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:
14678 Default is @code{new}.
14681 Set various debugging flags.
14686 Force libavfilter not to use any of the specified pixel formats for the
14687 input to the next filter.
14689 It accepts the following parameters:
14693 A '|'-separated list of pixel format names, such as
14694 pix_fmts=yuv420p|monow|rgb24".
14698 @subsection Examples
14702 Force libavfilter to use a format different from @var{yuv420p} for the
14703 input to the vflip filter:
14705 noformat=pix_fmts=yuv420p,vflip
14709 Convert the input video to any of the formats not contained in the list:
14711 noformat=yuv420p|yuv444p|yuv410p
14717 Add noise on video input frame.
14719 The filter accepts the following options:
14727 Set noise seed for specific pixel component or all pixel components in case
14728 of @var{all_seed}. Default value is @code{123457}.
14730 @item all_strength, alls
14731 @item c0_strength, c0s
14732 @item c1_strength, c1s
14733 @item c2_strength, c2s
14734 @item c3_strength, c3s
14735 Set noise strength for specific pixel component or all pixel components in case
14736 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14738 @item all_flags, allf
14739 @item c0_flags, c0f
14740 @item c1_flags, c1f
14741 @item c2_flags, c2f
14742 @item c3_flags, c3f
14743 Set pixel component flags or set flags for all components if @var{all_flags}.
14744 Available values for component flags are:
14747 averaged temporal noise (smoother)
14749 mix random noise with a (semi)regular pattern
14751 temporal noise (noise pattern changes between frames)
14753 uniform noise (gaussian otherwise)
14757 @subsection Examples
14759 Add temporal and uniform noise to input video:
14761 noise=alls=20:allf=t+u
14766 Normalize RGB video (aka histogram stretching, contrast stretching).
14767 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14769 For each channel of each frame, the filter computes the input range and maps
14770 it linearly to the user-specified output range. The output range defaults
14771 to the full dynamic range from pure black to pure white.
14773 Temporal smoothing can be used on the input range to reduce flickering (rapid
14774 changes in brightness) caused when small dark or bright objects enter or leave
14775 the scene. This is similar to the auto-exposure (automatic gain control) on a
14776 video camera, and, like a video camera, it may cause a period of over- or
14777 under-exposure of the video.
14779 The R,G,B channels can be normalized independently, which may cause some
14780 color shifting, or linked together as a single channel, which prevents
14781 color shifting. Linked normalization preserves hue. Independent normalization
14782 does not, so it can be used to remove some color casts. Independent and linked
14783 normalization can be combined in any ratio.
14785 The normalize filter accepts the following options:
14790 Colors which define the output range. The minimum input value is mapped to
14791 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14792 The defaults are black and white respectively. Specifying white for
14793 @var{blackpt} and black for @var{whitept} will give color-inverted,
14794 normalized video. Shades of grey can be used to reduce the dynamic range
14795 (contrast). Specifying saturated colors here can create some interesting
14799 The number of previous frames to use for temporal smoothing. The input range
14800 of each channel is smoothed using a rolling average over the current frame
14801 and the @var{smoothing} previous frames. The default is 0 (no temporal
14805 Controls the ratio of independent (color shifting) channel normalization to
14806 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14807 independent. Defaults to 1.0 (fully independent).
14810 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14811 expensive no-op. Defaults to 1.0 (full strength).
14815 @subsection Commands
14816 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14817 The command accepts the same syntax of the corresponding option.
14819 If the specified expression is not valid, it is kept at its current
14822 @subsection Examples
14824 Stretch video contrast to use the full dynamic range, with no temporal
14825 smoothing; may flicker depending on the source content:
14827 normalize=blackpt=black:whitept=white:smoothing=0
14830 As above, but with 50 frames of temporal smoothing; flicker should be
14831 reduced, depending on the source content:
14833 normalize=blackpt=black:whitept=white:smoothing=50
14836 As above, but with hue-preserving linked channel normalization:
14838 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14841 As above, but with half strength:
14843 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14846 Map the darkest input color to red, the brightest input color to cyan:
14848 normalize=blackpt=red:whitept=cyan
14853 Pass the video source unchanged to the output.
14856 Optical Character Recognition
14858 This filter uses Tesseract for optical character recognition. To enable
14859 compilation of this filter, you need to configure FFmpeg with
14860 @code{--enable-libtesseract}.
14862 It accepts the following options:
14866 Set datapath to tesseract data. Default is to use whatever was
14867 set at installation.
14870 Set language, default is "eng".
14873 Set character whitelist.
14876 Set character blacklist.
14879 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14880 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14884 Apply a video transform using libopencv.
14886 To enable this filter, install the libopencv library and headers and
14887 configure FFmpeg with @code{--enable-libopencv}.
14889 It accepts the following parameters:
14894 The name of the libopencv filter to apply.
14896 @item filter_params
14897 The parameters to pass to the libopencv filter. If not specified, the default
14898 values are assumed.
14902 Refer to the official libopencv documentation for more precise
14904 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14906 Several libopencv filters are supported; see the following subsections.
14911 Dilate an image by using a specific structuring element.
14912 It corresponds to the libopencv function @code{cvDilate}.
14914 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14916 @var{struct_el} represents a structuring element, and has the syntax:
14917 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14919 @var{cols} and @var{rows} represent the number of columns and rows of
14920 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14921 point, and @var{shape} the shape for the structuring element. @var{shape}
14922 must be "rect", "cross", "ellipse", or "custom".
14924 If the value for @var{shape} is "custom", it must be followed by a
14925 string of the form "=@var{filename}". The file with name
14926 @var{filename} is assumed to represent a binary image, with each
14927 printable character corresponding to a bright pixel. When a custom
14928 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14929 or columns and rows of the read file are assumed instead.
14931 The default value for @var{struct_el} is "3x3+0x0/rect".
14933 @var{nb_iterations} specifies the number of times the transform is
14934 applied to the image, and defaults to 1.
14938 # Use the default values
14941 # Dilate using a structuring element with a 5x5 cross, iterating two times
14942 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14944 # Read the shape from the file diamond.shape, iterating two times.
14945 # The file diamond.shape may contain a pattern of characters like this
14951 # The specified columns and rows are ignored
14952 # but the anchor point coordinates are not
14953 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14958 Erode an image by using a specific structuring element.
14959 It corresponds to the libopencv function @code{cvErode}.
14961 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14962 with the same syntax and semantics as the @ref{dilate} filter.
14966 Smooth the input video.
14968 The filter takes the following parameters:
14969 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14971 @var{type} is the type of smooth filter to apply, and must be one of
14972 the following values: "blur", "blur_no_scale", "median", "gaussian",
14973 or "bilateral". The default value is "gaussian".
14975 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14976 depends on the smooth type. @var{param1} and
14977 @var{param2} accept integer positive values or 0. @var{param3} and
14978 @var{param4} accept floating point values.
14980 The default value for @var{param1} is 3. The default value for the
14981 other parameters is 0.
14983 These parameters correspond to the parameters assigned to the
14984 libopencv function @code{cvSmooth}.
14986 @section oscilloscope
14988 2D Video Oscilloscope.
14990 Useful to measure spatial impulse, step responses, chroma delays, etc.
14992 It accepts the following parameters:
14996 Set scope center x position.
14999 Set scope center y position.
15002 Set scope size, relative to frame diagonal.
15005 Set scope tilt/rotation.
15011 Set trace center x position.
15014 Set trace center y position.
15017 Set trace width, relative to width of frame.
15020 Set trace height, relative to height of frame.
15023 Set which components to trace. By default it traces first three components.
15026 Draw trace grid. By default is enabled.
15029 Draw some statistics. By default is enabled.
15032 Draw scope. By default is enabled.
15035 @subsection Commands
15036 This filter supports same @ref{commands} as options.
15037 The command accepts the same syntax of the corresponding option.
15039 If the specified expression is not valid, it is kept at its current
15042 @subsection Examples
15046 Inspect full first row of video frame.
15048 oscilloscope=x=0.5:y=0:s=1
15052 Inspect full last row of video frame.
15054 oscilloscope=x=0.5:y=1:s=1
15058 Inspect full 5th line of video frame of height 1080.
15060 oscilloscope=x=0.5:y=5/1080:s=1
15064 Inspect full last column of video frame.
15066 oscilloscope=x=1:y=0.5:s=1:t=1
15074 Overlay one video on top of another.
15076 It takes two inputs and has one output. The first input is the "main"
15077 video on which the second input is overlaid.
15079 It accepts the following parameters:
15081 A description of the accepted options follows.
15086 Set the expression for the x and y coordinates of the overlaid video
15087 on the main video. Default value is "0" for both expressions. In case
15088 the expression is invalid, it is set to a huge value (meaning that the
15089 overlay will not be displayed within the output visible area).
15092 See @ref{framesync}.
15095 Set when the expressions for @option{x}, and @option{y} are evaluated.
15097 It accepts the following values:
15100 only evaluate expressions once during the filter initialization or
15101 when a command is processed
15104 evaluate expressions for each incoming frame
15107 Default value is @samp{frame}.
15110 See @ref{framesync}.
15113 Set the format for the output video.
15115 It accepts the following values:
15118 force YUV420 output
15121 force YUV420p10 output
15124 force YUV422 output
15127 force YUV422p10 output
15130 force YUV444 output
15133 force packed RGB output
15136 force planar RGB output
15139 automatically pick format
15142 Default value is @samp{yuv420}.
15145 See @ref{framesync}.
15148 Set format of alpha of the overlaid video, it can be @var{straight} or
15149 @var{premultiplied}. Default is @var{straight}.
15152 The @option{x}, and @option{y} expressions can contain the following
15158 The main input width and height.
15162 The overlay input width and height.
15166 The computed values for @var{x} and @var{y}. They are evaluated for
15171 horizontal and vertical chroma subsample values of the output
15172 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
15176 the number of input frame, starting from 0
15179 the position in the file of the input frame, NAN if unknown
15182 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
15186 This filter also supports the @ref{framesync} options.
15188 Note that the @var{n}, @var{pos}, @var{t} variables are available only
15189 when evaluation is done @emph{per frame}, and will evaluate to NAN
15190 when @option{eval} is set to @samp{init}.
15192 Be aware that frames are taken from each input video in timestamp
15193 order, hence, if their initial timestamps differ, it is a good idea
15194 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
15195 have them begin in the same zero timestamp, as the example for
15196 the @var{movie} filter does.
15198 You can chain together more overlays but you should test the
15199 efficiency of such approach.
