrait
19695 transpose=dir=1:passthrough=portrait
19698 The command above can also be specified as:
19700 transpose=1:portrait
19703 @section transpose_npp
19705 Transpose rows with columns in the input video and optionally flip it.
19706 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19708 It accepts the following parameters:
19713 Specify the transposition direction.
19715 Can assume the following values:
19718 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19721 Rotate by 90 degrees clockwise.
19724 Rotate by 90 degrees counterclockwise.
19727 Rotate by 90 degrees clockwise and vertically flip.
19731 Do not apply the transposition if the input geometry matches the one
19732 specified by the specified value. It accepts the following values:
19735 Always apply transposition. (default)
19737 Preserve portrait geometry (when @var{height} >= @var{width}).
19739 Preserve landscape geometry (when @var{width} >= @var{height}).
19745 Trim the input so that the output contains one continuous subpart of the input.
19747 It accepts the following parameters:
19750 Specify the time of the start of the kept section, i.e. the frame with the
19751 timestamp @var{start} will be the first frame in the output.
19754 Specify the time of the first frame that will be dropped, i.e. the frame
19755 immediately preceding the one with the timestamp @var{end} will be the last
19756 frame in the output.
19759 This is the same as @var{start}, except this option sets the start timestamp
19760 in timebase units instead of seconds.
19763 This is the same as @var{end}, except this option sets the end timestamp
19764 in timebase units instead of seconds.
19767 The maximum duration of the output in seconds.
19770 The number of the first frame that should be passed to the output.
19773 The number of the first frame that should be dropped.
19776 @option{start}, @option{end}, and @option{duration} are expressed as time
19777 duration specifications; see
19778 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19779 for the accepted syntax.
19781 Note that the first two sets of the start/end options and the @option{duration}
19782 option look at the frame timestamp, while the _frame variants simply count the
19783 frames that pass through the filter. Also note that this filter does not modify
19784 the timestamps. If you wish for the output timestamps to start at zero, insert a
19785 setpts filter after the trim filter.
19787 If multiple start or end options are set, this filter tries to be greedy and
19788 keep all the frames that match at least one of the specified constraints. To keep
19789 only the part that matches all the constraints at once, chain multiple trim
19792 The defaults are such that all the input is kept. So it is possible to set e.g.
19793 just the end values to keep everything before the specified time.
19798 Drop everything except the second minute of input:
19800 ffmpeg -i INPUT -vf trim=60:120
19804 Keep only the first second:
19806 ffmpeg -i INPUT -vf trim=duration=1
19811 @section unpremultiply
19812 Apply alpha unpremultiply effect to input video stream using first plane
19813 of second stream as alpha.
19815 Both streams must have same dimensions and same pixel format.
19817 The filter accepts the following option:
19821 Set which planes will be processed, unprocessed planes will be copied.
19822 By default value 0xf, all planes will be processed.
19824 If the format has 1 or 2 components, then luma is bit 0.
19825 If the format has 3 or 4 components:
19826 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19827 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19828 If present, the alpha channel is always the last bit.
19831 Do not require 2nd input for processing, instead use alpha plane from input stream.
19837 Sharpen or blur the input video.
19839 It accepts the following parameters:
19842 @item luma_msize_x, lx
19843 Set the luma matrix horizontal size. It must be an odd integer between
19844 3 and 23. The default value is 5.
19846 @item luma_msize_y, ly
19847 Set the luma matrix vertical size. It must be an odd integer between 3
19848 and 23. The default value is 5.
19850 @item luma_amount, la
19851 Set the luma effect strength. It must be a floating point number, reasonable
19852 values lay between -1.5 and 1.5.
19854 Negative values will blur the input video, while positive values will
19855 sharpen it, a value of zero will disable the effect.
19857 Default value is 1.0.
19859 @item chroma_msize_x, cx
19860 Set the chroma matrix horizontal size. It must be an odd integer
19861 between 3 and 23. The default value is 5.
19863 @item chroma_msize_y, cy
19864 Set the chroma matrix vertical size. It must be an odd integer
19865 between 3 and 23. The default value is 5.
19867 @item chroma_amount, ca
19868 Set the chroma effect strength. It must be a floating point number, reasonable
19869 values lay between -1.5 and 1.5.
19871 Negative values will blur the input video, while positive values will
19872 sharpen it, a value of zero will disable the effect.
19874 Default value is 0.0.
19878 All parameters are optional and default to the equivalent of the
19879 string '5:5:1.0:5:5:0.0'.
19881 @subsection Examples
19885 Apply strong luma sharpen effect:
19887 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19891 Apply a strong blur of both luma and chroma parameters:
19893 unsharp=7:7:-2:7:7:-2
19900 Decompose a video made of tiled images into the individual images.
19902 The frame rate of the output video is the frame rate of the input video
19903 multiplied by the number of tiles.
19905 This filter does the reverse of @ref{tile}.
19907 The filter accepts the following options:
19912 Set the grid size (i.e. the number of lines and columns). For the syntax of
19913 this option, check the
19914 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19917 @subsection Examples
19921 Produce a 1-second video from a still image file made of 25 frames stacked
19922 vertically, like an analogic film reel:
19924 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19930 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19931 the image at several (or - in the case of @option{quality} level @code{8} - all)
19932 shifts and average the results.
19934 The way this differs from the behavior of spp is that uspp actually encodes &
19935 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19936 DCT similar to MJPEG.
19938 The filter accepts the following options:
19942 Set quality. This option defines the number of levels for averaging. It accepts
19943 an integer in the range 0-8. If set to @code{0}, the filter will have no
19944 effect. A value of @code{8} means the higher quality. For each increment of
19945 that value the speed drops by a factor of approximately 2. Default value is
19949 Force a constant quantization parameter. If not set, the filter will use the QP
19950 from the video stream (if available).
19955 Convert 360 videos between various formats.
19957 The filter accepts the following options:
19963 Set format of the input/output video.
19971 Equirectangular projection.
19976 Cubemap with 3x2/6x1/1x6 layout.
19978 Format specific options:
19983 Set padding proportion for the input/output cubemap. Values in decimals.
19990 1% of face is padding. For example, with 1920x1280 resolution face size would be 640x640 and padding would be 3 pixels from each side. (640 * 0.01 = 6 pixels)
19993 Default value is @b{@samp{0}}.
19994 Maximum value is @b{@samp{0.1}}.
19998 Set fixed padding for the input/output cubemap. Values in pixels.
20000 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
20004 Set order of faces for the input/output cubemap. Choose one direction for each position.
20006 Designation of directions:
20022 Default value is @b{@samp{rludfb}}.
20026 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
20028 Designation of angles:
20031 0 degrees clockwise
20033 90 degrees clockwise
20035 180 degrees clockwise
20037 270 degrees clockwise
20040 Default value is @b{@samp{000000}}.
20044 Equi-Angular Cubemap.
20051 Format specific options:
20056 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20058 If diagonal field of view is set it overrides horizontal and vertical field of view.
20063 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20065 If diagonal field of view is set it overrides horizontal and vertical field of view.
20071 Format specific options:
20076 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20078 If diagonal field of view is set it overrides horizontal and vertical field of view.
20083 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20085 If diagonal field of view is set it overrides horizontal and vertical field of view.
20091 Facebook's 360 formats.
20094 Stereographic format.
20096 Format specific options:
20101 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20103 If diagonal field of view is set it overrides horizontal and vertical field of view.
20108 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20110 If diagonal field of view is set it overrides horizontal and vertical field of view.
20117 Ball format, gives significant distortion toward the back.
20120 Hammer-Aitoff map projection format.
20123 Sinusoidal map projection format.
20126 Fisheye projection.
20128 Format specific options:
20133 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20135 If diagonal field of view is set it overrides horizontal and vertical field of view.
20140 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20142 If diagonal field of view is set it overrides horizontal and vertical field of view.
20146 Pannini projection.
20148 Format specific options:
20151 Set output pannini parameter.
20154 Set input pannini parameter.
20158 Cylindrical projection.
20160 Format specific options:
20165 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20167 If diagonal field of view is set it overrides horizontal and vertical field of view.
20172 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20174 If diagonal field of view is set it overrides horizontal and vertical field of view.
20178 Perspective projection. @i{(output only)}
20180 Format specific options:
20183 Set perspective parameter.
20187 Tetrahedron projection.
20190 Truncated square pyramid projection.
20194 Half equirectangular projection.
20199 Format specific options:
20204 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20206 If diagonal field of view is set it overrides horizontal and vertical field of view.
20211 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20213 If diagonal field of view is set it overrides horizontal and vertical field of view.
20217 Orthographic format.
20219 Format specific options:
20224 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20226 If diagonal field of view is set it overrides horizontal and vertical field of view.
20231 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20233 If diagonal field of view is set it overrides horizontal and vertical field of view.
20237 Octahedron projection.
20241 Set interpolation method.@*
20242 @i{Note: more complex interpolation methods require much more memory to run.}
20252 Bilinear interpolation.
20254 Lagrange9 interpolation.
20257 Bicubic interpolation.
20260 Lanczos interpolation.
20263 Spline16 interpolation.
20266 Gaussian interpolation.
20268 Mitchell interpolation.
20271 Default value is @b{@samp{line}}.
20275 Set the output video resolution.
20277 Default resolution depends on formats.
20281 Set the input/output stereo format.
20292 Default value is @b{@samp{2d}} for input and output format.
20297 Set rotation for the output video. Values in degrees.
20300 Set rotation order for the output video. Choose one item for each position.
20311 Default value is @b{@samp{ypr}}.
20316 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
20320 Set if input video is flipped horizontally/vertically. Boolean values.
20323 Set if input video is transposed. Boolean value, by default disabled.
20326 Set if output video needs to be transposed. Boolean value, by default disabled.
20329 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
20332 @subsection Examples
20336 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
20338 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20341 Extract back view of Equi-Angular Cubemap:
20343 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20346 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
20348 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20352 @subsection Commands
20354 This filter supports subset of above options as @ref{commands}.
20356 @section vaguedenoiser
20358 Apply a wavelet based denoiser.
20360 It transforms each frame from the video input into the wavelet domain,
20361 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20362 the obtained coefficients. It does an inverse wavelet transform after.
20363 Due to wavelet properties, it should give a nice smoothed result, and
20364 reduced noise, without blurring picture features.
20366 This filter accepts the following options:
20370 The filtering strength. The higher, the more filtered the video will be.
20371 Hard thresholding can use a higher threshold than soft thresholding
20372 before the video looks overfiltered. Default value is 2.
20375 The filtering method the filter will use.
20377 It accepts the following values:
20380 All values under the threshold will be zeroed.
20383 All values under the threshold will be zeroed. All values above will be
20384 reduced by the threshold.
20387 Scales or nullifies coefficients - intermediary between (more) soft and
20388 (less) hard thresholding.
20391 Default is garrote.
20394 Number of times, the wavelet will decompose the picture. Picture can't
20395 be decomposed beyond a particular point (typically, 8 for a 640x480
20396 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
20399 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
20402 A list of the planes to process. By default all planes are processed.
20405 The threshold type the filter will use.
20407 It accepts the following values:
20410 Threshold used is same for all decompositions.
20413 Threshold used depends also on each decomposition coefficients.
20416 Default is universal.
20419 @section vectorscope
20421 Display 2 color component values in the two dimensional graph (which is called
20424 This filter accepts the following options:
20428 Set vectorscope mode.
20430 It accepts the following values:
20434 Gray values are displayed on graph, higher brightness means more pixels have
20435 same component color value on location in graph. This is the default mode.
20438 Gray values are displayed on graph. Surrounding pixels values which are not
20439 present in video frame are drawn in gradient of 2 color components which are
20440 set by option @code{x} and @code{y}. The 3rd color component is static.
20443 Actual color components values present in video frame are displayed on graph.
20446 Similar as color2 but higher frequency of same values @code{x} and @code{y}
20447 on graph increases value of another color component, which is luminance by
20448 default values of @code{x} and @code{y}.
20451 Actual colors present in video frame are displayed on graph. If two different
20452 colors map to same position on graph then color with higher value of component
20453 not present in graph is picked.
20456 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
20457 component picked from radial gradient.
20461 Set which color component will be represented on X-axis. Default is @code{1}.
20464 Set which color component will be represented on Y-axis. Default is @code{2}.
20467 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20468 of color component which represents frequency of (X, Y) location in graph.
20473 No envelope, this is default.
20476 Instant envelope, even darkest single pixel will be clearly highlighted.
20479 Hold maximum and minimum values presented in graph over time. This way you
20480 can still spot out of range values without constantly looking at vectorscope.
20483 Peak and instant envelope combined together.
20487 Set what kind of graticule to draw.
20496 Set graticule opacity.
20499 Set graticule flags.
20503 Draw graticule for white point.
20506 Draw graticule for black point.
20509 Draw color points short names.
20513 Set background opacity.
20515 @item lthreshold, l
20516 Set low threshold for color component not represented on X or Y axis.
20517 Values lower than this value will be ignored. Default is 0.
20518 Note this value is multiplied with actual max possible value one pixel component
20519 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20522 @item hthreshold, h
20523 Set high threshold for color component not represented on X or Y axis.
20524 Values higher than this value will be ignored. Default is 1.
20525 Note this value is multiplied with actual max possible value one pixel component
20526 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20527 is 0.9 * 255 = 230.
20529 @item colorspace, c
20530 Set what kind of colorspace to use when drawing graticule.
20540 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20541 This means no tint, and output will remain gray.
20544 @anchor{vidstabdetect}
20545 @section vidstabdetect
20547 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20548 @ref{vidstabtransform} for pass 2.
20550 This filter generates a file with relative translation and rotation
20551 transform information about subsequent frames, which is then used by
20552 the @ref{vidstabtransform} filter.
20554 To enable compilation of this filter you need to configure FFmpeg with
20555 @code{--enable-libvidstab}.
20557 This filter accepts the following options:
20561 Set the path to the file used to write the transforms information.
20562 Default value is @file{transforms.trf}.
20565 Set how shaky the video is and how quick the camera is. It accepts an
20566 integer in the range 1-10, a value of 1 means little shakiness, a
20567 value of 10 means strong shakiness. Default value is 5.
20570 Set the accuracy of the detection process. It must be a value in the
20571 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20572 accuracy. Default value is 15.
20575 Set stepsize of the search process. The region around minimum is
20576 scanned with 1 pixel resolution. Default value is 6.
20579 Set minimum contrast. Below this value a local measurement field is
20580 discarded. Must be a floating point value in the range 0-1. Default
20584 Set reference frame number for tripod mode.
20586 If enabled, the motion of the frames is compared to a reference frame
20587 in the filtered stream, identified by the specified number. The idea
20588 is to compensate all movements in a more-or-less static scene and keep
20589 the camera view absolutely still.
20591 If set to 0, it is disabled. The frames are counted starting from 1.
20594 Show fields and transforms in the resulting frames. It accepts an
20595 integer in the range 0-2. Default value is 0, which disables any
20599 @subsection Examples
20603 Use default values:
20609 Analyze strongly shaky movie and put the results in file
20610 @file{mytransforms.trf}:
20612 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20616 Visualize the result of internal transformations in the resulting
20619 vidstabdetect=show=1
20623 Analyze a video with medium shakiness using @command{ffmpeg}:
20625 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20629 @anchor{vidstabtransform}
20630 @section vidstabtransform
20632 Video stabilization/deshaking: pass 2 of 2,
20633 see @ref{vidstabdetect} for pass 1.
20635 Read a file with transform information for each frame and
20636 apply/compensate them. Together with the @ref{vidstabdetect}
20637 filter this can be used to deshake videos. See also
20638 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20639 the @ref{unsharp} filter, see below.
20641 To enable compilation of this filter you need to configure FFmpeg with
20642 @code{--enable-libvidstab}.
20644 @subsection Options
20648 Set path to the file used to read the transforms. Default value is
20649 @file{transforms.trf}.