15201 @subsection Commands
15203 This filter supports the following commands:
15207 Modify the x and y of the overlay input.
15208 The command accepts the same syntax of the corresponding option.
15210 If the specified expression is not valid, it is kept at its current
15214 @subsection Examples
15218 Draw the overlay at 10 pixels from the bottom right corner of the main
15221 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15224 Using named options the example above becomes:
15226 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15230 Insert a transparent PNG logo in the bottom left corner of the input,
15231 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
15233 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15237 Insert 2 different transparent PNG logos (second logo on bottom
15238 right corner) using the @command{ffmpeg} tool:
15240 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
15244 Add a transparent color layer on top of the main video; @code{WxH}
15245 must specify the size of the main input to the overlay filter:
15247 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
15251 Play an original video and a filtered version (here with the deshake
15252 filter) side by side using the @command{ffplay} tool:
15254 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15257 The above command is the same as:
15259 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15263 Make a sliding overlay appearing from the left to the right top part of the
15264 screen starting since time 2:
15266 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15270 Compose output by putting two input videos side to side:
15272 ffmpeg -i left.avi -i right.avi -filter_complex "
15273 nullsrc=size=200x100 [background];
15274 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15275 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15276 [background][left] overlay=shortest=1 [background+left];
15277 [background+left][right] overlay=shortest=1:x=100 [left+right]
15282 Mask 10-20 seconds of a video by applying the delogo filter to a section
15284 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15285 -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]'
15290 Chain several overlays in cascade:
15292 nullsrc=s=200x200 [bg];
15293 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15294 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15295 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15296 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15297 [in3] null, [mid2] overlay=100:100 [out0]
15302 @anchor{overlay_cuda}
15303 @section overlay_cuda
15305 Overlay one video on top of another.
15307 This is the CUDA variant of the @ref{overlay} filter.
15308 It only accepts CUDA frames. The underlying input pixel formats have to match.
15310 It takes two inputs and has one output. The first input is the "main"
15311 video on which the second input is overlaid.
15313 It accepts the following parameters:
15318 Set the x and y coordinates of the overlaid video on the main video.
15319 Default value is "0" for both expressions.
15322 See @ref{framesync}.
15325 See @ref{framesync}.
15328 See @ref{framesync}.
15332 This filter also supports the @ref{framesync} options.
15336 Apply Overcomplete Wavelet denoiser.
15338 The filter accepts the following options:
15344 Larger depth values will denoise lower frequency components more, but
15345 slow down filtering.
15347 Must be an int in the range 8-16, default is @code{8}.
15349 @item luma_strength, ls
15352 Must be a double value in the range 0-1000, default is @code{1.0}.
15354 @item chroma_strength, cs
15355 Set chroma strength.
15357 Must be a double value in the range 0-1000, default is @code{1.0}.
15363 Add paddings to the input image, and place the original input at the
15364 provided @var{x}, @var{y} coordinates.
15366 It accepts the following parameters:
15371 Specify an expression for the size of the output image with the
15372 paddings added. If the value for @var{width} or @var{height} is 0, the
15373 corresponding input size is used for the output.
15375 The @var{width} expression can reference the value set by the
15376 @var{height} expression, and vice versa.
15378 The default value of @var{width} and @var{height} is 0.
15382 Specify the offsets to place the input image at within the padded area,
15383 with respect to the top/left border of the output image.
15385 The @var{x} expression can reference the value set by the @var{y}
15386 expression, and vice versa.
15388 The default value of @var{x} and @var{y} is 0.
15390 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15391 so the input image is centered on the padded area.
15394 Specify the color of the padded area. For the syntax of this option,
15395 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15396 manual,ffmpeg-utils}.
15398 The default value of @var{color} is "black".
15401 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15403 It accepts the following values:
15407 Only evaluate expressions once during the filter initialization or when
15408 a command is processed.
15411 Evaluate expressions for each incoming frame.
15415 Default value is @samp{init}.
15418 Pad to aspect instead to a resolution.
15422 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15423 options are expressions containing the following constants:
15428 The input video width and height.
15432 These are the same as @var{in_w} and @var{in_h}.
15436 The output width and height (the size of the padded area), as
15437 specified by the @var{width} and @var{height} expressions.
15441 These are the same as @var{out_w} and @var{out_h}.
15445 The x and y offsets as specified by the @var{x} and @var{y}
15446 expressions, or NAN if not yet specified.
15449 same as @var{iw} / @var{ih}
15452 input sample aspect ratio
15455 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15459 The horizontal and vertical chroma subsample values. For example for the
15460 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15463 @subsection Examples
15467 Add paddings with the color "violet" to the input video. The output video
15468 size is 640x480, and the top-left corner of the input video is placed at
15471 pad=640:480:0:40:violet
15474 The example above is equivalent to the following command:
15476 pad=width=640:height=480:x=0:y=40:color=violet
15480 Pad the input to get an output with dimensions increased by 3/2,
15481 and put the input video at the center of the padded area:
15483 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15487 Pad the input to get a squared output with size equal to the maximum
15488 value between the input width and height, and put the input video at
15489 the center of the padded area:
15491 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15495 Pad the input to get a final w/h ratio of 16:9:
15497 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15501 In case of anamorphic video, in order to set the output display aspect
15502 correctly, it is necessary to use @var{sar} in the expression,
15503 according to the relation:
15505 (ih * X / ih) * sar = output_dar
15506 X = output_dar / sar
15509 Thus the previous example needs to be modified to:
15511 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15515 Double the output size and put the input video in the bottom-right
15516 corner of the output padded area:
15518 pad="2*iw:2*ih:ow-iw:oh-ih"
15522 @anchor{palettegen}
15523 @section palettegen
15525 Generate one palette for a whole video stream.
15527 It accepts the following options:
15531 Set the maximum number of colors to quantize in the palette.
15532 Note: the palette will still contain 256 colors; the unused palette entries
15535 @item reserve_transparent
15536 Create a palette of 255 colors maximum and reserve the last one for
15537 transparency. Reserving the transparency color is useful for GIF optimization.
15538 If not set, the maximum of colors in the palette will be 256. You probably want
15539 to disable this option for a standalone image.
15542 @item transparency_color
15543 Set the color that will be used as background for transparency.
15546 Set statistics mode.
15548 It accepts the following values:
15551 Compute full frame histograms.
15553 Compute histograms only for the part that differs from previous frame. This
15554 might be relevant to give more importance to the moving part of your input if
15555 the background is static.
15557 Compute new histogram for each frame.
15560 Default value is @var{full}.
15563 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15564 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15565 color quantization of the palette. This information is also visible at
15566 @var{info} logging level.
15568 @subsection Examples
15572 Generate a representative palette of a given video using @command{ffmpeg}:
15574 ffmpeg -i input.mkv -vf palettegen palette.png
15578 @section paletteuse
15580 Use a palette to downsample an input video stream.
15582 The filter takes two inputs: one video stream and a palette. The palette must
15583 be a 256 pixels image.
15585 It accepts the following options:
15589 Select dithering mode. Available algorithms are:
15592 Ordered 8x8 bayer dithering (deterministic)
15594 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15595 Note: this dithering is sometimes considered "wrong" and is included as a
15597 @item floyd_steinberg
15598 Floyd and Steingberg dithering (error diffusion)
15600 Frankie Sierra dithering v2 (error diffusion)
15602 Frankie Sierra dithering v2 "Lite" (error diffusion)
15605 Default is @var{sierra2_4a}.
15608 When @var{bayer} dithering is selected, this option defines the scale of the
15609 pattern (how much the crosshatch pattern is visible). A low value means more
15610 visible pattern for less banding, and higher value means less visible pattern
15611 at the cost of more banding.
15613 The option must be an integer value in the range [0,5]. Default is @var{2}.
15616 If set, define the zone to process
15620 Only the changing rectangle will be reprocessed. This is similar to GIF
15621 cropping/offsetting compression mechanism. This option can be useful for speed
15622 if only a part of the image is changing, and has use cases such as limiting the
15623 scope of the error diffusal @option{dither} to the rectangle that bounds the
15624 moving scene (it leads to more deterministic output if the scene doesn't change
15625 much, and as a result less moving noise and better GIF compression).
15628 Default is @var{none}.
15631 Take new palette for each output frame.
15633 @item alpha_threshold
15634 Sets the alpha threshold for transparency. Alpha values above this threshold
15635 will be treated as completely opaque, and values below this threshold will be
15636 treated as completely transparent.
15638 The option must be an integer value in the range [0,255]. Default is @var{128}.
15641 @subsection Examples
15645 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15646 using @command{ffmpeg}:
15648 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15652 @section perspective
15654 Correct perspective of video not recorded perpendicular to the screen.
15656 A description of the accepted parameters follows.
15667 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15668 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15669 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15670 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15671 then the corners of the source will be sent to the specified coordinates.
15673 The expressions can use the following variables:
15678 the width and height of video frame.
15682 Output frame count.
15685 @item interpolation
15686 Set interpolation for perspective correction.
15688 It accepts the following values:
15694 Default value is @samp{linear}.
15697 Set interpretation of coordinate options.
15699 It accepts the following values:
15703 Send point in the source specified by the given coordinates to
15704 the corners of the destination.
15706 @item 1, destination
15708 Send the corners of the source to the point in the destination specified
15709 by the given coordinates.
15711 Default value is @samp{source}.
15715 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15717 It accepts the following values:
15720 only evaluate expressions once during the filter initialization or
15721 when a command is processed
15724 evaluate expressions for each incoming frame
15727 Default value is @samp{init}.
15732 Delay interlaced video by one field time so that the field order changes.
15734 The intended use is to fix PAL movies that have been captured with the
15735 opposite field order to the film-to-video transfer.
15737 A description of the accepted parameters follows.
15743 It accepts the following values:
15746 Capture field order top-first, transfer bottom-first.
15747 Filter will delay the bottom field.
15750 Capture field order bottom-first, transfer top-first.
15751 Filter will delay the top field.
15754 Capture and transfer with the same field order. This mode only exists
15755 for the documentation of the other options to refer to, but if you
15756 actually select it, the filter will faithfully do nothing.
15759 Capture field order determined automatically by field flags, transfer
15761 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15762 basis using field flags. If no field information is available,
15763 then this works just like @samp{u}.
15766 Capture unknown or varying, transfer opposite.
15767 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15768 analyzing the images and selecting the alternative that produces best
15769 match between the fields.
15772 Capture top-first, transfer unknown or varying.
15773 Filter selects among @samp{t} and @samp{p} using image analysis.
15776 Capture bottom-first, transfer unknown or varying.
15777 Filter selects among @samp{b} and @samp{p} using image analysis.
15780 Capture determined by field flags, transfer unknown or varying.
15781 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15782 image analysis. If no field information is available, then this works just
15783 like @samp{U}. This is the default mode.
15786 Both capture and transfer unknown or varying.
15787 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15791 @subsection Commands
15793 This filter supports the all above options as @ref{commands}.
15795 @section photosensitivity
15796 Reduce various flashes in video, so to help users with epilepsy.
15798 It accepts the following options:
15801 Set how many frames to use when filtering. Default is 30.
15804 Set detection threshold factor. Default is 1.
15808 Set how many pixels to skip when sampling frames. Default is 1.
15809 Allowed range is from 1 to 1024.
15812 Leave frames unchanged. Default is disabled.
15815 @section pixdesctest
15817 Pixel format descriptor test filter, mainly useful for internal
15818 testing. The output video should be equal to the input video.
15822 format=monow, pixdesctest
15825 can be used to test the monowhite pixel format descriptor definition.
15829 Display sample values of color channels. Mainly useful for checking color
15830 and levels. Minimum supported resolution is 640x480.
15832 The filters accept the following options:
15836 Set scope X position, relative offset on X axis.
15839 Set scope Y position, relative offset on Y axis.
15848 Set window opacity. This window also holds statistics about pixel area.
15851 Set window X position, relative offset on X axis.
15854 Set window Y position, relative offset on Y axis.
15859 Enable the specified chain of postprocessing subfilters using libpostproc. This
15860 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15861 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15862 Each subfilter and some options have a short and a long name that can be used
15863 interchangeably, i.e. dr/dering are the same.
15865 The filters accept the following options:
15869 Set postprocessing subfilters string.
15872 All subfilters share common options to determine their scope:
15876 Honor the quality commands for this subfilter.
15879 Do chrominance filtering, too (default).
15882 Do luminance filtering only (no chrominance).
15885 Do chrominance filtering only (no luminance).
15888 These options can be appended after the subfilter name, separated by a '|'.
15890 Available subfilters are:
15893 @item hb/hdeblock[|difference[|flatness]]
15894 Horizontal deblocking filter
15897 Difference factor where higher values mean more deblocking (default: @code{32}).
15899 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15902 @item vb/vdeblock[|difference[|flatness]]
15903 Vertical deblocking filter
15906 Difference factor where higher values mean more deblocking (default: @code{32}).
15908 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15911 @item ha/hadeblock[|difference[|flatness]]
15912 Accurate horizontal deblocking filter
15915 Difference factor where higher values mean more deblocking (default: @code{32}).
15917 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15920 @item va/vadeblock[|difference[|flatness]]
15921 Accurate vertical deblocking filter
15924 Difference factor where higher values mean more deblocking (default: @code{32}).
15926 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15930 The horizontal and vertical deblocking filters share the difference and
15931 flatness values so you cannot set different horizontal and vertical
15935 @item h1/x1hdeblock
15936 Experimental horizontal deblocking filter
15938 @item v1/x1vdeblock
15939 Experimental vertical deblocking filter
15944 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15947 larger -> stronger filtering
15949 larger -> stronger filtering
15951 larger -> stronger filtering
15954 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15957 Stretch luminance to @code{0-255}.
15960 @item lb/linblenddeint
15961 Linear blend deinterlacing filter that deinterlaces the given block by
15962 filtering all lines with a @code{(1 2 1)} filter.
15964 @item li/linipoldeint
15965 Linear interpolating deinterlacing filter that deinterlaces the given block by
15966 linearly interpolating every second line.
15968 @item ci/cubicipoldeint
15969 Cubic interpolating deinterlacing filter deinterlaces the given block by
15970 cubically interpolating every second line.
15972 @item md/mediandeint
15973 Median deinterlacing filter that deinterlaces the given block by applying a
15974 median filter to every second line.
15976 @item fd/ffmpegdeint
15977 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15978 second line with a @code{(-1 4 2 4 -1)} filter.
15981 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15982 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15984 @item fq/forceQuant[|quantizer]
15985 Overrides the quantizer table from the input with the constant quantizer you
15993 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15996 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15999 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
16002 @subsection Examples
16006 Apply horizontal and vertical deblocking, deringing and automatic
16007 brightness/contrast:
16013 Apply default filters without brightness/contrast correction:
16019 Apply default filters and temporal denoiser:
16021 pp=default/tmpnoise|1|2|3
16025 Apply deblocking on luminance only, and switch vertical deblocking on or off
16026 automatically depending on available CPU time:
16033 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
16034 similar to spp = 6 with 7 point DCT, where only the center sample is
16037 The filter accepts the following options:
16041 Force a constant quantization parameter. It accepts an integer in range
16042 0 to 63. If not set, the filter will use the QP from the video stream
16046 Set thresholding mode. Available modes are:
16050 Set hard thresholding.
16052 Set soft thresholding (better de-ringing effect, but likely blurrier).
16054 Set medium thresholding (good results, default).
16058 @section premultiply
16059 Apply alpha premultiply effect to input video stream using first plane
16060 of second stream as alpha.
16062 Both streams must have same dimensions and same pixel format.
16064 The filter accepts the following option:
16068 Set which planes will be processed, unprocessed planes will be copied.
16069 By default value 0xf, all planes will be processed.
16072 Do not require 2nd input for processing, instead use alpha plane from input stream.
16076 Apply prewitt operator to input video stream.
16078 The filter accepts the following option:
16082 Set which planes will be processed, unprocessed planes will be copied.
16083 By default value 0xf, all planes will be processed.
16086 Set value which will be multiplied with filtered result.
16089 Set value which will be added to filtered result.
16092 @subsection Commands
16094 This filter supports the all above options as @ref{commands}.
16096 @section pseudocolor
16098 Alter frame colors in video with pseudocolors.
16100 This filter accepts the following options:
16104 set pixel first component expression
16107 set pixel second component expression
16110 set pixel third component expression
16113 set pixel fourth component expression, corresponds to the alpha component
16116 set component to use as base for altering colors
16119 Each of them specifies the expression to use for computing the lookup table for
16120 the corresponding pixel component values.
16122 The expressions can contain the following constants and functions:
16127 The input width and height.
16130 The input value for the pixel component.
16132 @item ymin, umin, vmin, amin
16133 The minimum allowed component value.
16135 @item ymax, umax, vmax, amax
16136 The maximum allowed component value.
16139 All expressions default to "val".
16141 @subsection Examples
16145 Change too high luma values to gradient:
16147 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'"
16153 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16154 Ratio) between two input videos.
16156 This filter takes in input two input videos, the first input is
16157 considered the "main" source and is passed unchanged to the
16158 output. The second input is used as a "reference" video for computing
16161 Both video inputs must have the same resolution and pixel format for
16162 this filter to work correctly. Also it assumes that both inputs
16163 have the same number of frames, which are compared one by one.
16165 The obtained average PSNR is printed through the logging system.
16167 The filter stores the accumulated MSE (mean squared error) of each
16168 frame, and at the end of the processing it is averaged across all frames
16169 equally, and the following formula is applied to obtain the PSNR:
16172 PSNR = 10*log10(MAX^2/MSE)
16175 Where MAX is the average of the maximum values of each component of the
16178 The description of the accepted parameters follows.
16181 @item stats_file, f
16182 If specified the filter will use the named file to save the PSNR of
16183 each individual frame. When filename equals "-" the data is sent to
16186 @item stats_version
16187 Specifies which version of the stats file format to use. Details of
16188 each format are written below.
16189 Default value is 1.
16191 @item stats_add_max
16192 Determines whether the max value is output to the stats log.
16193 Default value is 0.
16194 Requires stats_version >= 2. If this is set and stats_version < 2,
16195 the filter will return an error.
16198 This filter also supports the @ref{framesync} options.
16200 The file printed if @var{stats_file} is selected, contains a sequence of
16201 key/value pairs of the form @var{key}:@var{value} for each compared
16204 If a @var{stats_version} greater than 1 is specified, a header line precedes
16205 the list of per-frame-pair stats, with key value pairs following the frame
16206 format with the following parameters:
16209 @item psnr_log_version
16210 The version of the log file format. Will match @var{stats_version}.
16213 A comma separated list of the per-frame-pair parameters included in
16217 A description of each shown per-frame-pair parameter follows:
16221 sequential number of the input frame, starting from 1
16224 Mean Square Error pixel-by-pixel average difference of the compared
16225 frames, averaged over all the image components.
16227 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16228 Mean Square Error pixel-by-pixel average difference of the compared
16229 frames for the component specified by the suffix.
16231 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16232 Peak Signal to Noise ratio of the compared frames for the component
16233 specified by the suffix.
16235 @item max_avg, max_y, max_u, max_v
16236 Maximum allowed value for each channel, and average over all
16240 @subsection Examples
16245 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16246 [main][ref] psnr="stats_file=stats.log" [out]
16249 On this example the input file being processed is compared with the
16250 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
16251 is stored in @file{stats.log}.
16254 Another example with different containers:
16256 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 -
16263 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16264 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
16267 The pullup filter is designed to take advantage of future context in making
16268 its decisions. This filter is stateless in the sense that it does not lock
16269 onto a pattern to follow, but it instead looks forward to the following
16270 fields in order to identify matches and rebuild progressive frames.
16272 To produce content with an even framerate, insert the fps filter after
16273 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
16274 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
16276 The filter accepts the following options:
16283 These options set the amount of "junk" to ignore at the left, right, top, and
16284 bottom of the image, respectively. Left and right are in units of 8 pixels,
16285 while top and bottom are in units of 2 lines.
16286 The default is 8 pixels on each side.
16289 Set the strict breaks. Setting this option to 1 will reduce the chances of
16290 filter generating an occasional mismatched frame, but it may also cause an
16291 excessive number of frames to be dropped during high motion sequences.
16292 Conversely, setting it to -1 will make filter match fields more easily.
16293 This may help processing of video where there is slight blurring between
16294 the fields, but may also cause there to be interlaced frames in the output.
16295 Default value is @code{0}.
16298 Set the metric plane to use. It accepts the following values:
16304 Use chroma blue plane.
16307 Use chroma red plane.
16310 This option may be set to use chroma plane instead of the default luma plane
16311 for doing filter's computations. This may improve accuracy on very clean
16312 source material, but more likely will decrease accuracy, especially if there
16313 is chroma noise (rainbow effect) or any grayscale video.
16314 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
16315 load and make pullup usable in realtime on slow machines.
16318 For best results (without duplicated frames in the output file) it is
16319 necessary to change the output frame rate. For example, to inverse
16320 telecine NTSC input:
16322 ffmpeg -i input -vf pullup -r 24000/1001 ...
16327 Change video quantization parameters (QP).
16329 The filter accepts the following option:
16333 Set expression for quantization parameter.
16336 The expression is evaluated through the eval API and can contain, among others,
16337 the following constants:
16341 1 if index is not 129, 0 otherwise.
16344 Sequential index starting from -129 to 128.
16347 @subsection Examples
16351 Some equation like:
16359 Flush video frames from internal cache of frames into a random order.
16360 No frame is discarded.
16361 Inspired by @ref{frei0r} nervous filter.
16365 Set size in number of frames of internal cache, in range from @code{2} to
16366 @code{512}. Default is @code{30}.
16369 Set seed for random number generator, must be an integer included between
16370 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
16371 less than @code{0}, the filter will try to use a good random seed on a
16375 @section readeia608
16377 Read closed captioning (EIA-608) information from the top lines of a video frame.
16379 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
16380 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
16381 with EIA-608 data (starting from 0). A description of each metadata value follows:
16384 @item lavfi.readeia608.X.cc
16385 The two bytes stored as EIA-608 data (printed in hexadecimal).
16387 @item lavfi.readeia608.X.line
16388 The number of the line on which the EIA-608 data was identified and read.
16391 This filter accepts the following options:
16395 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16398 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16401 Set the ratio of width reserved for sync code detection.
16402 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16405 Enable checking the parity bit. In the event of a parity error, the filter will output
16406 @code{0x00} for that character. Default is false.
16409 Lowpass lines prior to further processing. Default is enabled.
16412 @subsection Commands
16414 This filter supports the all above options as @ref{commands}.
16416 @subsection Examples
16420 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16422 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
16428 Read vertical interval timecode (VITC) information from the top lines of a
16431 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16432 timecode value, if a valid timecode has been detected. Further metadata key
16433 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16434 timecode data has been found or not.