20652 Set the number of frames (value*2 + 1) used for lowpass filtering the
20653 camera movements. Default value is 10.
20655 For example a number of 10 means that 21 frames are used (10 in the
20656 past and 10 in the future) to smoothen the motion in the video. A
20657 larger value leads to a smoother video, but limits the acceleration of
20658 the camera (pan/tilt movements). 0 is a special case where a static
20659 camera is simulated.
20662 Set the camera path optimization algorithm.
20664 Accepted values are:
20667 gaussian kernel low-pass filter on camera motion (default)
20669 averaging on transformations
20673 Set maximal number of pixels to translate frames. Default value is -1,
20677 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20678 value is -1, meaning no limit.
20681 Specify how to deal with borders that may be visible due to movement
20684 Available values are:
20687 keep image information from previous frame (default)
20689 fill the border black
20693 Invert transforms if set to 1. Default value is 0.
20696 Consider transforms as relative to previous frame if set to 1,
20697 absolute if set to 0. Default value is 0.
20700 Set percentage to zoom. A positive value will result in a zoom-in
20701 effect, a negative value in a zoom-out effect. Default value is 0 (no
20705 Set optimal zooming to avoid borders.
20707 Accepted values are:
20712 optimal static zoom value is determined (only very strong movements
20713 will lead to visible borders) (default)
20715 optimal adaptive zoom value is determined (no borders will be
20716 visible), see @option{zoomspeed}
20719 Note that the value given at zoom is added to the one calculated here.
20722 Set percent to zoom maximally each frame (enabled when
20723 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20727 Specify type of interpolation.
20729 Available values are:
20734 linear only horizontal
20736 linear in both directions (default)
20738 cubic in both directions (slow)
20742 Enable virtual tripod mode if set to 1, which is equivalent to
20743 @code{relative=0:smoothing=0}. Default value is 0.
20745 Use also @code{tripod} option of @ref{vidstabdetect}.
20748 Increase log verbosity if set to 1. Also the detected global motions
20749 are written to the temporary file @file{global_motions.trf}. Default
20753 @subsection Examples
20757 Use @command{ffmpeg} for a typical stabilization with default values:
20759 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20762 Note the use of the @ref{unsharp} filter which is always recommended.
20765 Zoom in a bit more and load transform data from a given file:
20767 vidstabtransform=zoom=5:input="mytransforms.trf"
20771 Smoothen the video even more:
20773 vidstabtransform=smoothing=30
20779 Flip the input video vertically.
20781 For example, to vertically flip a video with @command{ffmpeg}:
20783 ffmpeg -i in.avi -vf "vflip" out.avi
20788 Detect variable frame rate video.
20790 This filter tries to detect if the input is variable or constant frame rate.
20792 At end it will output number of frames detected as having variable delta pts,
20793 and ones with constant delta pts.
20794 If there was frames with variable delta, than it will also show min, max and
20795 average delta encountered.
20799 Boost or alter saturation.
20801 The filter accepts the following options:
20804 Set strength of boost if positive value or strength of alter if negative value.
20805 Default is 0. Allowed range is from -2 to 2.
20808 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20811 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20814 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20817 Set the red luma coefficient.
20820 Set the green luma coefficient.
20823 Set the blue luma coefficient.
20826 If @code{intensity} is negative and this is set to 1, colors will change,
20827 otherwise colors will be less saturated, more towards gray.
20830 @subsection Commands
20832 This filter supports the all above options as @ref{commands}.
20837 Make or reverse a natural vignetting effect.
20839 The filter accepts the following options:
20843 Set lens angle expression as a number of radians.
20845 The value is clipped in the @code{[0,PI/2]} range.
20847 Default value: @code{"PI/5"}
20851 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20855 Set forward/backward mode.
20857 Available modes are:
20860 The larger the distance from the central point, the darker the image becomes.
20863 The larger the distance from the central point, the brighter the image becomes.
20864 This can be used to reverse a vignette effect, though there is no automatic
20865 detection to extract the lens @option{angle} and other settings (yet). It can
20866 also be used to create a burning effect.
20869 Default value is @samp{forward}.
20872 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20874 It accepts the following values:
20877 Evaluate expressions only once during the filter initialization.
20880 Evaluate expressions for each incoming frame. This is way slower than the
20881 @samp{init} mode since it requires all the scalers to be re-computed, but it
20882 allows advanced dynamic expressions.
20885 Default value is @samp{init}.
20888 Set dithering to reduce the circular banding effects. Default is @code{1}
20892 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20893 Setting this value to the SAR of the input will make a rectangular vignetting
20894 following the dimensions of the video.
20896 Default is @code{1/1}.
20899 @subsection Expressions
20901 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20902 following parameters.
20907 input width and height
20910 the number of input frame, starting from 0
20913 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20914 @var{TB} units, NAN if undefined
20917 frame rate of the input video, NAN if the input frame rate is unknown
20920 the PTS (Presentation TimeStamp) of the filtered video frame,
20921 expressed in seconds, NAN if undefined
20924 time base of the input video
20928 @subsection Examples
20932 Apply simple strong vignetting effect:
20938 Make a flickering vignetting:
20940 vignette='PI/4+random(1)*PI/50':eval=frame
20945 @section vmafmotion
20947 Obtain the average VMAF motion score of a video.
20948 It is one of the component metrics of VMAF.
20950 The obtained average motion score is printed through the logging system.
20952 The filter accepts the following options:
20956 If specified, the filter will use the named file to save the motion score of
20957 each frame with respect to the previous frame.
20958 When filename equals "-" the data is sent to standard output.
20963 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20967 Stack input videos vertically.
20969 All streams must be of same pixel format and of same width.
20971 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20972 to create same output.
20974 The filter accepts the following options:
20978 Set number of input streams. Default is 2.
20981 If set to 1, force the output to terminate when the shortest input
20982 terminates. Default value is 0.
20987 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20988 Deinterlacing Filter").
20990 Based on the process described by Martin Weston for BBC R&D, and
20991 implemented based on the de-interlace algorithm written by Jim
20992 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20993 uses filter coefficients calculated by BBC R&D.
20995 This filter uses field-dominance information in frame to decide which
20996 of each pair of fields to place first in the output.
20997 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20999 There are two sets of filter coefficients, so called "simple"
21000 and "complex". Which set of filter coefficients is used can
21001 be set by passing an optional parameter:
21005 Set the interlacing filter coefficients. Accepts one of the following values:
21009 Simple filter coefficient set.
21011 More-complex filter coefficient set.
21013 Default value is @samp{complex}.
21016 The interlacing mode to adopt. It accepts one of the following values:
21020 Output one frame for each frame.
21022 Output one frame for each field.
21025 The default value is @code{field}.
21028 The picture field parity assumed for the input interlaced video. It accepts one
21029 of the following values:
21033 Assume the top field is first.
21035 Assume the bottom field is first.
21037 Enable automatic detection of field parity.
21040 The default value is @code{auto}.
21041 If the interlacing is unknown or the decoder does not export this information,
21042 top field first will be assumed.
21045 Specify which frames to deinterlace. Accepts one of the following values:
21049 Deinterlace all frames,
21051 Only deinterlace frames marked as interlaced.
21054 Default value is @samp{all}.
21057 @subsection Commands
21058 This filter supports same @ref{commands} as options.
21061 Video waveform monitor.
21063 The waveform monitor plots color component intensity. By default luminance
21064 only. Each column of the waveform corresponds to a column of pixels in the
21067 It accepts the following options:
21071 Can be either @code{row}, or @code{column}. Default is @code{column}.
21072 In row mode, the graph on the left side represents color component value 0 and
21073 the right side represents value = 255. In column mode, the top side represents
21074 color component value = 0 and bottom side represents value = 255.
21077 Set intensity. Smaller values are useful to find out how many values of the same
21078 luminance are distributed across input rows/columns.
21079 Default value is @code{0.04}. Allowed range is [0, 1].
21082 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
21083 In mirrored mode, higher values will be represented on the left
21084 side for @code{row} mode and at the top for @code{column} mode. Default is
21085 @code{1} (mirrored).
21089 It accepts the following values:
21092 Presents information identical to that in the @code{parade}, except
21093 that the graphs representing color components are superimposed directly
21096 This display mode makes it easier to spot relative differences or similarities
21097 in overlapping areas of the color components that are supposed to be identical,
21098 such as neutral whites, grays, or blacks.
21101 Display separate graph for the color components side by side in
21102 @code{row} mode or one below the other in @code{column} mode.
21105 Display separate graph for the color components side by side in
21106 @code{column} mode or one below the other in @code{row} mode.
21108 Using this display mode makes it easy to spot color casts in the highlights
21109 and shadows of an image, by comparing the contours of the top and the bottom
21110 graphs of each waveform. Since whites, grays, and blacks are characterized
21111 by exactly equal amounts of red, green, and blue, neutral areas of the picture
21112 should display three waveforms of roughly equal width/height. If not, the
21113 correction is easy to perform by making level adjustments the three waveforms.
21115 Default is @code{stack}.
21117 @item components, c
21118 Set which color components to display. Default is 1, which means only luminance
21119 or red color component if input is in RGB colorspace. If is set for example to
21120 7 it will display all 3 (if) available color components.
21125 No envelope, this is default.
21128 Instant envelope, minimum and maximum values presented in graph will be easily
21129 visible even with small @code{step} value.
21132 Hold minimum and maximum values presented in graph across time. This way you
21133 can still spot out of range values without constantly looking at waveforms.
21136 Peak and instant envelope combined together.
21142 No filtering, this is default.
21145 Luma and chroma combined together.
21148 Similar as above, but shows difference between blue and red chroma.
21151 Similar as above, but use different colors.
21154 Similar as above, but again with different colors.
21157 Displays only chroma.
21160 Displays actual color value on waveform.
21163 Similar as above, but with luma showing frequency of chroma values.
21167 Set which graticule to display.
21171 Do not display graticule.
21174 Display green graticule showing legal broadcast ranges.
21177 Display orange graticule showing legal broadcast ranges.
21180 Display invert graticule showing legal broadcast ranges.
21184 Set graticule opacity.
21187 Set graticule flags.
21191 Draw numbers above lines. By default enabled.
21194 Draw dots instead of lines.
21198 Set scale used for displaying graticule.
21205 Default is digital.
21208 Set background opacity.
21212 Set tint for output.
21213 Only used with lowpass filter and when display is not overlay and input
21214 pixel formats are not RGB.
21217 @section weave, doubleweave
21219 The @code{weave} takes a field-based video input and join
21220 each two sequential fields into single frame, producing a new double
21221 height clip with half the frame rate and half the frame count.
21223 The @code{doubleweave} works same as @code{weave} but without
21224 halving frame rate and frame count.
21226 It accepts the following option:
21230 Set first field. Available values are:
21234 Set the frame as top-field-first.
21237 Set the frame as bottom-field-first.
21241 @subsection Examples
21245 Interlace video using @ref{select} and @ref{separatefields} filter:
21247 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21252 Apply the xBR high-quality magnification filter which is designed for pixel
21253 art. It follows a set of edge-detection rules, see
21254 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
21256 It accepts the following option:
21260 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
21261 @code{3xBR} and @code{4} for @code{4xBR}.
21262 Default is @code{3}.
21267 Apply cross fade from one input video stream to another input video stream.
21268 The cross fade is applied for specified duration.
21270 The filter accepts the following options:
21274 Set one of available transition effects:
21322 Default transition effect is fade.
21325 Set cross fade duration in seconds.
21326 Default duration is 1 second.
21329 Set cross fade start relative to first input stream in seconds.
21330 Default offset is 0.
21333 Set expression for custom transition effect.
21335 The expressions can use the following variables and functions:
21340 The coordinates of the current sample.
21344 The width and height of the image.
21347 Progress of transition effect.
21350 Currently processed plane.
21353 Return value of first input at current location and plane.
21356 Return value of second input at current location and plane.
21362 Return the value of the pixel at location (@var{x},@var{y}) of the
21363 first/second/third/fourth component of first input.
21369 Return the value of the pixel at location (@var{x},@var{y}) of the
21370 first/second/third/fourth component of second input.
21374 @subsection Examples
21378 Cross fade from one input video to another input video, with fade transition and duration of transition
21379 of 2 seconds starting at offset of 5 seconds:
21381 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21386 Pick median pixels from several input videos.
21388 The filter accepts the following options:
21392 Set number of inputs.
21393 Default is 3. Allowed range is from 3 to 255.
21394 If number of inputs is even number, than result will be mean value between two median values.
21397 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
21400 Set median percentile. Default value is @code{0.5}.
21401 Default value of @code{0.5} will pick always median values, while @code{0} will pick
21402 minimum values, and @code{1} maximum values.
21405 @subsection Commands
21407 This filter supports all above options as @ref{commands}, excluding option @code{inputs}.
21410 Stack video inputs into custom layout.
21412 All streams must be of same pixel format.
21414 The filter accepts the following options:
21418 Set number of input streams. Default is 2.
21421 Specify layout of inputs.
21422 This option requires the desired layout configuration to be explicitly set by the user.
21423 This sets position of each video input in output. Each input
21424 is separated by '|'.
21425 The first number represents the column, and the second number represents the row.
21426 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
21427 where X is video input from which to take width or height.
21428 Multiple values can be used when separated by '+'. In such
21429 case values are summed together.
21431 Note that if inputs are of different sizes gaps may appear, as not all of
21432 the output video frame will be filled. Similarly, videos can overlap each
21433 other if their position doesn't leave enough space for the full frame of
21436 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
21437 a layout must be set by the user.
21440 If set to 1, force the output to terminate when the shortest input
21441 terminates. Default value is 0.
21444 If set to valid color, all unused pixels will be filled with that color.
21445 By default fill is set to none, so it is disabled.
21448 @subsection Examples
21452 Display 4 inputs into 2x2 grid.
21456 input1(0, 0) | input3(w0, 0)
21457 input2(0, h0) | input4(w0, h0)
21461 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21464 Note that if inputs are of different sizes, gaps or overlaps may occur.
21467 Display 4 inputs into 1x4 grid.
21474 input4(0, h0+h1+h2)
21478 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21481 Note that if inputs are of different widths, unused space will appear.
21484 Display 9 inputs into 3x3 grid.
21488 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21489 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21490 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21494 xstack=inputs=9:layout=0_0|0_h0|0_h0+h1|w0_0|w0_h0|w0_h0+h1|w0+w3_0|w0+w3_h0|w0+w3_h0+h1
21497 Note that if inputs are of different sizes, gaps or overlaps may occur.
21500 Display 16 inputs into 4x4 grid.
21504 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21505 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21506 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21507 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21511 xstack=inputs=16:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2|w0_0|w0_h0|w0_h0+h1|w0_h0+h1+h2|w0+w4_0|
21512 w0+w4_h0|w0+w4_h0+h1|w0+w4_h0+h1+h2|w0+w4+w8_0|w0+w4+w8_h0|w0+w4+w8_h0+h1|w0+w4+w8_h0+h1+h2
21515 Note that if inputs are of different sizes, gaps or overlaps may occur.
21522 Deinterlace the input video ("yadif" means "yet another deinterlacing
21525 It accepts the following parameters:
21531 The interlacing mode to adopt. It accepts one of the following values:
21534 @item 0, send_frame
21535 Output one frame for each frame.
21536 @item 1, send_field
21537 Output one frame for each field.
21538 @item 2, send_frame_nospatial
21539 Like @code{send_frame}, but it skips the spatial interlacing check.
21540 @item 3, send_field_nospatial
21541 Like @code{send_field}, but it skips the spatial interlacing check.
21544 The default value is @code{send_frame}.
21547 The picture field parity assumed for the input interlaced video. It accepts one
21548 of the following values:
21552 Assume the top field is first.
21554 Assume the bottom field is first.
21556 Enable automatic detection of field parity.
21559 The default value is @code{auto}.
21560 If the interlacing is unknown or the decoder does not export this information,
21561 top field first will be assumed.