16436 This filter accepts the following options:
16440 Set the maximum number of lines to scan for VITC data. If the value is set to
16441 @code{-1} the full video frame is scanned. Default is @code{45}.
16444 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16445 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16448 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16449 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16452 @subsection Examples
16456 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16457 draw @code{--:--:--:--} as a placeholder:
16459 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16465 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16467 Destination pixel at position (X, Y) will be picked from source (x, y) position
16468 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16469 value for pixel will be used for destination pixel.
16471 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16472 will have Xmap/Ymap video stream dimensions.
16473 Xmap and Ymap input video streams are 16bit depth, single channel.
16477 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16478 Default is @code{color}.
16481 Specify the color of the unmapped pixels. For the syntax of this option,
16482 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16483 manual,ffmpeg-utils}. Default color is @code{black}.
16486 @section removegrain
16488 The removegrain filter is a spatial denoiser for progressive video.
16492 Set mode for the first plane.
16495 Set mode for the second plane.
16498 Set mode for the third plane.
16501 Set mode for the fourth plane.
16504 Range of mode is from 0 to 24. Description of each mode follows:
16508 Leave input plane unchanged. Default.
16511 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16514 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16517 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16520 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16521 This is equivalent to a median filter.
16524 Line-sensitive clipping giving the minimal change.
16527 Line-sensitive clipping, intermediate.
16530 Line-sensitive clipping, intermediate.
16533 Line-sensitive clipping, intermediate.
16536 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16539 Replaces the target pixel with the closest neighbour.
16542 [1 2 1] horizontal and vertical kernel blur.
16548 Bob mode, interpolates top field from the line where the neighbours
16549 pixels are the closest.
16552 Bob mode, interpolates bottom field from the line where the neighbours
16553 pixels are the closest.
16556 Bob mode, interpolates top field. Same as 13 but with a more complicated
16557 interpolation formula.
16560 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16561 interpolation formula.
16564 Clips the pixel with the minimum and maximum of respectively the maximum and
16565 minimum of each pair of opposite neighbour pixels.
16568 Line-sensitive clipping using opposite neighbours whose greatest distance from
16569 the current pixel is minimal.
16572 Replaces the pixel with the average of its 8 neighbours.
16575 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16578 Clips pixels using the averages of opposite neighbour.
16581 Same as mode 21 but simpler and faster.
16584 Small edge and halo removal, but reputed useless.
16590 @section removelogo
16592 Suppress a TV station logo, using an image file to determine which
16593 pixels comprise the logo. It works by filling in the pixels that
16594 comprise the logo with neighboring pixels.
16596 The filter accepts the following options:
16600 Set the filter bitmap file, which can be any image format supported by
16601 libavformat. The width and height of the image file must match those of the
16602 video stream being processed.
16605 Pixels in the provided bitmap image with a value of zero are not
16606 considered part of the logo, non-zero pixels are considered part of
16607 the logo. If you use white (255) for the logo and black (0) for the
16608 rest, you will be safe. For making the filter bitmap, it is
16609 recommended to take a screen capture of a black frame with the logo
16610 visible, and then using a threshold filter followed by the erode
16611 filter once or twice.
16613 If needed, little splotches can be fixed manually. Remember that if
16614 logo pixels are not covered, the filter quality will be much
16615 reduced. Marking too many pixels as part of the logo does not hurt as
16616 much, but it will increase the amount of blurring needed to cover over
16617 the image and will destroy more information than necessary, and extra
16618 pixels will slow things down on a large logo.
16620 @section repeatfields
16622 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16623 fields based on its value.
16627 Reverse a video clip.
16629 Warning: This filter requires memory to buffer the entire clip, so trimming
16632 @subsection Examples
16636 Take the first 5 seconds of a clip, and reverse it.
16643 Shift R/G/B/A pixels horizontally and/or vertically.
16645 The filter accepts the following options:
16648 Set amount to shift red horizontally.
16650 Set amount to shift red vertically.
16652 Set amount to shift green horizontally.
16654 Set amount to shift green vertically.
16656 Set amount to shift blue horizontally.
16658 Set amount to shift blue vertically.
16660 Set amount to shift alpha horizontally.
16662 Set amount to shift alpha vertically.
16664 Set edge mode, can be @var{smear}, default, or @var{warp}.
16667 @subsection Commands
16669 This filter supports the all above options as @ref{commands}.
16672 Apply roberts cross operator to input video stream.
16674 The filter accepts the following option:
16678 Set which planes will be processed, unprocessed planes will be copied.
16679 By default value 0xf, all planes will be processed.
16682 Set value which will be multiplied with filtered result.
16685 Set value which will be added to filtered result.
16688 @subsection Commands
16690 This filter supports the all above options as @ref{commands}.
16694 Rotate video by an arbitrary angle expressed in radians.
16696 The filter accepts the following options:
16698 A description of the optional parameters follows.
16701 Set an expression for the angle by which to rotate the input video
16702 clockwise, expressed as a number of radians. A negative value will
16703 result in a counter-clockwise rotation. By default it is set to "0".
16705 This expression is evaluated for each frame.
16708 Set the output width expression, default value is "iw".
16709 This expression is evaluated just once during configuration.
16712 Set the output height expression, default value is "ih".
16713 This expression is evaluated just once during configuration.
16716 Enable bilinear interpolation if set to 1, a value of 0 disables
16717 it. Default value is 1.
16720 Set the color used to fill the output area not covered by the rotated
16721 image. For the general syntax of this option, check the
16722 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16723 If the special value "none" is selected then no
16724 background is printed (useful for example if the background is never shown).
16726 Default value is "black".
16729 The expressions for the angle and the output size can contain the
16730 following constants and functions:
16734 sequential number of the input frame, starting from 0. It is always NAN
16735 before the first frame is filtered.
16738 time in seconds of the input frame, it is set to 0 when the filter is
16739 configured. It is always NAN before the first frame is filtered.
16743 horizontal and vertical chroma subsample values. For example for the
16744 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16748 the input video width and height
16752 the output width and height, that is the size of the padded area as
16753 specified by the @var{width} and @var{height} expressions
16757 the minimal width/height required for completely containing the input
16758 video rotated by @var{a} radians.
16760 These are only available when computing the @option{out_w} and
16761 @option{out_h} expressions.
16764 @subsection Examples
16768 Rotate the input by PI/6 radians clockwise:
16774 Rotate the input by PI/6 radians counter-clockwise:
16780 Rotate the input by 45 degrees clockwise:
16786 Apply a constant rotation with period T, starting from an angle of PI/3:
16788 rotate=PI/3+2*PI*t/T
16792 Make the input video rotation oscillating with a period of T
16793 seconds and an amplitude of A radians:
16795 rotate=A*sin(2*PI/T*t)
16799 Rotate the video, output size is chosen so that the whole rotating
16800 input video is always completely contained in the output:
16802 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16806 Rotate the video, reduce the output size so that no background is ever
16809 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16813 @subsection Commands
16815 The filter supports the following commands:
16819 Set the angle expression.
16820 The command accepts the same syntax of the corresponding option.
16822 If the specified expression is not valid, it is kept at its current
16828 Apply Shape Adaptive Blur.
16830 The filter accepts the following options:
16833 @item luma_radius, lr
16834 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16835 value is 1.0. A greater value will result in a more blurred image, and
16836 in slower processing.
16838 @item luma_pre_filter_radius, lpfr
16839 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16842 @item luma_strength, ls
16843 Set luma maximum difference between pixels to still be considered, must
16844 be a value in the 0.1-100.0 range, default value is 1.0.
16846 @item chroma_radius, cr
16847 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16848 greater value will result in a more blurred image, and in slower
16851 @item chroma_pre_filter_radius, cpfr
16852 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16854 @item chroma_strength, cs
16855 Set chroma maximum difference between pixels to still be considered,
16856 must be a value in the -0.9-100.0 range.
16859 Each chroma option value, if not explicitly specified, is set to the
16860 corresponding luma option value.
16865 Scale (resize) the input video, using the libswscale library.
16867 The scale filter forces the output display aspect ratio to be the same
16868 of the input, by changing the output sample aspect ratio.
16870 If the input image format is different from the format requested by
16871 the next filter, the scale filter will convert the input to the
16874 @subsection Options
16875 The filter accepts the following options, or any of the options
16876 supported by the libswscale scaler.
16878 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16879 the complete list of scaler options.
16884 Set the output video dimension expression. Default value is the input
16887 If the @var{width} or @var{w} value is 0, the input width is used for
16888 the output. If the @var{height} or @var{h} value is 0, the input height
16889 is used for the output.
16891 If one and only one of the values is -n with n >= 1, the scale filter
16892 will use a value that maintains the aspect ratio of the input image,
16893 calculated from the other specified dimension. After that it will,
16894 however, make sure that the calculated dimension is divisible by n and
16895 adjust the value if necessary.
16897 If both values are -n with n >= 1, the behavior will be identical to
16898 both values being set to 0 as previously detailed.
16900 See below for the list of accepted constants for use in the dimension
16904 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16908 Only evaluate expressions once during the filter initialization or when a command is processed.
16911 Evaluate expressions for each incoming frame.
16915 Default value is @samp{init}.
16919 Set the interlacing mode. It accepts the following values:
16923 Force interlaced aware scaling.
16926 Do not apply interlaced scaling.
16929 Select interlaced aware scaling depending on whether the source frames
16930 are flagged as interlaced or not.
16933 Default value is @samp{0}.
16936 Set libswscale scaling flags. See
16937 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16938 complete list of values. If not explicitly specified the filter applies
16942 @item param0, param1
16943 Set libswscale input parameters for scaling algorithms that need them. See
16944 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16945 complete documentation. If not explicitly specified the filter applies
16951 Set the video size. For the syntax of this option, check the
16952 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16954 @item in_color_matrix
16955 @item out_color_matrix
16956 Set in/output YCbCr color space type.
16958 This allows the autodetected value to be overridden as well as allows forcing
16959 a specific value used for the output and encoder.
16961 If not specified, the color space type depends on the pixel format.
16967 Choose automatically.
16970 Format conforming to International Telecommunication Union (ITU)
16971 Recommendation BT.709.
16974 Set color space conforming to the United States Federal Communications
16975 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16980 Set color space conforming to:
16984 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16987 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16990 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16995 Set color space conforming to SMPTE ST 240:1999.
16998 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
17003 Set in/output YCbCr sample range.
17005 This allows the autodetected value to be overridden as well as allows forcing
17006 a specific value used for the output and encoder. If not specified, the
17007 range depends on the pixel format. Possible values:
17011 Choose automatically.
17014 Set full range (0-255 in case of 8-bit luma).
17016 @item mpeg/limited/tv
17017 Set "MPEG" range (16-235 in case of 8-bit luma).
17020 @item force_original_aspect_ratio
17021 Enable decreasing or increasing output video width or height if necessary to
17022 keep the original aspect ratio. Possible values:
17026 Scale the video as specified and disable this feature.
17029 The output video dimensions will automatically be decreased if needed.
17032 The output video dimensions will automatically be increased if needed.
17036 One useful instance of this option is that when you know a specific device's
17037 maximum allowed resolution, you can use this to limit the output video to
17038 that, while retaining the aspect ratio. For example, device A allows
17039 1280x720 playback, and your video is 1920x800. Using this option (set it to
17040 decrease) and specifying 1280x720 to the command line makes the output
17043 Please note that this is a different thing than specifying -1 for @option{w}
17044 or @option{h}, you still need to specify the output resolution for this option
17047 @item force_divisible_by
17048 Ensures that both the output dimensions, width and height, are divisible by the
17049 given integer when used together with @option{force_original_aspect_ratio}. This
17050 works similar to using @code{-n} in the @option{w} and @option{h} options.
17052 This option respects the value set for @option{force_original_aspect_ratio},
17053 increasing or decreasing the resolution accordingly. The video's aspect ratio
17054 may be slightly modified.
17056 This option can be handy if you need to have a video fit within or exceed
17057 a defined resolution using @option{force_original_aspect_ratio} but also have
17058 encoder restrictions on width or height divisibility.
17062 The values of the @option{w} and @option{h} options are expressions
17063 containing the following constants:
17068 The input width and height
17072 These are the same as @var{in_w} and @var{in_h}.
17076 The output (scaled) width and height
17080 These are the same as @var{out_w} and @var{out_h}
17083 The same as @var{iw} / @var{ih}
17086 input sample aspect ratio
17089 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
17093 horizontal and vertical input chroma subsample values. For example for the
17094 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17098 horizontal and vertical output chroma subsample values. For example for the
17099 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17102 The (sequential) number of the input frame, starting from 0.
17103 Only available with @code{eval=frame}.
17106 The presentation timestamp of the input frame, expressed as a number of
17107 seconds. Only available with @code{eval=frame}.
17110 The position (byte offset) of the frame in the input stream, or NaN if
17111 this information is unavailable and/or meaningless (for example in case of synthetic video).
17112 Only available with @code{eval=frame}.
17115 @subsection Examples
17119 Scale the input video to a size of 200x100
17124 This is equivalent to:
17135 Specify a size abbreviation for the output size:
17140 which can also be written as:
17146 Scale the input to 2x:
17148 scale=w=2*iw:h=2*ih
17152 The above is the same as:
17154 scale=2*in_w:2*in_h
17158 Scale the input to 2x with forced interlaced scaling:
17160 scale=2*iw:2*ih:interl=1
17164 Scale the input to half size:
17166 scale=w=iw/2:h=ih/2
17170 Increase the width, and set the height to the same size:
17176 Seek Greek harmony:
17183 Increase the height, and set the width to 3/2 of the height:
17185 scale=w=3/2*oh:h=3/5*ih
17189 Increase the size, making the size a multiple of the chroma
17192 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17196 Increase the width to a maximum of 500 pixels,
17197 keeping the same aspect ratio as the input:
17199 scale=w='min(500\, iw*3/2):h=-1'
17203 Make pixels square by combining scale and setsar:
17205 scale='trunc(ih*dar):ih',setsar=1/1
17209 Make pixels square by combining scale and setsar,
17210 making sure the resulting resolution is even (required by some codecs):
17212 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17216 @subsection Commands
17218 This filter supports the following commands:
17222 Set the output video dimension expression.
17223 The command accepts the same syntax of the corresponding option.
17225 If the specified expression is not valid, it is kept at its current
17231 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
17232 format conversion on CUDA video frames. Setting the output width and height
17233 works in the same way as for the @var{scale} filter.
17235 The following additional options are accepted:
17238 The pixel format of the output CUDA frames. If set to the string "same" (the
17239 default), the input format will be kept. Note that automatic format negotiation
17240 and conversion is not yet supported for hardware frames
17243 The interpolation algorithm used for resizing. One of the following:
17250 @item cubic2p_bspline
17251 2-parameter cubic (B=1, C=0)
17253 @item cubic2p_catmullrom
17254 2-parameter cubic (B=0, C=1/2)
17256 @item cubic2p_b05c03
17257 2-parameter cubic (B=1/2, C=3/10)
17265 @item force_original_aspect_ratio
17266 Enable decreasing or increasing output video width or height if necessary to
17267 keep the original aspect ratio. Possible values:
17271 Scale the video as specified and disable this feature.
17274 The output video dimensions will automatically be decreased if needed.
17277 The output video dimensions will automatically be increased if needed.
17281 One useful instance of this option is that when you know a specific device's
17282 maximum allowed resolution, you can use this to limit the output video to
17283 that, while retaining the aspect ratio. For example, device A allows
17284 1280x720 playback, and your video is 1920x800. Using this option (set it to
17285 decrease) and specifying 1280x720 to the command line makes the output
17288 Please note that this is a different thing than specifying -1 for @option{w}
17289 or @option{h}, you still need to specify the output resolution for this option
17292 @item force_divisible_by
17293 Ensures that both the output dimensions, width and height, are divisible by the
17294 given integer when used together with @option{force_original_aspect_ratio}. This
17295 works similar to using @code{-n} in the @option{w} and @option{h} options.
17297 This option respects the value set for @option{force_original_aspect_ratio},
17298 increasing or decreasing the resolution accordingly. The video's aspect ratio
17299 may be slightly modified.
17301 This option can be handy if you need to have a video fit within or exceed
17302 a defined resolution using @option{force_original_aspect_ratio} but also have
17303 encoder restrictions on width or height divisibility.
17309 Scale (resize) the input video, based on a reference video.
17311 See the scale filter for available options, scale2ref supports the same but
17312 uses the reference video instead of the main input as basis. scale2ref also
17313 supports the following additional constants for the @option{w} and
17314 @option{h} options:
17319 The main input video's width and height
17322 The same as @var{main_w} / @var{main_h}
17325 The main input video's sample aspect ratio
17327 @item main_dar, mdar
17328 The main input video's display aspect ratio. Calculated from
17329 @code{(main_w / main_h) * main_sar}.
17333 The main input video's horizontal and vertical chroma subsample values.
17334 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
17338 The (sequential) number of the main input frame, starting from 0.
17339 Only available with @code{eval=frame}.
17342 The presentation timestamp of the main input frame, expressed as a number of
17343 seconds. Only available with @code{eval=frame}.
17346 The position (byte offset) of the frame in the main input stream, or NaN if
17347 this information is unavailable and/or meaningless (for example in case of synthetic video).
17348 Only available with @code{eval=frame}.
17351 @subsection Examples
17355 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
17357 'scale2ref[b][a];[a][b]overlay'
17361 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
17363 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17367 @subsection Commands
17369 This filter supports the following commands:
17373 Set the output video dimension expression.
17374 The command accepts the same syntax of the corresponding option.
17376 If the specified expression is not valid, it is kept at its current
17381 Scroll input video horizontally and/or vertically by constant speed.
17383 The filter accepts the following options:
17385 @item horizontal, h
17386 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
17387 Negative values changes scrolling direction.
17390 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
17391 Negative values changes scrolling direction.
17394 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
17397 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
17400 @subsection Commands
17402 This filter supports the following @ref{commands}:
17404 @item horizontal, h
17405 Set the horizontal scrolling speed.
17407 Set the vertical scrolling speed.
17413 Detect video scene change.
17415 This filter sets frame metadata with mafd between frame, the scene score, and
17416 forward the frame to the next filter, so they can use these metadata to detect
17417 scene change or others.
17419 In addition, this filter logs a message and sets frame metadata when it detects
17420 a scene change by @option{threshold}.
17422 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17424 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17425 to detect scene change.
17427 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17428 detect scene change with @option{threshold}.
17430 The filter accepts the following options:
17434 Set the scene change detection threshold as a percentage of maximum change. Good
17435 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17438 Default value is @code{10.}.
17441 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17442 You can enable it if you want to get snapshot of scene change frames only.
17445 @anchor{selectivecolor}
17446 @section selectivecolor
17448 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17449 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17450 by the "purity" of the color (that is, how saturated it already is).
17452 This filter is similar to the Adobe Photoshop Selective Color tool.
17454 The filter accepts the following options:
17457 @item correction_method
17458 Select color correction method.
17460 Available values are:
17463 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17466 Specified adjustments are relative to the original component value.
17468 Default is @code{absolute}.
17470 Adjustments for red pixels (pixels where the red component is the maximum)
17472 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17474 Adjustments for green pixels (pixels where the green component is the maximum)
17476 Adjustments for cyan pixels (pixels where the red component is the minimum)
17478 Adjustments for blue pixels (pixels where the blue component is the maximum)
17480 Adjustments for magenta pixels (pixels where the green component is the minimum)
17482 Adjustments for white pixels (pixels where all components are greater than 128)
17484 Adjustments for all pixels except pure black and pure white
17486 Adjustments for black pixels (pixels where all components are lesser than 128)
17488 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17491 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17492 4 space separated floating point adjustment values in the [-1,1] range,
17493 respectively to adjust the amount of cyan, magenta, yellow and black for the
17494 pixels of its range.
17496 @subsection Examples
17500 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17501 increase magenta by 27% in blue areas:
17503 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17507 Use a Photoshop selective color preset:
17509 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17513 @anchor{separatefields}
17514 @section separatefields
17516 The @code{separatefields} takes a frame-based video input and splits
17517 each frame into its components fields, producing a new half height clip
17518 with twice the frame rate and twice the frame count.
17520 This filter use field-dominance information in frame to decide which
17521 of each pair of fields to place first in the output.
17522 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17524 @section setdar, setsar
17526 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17529 This is done by changing the specified Sample (aka Pixel) Aspect
17530 Ratio, according to the following equation:
17532 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17535 Keep in mind that the @code{setdar} filter does not modify the pixel
17536 dimensions of the video frame. Also, the display aspect ratio set by
17537 this filter may be changed by later filters in the filterchain,
17538 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17541 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17542 the filter output video.
17544 Note that as a consequence of the application of this filter, the
17545 output display aspect ratio will change according to the equation
17548 Keep in mind that the sample aspect ratio set by the @code{setsar}
17549 filter may be changed by later filters in the filterchain, e.g. if
17550 another "setsar" or a "setdar" filter is applied.
17552 It accepts the following parameters:
17555 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17556 Set the aspect ratio used by the filter.
17558 The parameter can be a floating point number string, an expression, or
17559 a string of the form @var{num}:@var{den}, where @var{num} and
17560 @var{den} are the numerator and denominator of the aspect ratio. If
17561 the parameter is not specified, it is assumed the value "0".
17562 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17566 Set the maximum integer value to use for expressing numerator and
17567 denominator when reducing the expressed aspect ratio to a rational.
17568 Default value is @code{100}.
17572 The parameter @var{sar} is an expression containing
17573 the following constants:
17577 These are approximated values for the mathematical constants e
17578 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17581 The input width and height.
17584 These are the same as @var{w} / @var{h}.
17587 The input sample aspect ratio.
17590 The input display aspect ratio. It is the same as
17591 (@var{w} / @var{h}) * @var{sar}.