21564 Specify which frames to deinterlace. Accepts one of the following
21569 Deinterlace all frames.
21570 @item 1, interlaced
21571 Only deinterlace frames marked as interlaced.
21574 The default value is @code{all}.
21577 @section yadif_cuda
21579 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21580 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21583 It accepts the following parameters:
21589 The interlacing mode to adopt. It accepts one of the following values:
21592 @item 0, send_frame
21593 Output one frame for each frame.
21594 @item 1, send_field
21595 Output one frame for each field.
21596 @item 2, send_frame_nospatial
21597 Like @code{send_frame}, but it skips the spatial interlacing check.
21598 @item 3, send_field_nospatial
21599 Like @code{send_field}, but it skips the spatial interlacing check.
21602 The default value is @code{send_frame}.
21605 The picture field parity assumed for the input interlaced video. It accepts one
21606 of the following values:
21610 Assume the top field is first.
21612 Assume the bottom field is first.
21614 Enable automatic detection of field parity.
21617 The default value is @code{auto}.
21618 If the interlacing is unknown or the decoder does not export this information,
21619 top field first will be assumed.
21622 Specify which frames to deinterlace. Accepts one of the following
21627 Deinterlace all frames.
21628 @item 1, interlaced
21629 Only deinterlace frames marked as interlaced.
21632 The default value is @code{all}.
21637 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21638 The algorithm is described in
21639 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21641 It accepts the following parameters:
21645 Set the window radius. Default value is 3.
21648 Set which planes to filter. Default is only the first plane.
21651 Set blur strength. Default value is 128.
21654 @subsection Commands
21655 This filter supports same @ref{commands} as options.
21659 Apply Zoom & Pan effect.
21661 This filter accepts the following options:
21665 Set the zoom expression. Range is 1-10. Default is 1.
21669 Set the x and y expression. Default is 0.
21672 Set the duration expression in number of frames.
21673 This sets for how many number of frames effect will last for
21674 single input image.
21677 Set the output image size, default is 'hd720'.
21680 Set the output frame rate, default is '25'.
21683 Each expression can contain the following constants:
21702 Output frame count.
21705 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21707 @item out_time, time, ot
21708 The output timestamp expressed in seconds.
21712 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21713 for current input frame.
21717 'x' and 'y' of last output frame of previous input frame or 0 when there was
21718 not yet such frame (first input frame).
21721 Last calculated zoom from 'z' expression for current input frame.
21724 Last calculated zoom of last output frame of previous input frame.
21727 Number of output frames for current input frame. Calculated from 'd' expression
21728 for each input frame.
21731 number of output frames created for previous input frame
21734 Rational number: input width / input height
21737 sample aspect ratio
21740 display aspect ratio
21744 @subsection Examples
21748 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21750 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='if(gte(zoom,1.5),x,x+1/a)':y='if(gte(zoom,1.5),y,y+1)':s=640x360
21754 Zoom in up to 1.5x and pan always at center of picture:
21756 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21760 Same as above but without pausing:
21762 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21766 Zoom in 2x into center of picture only for the first second of the input video:
21768 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21775 Scale (resize) the input video, using the z.lib library:
21776 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21777 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21779 The zscale filter forces the output display aspect ratio to be the same
21780 as the input, by changing the output sample aspect ratio.
21782 If the input image format is different from the format requested by
21783 the next filter, the zscale filter will convert the input to the
21786 @subsection Options
21787 The filter accepts the following options.
21792 Set the output video dimension expression. Default value is the input
21795 If the @var{width} or @var{w} value is 0, the input width is used for
21796 the output. If the @var{height} or @var{h} value is 0, the input height
21797 is used for the output.
21799 If one and only one of the values is -n with n >= 1, the zscale filter
21800 will use a value that maintains the aspect ratio of the input image,
21801 calculated from the other specified dimension. After that it will,
21802 however, make sure that the calculated dimension is divisible by n and
21803 adjust the value if necessary.
21805 If both values are -n with n >= 1, the behavior will be identical to
21806 both values being set to 0 as previously detailed.
21808 See below for the list of accepted constants for use in the dimension
21812 Set the video size. For the syntax of this option, check the
21813 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21816 Set the dither type.
21818 Possible values are:
21823 @item error_diffusion
21829 Set the resize filter type.
21831 Possible values are:
21841 Default is bilinear.
21844 Set the color range.
21846 Possible values are:
21853 Default is same as input.
21856 Set the color primaries.
21858 Possible values are:
21868 Default is same as input.
21871 Set the transfer characteristics.
21873 Possible values are:
21887 Default is same as input.
21890 Set the colorspace matrix.
21892 Possible value are:
21903 Default is same as input.
21906 Set the input color range.
21908 Possible values are:
21915 Default is same as input.
21917 @item primariesin, pin
21918 Set the input color primaries.
21920 Possible values are:
21930 Default is same as input.
21932 @item transferin, tin
21933 Set the input transfer characteristics.
21935 Possible values are:
21946 Default is same as input.
21948 @item matrixin, min
21949 Set the input colorspace matrix.
21951 Possible value are:
21963 Set the output chroma location.
21965 Possible values are:
21976 @item chromalin, cin
21977 Set the input chroma location.
21979 Possible values are:
21991 Set the nominal peak luminance.
21994 The values of the @option{w} and @option{h} options are expressions
21995 containing the following constants:
22000 The input width and height
22004 These are the same as @var{in_w} and @var{in_h}.
22008 The output (scaled) width and height
22012 These are the same as @var{out_w} and @var{out_h}
22015 The same as @var{iw} / @var{ih}
22018 input sample aspect ratio
22021 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
22025 horizontal and vertical input chroma subsample values. For example for the
22026 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
22030 horizontal and vertical output chroma subsample values. For example for the
22031 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
22034 @subsection Commands
22036 This filter supports the following commands:
22040 Set the output video dimension expression.
22041 The command accepts the same syntax of the corresponding option.
22043 If the specified expression is not valid, it is kept at its current
22047 @c man end VIDEO FILTERS
22049 @chapter OpenCL Video Filters
22050 @c man begin OPENCL VIDEO FILTERS
22052 Below is a description of the currently available OpenCL video filters.
22054 To enable compilation of these filters you need to configure FFmpeg with
22055 @code{--enable-opencl}.
22057 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
22060 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
22061 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
22062 given device parameters.
22064 @item -filter_hw_device @var{name}
22065 Pass the hardware device called @var{name} to all filters in any filter graph.
22069 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
22073 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
22075 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
22079 Since OpenCL filters are not able to access frame data in normal memory, all frame data needs to be uploaded(@ref{hwupload}) to hardware surfaces connected to the appropriate device before being used and then downloaded(@ref{hwdownload}) back to normal memory. Note that @ref{hwupload} will upload to a surface with the same layout as the software frame, so it may be necessary to add a @ref{format} filter immediately before to get the input into the right format and @ref{hwdownload} does not support all formats on the output - it may be necessary to insert an additional @ref{format} filter immediately following in the graph to get the output in a supported format.
22081 @section avgblur_opencl
22083 Apply average blur filter.
22085 The filter accepts the following options:
22089 Set horizontal radius size.
22090 Range is @code{[1, 1024]} and default value is @code{1}.
22093 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22096 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
22099 @subsection Example
22103 Apply average blur filter with horizontal and vertical size of 3, setting each pixel of the output to the average value of the 7x7 region centered on it in the input. For pixels on the edges of the image, the region does not extend beyond the image boundaries, and so out-of-range coordinates are not used in the calculations.
22105 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
22109 @section boxblur_opencl
22111 Apply a boxblur algorithm to the input video.
22113 It accepts the following parameters:
22117 @item luma_radius, lr
22118 @item luma_power, lp
22119 @item chroma_radius, cr
22120 @item chroma_power, cp
22121 @item alpha_radius, ar
22122 @item alpha_power, ap
22126 A description of the accepted options follows.
22129 @item luma_radius, lr
22130 @item chroma_radius, cr
22131 @item alpha_radius, ar
22132 Set an expression for the box radius in pixels used for blurring the
22133 corresponding input plane.
22135 The radius value must be a non-negative number, and must not be
22136 greater than the value of the expression @code{min(w,h)/2} for the
22137 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
22140 Default value for @option{luma_radius} is "2". If not specified,
22141 @option{chroma_radius} and @option{alpha_radius} default to the
22142 corresponding value set for @option{luma_radius}.
22144 The expressions can contain the following constants:
22148 The input width and height in pixels.
22152 The input chroma image width and height in pixels.
22156 The horizontal and vertical chroma subsample values. For example, for the
22157 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
22160 @item luma_power, lp
22161 @item chroma_power, cp
22162 @item alpha_power, ap
22163 Specify how many times the boxblur filter is applied to the
22164 corresponding plane.
22166 Default value for @option{luma_power} is 2. If not specified,
22167 @option{chroma_power} and @option{alpha_power} default to the
22168 corresponding value set for @option{luma_power}.
22170 A value of 0 will disable the effect.
22173 @subsection Examples
22175 Apply boxblur filter, setting each pixel of the output to the average value of box-radiuses @var{luma_radius}, @var{chroma_radius}, @var{alpha_radius} for each plane respectively. The filter will apply @var{luma_power}, @var{chroma_power}, @var{alpha_power} times onto the corresponding plane. For pixels on the edges of the image, the radius does not extend beyond the image boundaries, and so out-of-range coordinates are not used in the calculations.
22179 Apply a boxblur filter with the luma, chroma, and alpha radius
22180 set to 2 and luma, chroma, and alpha power set to 3. The filter will run 3 times with box-radius set to 2 for every plane of the image.
22182 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
22183 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
22187 Apply a boxblur filter with luma radius set to 2, luma_power to 1, chroma_radius to 4, chroma_power to 5, alpha_radius to 3 and alpha_power to 7.
22189 For the luma plane, a 2x2 box radius will be run once.
22191 For the chroma plane, a 4x4 box radius will be run 5 times.
22193 For the alpha plane, a 3x3 box radius will be run 7 times.
22195 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
22199 @section colorkey_opencl
22200 RGB colorspace color keying.
22202 The filter accepts the following options:
22206 The color which will be replaced with transparency.
22209 Similarity percentage with the key color.
22211 0.01 matches only the exact key color, while 1.0 matches everything.
22216 0.0 makes pixels either fully transparent, or not transparent at all.
22218 Higher values result in semi-transparent pixels, with a higher transparency
22219 the more similar the pixels color is to the key color.
22222 @subsection Examples
22226 Make every semi-green pixel in the input transparent with some slight blending:
22228 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22232 @section convolution_opencl
22234 Apply convolution of 3x3, 5x5, 7x7 matrix.
22236 The filter accepts the following options:
22243 Set matrix for each plane.
22244 Matrix is sequence of 9, 25 or 49 signed numbers.
22245 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
22251 Set multiplier for calculated value for each plane.
22252 If unset or 0, it will be sum of all matrix elements.
22253 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
22259 Set bias for each plane. This value is added to the result of the multiplication.
22260 Useful for making the overall image brighter or darker.
22261 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
22265 @subsection Examples
22271 -i INPUT -vf "hwupload, convolution_opencl=0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0, hwdownload" OUTPUT
22277 -i INPUT -vf "hwupload, convolution_opencl=1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1 1 1 1 1 1 1 1 1:1/9:1/9:1/9:1/9, hwdownload" OUTPUT
22281 Apply edge enhance:
22283 -i INPUT -vf "hwupload, convolution_opencl=0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:0 0 0 -1 1 0 0 0 0:5:1:1:1:0:128:128:128, hwdownload" OUTPUT
22289 -i INPUT -vf "hwupload, convolution_opencl=0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:0 1 0 1 -4 1 0 1 0:5:5:5:1:0:128:128:128, hwdownload" OUTPUT
22293 Apply laplacian edge detector which includes diagonals:
22295 -i INPUT -vf "hwupload, convolution_opencl=1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:1 1 1 1 -8 1 1 1 1:5:5:5:1:0:128:128:0, hwdownload" OUTPUT
22301 -i INPUT -vf "hwupload, convolution_opencl=-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2:-2 -1 0 -1 1 1 0 1 2, hwdownload" OUTPUT
22305 @section erosion_opencl
22307 Apply erosion effect to the video.
22309 This filter replaces the pixel by the local(3x3) minimum.
22311 It accepts the following options:
22318 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22319 If @code{0}, plane will remain unchanged.
22322 Flag which specifies the pixel to refer to.
22323 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22325 Flags to local 3x3 coordinates region centered on @code{x}:
22334 @subsection Example
22338 Apply erosion filter with threshold0 set to 30, threshold1 set 40, threshold2 set to 50 and coordinates set to 231, setting each pixel of the output to the local minimum between pixels: 1, 2, 3, 6, 7, 8 of the 3x3 region centered on it in the input. If the difference between input pixel and local minimum is more then threshold of the corresponding plane, output pixel will be set to input pixel - threshold of corresponding plane.
22340 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22344 @section deshake_opencl
22345 Feature-point based video stabilization filter.
22347 The filter accepts the following options:
22351 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
22354 Whether or not additional debug info should be displayed, both in the processed output and in the console.
22356 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
22358 Viewing point matches in the output video is only supported for RGB input.
22360 Defaults to @code{0}.
22362 @item adaptive_crop
22363 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
22365 Defaults to @code{1}.
22367 @item refine_features
22368 Whether or not feature points should be refined at a sub-pixel level.
22370 This can be turned off for a slight performance gain at the cost of precision.
22372 Defaults to @code{1}.
22374 @item smooth_strength
22375 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
22377 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
22379 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
22381 Defaults to @code{0.0}.
22383 @item smooth_window_multiplier
22384 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
22386 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
22388 Acceptable values range from @code{0.1} to @code{10.0}.
22390 Larger values increase the amount of motion data available for determining how to smooth the camera path,
22391 potentially improving smoothness, but also increase latency and memory usage.
22393 Defaults to @code{2.0}.
22397 @subsection Examples
22401 Stabilize a video with a fixed, medium smoothing strength:
22403 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22407 Stabilize a video with debugging (both in console and in rendered video):
22409 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22413 @section dilation_opencl
22415 Apply dilation effect to the video.
22417 This filter replaces the pixel by the local(3x3) maximum.
22419 It accepts the following options:
22426 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22427 If @code{0}, plane will remain unchanged.
22430 Flag which specifies the pixel to refer to.
22431 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22433 Flags to local 3x3 coordinates region centered on @code{x}:
22442 @subsection Example
22446 Apply dilation filter with threshold0 set to 30, threshold1 set 40, threshold2 set to 50 and coordinates set to 231, setting each pixel of the output to the local maximum between pixels: 1, 2, 3, 6, 7, 8 of the 3x3 region centered on it in the input. If the difference between input pixel and local maximum is more then threshold of the corresponding plane, output pixel will be set to input pixel + threshold of corresponding plane.
22448 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22452 @section nlmeans_opencl
22454 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
22456 @section overlay_opencl
22458 Overlay one video on top of another.
22460 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
22461 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
22463 The filter accepts the following options:
22468 Set the x coordinate of the overlaid video on the main video.
22469 Default value is @code{0}.
22472 Set the y coordinate of the overlaid video on the main video.
22473 Default value is @code{0}.
22477 @subsection Examples
22481 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
22483 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22486 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
22488 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22493 @section pad_opencl
22495 Add paddings to the input image, and place the original input at the
22496 provided @var{x}, @var{y} coordinates.
22498 It accepts the following options:
22503 Specify an expression for the size of the output image with the
22504 paddings added. If the value for @var{width} or @var{height} is 0, the
22505 corresponding input size is used for the output.
22507 The @var{width} expression can reference the value set by the
22508 @var{height} expression, and vice versa.