17594 Horizontal and vertical chroma subsample values. For example, for the
17595 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17598 @subsection Examples
17603 To change the display aspect ratio to 16:9, specify one of the following:
17610 To change the sample aspect ratio to 10:11, specify:
17616 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17617 1000 in the aspect ratio reduction, use the command:
17619 setdar=ratio=16/9:max=1000
17627 Force field for the output video frame.
17629 The @code{setfield} filter marks the interlace type field for the
17630 output frames. It does not change the input frame, but only sets the
17631 corresponding property, which affects how the frame is treated by
17632 following filters (e.g. @code{fieldorder} or @code{yadif}).
17634 The filter accepts the following options:
17639 Available values are:
17643 Keep the same field property.
17646 Mark the frame as bottom-field-first.
17649 Mark the frame as top-field-first.
17652 Mark the frame as progressive.
17659 Force frame parameter for the output video frame.
17661 The @code{setparams} filter marks interlace and color range for the
17662 output frames. It does not change the input frame, but only sets the
17663 corresponding property, which affects how the frame is treated by
17668 Available values are:
17672 Keep the same field property (default).
17675 Mark the frame as bottom-field-first.
17678 Mark the frame as top-field-first.
17681 Mark the frame as progressive.
17685 Available values are:
17689 Keep the same color range property (default).
17691 @item unspecified, unknown
17692 Mark the frame as unspecified color range.
17694 @item limited, tv, mpeg
17695 Mark the frame as limited range.
17697 @item full, pc, jpeg
17698 Mark the frame as full range.
17701 @item color_primaries
17702 Set the color primaries.
17703 Available values are:
17707 Keep the same color primaries property (default).
17724 Set the color transfer.
17725 Available values are:
17729 Keep the same color trc property (default).
17751 Set the colorspace.
17752 Available values are:
17756 Keep the same colorspace property (default).
17769 @item chroma-derived-nc
17770 @item chroma-derived-c
17777 Show a line containing various information for each input video frame.
17778 The input video is not modified.
17780 This filter supports the following options:
17784 Calculate checksums of each plane. By default enabled.
17787 The shown line contains a sequence of key/value pairs of the form
17788 @var{key}:@var{value}.
17790 The following values are shown in the output:
17794 The (sequential) number of the input frame, starting from 0.
17797 The Presentation TimeStamp of the input frame, expressed as a number of
17798 time base units. The time base unit depends on the filter input pad.
17801 The Presentation TimeStamp of the input frame, expressed as a number of
17805 The position of the frame in the input stream, or -1 if this information is
17806 unavailable and/or meaningless (for example in case of synthetic video).
17809 The pixel format name.
17812 The sample aspect ratio of the input frame, expressed in the form
17813 @var{num}/@var{den}.
17816 The size of the input frame. For the syntax of this option, check the
17817 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17820 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17821 for bottom field first).
17824 This is 1 if the frame is a key frame, 0 otherwise.
17827 The picture type of the input frame ("I" for an I-frame, "P" for a
17828 P-frame, "B" for a B-frame, or "?" for an unknown type).
17829 Also refer to the documentation of the @code{AVPictureType} enum and of
17830 the @code{av_get_picture_type_char} function defined in
17831 @file{libavutil/avutil.h}.
17834 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17836 @item plane_checksum
17837 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17838 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17841 The mean value of pixels in each plane of the input frame, expressed in the form
17842 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17845 The standard deviation of pixel values in each plane of the input frame, expressed
17846 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17850 @section showpalette
17852 Displays the 256 colors palette of each frame. This filter is only relevant for
17853 @var{pal8} pixel format frames.
17855 It accepts the following option:
17859 Set the size of the box used to represent one palette color entry. Default is
17860 @code{30} (for a @code{30x30} pixel box).
17863 @section shuffleframes
17865 Reorder and/or duplicate and/or drop video frames.
17867 It accepts the following parameters:
17871 Set the destination indexes of input frames.
17872 This is space or '|' separated list of indexes that maps input frames to output
17873 frames. Number of indexes also sets maximal value that each index may have.
17874 '-1' index have special meaning and that is to drop frame.
17877 The first frame has the index 0. The default is to keep the input unchanged.
17879 @subsection Examples
17883 Swap second and third frame of every three frames of the input:
17885 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17889 Swap 10th and 1st frame of every ten frames of the input:
17891 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17895 @section shufflepixels
17897 Reorder pixels in video frames.
17899 This filter accepts the following options:
17903 Set shuffle direction. Can be forward or inverse direction.
17904 Default direction is forward.
17907 Set shuffle mode. Can be horizontal, vertical or block mode.
17911 Set shuffle block_size. In case of horizontal shuffle mode only width
17912 part of size is used, and in case of vertical shuffle mode only height
17913 part of size is used.
17916 Set random seed used with shuffling pixels. Mainly useful to set to be able
17917 to reverse filtering process to get original input.
17918 For example, to reverse forward shuffle you need to use same parameters
17919 and exact same seed and to set direction to inverse.
17922 @section shuffleplanes
17924 Reorder and/or duplicate video planes.
17926 It accepts the following parameters:
17931 The index of the input plane to be used as the first output plane.
17934 The index of the input plane to be used as the second output plane.
17937 The index of the input plane to be used as the third output plane.
17940 The index of the input plane to be used as the fourth output plane.
17944 The first plane has the index 0. The default is to keep the input unchanged.
17946 @subsection Examples
17950 Swap the second and third planes of the input:
17952 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17956 @anchor{signalstats}
17957 @section signalstats
17958 Evaluate various visual metrics that assist in determining issues associated
17959 with the digitization of analog video media.
17961 By default the filter will log these metadata values:
17965 Display the minimal Y value contained within the input frame. Expressed in
17969 Display the Y value at the 10% percentile within the input frame. Expressed in
17973 Display the average Y value within the input frame. Expressed in range of
17977 Display the Y value at the 90% percentile within the input frame. Expressed in
17981 Display the maximum Y value contained within the input frame. Expressed in
17985 Display the minimal U value contained within the input frame. Expressed in
17989 Display the U value at the 10% percentile within the input frame. Expressed in
17993 Display the average U value within the input frame. Expressed in range of
17997 Display the U value at the 90% percentile within the input frame. Expressed in
18001 Display the maximum U value contained within the input frame. Expressed in
18005 Display the minimal V value contained within the input frame. Expressed in
18009 Display the V value at the 10% percentile within the input frame. Expressed in
18013 Display the average V value within the input frame. Expressed in range of
18017 Display the V value at the 90% percentile within the input frame. Expressed in
18021 Display the maximum V value contained within the input frame. Expressed in
18025 Display the minimal saturation value contained within the input frame.
18026 Expressed in range of [0-~181.02].
18029 Display the saturation value at the 10% percentile within the input frame.
18030 Expressed in range of [0-~181.02].
18033 Display the average saturation value within the input frame. Expressed in range
18037 Display the saturation value at the 90% percentile within the input frame.
18038 Expressed in range of [0-~181.02].
18041 Display the maximum saturation value contained within the input frame.
18042 Expressed in range of [0-~181.02].
18045 Display the median value for hue within the input frame. Expressed in range of
18049 Display the average value for hue within the input frame. Expressed in range of
18053 Display the average of sample value difference between all values of the Y
18054 plane in the current frame and corresponding values of the previous input frame.
18055 Expressed in range of [0-255].
18058 Display the average of sample value difference between all values of the U
18059 plane in the current frame and corresponding values of the previous input frame.
18060 Expressed in range of [0-255].
18063 Display the average of sample value difference between all values of the V
18064 plane in the current frame and corresponding values of the previous input frame.
18065 Expressed in range of [0-255].
18068 Display bit depth of Y plane in current frame.
18069 Expressed in range of [0-16].
18072 Display bit depth of U plane in current frame.
18073 Expressed in range of [0-16].
18076 Display bit depth of V plane in current frame.
18077 Expressed in range of [0-16].
18080 The filter accepts the following options:
18086 @option{stat} specify an additional form of image analysis.
18087 @option{out} output video with the specified type of pixel highlighted.
18089 Both options accept the following values:
18093 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
18094 unlike the neighboring pixels of the same field. Examples of temporal outliers
18095 include the results of video dropouts, head clogs, or tape tracking issues.
18098 Identify @var{vertical line repetition}. Vertical line repetition includes
18099 similar rows of pixels within a frame. In born-digital video vertical line
18100 repetition is common, but this pattern is uncommon in video digitized from an
18101 analog source. When it occurs in video that results from the digitization of an
18102 analog source it can indicate concealment from a dropout compensator.
18105 Identify pixels that fall outside of legal broadcast range.
18109 Set the highlight color for the @option{out} option. The default color is
18113 @subsection Examples
18117 Output data of various video metrics:
18119 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
18123 Output specific data about the minimum and maximum values of the Y plane per frame:
18125 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
18129 Playback video while highlighting pixels that are outside of broadcast range in red.
18131 ffplay example.mov -vf signalstats="out=brng:color=red"
18135 Playback video with signalstats metadata drawn over the frame.
18137 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18140 The contents of signalstat_drawtext.txt used in the command are:
18143 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
18144 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
18145 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
18146 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
18154 Calculates the MPEG-7 Video Signature. The filter can handle more than one
18155 input. In this case the matching between the inputs can be calculated additionally.
18156 The filter always passes through the first input. The signature of each stream can
18157 be written into a file.
18159 It accepts the following options:
18163 Enable or disable the matching process.
18165 Available values are:
18169 Disable the calculation of a matching (default).
18171 Calculate the matching for the whole video and output whether the whole video
18172 matches or only parts.
18174 Calculate only until a matching is found or the video ends. Should be faster in
18179 Set the number of inputs. The option value must be a non negative integer.
18180 Default value is 1.
18183 Set the path to which the output is written. If there is more than one input,
18184 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
18185 integer), that will be replaced with the input number. If no filename is
18186 specified, no output will be written. This is the default.
18189 Choose the output format.
18191 Available values are:
18195 Use the specified binary representation (default).
18197 Use the specified xml representation.
18201 Set threshold to detect one word as similar. The option value must be an integer
18202 greater than zero. The default value is 9000.
18205 Set threshold to detect all words as similar. The option value must be an integer
18206 greater than zero. The default value is 60000.
18209 Set threshold to detect frames as similar. The option value must be an integer
18210 greater than zero. The default value is 116.
18213 Set the minimum length of a sequence in frames to recognize it as matching
18214 sequence. The option value must be a non negative integer value.
18215 The default value is 0.
18218 Set the minimum relation, that matching frames to all frames must have.
18219 The option value must be a double value between 0 and 1. The default value is 0.5.
18222 @subsection Examples
18226 To calculate the signature of an input video and store it in signature.bin:
18228 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18232 To detect whether two videos match and store the signatures in XML format in
18233 signature0.xml and signature1.xml:
18235 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 -
18243 Blur the input video without impacting the outlines.
18245 It accepts the following options:
18248 @item luma_radius, lr
18249 Set the luma radius. The option value must be a float number in
18250 the range [0.1,5.0] that specifies the variance of the gaussian filter
18251 used to blur the image (slower if larger). Default value is 1.0.
18253 @item luma_strength, ls
18254 Set the luma strength. The option value must be a float number
18255 in the range [-1.0,1.0] that configures the blurring. A value included
18256 in [0.0,1.0] will blur the image whereas a value included in
18257 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18259 @item luma_threshold, lt
18260 Set the luma threshold used as a coefficient to determine
18261 whether a pixel should be blurred or not. The option value must be an
18262 integer in the range [-30,30]. A value of 0 will filter all the image,
18263 a value included in [0,30] will filter flat areas and a value included
18264 in [-30,0] will filter edges. Default value is 0.
18266 @item chroma_radius, cr
18267 Set the chroma radius. The option value must be a float number in
18268 the range [0.1,5.0] that specifies the variance of the gaussian filter
18269 used to blur the image (slower if larger). Default value is @option{luma_radius}.
18271 @item chroma_strength, cs
18272 Set the chroma strength. The option value must be a float number
18273 in the range [-1.0,1.0] that configures the blurring. A value included
18274 in [0.0,1.0] will blur the image whereas a value included in
18275 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
18277 @item chroma_threshold, ct
18278 Set the chroma threshold used as a coefficient to determine
18279 whether a pixel should be blurred or not. The option value must be an
18280 integer in the range [-30,30]. A value of 0 will filter all the image,
18281 a value included in [0,30] will filter flat areas and a value included
18282 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
18285 If a chroma option is not explicitly set, the corresponding luma value
18289 Apply sobel operator to input video stream.
18291 The filter accepts the following option:
18295 Set which planes will be processed, unprocessed planes will be copied.
18296 By default value 0xf, all planes will be processed.
18299 Set value which will be multiplied with filtered result.
18302 Set value which will be added to filtered result.
18305 @subsection Commands
18307 This filter supports the all above options as @ref{commands}.
18312 Apply a simple postprocessing filter that compresses and decompresses the image
18313 at several (or - in the case of @option{quality} level @code{6} - all) shifts
18314 and average the results.
18316 The filter accepts the following options:
18320 Set quality. This option defines the number of levels for averaging. It accepts
18321 an integer in the range 0-6. If set to @code{0}, the filter will have no
18322 effect. A value of @code{6} means the higher quality. For each increment of
18323 that value the speed drops by a factor of approximately 2. Default value is
18327 Force a constant quantization parameter. If not set, the filter will use the QP
18328 from the video stream (if available).
18331 Set thresholding mode. Available modes are:
18335 Set hard thresholding (default).
18337 Set soft thresholding (better de-ringing effect, but likely blurrier).
18340 @item use_bframe_qp
18341 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
18342 option may cause flicker since the B-Frames have often larger QP. Default is
18343 @code{0} (not enabled).
18346 @subsection Commands
18348 This filter supports the following commands:
18350 @item quality, level
18351 Set quality level. The value @code{max} can be used to set the maximum level,
18352 currently @code{6}.
18358 Scale the input by applying one of the super-resolution methods based on
18359 convolutional neural networks. Supported models:
18363 Super-Resolution Convolutional Neural Network model (SRCNN).
18364 See @url{https://arxiv.org/abs/1501.00092}.
18367 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18368 See @url{https://arxiv.org/abs/1609.05158}.
18371 Training scripts as well as scripts for model file (.pb) saving can be found at
18372 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
18373 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
18375 Native model files (.model) can be generated from TensorFlow model
18376 files (.pb) by using tools/python/convert.py
18378 The filter accepts the following options:
18382 Specify which DNN backend to use for model loading and execution. This option accepts
18383 the following values:
18387 Native implementation of DNN loading and execution.
18390 TensorFlow backend. To enable this backend you
18391 need to install the TensorFlow for C library (see
18392 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
18393 @code{--enable-libtensorflow}
18396 Default value is @samp{native}.
18399 Set path to model file specifying network architecture and its parameters.
18400 Note that different backends use different file formats. TensorFlow backend
18401 can load files for both formats, while native backend can load files for only
18405 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
18406 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
18407 input upscaled using bicubic upscaling with proper scale factor.
18410 This feature can also be finished with @ref{dnn_processing} filter.
18414 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
18416 This filter takes in input two input videos, the first input is
18417 considered the "main" source and is passed unchanged to the
18418 output. The second input is used as a "reference" video for computing
18421 Both video inputs must have the same resolution and pixel format for
18422 this filter to work correctly. Also it assumes that both inputs
18423 have the same number of frames, which are compared one by one.
18425 The filter stores the calculated SSIM of each frame.
18427 The description of the accepted parameters follows.
18430 @item stats_file, f
18431 If specified the filter will use the named file to save the SSIM of
18432 each individual frame. When filename equals "-" the data is sent to
18436 The file printed if @var{stats_file} is selected, contains a sequence of
18437 key/value pairs of the form @var{key}:@var{value} for each compared
18440 A description of each shown parameter follows:
18444 sequential number of the input frame, starting from 1
18446 @item Y, U, V, R, G, B
18447 SSIM of the compared frames for the component specified by the suffix.
18450 SSIM of the compared frames for the whole frame.
18453 Same as above but in dB representation.
18456 This filter also supports the @ref{framesync} options.
18458 @subsection Examples
18463 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18464 [main][ref] ssim="stats_file=stats.log" [out]
18467 On this example the input file being processed is compared with the
18468 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18469 is stored in @file{stats.log}.
18472 Another example with both psnr and ssim at same time:
18474 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18478 Another example with different containers:
18480 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 -
18486 Convert between different stereoscopic image formats.
18488 The filters accept the following options:
18492 Set stereoscopic image format of input.
18494 Available values for input image formats are:
18497 side by side parallel (left eye left, right eye right)
18500 side by side crosseye (right eye left, left eye right)
18503 side by side parallel with half width resolution
18504 (left eye left, right eye right)
18507 side by side crosseye with half width resolution
18508 (right eye left, left eye right)
18512 above-below (left eye above, right eye below)
18516 above-below (right eye above, left eye below)
18520 above-below with half height resolution
18521 (left eye above, right eye below)
18525 above-below with half height resolution
18526 (right eye above, left eye below)
18529 alternating frames (left eye first, right eye second)
18532 alternating frames (right eye first, left eye second)
18535 interleaved rows (left eye has top row, right eye starts on next row)
18538 interleaved rows (right eye has top row, left eye starts on next row)
18541 interleaved columns, left eye first
18544 interleaved columns, right eye first
18546 Default value is @samp{sbsl}.
18550 Set stereoscopic image format of output.
18554 side by side parallel (left eye left, right eye right)
18557 side by side crosseye (right eye left, left eye right)
18560 side by side parallel with half width resolution
18561 (left eye left, right eye right)
18564 side by side crosseye with half width resolution
18565 (right eye left, left eye right)
18569 above-below (left eye above, right eye below)
18573 above-below (right eye above, left eye below)
18577 above-below with half height resolution
18578 (left eye above, right eye below)
18582 above-below with half height resolution
18583 (right eye above, left eye below)
18586 alternating frames (left eye first, right eye second)
18589 alternating frames (right eye first, left eye second)
18592 interleaved rows (left eye has top row, right eye starts on next row)
18595 interleaved rows (right eye has top row, left eye starts on next row)
18598 anaglyph red/blue gray
18599 (red filter on left eye, blue filter on right eye)
18602 anaglyph red/green gray
18603 (red filter on left eye, green filter on right eye)
18606 anaglyph red/cyan gray
18607 (red filter on left eye, cyan filter on right eye)
18610 anaglyph red/cyan half colored
18611 (red filter on left eye, cyan filter on right eye)
18614 anaglyph red/cyan color
18615 (red filter on left eye, cyan filter on right eye)
18618 anaglyph red/cyan color optimized with the least squares projection of dubois
18619 (red filter on left eye, cyan filter on right eye)
18622 anaglyph green/magenta gray
18623 (green filter on left eye, magenta filter on right eye)
18626 anaglyph green/magenta half colored
18627 (green filter on left eye, magenta filter on right eye)
18630 anaglyph green/magenta colored
18631 (green filter on left eye, magenta filter on right eye)
18634 anaglyph green/magenta color optimized with the least squares projection of dubois
18635 (green filter on left eye, magenta filter on right eye)
18638 anaglyph yellow/blue gray
18639 (yellow filter on left eye, blue filter on right eye)
18642 anaglyph yellow/blue half colored
18643 (yellow filter on left eye, blue filter on right eye)
18646 anaglyph yellow/blue colored
18647 (yellow filter on left eye, blue filter on right eye)
18650 anaglyph yellow/blue color optimized with the least squares projection of dubois
18651 (yellow filter on left eye, blue filter on right eye)
18654 mono output (left eye only)
18657 mono output (right eye only)
18660 checkerboard, left eye first
18663 checkerboard, right eye first
18666 interleaved columns, left eye first
18669 interleaved columns, right eye first
18675 Default value is @samp{arcd}.
18678 @subsection Examples
18682 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18688 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18694 @section streamselect, astreamselect
18695 Select video or audio streams.
18697 The filter accepts the following options:
18701 Set number of inputs. Default is 2.
18704 Set input indexes to remap to outputs.
18707 @subsection Commands
18709 The @code{streamselect} and @code{astreamselect} filter supports the following
18714 Set input indexes to remap to outputs.
18717 @subsection Examples
18721 Select first 5 seconds 1st stream and rest of time 2nd stream:
18723 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18727 Same as above, but for audio:
18729 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18736 Draw subtitles on top of input video using the libass library.
18738 To enable compilation of this filter you need to configure FFmpeg with
18739 @code{--enable-libass}. This filter also requires a build with libavcodec and
18740 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18741 Alpha) subtitles format.
18743 The filter accepts the following options:
18747 Set the filename of the subtitle file to read. It must be specified.
18749 @item original_size
18750 Specify the size of the original video, the video for which the ASS file
18751 was composed. For the syntax of this option, check the
18752 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18753 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18754 correctly scale the fonts if the aspect ratio has been changed.
18757 Set a directory path containing fonts that can be used by the filter.
18758 These fonts will be used in addition to whatever the font provider uses.
18761 Process alpha channel, by default alpha channel is untouched.
18764 Set subtitles input character encoding. @code{subtitles} filter only. Only
18765 useful if not UTF-8.
18767 @item stream_index, si
18768 Set subtitles stream index. @code{subtitles} filter only.
18771 Override default style or script info parameters of the subtitles. It accepts a
18772 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18775 If the first key is not specified, it is assumed that the first value
18776 specifies the @option{filename}.
18778 For example, to render the file @file{sub.srt} on top of the input
18779 video, use the command:
18784 which is equivalent to:
18786 subtitles=filename=sub.srt
18789 To render the default subtitles stream from file @file{video.mkv}, use:
18791 subtitles=video.mkv
18794 To render the second subtitles stream from that file, use:
18796 subtitles=video.mkv:si=1
18799 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18800 @code{DejaVu Serif}, use:
18802 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18805 @section super2xsai
18807 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18808 Interpolate) pixel art scaling algorithm.