22510 The default value of @var{width} and @var{height} is 0.
22514 Specify the offsets to place the input image at within the padded area,
22515 with respect to the top/left border of the output image.
22517 The @var{x} expression can reference the value set by the @var{y}
22518 expression, and vice versa.
22520 The default value of @var{x} and @var{y} is 0.
22522 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22523 so the input image is centered on the padded area.
22526 Specify the color of the padded area. For the syntax of this option,
22527 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22528 manual,ffmpeg-utils}.
22531 Pad to an aspect instead to a resolution.
22534 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22535 options are expressions containing the following constants:
22540 The input video width and height.
22544 These are the same as @var{in_w} and @var{in_h}.
22548 The output width and height (the size of the padded area), as
22549 specified by the @var{width} and @var{height} expressions.
22553 These are the same as @var{out_w} and @var{out_h}.
22557 The x and y offsets as specified by the @var{x} and @var{y}
22558 expressions, or NAN if not yet specified.
22561 same as @var{iw} / @var{ih}
22564 input sample aspect ratio
22567 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22570 @section prewitt_opencl
22572 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22574 The filter accepts the following option:
22578 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22581 Set value which will be multiplied with filtered result.
22582 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22585 Set value which will be added to filtered result.
22586 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22589 @subsection Example
22593 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22595 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22599 @anchor{program_opencl}
22600 @section program_opencl
22602 Filter video using an OpenCL program.
22607 OpenCL program source file.
22610 Kernel name in program.
22613 Number of inputs to the filter. Defaults to 1.
22616 Size of output frames. Defaults to the same as the first input.
22620 The @code{program_opencl} filter also supports the @ref{framesync} options.
22622 The program source file must contain a kernel function with the given name,
22623 which will be run once for each plane of the output. Each run on a plane
22624 gets enqueued as a separate 2D global NDRange with one work-item for each
22625 pixel to be generated. The global ID offset for each work-item is therefore
22626 the coordinates of a pixel in the destination image.
22628 The kernel function needs to take the following arguments:
22631 Destination image, @var{__write_only image2d_t}.
22633 This image will become the output; the kernel should write all of it.
22635 Frame index, @var{unsigned int}.
22637 This is a counter starting from zero and increasing by one for each frame.
22639 Source images, @var{__read_only image2d_t}.
22641 These are the most recent images on each input. The kernel may read from
22642 them to generate the output, but they can't be written to.
22649 Copy the input to the output (output must be the same size as the input).
22651 __kernel void copy(__write_only image2d_t destination,
22652 unsigned int index,
22653 __read_only image2d_t source)
22655 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22657 int2 location = (int2)(get_global_id(0), get_global_id(1));
22659 float4 value = read_imagef(source, sampler, location);
22661 write_imagef(destination, location, value);
22666 Apply a simple transformation, rotating the input by an amount increasing
22667 with the index counter. Pixel values are linearly interpolated by the
22668 sampler, and the output need not have the same dimensions as the input.
22670 __kernel void rotate_image(__write_only image2d_t dst,
22671 unsigned int index,
22672 __read_only image2d_t src)
22674 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22675 CLK_FILTER_LINEAR);
22677 float angle = (float)index / 100.0f;
22679 float2 dst_dim = convert_float2(get_image_dim(dst));
22680 float2 src_dim = convert_float2(get_image_dim(src));
22682 float2 dst_cen = dst_dim / 2.0f;
22683 float2 src_cen = src_dim / 2.0f;
22685 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22687 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22689 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22690 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22692 src_pos = src_pos * src_dim / dst_dim;
22694 float2 src_loc = src_pos + src_cen;
22696 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22697 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22698 write_imagef(dst, dst_loc, 0.5f);
22700 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22705 Blend two inputs together, with the amount of each input used varying
22706 with the index counter.
22708 __kernel void blend_images(__write_only image2d_t dst,
22709 unsigned int index,
22710 __read_only image2d_t src1,
22711 __read_only image2d_t src2)
22713 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22714 CLK_FILTER_LINEAR);
22716 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22718 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22719 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22720 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22722 float4 val1 = read_imagef(src1, sampler, src1_loc);
22723 float4 val2 = read_imagef(src2, sampler, src2_loc);
22725 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22731 @section roberts_opencl
22732 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22734 The filter accepts the following option:
22738 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22741 Set value which will be multiplied with filtered result.
22742 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22745 Set value which will be added to filtered result.
22746 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22749 @subsection Example
22753 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22755 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22759 @section sobel_opencl
22761 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22763 The filter accepts the following option:
22767 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22770 Set value which will be multiplied with filtered result.
22771 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22774 Set value which will be added to filtered result.
22775 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22778 @subsection Example
22782 Apply sobel operator with scale set to 2 and delta set to 10
22784 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22788 @section tonemap_opencl
22790 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22792 It accepts the following parameters:
22796 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22799 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22802 Apply desaturation for highlights that exceed this level of brightness. The
22803 higher the parameter, the more color information will be preserved. This
22804 setting helps prevent unnaturally blown-out colors for super-highlights, by
22805 (smoothly) turning into white instead. This makes images feel more natural,
22806 at the cost of reducing information about out-of-range colors.
22808 The default value is 0.5, and the algorithm here is a little different from
22809 the cpu version tonemap currently. A setting of 0.0 disables this option.
22812 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22813 is used to detect whether the scene has changed or not. If the distance between
22814 the current frame average brightness and the current running average exceeds
22815 a threshold value, we would re-calculate scene average and peak brightness.
22816 The default value is 0.2.
22819 Specify the output pixel format.
22821 Currently supported formats are:
22828 Set the output color range.
22830 Possible values are:
22836 Default is same as input.
22839 Set the output color primaries.
22841 Possible values are:
22847 Default is same as input.
22850 Set the output transfer characteristics.
22852 Possible values are:
22861 Set the output colorspace matrix.
22863 Possible value are:
22869 Default is same as input.
22873 @subsection Example
22877 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22879 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22883 @section unsharp_opencl
22885 Sharpen or blur the input video.
22887 It accepts the following parameters:
22890 @item luma_msize_x, lx
22891 Set the luma matrix horizontal size.
22892 Range is @code{[1, 23]} and default value is @code{5}.
22894 @item luma_msize_y, ly
22895 Set the luma matrix vertical size.
22896 Range is @code{[1, 23]} and default value is @code{5}.
22898 @item luma_amount, la
22899 Set the luma effect strength.
22900 Range is @code{[-10, 10]} and default value is @code{1.0}.
22902 Negative values will blur the input video, while positive values will
22903 sharpen it, a value of zero will disable the effect.
22905 @item chroma_msize_x, cx
22906 Set the chroma matrix horizontal size.
22907 Range is @code{[1, 23]} and default value is @code{5}.
22909 @item chroma_msize_y, cy
22910 Set the chroma matrix vertical size.
22911 Range is @code{[1, 23]} and default value is @code{5}.
22913 @item chroma_amount, ca
22914 Set the chroma effect strength.
22915 Range is @code{[-10, 10]} and default value is @code{0.0}.
22917 Negative values will blur the input video, while positive values will
22918 sharpen it, a value of zero will disable the effect.
22922 All parameters are optional and default to the equivalent of the
22923 string '5:5:1.0:5:5:0.0'.
22925 @subsection Examples
22929 Apply strong luma sharpen effect:
22931 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22935 Apply a strong blur of both luma and chroma parameters:
22937 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22941 @section xfade_opencl
22943 Cross fade two videos with custom transition effect by using OpenCL.
22945 It accepts the following options:
22949 Set one of possible transition effects.
22953 Select custom transition effect, the actual transition description
22954 will be picked from source and kernel options.
22966 Default transition is fade.
22970 OpenCL program source file for custom transition.
22973 Set name of kernel to use for custom transition from program source file.
22976 Set duration of video transition.
22979 Set time of start of transition relative to first video.
22982 The program source file must contain a kernel function with the given name,
22983 which will be run once for each plane of the output. Each run on a plane
22984 gets enqueued as a separate 2D global NDRange with one work-item for each
22985 pixel to be generated. The global ID offset for each work-item is therefore
22986 the coordinates of a pixel in the destination image.
22988 The kernel function needs to take the following arguments:
22991 Destination image, @var{__write_only image2d_t}.
22993 This image will become the output; the kernel should write all of it.
22996 First Source image, @var{__read_only image2d_t}.
22997 Second Source image, @var{__read_only image2d_t}.
22999 These are the most recent images on each input. The kernel may read from
23000 them to generate the output, but they can't be written to.
23003 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
23010 Apply dots curtain transition effect:
23012 __kernel void blend_images(__write_only image2d_t dst,
23013 __read_only image2d_t src1,
23014 __read_only image2d_t src2,
23017 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
23018 CLK_FILTER_LINEAR);
23019 int2 p = (int2)(get_global_id(0), get_global_id(1));
23020 float2 rp = (float2)(get_global_id(0), get_global_id(1));
23021 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
23024 float2 dots = (float2)(20.0, 20.0);
23025 float2 center = (float2)(0,0);
23028 float4 val1 = read_imagef(src1, sampler, p);
23029 float4 val2 = read_imagef(src2, sampler, p);
23030 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
23032 write_imagef(dst, p, next ? val1 : val2);
23038 @c man end OPENCL VIDEO FILTERS
23040 @chapter VAAPI Video Filters
23041 @c man begin VAAPI VIDEO FILTERS
23043 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
23045 To enable compilation of these filters you need to configure FFmpeg with
23046 @code{--enable-vaapi}.
23048 To use vaapi filters, you need to setup the vaapi device correctly. For more information, please read @url{https://trac.ffmpeg.org/wiki/Hardware/VAAPI}
23050 @section tonemap_vaapi
23052 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
23053 It maps the dynamic range of HDR10 content to the SDR content.
23054 It currently only accepts HDR10 as input.
23056 It accepts the following parameters:
23060 Specify the output pixel format.
23062 Currently supported formats are:
23071 Set the output color primaries.
23073 Default is same as input.
23076 Set the output transfer characteristics.
23081 Set the output colorspace matrix.
23083 Default is same as input.
23087 @subsection Example
23091 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
23093 tonemap_vaapi=format=p010:t=bt2020-10
23097 @c man end VAAPI VIDEO FILTERS
23099 @chapter Video Sources
23100 @c man begin VIDEO SOURCES
23102 Below is a description of the currently available video sources.
23106 Buffer video frames, and make them available to the filter chain.
23108 This source is mainly intended for a programmatic use, in particular
23109 through the interface defined in @file{libavfilter/buffersrc.h}.
23111 It accepts the following parameters:
23116 Specify the size (width and height) of the buffered video frames. For the
23117 syntax of this option, check the
23118 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23121 The input video width.
23124 The input video height.
23127 A string representing the pixel format of the buffered video frames.
23128 It may be a number corresponding to a pixel format, or a pixel format
23132 Specify the timebase assumed by the timestamps of the buffered frames.
23135 Specify the frame rate expected for the video stream.
23137 @item pixel_aspect, sar
23138 The sample (pixel) aspect ratio of the input video.
23141 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
23142 to the filtergraph description to specify swscale flags for automatically
23143 inserted scalers. See @ref{Filtergraph syntax}.
23145 @item hw_frames_ctx
23146 When using a hardware pixel format, this should be a reference to an
23147 AVHWFramesContext describing input frames.
23152 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
23155 will instruct the source to accept video frames with size 320x240 and
23156 with format "yuv410p", assuming 1/24 as the timestamps timebase and
23157 square pixels (1:1 sample aspect ratio).
23158 Since the pixel format with name "yuv410p" corresponds to the number 6
23159 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
23160 this example corresponds to:
23162 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
23165 Alternatively, the options can be specified as a flat string, but this
23166 syntax is deprecated:
23168 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
23172 Create a pattern generated by an elementary cellular automaton.
23174 The initial state of the cellular automaton can be defined through the
23175 @option{filename} and @option{pattern} options. If such options are
23176 not specified an initial state is created randomly.
23178 At each new frame a new row in the video is filled with the result of
23179 the cellular automaton next generation. The behavior when the whole
23180 frame is filled is defined by the @option{scroll} option.
23182 This source accepts the following options:
23186 Read the initial cellular automaton state, i.e. the starting row, from
23187 the specified file.
23188 In the file, each non-whitespace character is considered an alive
23189 cell, a newline will terminate the row, and further characters in the
23190 file will be ignored.
23193 Read the initial cellular automaton state, i.e. the starting row, from
23194 the specified string.
23196 Each non-whitespace character in the string is considered an alive
23197 cell, a newline will terminate the row, and further characters in the
23198 string will be ignored.
23201 Set the video rate, that is the number of frames generated per second.
23204 @item random_fill_ratio, ratio
23205 Set the random fill ratio for the initial cellular automaton row. It
23206 is a floating point number value ranging from 0 to 1, defaults to
23209 This option is ignored when a file or a pattern is specified.
23211 @item random_seed, seed
23212 Set the seed for filling randomly the initial row, must be an integer
23213 included between 0 and UINT32_MAX. If not specified, or if explicitly
23214 set to -1, the filter will try to use a good random seed on a best
23218 Set the cellular automaton rule, it is a number ranging from 0 to 255.
23219 Default value is 110.
23222 Set the size of the output video. For the syntax of this option, check the
23223 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23225 If @option{filename} or @option{pattern} is specified, the size is set
23226 by default to the width of the specified initial state row, and the
23227 height is set to @var{width} * PHI.
23229 If @option{size} is set, it must contain the width of the specified
23230 pattern string, and the specified pattern will be centered in the
23233 If a filename or a pattern string is not specified, the size value
23234 defaults to "320x518" (used for a randomly generated initial state).
23237 If set to 1, scroll the output upward when all the rows in the output
23238 have been already filled. If set to 0, the new generated row will be
23239 written over the top row just after the bottom row is filled.
23242 @item start_full, full
23243 If set to 1, completely fill the output with generated rows before
23244 outputting the first frame.
23245 This is the default behavior, for disabling set the value to 0.
23248 If set to 1, stitch the left and right row edges together.
23249 This is the default behavior, for disabling set the value to 0.
23252 @subsection Examples
23256 Read the initial state from @file{pattern}, and specify an output of
23259 cellauto=f=pattern:s=200x400
23263 Generate a random initial row with a width of 200 cells, with a fill
23266 cellauto=ratio=2/3:s=200x200
23270 Create a pattern generated by rule 18 starting by a single alive cell
23271 centered on an initial row with width 100:
23273 cellauto=p=@@:s=100x400:full=0:rule=18
23277 Specify a more elaborated initial pattern:
23279 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
23284 @anchor{coreimagesrc}
23285 @section coreimagesrc
23286 Video source generated on GPU using Apple's CoreImage API on OSX.
23288 This video source is a specialized version of the @ref{coreimage} video filter.
23289 Use a core image generator at the beginning of the applied filterchain to
23290 generate the content.
23292 The coreimagesrc video source accepts the following options:
23294 @item list_generators
23295 List all available generators along with all their respective options as well as
23296 possible minimum and maximum values along with the default values.
23298 list_generators=true
23302 Specify the size of the sourced video. For the syntax of this option, check the
23303 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23304 The default value is @code{320x240}.
23307 Specify the frame rate of the sourced video, as the number of frames
23308 generated per second. It has to be a string in the format
23309 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23310 number or a valid video frame rate abbreviation. The default value is
23314 Set the sample aspect ratio of the sourced video.
23317 Set the duration of the sourced video. See
23318 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23319 for the accepted syntax.
23321 If not specified, or the expressed duration is negative, the video is
23322 supposed to be generated forever.
23325 Additionally, all options of the @ref{coreimage} video filter are accepted.
23326 A complete filterchain can be used for further processing of the
23327 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
23328 and examples for details.
23330 @subsection Examples
23335 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23336 given as complete and escaped command-line for Apple's standard bash shell:
23338 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
23340 This example is equivalent to the QRCode example of @ref{coreimage} without the
23341 need for a nullsrc video source.
23346 Generate several gradients.
23350 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23351 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23354 Set frame rate, expressed as number of frames per second. Default
23357 @item c0, c1, c2, c3, c4, c5, c6, c7
23358 Set 8 colors. Default values for colors is to pick random one.
23360 @item x0, y0, y0, y1
23361 Set gradient line source and destination points. If negative or out of range, random ones
23365 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
23368 Set seed for picking gradient line points.
23371 Set the duration of the sourced video. See
23372 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23373 for the accepted syntax.
23375 If not specified, or the expressed duration is negative, the video is
23376 supposed to be generated forever.
23379 Set speed of gradients rotation.
23383 @section mandelbrot
23385 Generate a Mandelbrot set fractal, and progressively zoom towards the
23386 point specified with @var{start_x} and @var{start_y}.
23388 This source accepts the following options:
23393 Set the terminal pts value. Default value is 400.
23396 Set the terminal scale value.
23397 Must be a floating point value. Default value is 0.3.
23400 Set the inner coloring mode, that is the algorithm used to draw the
23401 Mandelbrot fractal internal region.
23403 It shall assume one of the following values:
23408 Show time until convergence.
23410 Set color based on point closest to the origin of the iterations.
23415 Default value is @var{mincol}.
23418 Set the bailout value. Default value is 10.0.
23421 Set the maximum of iterations performed by the rendering
23422 algorithm. Default value is 7189.
23425 Set outer coloring mode.
23426 It shall assume one of following values:
23428 @item iteration_count
23429 Set iteration count mode.
23430 @item normalized_iteration_count
23431 set normalized iteration count mode.
23433 Default value is @var{normalized_iteration_count}.
23436 Set frame rate, expressed as number of frames per second. Default
23440 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23441 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23444 Set the initial scale value. Default value is 3.0.
23447 Set the initial x position. Must be a floating point value between
23448 -100 and 100. Default value is -0.743643887037158704752191506114774.
23451 Set the initial y position. Must be a floating point value between
23452 -100 and 100. Default value is -0.131825904205311970493132056385139.
23457 Generate various test patterns, as generated by the MPlayer test filter.
23459 The size of the generated video is fixed, and is 256x256.
23460 This source is useful in particular for testing encoding features.
23462 This source accepts the following options:
23467 Specify the frame rate of the sourced video, as the number of frames
23468 generated per second. It has to be a string in the format
23469 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23470 number or a valid video frame rate abbreviation. The default value is
23474 Set the duration of the sourced video. See
23475 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23476 for the accepted syntax.
23478 If not specified, or the expressed duration is negative, the video is
23479 supposed to be generated forever.
23483 Set the number or the name of the test to perform. Supported tests are:
23497 @item max_frames, m
23498 Set the maximum number of frames generated for each test, default value is 30.
23502 Default value is "all", which will cycle through the list of all tests.
23507 mptestsrc=t=dc_luma
23510 will generate a "dc_luma" test pattern.
23512 @section frei0r_src
23514 Provide a frei0r source.
23516 To enable compilation of this filter you need to install the frei0r
23517 header and configure FFmpeg with @code{--enable-frei0r}.
23519 This source accepts the following parameters:
23524 The size of the video to generate. For the syntax of this option, check the
23525 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23528 The framerate of the generated video. It may be a string of the form
23529 @var{num}/@var{den} or a frame rate abbreviation.
23532 The name to the frei0r source to load. For more information regarding frei0r and
23533 how to set the parameters, read the @ref{frei0r} section in the video filters
23536 @item filter_params
23537 A '|'-separated list of parameters to pass to the frei0r source.
23541 For example, to generate a frei0r partik0l source with size 200x200
23542 and frame rate 10 which is overlaid on the overlay filter main input:
23544 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23549 Generate a life pattern.
23551 This source is based on a generalization of John Conway's life game.
23553 The sourced input represents a life grid, each pixel represents a cell
23554 which can be in one of two possible states, alive or dead. Every cell
23555 interacts with its eight neighbours, which are the cells that are
23556 horizontally, vertically, or diagonally adjacent.
23558 At each interaction the grid evolves according to the adopted rule,
23559 which specifies the number of neighbor alive cells which will make a
23560 cell stay alive or born. The @option{rule} option allows one to specify
23563 This source accepts the following options:
23567 Set the file from which to read the initial grid state. In the file,
23568 each non-whitespace character is considered an alive cell, and newline
23569 is used to delimit the end of each row.
23571 If this option is not specified, the initial grid is generated
23575 Set the video rate, that is the number of frames generated per second.
23578 @item random_fill_ratio, ratio
23579 Set the random fill ratio for the initial random grid. It is a
23580 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23581 It is ignored when a file is specified.
23583 @item random_seed, seed
23584 Set the seed for filling the initial random grid, must be an integer
23585 included between 0 and UINT32_MAX. If not specified, or if explicitly
23586 set to -1, the filter will try to use a good random seed on a best
23592 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23593 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23594 @var{NS} specifies the number of alive neighbor cells which make a
23595 live cell stay alive, and @var{NB} the number of alive neighbor cells
23596 which make a dead cell to become alive (i.e. to "born").
23597 "s" and "b" can be used in place of "S" and "B", respectively.
23599 Alternatively a rule can be specified by an 18-bits integer. The 9
23600 high order bits are used to encode the next cell state if it is alive
23601 for each number of neighbor alive cells, the low order bits specify
23602 the rule for "borning" new cells. Higher order bits encode for an
23603 higher number of neighbor cells.
23604 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23605 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23607 Default value is "S23/B3", which is the original Conway's game of life
23608 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23609 cells, and will born a new cell if there are three alive cells around
23613 Set the size of the output video. For the syntax of this option, check the
23614 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23616 If @option{filename} is specified, the size is set by default to the
23617 same size of the input file. If @option{size} is set, it must contain
23618 the size specified in the input file, and the initial grid defined in
23619 that file is centered in the larger resulting area.
23621 If a filename is not specified, the size value defaults to "320x240"
23622 (used for a randomly generated initial grid).
23625 If set to 1, stitch the left and right grid edges together, and the
23626 top and bottom edges also. Defaults to 1.
23629 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23630 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23631 value from 0 to 255.
23634 Set the color of living (or new born) cells.
23637 Set the color of dead cells. If @option{mold} is set, this is the first color
23638 used to represent a dead cell.
23641 Set mold color, for definitely dead and moldy cells.
23643 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23644 ffmpeg-utils manual,ffmpeg-utils}.
23647 @subsection Examples
23651 Read a grid from @file{pattern}, and center it on a grid of size
23654 life=f=pattern:s=300x300
23658 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23660 life=ratio=2/3:s=200x200
23664 Specify a custom rule for evolving a randomly generated grid:
23670 Full example with slow death effect (mold) using @command{ffplay}:
23672 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23679 @anchor{haldclutsrc}
23682 @anchor{pal100bars}
23683 @anchor{rgbtestsrc}
23685 @anchor{smptehdbars}
23688 @anchor{yuvtestsrc}
23689 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23691 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23693 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23695 The @code{color} source provides an uniformly colored input.
23697 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23698 @ref{haldclut} filter.
23700 The @code{nullsrc} source returns unprocessed video frames. It is
23701 mainly useful to be employed in analysis / debugging tools, or as the
23702 source for filters which ignore the input data.
23704 The @code{pal75bars} source generates a color bars pattern, based on
23705 EBU PAL recommendations with 75% color levels.
23707 The @code{pal100bars} source generates a color bars pattern, based on
23708 EBU PAL recommendations with 100% color levels.
23710 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23711 detecting RGB vs BGR issues. You should see a red, green and blue
23712 stripe from top to bottom.
23714 The @code{smptebars} source generates a color bars pattern, based on
23715 the SMPTE Engineering Guideline EG 1-1990.
23717 The @code{smptehdbars} source generates a color bars pattern, based on
23718 the SMPTE RP 219-2002.
23720 The @code{testsrc} source generates a test video pattern, showing a
23721 color pattern, a scrolling gradient and a timestamp. This is mainly
23722 intended for testing purposes.
23724 The @code{testsrc2} source is similar to testsrc, but supports more
23725 pixel formats instead of just @code{rgb24}. This allows using it as an
23726 input for other tests without requiring a format conversion.
23728 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23729 see a y, cb and cr stripe from top to bottom.
23731 The sources accept the following parameters:
23736 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23737 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23738 pixels to be used as identity matrix for 3D lookup tables. Each component is
23739 coded on a @code{1/(N*N)} scale.
23742 Specify the color of the source, only available in the @code{color}
23743 source. For the syntax of this option, check the
23744 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23747 Specify the size of the sourced video. For the syntax of this option, check the
23748 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23749 The default value is @code{320x240}.
23751 This option is not available with the @code{allrgb}, @code{allyuv}, and
23752 @code{haldclutsrc} filters.
23755 Specify the frame rate of the sourced video, as the number of frames
23756 generated per second. It has to be a string in the format
23757 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23758 number or a valid video frame rate abbreviation. The default value is
23762 Set the duration of the sourced video. See
23763 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23764 for the accepted syntax.
23766 If not specified, or the expressed duration is negative, the video is
23767 supposed to be generated forever.
23769 Since the frame rate is used as time base, all frames including the last one
23770 will have their full duration. If the specified duration is not a multiple
23771 of the frame duration, it will be rounded up.
23774 Set the sample aspect ratio of the sourced video.
23777 Specify the alpha (opacity) of the background, only available in the
23778 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23779 255 (fully opaque, the default).
23782 Set the number of decimals to show in the timestamp, only available in the
23783 @code{testsrc} source.
23785 The displayed timestamp value will correspond to the original
23786 timestamp value multiplied by the power of 10 of the specified
23787 value. Default value is 0.
23790 @subsection Examples
23794 Generate a video with a duration of 5.3 seconds, with size
23795 176x144 and a frame rate of 10 frames per second:
23797 testsrc=duration=5.3:size=qcif:rate=10
23801 The following graph description will generate a red source
23802 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23805 color=c=red@@0.2:s=qcif:r=10
23809 If the input content is to be ignored, @code{nullsrc} can be used. The
23810 following command generates noise in the luminance plane by employing
23811 the @code{geq} filter:
23813 nullsrc=s=256x256, geq=random(1)*255:128:128
23817 @subsection Commands
23819 The @code{color} source supports the following commands:
23823 Set the color of the created image. Accepts the same syntax of the
23824 corresponding @option{color} option.
23829 Generate video using an OpenCL program.
23834 OpenCL program source file.
23837 Kernel name in program.
23840 Size of frames to generate. This must be set.
23843 Pixel format to use for the generated frames. This must be set.
23846 Number of frames generated every second. Default value is '25'.
23850 For details of how the program loading works, see the @ref{program_opencl}
23857 Generate a colour ramp by setting pixel values from the position of the pixel
23858 in the output image. (Note that this will work with all pixel formats, but
23859 the generated output will not be the same.)
23861 __kernel void ramp(__write_only image2d_t dst,
23862 unsigned int index)
23864 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23867 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23869 write_imagef(dst, loc, val);
23874 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23876 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23877 unsigned int index)
23879 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23881 float4 value = 0.0f;
23882 int x = loc.x + index;
23883 int y = loc.y + index;
23884 while (x > 0 || y > 0) {
23885 if (x % 3 == 1 && y % 3 == 1) {
23893 write_imagef(dst, loc, value);
23899 @section sierpinski
23901 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23903 This source accepts the following options:
23907 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23908 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23911 Set frame rate, expressed as number of frames per second. Default
23915 Set seed which is used for random panning.
23918 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23921 Set fractal type, can be default @code{carpet} or @code{triangle}.
23924 @c man end VIDEO SOURCES
23926 @chapter Video Sinks
23927 @c man begin VIDEO SINKS
23929 Below is a description of the currently available video sinks.
23931 @section buffersink
23933 Buffer video frames, and make them available to the end of the filter
23936 This sink is mainly intended for programmatic use, in particular
23937 through the interface defined in @file{libavfilter/buffersink.h}
23938 or the options system.
23940 It accepts a pointer to an AVBufferSinkContext structure, which
23941 defines the incoming buffers' formats, to be passed as the opaque
23942 parameter to @code{avfilter_init_filter} for initialization.
23946 Null video sink: do absolutely nothing with the input video. It is
23947 mainly useful as a template and for use in analysis / debugging
23950 @c man end VIDEO SINKS
23952 @chapter Multimedia Filters
23953 @c man begin MULTIMEDIA FILTERS
23955 Below is a description of the currently available multimedia filters.
23959 Convert input audio to a video output, displaying the audio bit scope.
23961 The filter accepts the following options:
23965 Set frame rate, expressed as number of frames per second. Default
23969 Specify the video size for the output. For the syntax of this option, check the
23970 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23971 Default value is @code{1024x256}.
23974 Specify list of colors separated by space or by '|' which will be used to
23975 draw channels. Unrecognized or missing colors will be replaced
23979 @section adrawgraph
23980 Draw a graph using input audio metadata.
23982 See @ref{drawgraph}
23984 @section agraphmonitor
23986 See @ref{graphmonitor}.
23988 @section ahistogram
23990 Convert input audio to a video output, displaying the volume histogram.
23992 The filter accepts the following options:
23996 Specify how histogram is calculated.
23998 It accepts the following values:
24001 Use single histogram for all channels.
24003 Use separate histogram for each channel.
24005 Default is @code{single}.
24008 Set frame rate, expressed as number of frames per second. Default
24012 Specify the video size for the output. For the syntax of this option, check the
24013 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24014 Default value is @code{hd720}.
24019 It accepts the following values:
24030 reverse logarithmic
24032 Default is @code{log}.
24035 Set amplitude scale.
24037 It accepts the following values:
24044 Default is @code{log}.
24047 Set how much frames to accumulate in histogram.
24048 Default is 1. Setting this to -1 accumulates all frames.
24051 Set histogram ratio of window height.
24054 Set sonogram sliding.
24056 It accepts the following values:
24059 replace old rows with new ones.
24061 scroll from top to bottom.
24063 Default is @code{replace}.
24066 @section aphasemeter
24068 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
24069 representing mean phase of current audio frame. A video output can also be produced and is
24070 enabled by default. The audio is passed through as first output.
24072 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
24073 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
24074 and @code{1} means channels are in phase.
24076 The filter accepts the following options, all related to its video output:
24080 Set the output frame rate. Default value is @code{25}.
24083 Set the video size for the output. For the syntax of this option, check the
24084 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24085 Default value is @code{800x400}.
24090 Specify the red, green, blue contrast. Default values are @code{2},
24091 @code{7} and @code{1}.
24092 Allowed range is @code{[0, 255]}.
24095 Set color which will be used for drawing median phase. If color is
24096 @code{none} which is default, no median phase value will be drawn.
24099 Enable video output. Default is enabled.
24102 @subsection phasing detection
24104 The filter also detects out of phase and mono sequences in stereo streams.
24105 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
24107 The filter accepts the following options for this detection:
24111 Enable mono and out of phase detection. Default is disabled.
24114 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
24115 Allowed range is @code{[0, 1]}.
24118 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
24119 Allowed range is @code{[90, 180]}.
24122 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
24125 @subsection Examples
24129 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
24131 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
24135 @section avectorscope
24137 Convert input audio to a video output, representing the audio vector
24140 The filter is used to measure the difference between channels of stereo
24141 audio stream. A monaural signal, consisting of identical left and right
24142 signal, results in straight vertical line. Any stereo separation is visible
24143 as a deviation from this line, creating a Lissajous figure.
24144 If the straight (or deviation from it) but horizontal line appears this
24145 indicates that the left and right channels are out of phase.
24147 The filter accepts the following options:
24151 Set the vectorscope mode.
24153 Available values are:
24156 Lissajous rotated by 45 degrees.
24159 Same as above but not rotated.
24162 Shape resembling half of circle.
24165 Default value is @samp{lissajous}.
24168 Set the video size for the output. For the syntax of this option, check the
24169 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24170 Default value is @code{400x400}.
24173 Set the output frame rate. Default value is @code{25}.
24179 Specify the red, green, blue and alpha contrast. Default values are @code{40},
24180 @code{160}, @code{80} and @code{255}.
24181 Allowed range is @code{[0, 255]}.
24187 Specify the red, green, blue and alpha fade. Default values are @code{15},
24188 @code{10}, @code{5} and @code{5}.
24189 Allowed range is @code{[0, 255]}.
24192 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
24193 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
24196 Set the vectorscope drawing mode.
24198 Available values are:
24201 Draw dot for each sample.
24204 Draw line between previous and current sample.
24207 Default value is @samp{dot}.
24210 Specify amplitude scale of audio samples.
24212 Available values are:
24228 Swap left channel axis with right channel axis.
24238 Mirror only x axis.
24241 Mirror only y axis.
24249 @subsection Examples
24253 Complete example using @command{ffplay}:
24255 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24256 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24260 @section bench, abench
24262 Benchmark part of a filtergraph.
24264 The filter accepts the following options:
24268 Start or stop a timer.
24270 Available values are:
24273 Get the current time, set it as frame metadata (using the key
24274 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
24277 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
24278 the input frame metadata to get the time difference. Time difference, average,
24279 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
24280 @code{min}) are then printed. The timestamps are expressed in seconds.
24284 @subsection Examples
24288 Benchmark @ref{selectivecolor} filter:
24290 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24296 Concatenate audio and video streams, joining them together one after the
24299 The filter works on segments of synchronized video and audio streams. All
24300 segments must have the same number of streams of each type, and that will
24301 also be the number of streams at output.
24303 The filter accepts the following options:
24308 Set the number of segments. Default is 2.
24311 Set the number of output video streams, that is also the number of video
24312 streams in each segment. Default is 1.
24315 Set the number of output audio streams, that is also the number of audio
24316 streams in each segment. Default is 0.
24319 Activate unsafe mode: do not fail if segments have a different format.
24323 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
24324 @var{a} audio outputs.
24326 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
24327 segment, in the same order as the outputs, then the inputs for the second
24330 Related streams do not always have exactly the same duration, for various
24331 reasons including codec frame size or sloppy authoring. For that reason,
24332 related synchronized streams (e.g. a video and its audio track) should be
24333 concatenated at once. The concat filter will use the duration of the longest
24334 stream in each segment (except the last one), and if necessary pad shorter
24335 audio streams with silence.
24337 For this filter to work correctly, all segments must start at timestamp 0.
24339 All corresponding streams must have the same parameters in all segments; the
24340 filtering system will automatically select a common pixel format for video
24341 streams, and a common sample format, sample rate and channel layout for
24342 audio streams, but other settings, such as resolution, must be converted
24343 explicitly by the user.
24345 Different frame rates are acceptable but will result in variable frame rate
24346 at output; be sure to configure the output file to handle it.
24348 @subsection Examples
24352 Concatenate an opening, an episode and an ending, all in bilingual version
24353 (video in stream 0, audio in streams 1 and 2):
24355 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24356 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24357 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24358 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24362 Concatenate two parts, handling audio and video separately, using the
24363 (a)movie sources, and adjusting the resolution:
24365 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24366 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24367 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24369 Note that a desync will happen at the stitch if the audio and video streams
24370 do not have exactly the same duration in the first file.
24374 @subsection Commands
24376 This filter supports the following commands:
24379 Close the current segment and step to the next one
24385 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
24386 level. By default, it logs a message at a frequency of 10Hz with the
24387 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
24388 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
24390 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
24391 sample format is double-precision floating point. The input stream will be converted to
24392 this specification, if needed. Users may need to insert aformat and/or aresample filters
24393 after this filter to obtain the original parameters.
24395 The filter also has a video output (see the @var{video} option) with a real
24396 time graph to observe the loudness evolution. The graphic contains the logged
24397 message mentioned above, so it is not printed anymore when this option is set,
24398 unless the verbose logging is set. The main graphing area contains the
24399 short-term loudness (3 seconds of analysis), and the gauge on the right is for
24400 the momentary loudness (400 milliseconds), but can optionally be configured
24401 to instead display short-term loudness (see @var{gauge}).
24403 The green area marks a +/- 1LU target range around the target loudness
24404 (-23LUFS by default, unless modified through @var{target}).
24406 More information about the Loudness Recommendation EBU R128 on
24407 @url{http://tech.ebu.ch/loudness}.
24409 The filter accepts the following options:
24414 Activate the video output. The audio stream is passed unchanged whether this
24415 option is set or no. The video stream will be the first output stream if
24416 activated. Default is @code{0}.
24419 Set the video size. This option is for video only. For the syntax of this
24421 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24422 Default and minimum resolution is @code{640x480}.
24425 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
24426 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
24427 other integer value between this range is allowed.
24430 Set metadata injection. If set to @code{1}, the audio input will be segmented
24431 into 100ms output frames, each of them containing various loudness information
24432 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
24434 Default is @code{0}.
24437 Force the frame logging level.
24439 Available values are:
24442 information logging level
24444 verbose logging level
24447 By default, the logging level is set to @var{info}. If the @option{video} or
24448 the @option{metadata} options are set, it switches to @var{verbose}.
24453 Available modes can be cumulated (the option is a @code{flag} type). Possible
24457 Disable any peak mode (default).
24459 Enable sample-peak mode.
24461 Simple peak mode looking for the higher sample value. It logs a message
24462 for sample-peak (identified by @code{SPK}).
24464 Enable true-peak mode.
24466 If enabled, the peak lookup is done on an over-sampled version of the input
24467 stream for better peak accuracy. It logs a message for true-peak.
24468 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
24469 This mode requires a build with @code{libswresample}.
24473 Treat mono input files as "dual mono". If a mono file is intended for playback
24474 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
24475 If set to @code{true}, this option will compensate for this effect.
24476 Multi-channel input files are not affected by this option.
24479 Set a specific pan law to be used for the measurement of dual mono files.
24480 This parameter is optional, and has a default value of -3.01dB.
24483 Set a specific target level (in LUFS) used as relative zero in the visualization.
24484 This parameter is optional and has a default value of -23LUFS as specified
24485 by EBU R128. However, material published online may prefer a level of -16LUFS
24486 (e.g. for use with podcasts or video platforms).
24489 Set the value displayed by the gauge. Valid values are @code{momentary} and s
24490 @code{shortterm}. By default the momentary value will be used, but in certain
24491 scenarios it may be more useful to observe the short term value instead (e.g.
24495 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24496 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24497 video output, not the summary or continuous log output.
24500 @subsection Examples
24504 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24506 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24510 Run an analysis with @command{ffmpeg}:
24512 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24516 @section interleave, ainterleave
24518 Temporally interleave frames from several inputs.
24520 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24522 These filters read frames from several inputs and send the oldest
24523 queued frame to the output.
24525 Input streams must have well defined, monotonically increasing frame
24528 In order to submit one frame to output, these filters need to enqueue
24529 at least one frame for each input, so they cannot work in case one
24530 input is not yet terminated and will not receive incoming frames.
24532 For example consider the case when one input is a @code{select} filter
24533 which always drops input frames. The @code{interleave} filter will keep
24534 reading from that input, but it will never be able to send new frames
24535 to output until the input sends an end-of-stream signal.
24537 Also, depending on inputs synchronization, the filters will drop
24538 frames in case one input receives more frames than the other ones, and
24539 the queue is already filled.
24541 These filters accept the following options:
24545 Set the number of different inputs, it is 2 by default.
24548 How to determine the end-of-stream.
24552 The duration of the longest input. (default)
24555 The duration of the shortest input.
24558 The duration of the first input.
24563 @subsection Examples
24567 Interleave frames belonging to different streams using @command{ffmpeg}:
24569 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24573 Add flickering blur effect:
24575 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24579 @section metadata, ametadata
24581 Manipulate frame metadata.
24583 This filter accepts the following options:
24587 Set mode of operation of the filter.
24589 Can be one of the following:
24593 If both @code{value} and @code{key} is set, select frames
24594 which have such metadata. If only @code{key} is set, select
24595 every frame that has such key in metadata.
24598 Add new metadata @code{key} and @code{value}. If key is already available
24602 Modify value of already present key.
24605 If @code{value} is set, delete only keys that have such value.
24606 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24610 Print key and its value if metadata was found. If @code{key} is not set print all
24611 metadata values available in frame.
24615 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24618 Set metadata value which will be used. This option is mandatory for
24619 @code{modify} and @code{add} mode.
24622 Which function to use when comparing metadata value and @code{value}.
24624 Can be one of following:
24628 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24631 Values are interpreted as strings, returns true if metadata value starts with
24632 the @code{value} option string.
24635 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24638 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24641 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24644 Values are interpreted as floats, returns true if expression from option @code{expr}
24648 Values are interpreted as strings, returns true if metadata value ends with
24649 the @code{value} option string.
24653 Set expression which is used when @code{function} is set to @code{expr}.
24654 The expression is evaluated through the eval API and can contain the following
24659 Float representation of @code{value} from metadata key.
24662 Float representation of @code{value} as supplied by user in @code{value} option.
24666 If specified in @code{print} mode, output is written to the named file. Instead of
24667 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24668 for standard output. If @code{file} option is not set, output is written to the log
24669 with AV_LOG_INFO loglevel.
24672 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24676 @subsection Examples
24680 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24683 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24686 Print silencedetect output to file @file{metadata.txt}.
24688 silencedetect,ametadata=mode=print:file=metadata.txt
24691 Direct all metadata to a pipe with file descriptor 4.
24693 metadata=mode=print:file='pipe\:4'
24697 @section perms, aperms
24699 Set read/write permissions for the output frames.
24701 These filters are mainly aimed at developers to test direct path in the
24702 following filter in the filtergraph.
24704 The filters accept the following options:
24708 Select the permissions mode.
24710 It accepts the following values:
24713 Do nothing. This is the default.
24715 Set all the output frames read-only.
24717 Set all the output frames directly writable.
24719 Make the frame read-only if writable, and writable if read-only.
24721 Set each output frame read-only or writable randomly.
24725 Set the seed for the @var{random} mode, must be an integer included between
24726 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24727 @code{-1}, the filter will try to use a good random seed on a best effort
24731 Note: in case of auto-inserted filter between the permission filter and the
24732 following one, the permission might not be received as expected in that
24733 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24734 perms/aperms filter can avoid this problem.
24736 @section realtime, arealtime
24738 Slow down filtering to match real time approximately.
24740 These filters will pause the filtering for a variable amount of time to
24741 match the output rate with the input timestamps.
24742 They are similar to the @option{re} option to @code{ffmpeg}.
24744 They accept the following options:
24748 Time limit for the pauses. Any pause longer than that will be considered
24749 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24751 Speed factor for processing. The value must be a float larger than zero.
24752 Values larger than 1.0 will result in faster than realtime processing,
24753 smaller will slow processing down. The @var{limit} is automatically adapted
24754 accordingly. Default is 1.0.
24756 A processing speed faster than what is possible without these filters cannot
24761 @section select, aselect
24763 Select frames to pass in output.
24765 This filter accepts the following options:
24770 Set expression, which is evaluated for each input frame.
24772 If the expression is evaluated to zero, the frame is discarded.
24774 If the evaluation result is negative or NaN, the frame is sent to the
24775 first output; otherwise it is sent to the output with index
24776 @code{ceil(val)-1}, assuming that the input index starts from 0.
24778 For example a value of @code{1.2} corresponds to the output with index
24779 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24782 Set the number of outputs. The output to which to send the selected
24783 frame is based on the result of the evaluation. Default value is 1.
24786 The expression can contain the following constants:
24790 The (sequential) number of the filtered frame, starting from 0.
24793 The (sequential) number of the selected frame, starting from 0.
24795 @item prev_selected_n
24796 The sequential number of the last selected frame. It's NAN if undefined.
24799 The timebase of the input timestamps.
24802 The PTS (Presentation TimeStamp) of the filtered video frame,
24803 expressed in @var{TB} units. It's NAN if undefined.
24806 The PTS of the filtered video frame,
24807 expressed in seconds. It's NAN if undefined.
24810 The PTS of the previously filtered video frame. It's NAN if undefined.
24812 @item prev_selected_pts
24813 The PTS of the last previously filtered video frame. It's NAN if undefined.
24815 @item prev_selected_t
24816 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24819 The PTS of the first video frame in the video. It's NAN if undefined.
24822 The time of the first video frame in the video. It's NAN if undefined.
24824 @item pict_type @emph{(video only)}
24825 The type of the filtered frame. It can assume one of the following
24837 @item interlace_type @emph{(video only)}
24838 The frame interlace type. It can assume one of the following values:
24841 The frame is progressive (not interlaced).
24843 The frame is top-field-first.
24845 The frame is bottom-field-first.
24848 @item consumed_sample_n @emph{(audio only)}
24849 the number of selected samples before the current frame
24851 @item samples_n @emph{(audio only)}
24852 the number of samples in the current frame
24854 @item sample_rate @emph{(audio only)}
24855 the input sample rate
24858 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24861 the position in the file of the filtered frame, -1 if the information
24862 is not available (e.g. for synthetic video)
24864 @item scene @emph{(video only)}
24865 value between 0 and 1 to indicate a new scene; a low value reflects a low
24866 probability for the current frame to introduce a new scene, while a higher
24867 value means the current frame is more likely to be one (see the example below)
24869 @item concatdec_select
24870 The concat demuxer can select only part of a concat input file by setting an
24871 inpoint and an outpoint, but the output packets may not be entirely contained
24872 in the selected interval. By using this variable, it is possible to skip frames
24873 generated by the concat demuxer which are not exactly contained in the selected
24876 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24877 and the @var{lavf.concat.duration} packet metadata values which are also
24878 present in the decoded frames.
24880 The @var{concatdec_select} variable is -1 if the frame pts is at least
24881 start_time and either the duration metadata is missing or the frame pts is less
24882 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24885 That basically means that an input frame is selected if its pts is within the
24886 interval set by the concat demuxer.
24890 The default value of the select expression is "1".
24892 @subsection Examples
24896 Select all frames in input:
24901 The example above is the same as:
24913 Select only I-frames:
24915 select='eq(pict_type\,I)'
24919 Select one frame every 100:
24921 select='not(mod(n\,100))'
24925 Select only frames contained in the 10-20 time interval:
24927 select=between(t\,10\,20)
24931 Select only I-frames contained in the 10-20 time interval:
24933 select=between(t\,10\,20)*eq(pict_type\,I)
24937 Select frames with a minimum distance of 10 seconds:
24939 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24943 Use aselect to select only audio frames with samples number > 100:
24945 aselect='gt(samples_n\,100)'
24949 Create a mosaic of the first scenes:
24951 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24954 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24958 Send even and odd frames to separate outputs, and compose them:
24960 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24964 Select useful frames from an ffconcat file which is using inpoints and
24965 outpoints but where the source files are not intra frame only.
24967 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24971 @section sendcmd, asendcmd
24973 Send commands to filters in the filtergraph.
24975 These filters read commands to be sent to other filters in the
24978 @code{sendcmd} must be inserted between two video filters,
24979 @code{asendcmd} must be inserted between two audio filters, but apart
24980 from that they act the same way.
24982 The specification of commands can be provided in the filter arguments
24983 with the @var{commands} option, or in a file specified by the
24984 @var{filename} option.
24986 These filters accept the following options:
24989 Set the commands to be read and sent to the other filters.
24991 Set the filename of the commands to be read and sent to the other
24995 @subsection Commands syntax
24997 A commands description consists of a sequence of interval
24998 specifications, comprising a list of commands to be executed when a
24999 particular event related to that interval occurs. The occurring event
25000 is typically the current frame time entering or leaving a given time
25003 An interval is specified by the following syntax:
25005 @var{START}[-@var{END}] @var{COMMANDS};
25008 The time interval is specified by the @var{START} and @var{END} times.
25009 @var{END} is optional and defaults to the maximum time.
25011 The current frame time is considered within the specified interval if
25012 it is included in the interval [@var{START}, @var{END}), that is when
25013 the time is greater or equal to @var{START} and is lesser than
25016 @var{COMMANDS} consists of a sequence of one or more command
25017 specifications, separated by ",", relating to that interval. The
25018 syntax of a command specification is given by:
25020 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
25023 @var{FLAGS} is optional and specifies the type of events relating to
25024 the time interval which enable sending the specified command, and must
25025 be a non-null sequence of identifier flags separated by "+" or "|" and
25026 enclosed between "[" and "]".
25028 The following flags are recognized:
25031 The command is sent when the current frame timestamp enters the
25032 specified interval. In other words, the command is sent when the
25033 previous frame timestamp was not in the given interval, and the
25037 The command is sent when the current frame timestamp leaves the
25038 specified interval. In other words, the command is sent when the
25039 previous frame timestamp was in the given interval, and the
25043 The command @var{ARG} is interpreted as expression and result of
25044 expression is passed as @var{ARG}.
25046 The expression is evaluated through the eval API and can contain the following
25051 Original position in the file of the frame, or undefined if undefined
25052 for the current frame.
25055 The presentation timestamp in input.
25058 The count of the input frame for video or audio, starting from 0.
25061 The time in seconds of the current frame.
25064 The start time in seconds of the current command interval.
25067 The end time in seconds of the current command interval.
25070 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
25075 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
25078 @var{TARGET} specifies the target of the command, usually the name of
25079 the filter class or a specific filter instance name.
25081 @var{COMMAND} specifies the name of the command for the target filter.
25083 @var{ARG} is optional and specifies the optional list of argument for
25084 the given @var{COMMAND}.
25086 Between one interval specification and another, whitespaces, or
25087 sequences of characters starting with @code{#} until the end of line,
25088 are ignored and can be used to annotate comments.
25090 A simplified BNF description of the commands specification syntax
25093 @var{COMMAND_FLAG} ::= "enter" | "leave"
25094 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
25095 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
25096 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
25097 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
25098 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
25101 @subsection Examples
25105 Specify audio tempo change at second 4:
25107 asendcmd=c='4.0 atempo tempo 1.5',atempo
25111 Target a specific filter instance:
25113 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
25117 Specify a list of drawtext and hue commands in a file.
25119 # show text in the interval 5-10
25120 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
25121 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
25123 # desaturate the image in the interval 15-20
25124 15.0-20.0 [enter] hue s 0,
25125 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
25127 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
25129 # apply an exponential saturation fade-out effect, starting from time 25
25130 25 [enter] hue s exp(25-t)
25133 A filtergraph allowing to read and process the above command list
25134 stored in a file @file{test.cmd}, can be specified with:
25136 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
25141 @section setpts, asetpts
25143 Change the PTS (presentation timestamp) of the input frames.
25145 @code{setpts} works on video frames, @code{asetpts} on audio frames.
25147 This filter accepts the following options:
25152 The expression which is evaluated for each frame to construct its timestamp.
25156 The expression is evaluated through the eval API and can contain the following
25160 @item FRAME_RATE, FR
25161 frame rate, only defined for constant frame-rate video
25164 The presentation timestamp in input
25167 The count of the input frame for video or the number of consumed samples,
25168 not including the current frame for audio, starting from 0.
25170 @item NB_CONSUMED_SAMPLES
25171 The number of consumed samples, not including the current frame (only
25174 @item NB_SAMPLES, S
25175 The number of samples in the current frame (only audio)
25177 @item SAMPLE_RATE, SR
25178 The audio sample rate.
25181 The PTS of the first frame.
25184 the time in seconds of the first frame
25187 State whether the current frame is interlaced.
25190 the time in seconds of the current frame
25193 original position in the file of the frame, or undefined if undefined
25194 for the current frame
25197 The previous input PTS.
25200 previous input time in seconds
25203 The previous output PTS.
25206 previous output time in seconds
25209 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
25213 The wallclock (RTC) time at the start of the movie in microseconds.
25216 The timebase of the input timestamps.
25220 @subsection Examples
25224 Start counting PTS from zero
25226 setpts=PTS-STARTPTS
25230 Apply fast motion effect:
25236 Apply slow motion effect:
25242 Set fixed rate of 25 frames per second:
25248 Set fixed rate 25 fps with some jitter:
25250 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25254 Apply an offset of 10 seconds to the input PTS:
25260 Generate timestamps from a "live source" and rebase onto the current timebase:
25262 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25266 Generate timestamps by counting samples:
25275 Force color range for the output video frame.
25277 The @code{setrange} filter marks the color range property for the
25278 output frames. It does not change the input frame, but only sets the
25279 corresponding property, which affects how the frame is treated by
25282 The filter accepts the following options:
25287 Available values are:
25291 Keep the same color range property.
25293 @item unspecified, unknown
25294 Set the color range as unspecified.
25296 @item limited, tv, mpeg
25297 Set the color range as limited.
25299 @item full, pc, jpeg
25300 Set the color range as full.
25304 @section settb, asettb
25306 Set the timebase to use for the output frames timestamps.
25307 It is mainly useful for testing timebase configuration.
25309 It accepts the following parameters:
25314 The expression which is evaluated into the output timebase.
25318 The value for @option{tb} is an arithmetic expression representing a
25319 rational. The expression can contain the constants "AVTB" (the default
25320 timebase), "intb" (the input timebase) and "sr" (the sample rate,
25321 audio only). Default value is "intb".
25323 @subsection Examples
25327 Set the timebase to 1/25:
25333 Set the timebase to 1/10:
25339 Set the timebase to 1001/1000:
25345 Set the timebase to 2*intb:
25351 Set the default timebase value:
25358 Convert input audio to a video output representing frequency spectrum
25359 logarithmically using Brown-Puckette constant Q transform algorithm with
25360 direct frequency domain coefficient calculation (but the transform itself
25361 is not really constant Q, instead the Q factor is actually variable/clamped),
25362 with musical tone scale, from E0 to D#10.
25364 The filter accepts the following options:
25368 Specify the video size for the output. It must be even. For the syntax of this option,
25369 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25370 Default value is @code{1920x1080}.
25373 Set the output frame rate. Default value is @code{25}.
25376 Set the bargraph height. It must be even. Default value is @code{-1} which
25377 computes the bargraph height automatically.
25380 Set the axis height. It must be even. Default value is @code{-1} which computes
25381 the axis height automatically.
25384 Set the sonogram height. It must be even. Default value is @code{-1} which
25385 computes the sonogram height automatically.
25388 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
25389 instead. Default value is @code{1}.
25391 @item sono_v, volume
25392 Specify the sonogram volume expression. It can contain variables:
25395 the @var{bar_v} evaluated expression
25396 @item frequency, freq, f
25397 the frequency where it is evaluated
25398 @item timeclamp, tc
25399 the value of @var{timeclamp} option
25403 @item a_weighting(f)
25404 A-weighting of equal loudness
25405 @item b_weighting(f)
25406 B-weighting of equal loudness
25407 @item c_weighting(f)
25408 C-weighting of equal loudness.
25410 Default value is @code{16}.
25412 @item bar_v, volume2
25413 Specify the bargraph volume expression. It can contain variables:
25416 the @var{sono_v} evaluated expression
25417 @item frequency, freq, f
25418 the frequency where it is evaluated
25419 @item timeclamp, tc
25420 the value of @var{timeclamp} option
25424 @item a_weighting(f)
25425 A-weighting of equal loudness
25426 @item b_weighting(f)
25427 B-weighting of equal loudness
25428 @item c_weighting(f)
25429 C-weighting of equal loudness.
25431 Default value is @code{sono_v}.
25433 @item sono_g, gamma
25434 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
25435 higher gamma makes the spectrum having more range. Default value is @code{3}.
25436 Acceptable range is @code{[1, 7]}.
25438 @item bar_g, gamma2
25439 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
25443 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
25444 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
25446 @item timeclamp, tc
25447 Specify the transform timeclamp. At low frequency, there is trade-off between
25448 accuracy in time domain and frequency domain. If timeclamp is lower,
25449 event in time domain is represented more accurately (such as fast bass drum),
25450 otherwise event in frequency domain is represented more accurately
25451 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
25454 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
25455 limits future samples by applying asymmetric windowing in time domain, useful
25456 when low latency is required. Accepted range is @code{[0, 1]}.
25459 Specify the transform base frequency. Default value is @code{20.01523126408007475},
25460 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
25463 Specify the transform end frequency. Default value is @code{20495.59681441799654},
25464 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
25467 This option is deprecated and ignored.
25470 Specify the transform length in time domain. Use this option to control accuracy
25471 trade-off between time domain and frequency domain at every frequency sample.
25472 It can contain variables:
25474 @item frequency, freq, f
25475 the frequency where it is evaluated
25476 @item timeclamp, tc
25477 the value of @var{timeclamp} option.
25479 Default value is @code{384*tc/(384+tc*f)}.
25482 Specify the transform count for every video frame. Default value is @code{6}.
25483 Acceptable range is @code{[1, 30]}.
25486 Specify the transform count for every single pixel. Default value is @code{0},
25487 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
25490 Specify font file for use with freetype to draw the axis. If not specified,
25491 use embedded font. Note that drawing with font file or embedded font is not
25492 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25496 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25497 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25501 Specify font color expression. This is arithmetic expression that should return
25502 integer value 0xRRGGBB. It can contain variables:
25504 @item frequency, freq, f
25505 the frequency where it is evaluated
25506 @item timeclamp, tc
25507 the value of @var{timeclamp} option
25512 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25513 @item r(x), g(x), b(x)
25514 red, green, and blue value of intensity x.
25516 Default value is @code{st(0, (midi(f)-59.5)/12);
25517 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25518 r(1-ld(1)) + b(ld(1))}.
25521 Specify image file to draw the axis. This option override @var{fontfile} and
25522 @var{fontcolor} option.
25525 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25526 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25527 Default value is @code{1}.
25530 Set colorspace. The accepted values are:
25533 Unspecified (default)
25542 BT.470BG or BT.601-6 625
25545 SMPTE-170M or BT.601-6 525
25551 BT.2020 with non-constant luminance
25556 Set spectrogram color scheme. This is list of floating point values with format
25557 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25558 The default is @code{1|0.5|0|0|0.5|1}.
25562 @subsection Examples
25566 Playing audio while showing the spectrum:
25568 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25572 Same as above, but with frame rate 30 fps:
25574 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25578 Playing at 1280x720:
25580 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25584 Disable sonogram display:
25590 A1 and its harmonics: A1, A2, (near)E3, A3:
25592 ffplay -f lavfi 'aevalsrc=0.1*sin(2*PI*55*t)+0.1*sin(4*PI*55*t)+0.1*sin(6*PI*55*t)+0.1*sin(8*PI*55*t),
25593 asplit[a][out1]; [a] showcqt [out0]'
25597 Same as above, but with more accuracy in frequency domain:
25599 ffplay -f lavfi 'aevalsrc=0.1*sin(2*PI*55*t)+0.1*sin(4*PI*55*t)+0.1*sin(6*PI*55*t)+0.1*sin(8*PI*55*t),
25600 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25606 bar_v=10:sono_v=bar_v*a_weighting(f)
25610 Custom gamma, now spectrum is linear to the amplitude.
25616 Custom tlength equation:
25618 tc=0.33:tlength='st(0,0.17); 384*tc / (384 / ld(0) + tc*f /(1-ld(0))) + 384*tc / (tc*f / ld(0) + 384 /(1-ld(0)))'
25622 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25624 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25628 Custom font using fontconfig:
25630 font='Courier New,Monospace,mono|bold'
25634 Custom frequency range with custom axis using image file:
25636 axisfile=myaxis.png:basefreq=40:endfreq=10000
25642 Convert input audio to video output representing the audio power spectrum.
25643 Audio amplitude is on Y-axis while frequency is on X-axis.
25645 The filter accepts the following options:
25649 Specify size of video. For the syntax of this option, check the
25650 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25651 Default is @code{1024x512}.
25655 This set how each frequency bin will be represented.
25657 It accepts the following values:
25663 Default is @code{bar}.
25666 Set amplitude scale.
25668 It accepts the following values:
25682 Default is @code{log}.
25685 Set frequency scale.
25687 It accepts the following values:
25696 Reverse logarithmic scale.
25698 Default is @code{lin}.
25701 Set window size. Allowed range is from 16 to 65536.
25703 Default is @code{2048}
25706 Set windowing function.
25708 It accepts the following values:
25731 Default is @code{hanning}.
25734 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25735 which means optimal overlap for selected window function will be picked.
25738 Set time averaging. Setting this to 0 will display current maximal peaks.
25739 Default is @code{1}, which means time averaging is disabled.
25742 Specify list of colors separated by space or by '|' which will be used to
25743 draw channel frequencies. Unrecognized or missing colors will be replaced
25747 Set channel display mode.
25749 It accepts the following values:
25754 Default is @code{combined}.
25757 Set minimum amplitude used in @code{log} amplitude scaler.
25760 Set data display mode.
25762 It accepts the following values:
25768 Default is @code{magnitude}.
25771 @section showspatial
25773 Convert stereo input audio to a video output, representing the spatial relationship
25774 between two channels.
25776 The filter accepts the following options:
25780 Specify the video size for the output. For the syntax of this option, check the
25781 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25782 Default value is @code{512x512}.
25785 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25788 Set window function.
25790 It accepts the following values:
25815 Default value is @code{hann}.
25818 Set ratio of overlap window. Default value is @code{0.5}.
25819 When value is @code{1} overlap is set to recommended size for specific
25820 window function currently used.
25823 @anchor{showspectrum}
25824 @section showspectrum
25826 Convert input audio to a video output, representing the audio frequency
25829 The filter accepts the following options:
25833 Specify the video size for the output. For the syntax of this option, check the
25834 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25835 Default value is @code{640x512}.
25838 Specify how the spectrum should slide along the window.
25840 It accepts the following values:
25843 the samples start again on the left when they reach the right
25845 the samples scroll from right to left
25847 frames are only produced when the samples reach the right
25849 the samples scroll from left to right
25852 Default value is @code{replace}.
25855 Specify display mode.
25857 It accepts the following values:
25860 all channels are displayed in the same row
25862 all channels are displayed in separate rows
25865 Default value is @samp{combined}.
25868 Specify display color mode.
25870 It accepts the following values:
25873 each channel is displayed in a separate color
25875 each channel is displayed using the same color scheme
25877 each channel is displayed using the rainbow color scheme
25879 each channel is displayed using the moreland color scheme
25881 each channel is displayed using the nebulae color scheme
25883 each channel is displayed using the fire color scheme
25885 each channel is displayed using the fiery color scheme
25887 each channel is displayed using the fruit color scheme
25889 each channel is displayed using the cool color scheme
25891 each channel is displayed using the magma color scheme
25893 each channel is displayed using the green color scheme
25895 each channel is displayed using the viridis color scheme
25897 each channel is displayed using the plasma color scheme
25899 each channel is displayed using the cividis color scheme
25901 each channel is displayed using the terrain color scheme
25904 Default value is @samp{channel}.
25907 Specify scale used for calculating intensity color values.
25909 It accepts the following values:
25914 square root, default
25925 Default value is @samp{sqrt}.
25928 Specify frequency scale.
25930 It accepts the following values:
25938 Default value is @samp{lin}.
25941 Set saturation modifier for displayed colors. Negative values provide
25942 alternative color scheme. @code{0} is no saturation at all.
25943 Saturation must be in [-10.0, 10.0] range.
25944 Default value is @code{1}.
25947 Set window function.
25949 It accepts the following values:
25974 Default value is @code{hann}.
25977 Set orientation of time vs frequency axis. Can be @code{vertical} or
25978 @code{horizontal}. Default is @code{vertical}.
25981 Set ratio of overlap window. Default value is @code{0}.
25982 When value is @code{1} overlap is set to recommended size for specific
25983 window function currently used.
25986 Set scale gain for calculating intensity color values.
25987 Default value is @code{1}.
25990 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25993 Set color rotation, must be in [-1.0, 1.0] range.
25994 Default value is @code{0}.
25997 Set start frequency from which to display spectrogram. Default is @code{0}.
26000 Set stop frequency to which to display spectrogram. Default is @code{0}.
26003 Set upper frame rate limit. Default is @code{auto}, unlimited.
26006 Draw time and frequency axes and legends. Default is disabled.
26009 The usage is very similar to the showwaves filter; see the examples in that
26012 @subsection Examples
26016 Large window with logarithmic color scaling:
26018 showspectrum=s=1280x480:scale=log
26022 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
26024 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
26025 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
26029 @section showspectrumpic
26031 Convert input audio to a single video frame, representing the audio frequency
26034 The filter accepts the following options:
26038 Specify the video size for the output. For the syntax of this option, check the
26039 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26040 Default value is @code{4096x2048}.
26043 Specify display mode.
26045 It accepts the following values:
26048 all channels are displayed in the same row
26050 all channels are displayed in separate rows
26052 Default value is @samp{combined}.
26055 Specify display color mode.
26057 It accepts the following values:
26060 each channel is displayed in a separate color
26062 each channel is displayed using the same color scheme
26064 each channel is displayed using the rainbow color scheme
26066 each channel is displayed using the moreland color scheme
26068 each channel is displayed using the nebulae color scheme
26070 each channel is displayed using the fire color scheme
26072 each channel is displayed using the fiery color scheme
26074 each channel is displayed using the fruit color scheme
26076 each channel is displayed using the cool color scheme
26078 each channel is displayed using the magma color scheme
26080 each channel is displayed using the green color scheme
26082 each channel is displayed using the viridis color scheme
26084 each channel is displayed using the plasma color scheme
26086 each channel is displayed using the cividis color scheme
26088 each channel is displayed using the terrain color scheme
26090 Default value is @samp{intensity}.
26093 Specify scale used for calculating intensity color values.
26095 It accepts the following values:
26100 square root, default
26110 Default value is @samp{log}.
26113 Specify frequency scale.
26115 It accepts the following values:
26123 Default value is @samp{lin}.
26126 Set saturation modifier for displayed colors. Negative values provide
26127 alternative color scheme. @code{0} is no saturation at all.
26128 Saturation must be in [-10.0, 10.0] range.
26129 Default value is @code{1}.
26132 Set window function.
26134 It accepts the following values:
26158 Default value is @code{hann}.
26161 Set orientation of time vs frequency axis. Can be @code{vertical} or
26162 @code{horizontal}. Default is @code{vertical}.
26165 Set scale gain for calculating intensity color values.
26166 Default value is @code{1}.
26169 Draw time and frequency axes and legends. Default is enabled.
26172 Set color rotation, must be in [-1.0, 1.0] range.
26173 Default value is @code{0}.
26176 Set start frequency from which to display spectrogram. Default is @code{0}.
26179 Set stop frequency to which to display spectrogram. Default is @code{0}.
26182 @subsection Examples
26186 Extract an audio spectrogram of a whole audio track
26187 in a 1024x1024 picture using @command{ffmpeg}:
26189 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
26193 @section showvolume
26195 Convert input audio volume to a video output.
26197 The filter accepts the following options:
26204 Set border width, allowed range is [0, 5]. Default is 1.
26207 Set channel width, allowed range is [80, 8192]. Default is 400.
26210 Set channel height, allowed range is [1, 900]. Default is 20.
26213 Set fade, allowed range is [0, 1]. Default is 0.95.
26216 Set volume color expression.
26218 The expression can use the following variables:
26222 Current max volume of channel in dB.
26228 Current channel number, starting from 0.
26232 If set, displays channel names. Default is enabled.
26235 If set, displays volume values. Default is enabled.
26238 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
26239 default is @code{h}.
26242 Set step size, allowed range is [0, 5]. Default is 0, which means
26246 Set background opacity, allowed range is [0, 1]. Default is 0.
26249 Set metering mode, can be peak: @code{p} or rms: @code{r},
26250 default is @code{p}.
26253 Set display scale, can be linear: @code{lin} or log: @code{log},
26254 default is @code{lin}.
26258 If set to > 0., display a line for the max level
26259 in the previous seconds.
26260 default is disabled: @code{0.}
26263 The color of the max line. Use when @code{dm} option is set to > 0.
26264 default is: @code{orange}
26269 Convert input audio to a video output, representing the samples waves.
26271 The filter accepts the following options:
26275 Specify the video size for the output. For the syntax of this option, check the
26276 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26277 Default value is @code{600x240}.
26282 Available values are:
26285 Draw a point for each sample.
26288 Draw a vertical line for each sample.
26291 Draw a point for each sample and a line between them.
26294 Draw a centered vertical line for each sample.
26297 Default value is @code{point}.
26300 Set the number of samples which are printed on the same column. A
26301 larger value will decrease the frame rate. Must be a positive
26302 integer. This option can be set only if the value for @var{rate}
26303 is not explicitly specified.
26306 Set the (approximate) output frame rate. This is done by setting the
26307 option @var{n}. Default value is "25".
26309 @item split_channels
26310 Set if channels should be drawn separately or overlap. Default value is 0.
26313 Set colors separated by '|' which are going to be used for drawing of each channel.
26316 Set amplitude scale.
26318 Available values are:
26336 Set the draw mode. This is mostly useful to set for high @var{n}.
26338 Available values are:
26341 Scale pixel values for each drawn sample.
26344 Draw every sample directly.
26347 Default value is @code{scale}.
26350 @subsection Examples
26354 Output the input file audio and the corresponding video representation
26357 amovie=a.mp3,asplit[out0],showwaves[out1]
26361 Create a synthetic signal and show it with showwaves, forcing a
26362 frame rate of 30 frames per second:
26364 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26368 @section showwavespic
26370 Convert input audio to a single video frame, representing the samples waves.
26372 The filter accepts the following options:
26376 Specify the video size for the output. For the syntax of this option, check the
26377 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26378 Default value is @code{600x240}.
26380 @item split_channels
26381 Set if channels should be drawn separately or overlap. Default value is 0.
26384 Set colors separated by '|' which are going to be used for drawing of each channel.
26387 Set amplitude scale.
26389 Available values are:
26409 Available values are:
26412 Scale pixel values for each drawn sample.
26415 Draw every sample directly.
26418 Default value is @code{scale}.
26421 Set the filter mode.
26423 Available values are:
26426 Use average samples values for each drawn sample.
26429 Use peak samples values for each drawn sample.
26432 Default value is @code{average}.
26435 @subsection Examples
26439 Extract a channel split representation of the wave form of a whole audio track
26440 in a 1024x800 picture using @command{ffmpeg}:
26442 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26446 @section sidedata, asidedata
26448 Delete frame side data, or select frames based on it.
26450 This filter accepts the following options:
26454 Set mode of operation of the filter.
26456 Can be one of the following:
26460 Select every frame with side data of @code{type}.
26463 Delete side data of @code{type}. If @code{type} is not set, delete all side
26469 Set side data type used with all modes. Must be set for @code{select} mode. For
26470 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
26471 in @file{libavutil/frame.h}. For example, to choose
26472 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
26476 @section spectrumsynth
26478 Synthesize audio from 2 input video spectrums, first input stream represents
26479 magnitude across time and second represents phase across time.
26480 The filter will transform from frequency domain as displayed in videos back
26481 to time domain as presented in audio output.
26483 This filter is primarily created for reversing processed @ref{showspectrum}
26484 filter outputs, but can synthesize sound from other spectrograms too.
26485 But in such case results are going to be poor if the phase data is not
26486 available, because in such cases phase data need to be recreated, usually
26487 it's just recreated from random noise.
26488 For best results use gray only output (@code{channel} color mode in
26489 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
26490 @code{lin} scale for phase video. To produce phase, for 2nd video, use
26491 @code{data} option. Inputs videos should generally use @code{fullframe}
26492 slide mode as that saves resources needed for decoding video.
26494 The filter accepts the following options:
26498 Specify sample rate of output audio, the sample rate of audio from which
26499 spectrum was generated may differ.
26502 Set number of channels represented in input video spectrums.
26505 Set scale which was used when generating magnitude input spectrum.
26506 Can be @code{lin} or @code{log}. Default is @code{log}.
26509 Set slide which was used when generating inputs spectrums.
26510 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26511 Default is @code{fullframe}.
26514 Set window function used for resynthesis.
26517 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26518 which means optimal overlap for selected window function will be picked.
26521 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26522 Default is @code{vertical}.
26525 @subsection Examples
26529 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26530 then resynthesize videos back to audio with spectrumsynth:
26532 ffmpeg -i input.flac -lavfi showspectrum=mode=separate:scale=log:overlap=0.875:color=channel:slide=fullframe:data=magnitude -an -c:v rawvideo magnitude.nut
26533 ffmpeg -i input.flac -lavfi showspectrum=mode=separate:scale=lin:overlap=0.875:color=channel:slide=fullframe:data=phase -an -c:v rawvideo phase.nut
26534 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26538 @section split, asplit
26540 Split input into several identical outputs.
26542 @code{asplit} works with audio input, @code{split} with video.
26544 The filter accepts a single parameter which specifies the number of outputs. If
26545 unspecified, it defaults to 2.
26547 @subsection Examples
26551 Create two separate outputs from the same input:
26553 [in] split [out0][out1]
26557 To create 3 or more outputs, you need to specify the number of
26560 [in] asplit=3 [out0][out1][out2]
26564 Create two separate outputs from the same input, one cropped and
26567 [in] split [splitout1][splitout2];
26568 [splitout1] crop=100:100:0:0 [cropout];
26569 [splitout2] pad=200:200:100:100 [padout];
26573 Create 5 copies of the input audio with @command{ffmpeg}:
26575 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26581 Receive commands sent through a libzmq client, and forward them to
26582 filters in the filtergraph.
26584 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26585 must be inserted between two video filters, @code{azmq} between two
26586 audio filters. Both are capable to send messages to any filter type.
26588 To enable these filters you need to install the libzmq library and
26589 headers and configure FFmpeg with @code{--enable-libzmq}.
26591 For more information about libzmq see:
26592 @url{http://www.zeromq.org/}
26594 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26595 receives messages sent through a network interface defined by the
26596 @option{bind_address} (or the abbreviation "@option{b}") option.
26597 Default value of this option is @file{tcp://localhost:5555}. You may
26598 want to alter this value to your needs, but do not forget to escape any
26599 ':' signs (see @ref{filtergraph escaping}).
26601 The received message must be in the form:
26603 @var{TARGET} @var{COMMAND} [@var{ARG}]
26606 @var{TARGET} specifies the target of the command, usually the name of
26607 the filter class or a specific filter instance name. The default
26608 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26609 but you can override this by using the @samp{filter_name@@id} syntax
26610 (see @ref{Filtergraph syntax}).
26612 @var{COMMAND} specifies the name of the command for the target filter.
26614 @var{ARG} is optional and specifies the optional argument list for the
26615 given @var{COMMAND}.
26617 Upon reception, the message is processed and the corresponding command
26618 is injected into the filtergraph. Depending on the result, the filter
26619 will send a reply to the client, adopting the format:
26621 @var{ERROR_CODE} @var{ERROR_REASON}
26625 @var{MESSAGE} is optional.
26627 @subsection Examples
26629 Look at @file{tools/zmqsend} for an example of a zmq client which can
26630 be used to send commands processed by these filters.
26632 Consider the following filtergraph generated by @command{ffplay}.
26633 In this example the last overlay filter has an instance name. All other
26634 filters will have default instance names.
26637 ffplay -dumpgraph 1 -f lavfi "
26638 color=s=100x100:c=red [l];
26639 color=s=100x100:c=blue [r];
26640 nullsrc=s=200x100, zmq [bg];
26641 [bg][l] overlay [bg+l];
26642 [bg+l][r] overlay@@my=x=100 "
26645 To change the color of the left side of the video, the following
26646 command can be used:
26648 echo Parsed_color_0 c yellow | tools/zmqsend
26651 To change the right side:
26653 echo Parsed_color_1 c pink | tools/zmqsend
26656 To change the position of the right side:
26658 echo overlay@@my x 150 | tools/zmqsend
26662 @c man end MULTIMEDIA FILTERS
26664 @chapter Multimedia Sources
26665 @c man begin MULTIMEDIA SOURCES
26667 Below is a description of the currently available multimedia sources.
26671 This is the same as @ref{movie} source, except it selects an audio
26677 Read audio and/or video stream(s) from a movie container.
26679 It accepts the following parameters:
26683 The name of the resource to read (not necessarily a file; it can also be a
26684 device or a stream accessed through some protocol).
26686 @item format_name, f
26687 Specifies the format assumed for the movie to read, and can be either
26688 the name of a container or an input device. If not specified, the
26689 format is guessed from @var{movie_name} or by probing.
26691 @item seek_point, sp
26692 Specifies the seek point in seconds. The frames will be output
26693 starting from this seek point. The parameter is evaluated with
26694 @code{av_strtod}, so the numerical value may be suffixed by an IS
26695 postfix. The default value is "0".
26698 Specifies the streams to read. Several streams can be specified,
26699 separated by "+". The source will then have as many outputs, in the
26700 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26701 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26702 respectively the default (best suited) video and audio stream. Default
26703 is "dv", or "da" if the filter is called as "amovie".
26705 @item stream_index, si
26706 Specifies the index of the video stream to read. If the value is -1,
26707 the most suitable video stream will be automatically selected. The default
26708 value is "-1". Deprecated. If the filter is called "amovie", it will select
26709 audio instead of video.
26712 Specifies how many times to read the stream in sequence.
26713 If the value is 0, the stream will be looped infinitely.
26714 Default value is "1".
26716 Note that when the movie is looped the source timestamps are not
26717 changed, so it will generate non monotonically increasing timestamps.
26719 @item discontinuity
26720 Specifies the time difference between frames above which the point is
26721 considered a timestamp discontinuity which is removed by adjusting the later
26725 It allows overlaying a second video on top of the main input of
26726 a filtergraph, as shown in this graph:
26728 input -----------> deltapts0 --> overlay --> output
26731 movie --> scale--> deltapts1 -------+
26733 @subsection Examples
26737 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26738 on top of the input labelled "in":
26740 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26741 [in] setpts=PTS-STARTPTS [main];
26742 [main][over] overlay=16:16 [out]
26746 Read from a video4linux2 device, and overlay it on top of the input
26749 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26750 [in] setpts=PTS-STARTPTS [main];
26751 [main][over] overlay=16:16 [out]
26755 Read the first video stream and the audio stream with id 0x81 from
26756 dvd.vob; the video is connected to the pad named "video" and the audio is
26757 connected to the pad named "audio":
26759 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26763 @subsection Commands
26765 Both movie and amovie support the following commands:
26768 Perform seek using "av_seek_frame".
26769 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26772 @var{stream_index}: If stream_index is -1, a default
26773 stream is selected, and @var{timestamp} is automatically converted
26774 from AV_TIME_BASE units to the stream specific time_base.
26776 @var{timestamp}: Timestamp in AVStream.time_base units
26777 or, if no stream is specified, in AV_TIME_BASE units.
26779 @var{flags}: Flags which select direction and seeking mode.
26783 Get movie duration in AV_TIME_BASE units.
26787 @c man end MULTIMEDIA SOURCES