18810 Useful for enlarging pixel art images without reducing sharpness.
18814 Swap two rectangular objects in video.
18816 This filter accepts the following options:
18826 Set 1st rect x coordinate.
18829 Set 1st rect y coordinate.
18832 Set 2nd rect x coordinate.
18835 Set 2nd rect y coordinate.
18837 All expressions are evaluated once for each frame.
18840 The all options are expressions containing the following constants:
18845 The input width and height.
18848 same as @var{w} / @var{h}
18851 input sample aspect ratio
18854 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18857 The number of the input frame, starting from 0.
18860 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18863 the position in the file of the input frame, NAN if unknown
18870 Blend successive video frames.
18876 Apply telecine process to the video.
18878 This filter accepts the following options:
18887 The default value is @code{top}.
18891 A string of numbers representing the pulldown pattern you wish to apply.
18892 The default value is @code{23}.
18896 Some typical patterns:
18901 24p: 2332 (preferred)
18908 24p: 222222222223 ("Euro pulldown")
18913 @section thistogram
18915 Compute and draw a color distribution histogram for the input video across time.
18917 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18918 at certain time, this filter shows also past histograms of number of frames defined
18919 by @code{width} option.
18921 The computed histogram is a representation of the color component
18922 distribution in an image.
18924 The filter accepts the following options:
18928 Set width of single color component output. Default value is @code{0}.
18929 Value of @code{0} means width will be picked from input video.
18930 This also set number of passed histograms to keep.
18931 Allowed range is [0, 8192].
18933 @item display_mode, d
18935 It accepts the following values:
18938 Per color component graphs are placed below each other.
18941 Per color component graphs are placed side by side.
18944 Presents information identical to that in the @code{parade}, except
18945 that the graphs representing color components are superimposed directly
18948 Default is @code{stack}.
18950 @item levels_mode, m
18951 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18952 Default is @code{linear}.
18954 @item components, c
18955 Set what color components to display.
18956 Default is @code{7}.
18959 Set background opacity. Default is @code{0.9}.
18962 Show envelope. Default is disabled.
18965 Set envelope color. Default is @code{gold}.
18970 Available values for slide is:
18973 Draw new frame when right border is reached.
18976 Replace old columns with new ones.
18979 Scroll from right to left.
18982 Scroll from left to right.
18985 Draw single picture.
18988 Default is @code{replace}.
18993 Apply threshold effect to video stream.
18995 This filter needs four video streams to perform thresholding.
18996 First stream is stream we are filtering.
18997 Second stream is holding threshold values, third stream is holding min values,
18998 and last, fourth stream is holding max values.
19000 The filter accepts the following option:
19004 Set which planes will be processed, unprocessed planes will be copied.
19005 By default value 0xf, all planes will be processed.
19008 For example if first stream pixel's component value is less then threshold value
19009 of pixel component from 2nd threshold stream, third stream value will picked,
19010 otherwise fourth stream pixel component value will be picked.
19012 Using color source filter one can perform various types of thresholding:
19014 @subsection Examples
19018 Binary threshold, using gray color as threshold:
19020 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
19024 Inverted binary threshold, using gray color as threshold:
19026 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
19030 Truncate binary threshold, using gray color as threshold:
19032 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
19036 Threshold to zero, using gray color as threshold:
19038 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
19042 Inverted threshold to zero, using gray color as threshold:
19044 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
19049 Select the most representative frame in a given sequence of consecutive frames.
19051 The filter accepts the following options:
19055 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
19056 will pick one of them, and then handle the next batch of @var{n} frames until
19057 the end. Default is @code{100}.
19060 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
19061 value will result in a higher memory usage, so a high value is not recommended.
19063 @subsection Examples
19067 Extract one picture each 50 frames:
19073 Complete example of a thumbnail creation with @command{ffmpeg}:
19075 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
19082 Tile several successive frames together.
19084 The @ref{untile} filter can do the reverse.
19086 The filter accepts the following options:
19091 Set the grid size (i.e. the number of lines and columns). For the syntax of
19092 this option, check the
19093 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19096 Set the maximum number of frames to render in the given area. It must be less
19097 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
19098 the area will be used.
19101 Set the outer border margin in pixels.
19104 Set the inner border thickness (i.e. the number of pixels between frames). For
19105 more advanced padding options (such as having different values for the edges),
19106 refer to the pad video filter.
19109 Specify the color of the unused area. For the syntax of this option, check the
19110 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
19111 The default value of @var{color} is "black".
19114 Set the number of frames to overlap when tiling several successive frames together.
19115 The value must be between @code{0} and @var{nb_frames - 1}.
19118 Set the number of frames to initially be empty before displaying first output frame.
19119 This controls how soon will one get first output frame.
19120 The value must be between @code{0} and @var{nb_frames - 1}.
19123 @subsection Examples
19127 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
19129 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
19131 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
19132 duplicating each output frame to accommodate the originally detected frame
19136 Display @code{5} pictures in an area of @code{3x2} frames,
19137 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
19138 mixed flat and named options:
19140 tile=3x2:nb_frames=5:padding=7:margin=2
19144 @section tinterlace
19146 Perform various types of temporal field interlacing.
19148 Frames are counted starting from 1, so the first input frame is
19151 The filter accepts the following options:
19156 Specify the mode of the interlacing. This option can also be specified
19157 as a value alone. See below for a list of values for this option.
19159 Available values are:
19163 Move odd frames into the upper field, even into the lower field,
19164 generating a double height frame at half frame rate.
19168 Frame 1 Frame 2 Frame 3 Frame 4
19170 11111 22222 33333 44444
19171 11111 22222 33333 44444
19172 11111 22222 33333 44444
19173 11111 22222 33333 44444
19187 Only output odd frames, even frames are dropped, generating a frame with
19188 unchanged height at half frame rate.
19193 Frame 1 Frame 2 Frame 3 Frame 4
19195 11111 22222 33333 44444
19196 11111 22222 33333 44444
19197 11111 22222 33333 44444
19198 11111 22222 33333 44444
19208 Only output even frames, odd frames are dropped, generating a frame with
19209 unchanged height at half frame rate.
19214 Frame 1 Frame 2 Frame 3 Frame 4
19216 11111 22222 33333 44444
19217 11111 22222 33333 44444
19218 11111 22222 33333 44444
19219 11111 22222 33333 44444
19229 Expand each frame to full height, but pad alternate lines with black,
19230 generating a frame with double height at the same input frame rate.
19235 Frame 1 Frame 2 Frame 3 Frame 4
19237 11111 22222 33333 44444
19238 11111 22222 33333 44444
19239 11111 22222 33333 44444
19240 11111 22222 33333 44444
19243 11111 ..... 33333 .....
19244 ..... 22222 ..... 44444
19245 11111 ..... 33333 .....
19246 ..... 22222 ..... 44444
19247 11111 ..... 33333 .....
19248 ..... 22222 ..... 44444
19249 11111 ..... 33333 .....
19250 ..... 22222 ..... 44444
19254 @item interleave_top, 4
19255 Interleave the upper field from odd frames with the lower field from
19256 even frames, generating a frame with unchanged height at half frame rate.
19261 Frame 1 Frame 2 Frame 3 Frame 4
19263 11111<- 22222 33333<- 44444
19264 11111 22222<- 33333 44444<-
19265 11111<- 22222 33333<- 44444
19266 11111 22222<- 33333 44444<-
19276 @item interleave_bottom, 5
19277 Interleave the lower field from odd frames with the upper field from
19278 even frames, generating a frame with unchanged height at half frame rate.
19283 Frame 1 Frame 2 Frame 3 Frame 4
19285 11111 22222<- 33333 44444<-
19286 11111<- 22222 33333<- 44444
19287 11111 22222<- 33333 44444<-
19288 11111<- 22222 33333<- 44444
19298 @item interlacex2, 6
19299 Double frame rate with unchanged height. Frames are inserted each
19300 containing the second temporal field from the previous input frame and
19301 the first temporal field from the next input frame. This mode relies on
19302 the top_field_first flag. Useful for interlaced video displays with no
19303 field synchronisation.
19308 Frame 1 Frame 2 Frame 3 Frame 4
19310 11111 22222 33333 44444
19311 11111 22222 33333 44444
19312 11111 22222 33333 44444
19313 11111 22222 33333 44444
19316 11111 22222 22222 33333 33333 44444 44444
19317 11111 11111 22222 22222 33333 33333 44444
19318 11111 22222 22222 33333 33333 44444 44444
19319 11111 11111 22222 22222 33333 33333 44444
19324 Move odd frames into the upper field, even into the lower field,
19325 generating a double height frame at same frame rate.
19330 Frame 1 Frame 2 Frame 3 Frame 4
19332 11111 22222 33333 44444
19333 11111 22222 33333 44444
19334 11111 22222 33333 44444
19335 11111 22222 33333 44444
19338 11111 33333 33333 55555
19339 22222 22222 44444 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
19350 Numeric values are deprecated but are accepted for backward
19351 compatibility reasons.
19353 Default mode is @code{merge}.
19356 Specify flags influencing the filter process.
19358 Available value for @var{flags} is:
19361 @item low_pass_filter, vlpf
19362 Enable linear vertical low-pass filtering in the filter.
19363 Vertical low-pass filtering is required when creating an interlaced
19364 destination from a progressive source which contains high-frequency
19365 vertical detail. Filtering will reduce interlace 'twitter' and Moire
19368 @item complex_filter, cvlpf
19369 Enable complex vertical low-pass filtering.
19370 This will slightly less reduce interlace 'twitter' and Moire
19371 patterning but better retain detail and subjective sharpness impression.
19374 Bypass already interlaced frames, only adjust the frame rate.
19377 Vertical low-pass filtering and bypassing already interlaced frames can only be
19378 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
19383 Pick median pixels from several successive input video frames.
19385 The filter accepts the following options:
19389 Set radius of median filter.
19390 Default is 1. Allowed range is from 1 to 127.
19393 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19396 Set median percentile. Default value is @code{0.5}.
19397 Default value of @code{0.5} will pick always median values, while @code{0} will pick
19398 minimum values, and @code{1} maximum values.
19401 @subsection Commands
19403 This filter supports all above options as @ref{commands}, excluding option @code{radius}.
19405 @section tmidequalizer
19407 Apply Temporal Midway Video Equalization effect.
19409 Midway Video Equalization adjusts a sequence of video frames to have the same
19410 histograms, while maintaining their dynamics as much as possible. It's
19411 useful for e.g. matching exposures from a video frames sequence.
19413 This filter accepts the following option:
19417 Set filtering radius. Default is @code{5}. Allowed range is from 1 to 127.
19420 Set filtering sigma. Default is @code{0.5}. This controls strength of filtering.
19421 Setting this option to 0 effectively does nothing.
19424 Set which planes to process. Default is @code{15}, which is all available planes.
19429 Mix successive video frames.
19431 A description of the accepted options follows.
19435 The number of successive frames to mix. If unspecified, it defaults to 3.
19438 Specify weight of each input video frame.
19439 Each weight is separated by space. If number of weights is smaller than
19440 number of @var{frames} last specified weight will be used for all remaining
19444 Specify scale, if it is set it will be multiplied with sum
19445 of each weight multiplied with pixel values to give final destination
19446 pixel value. By default @var{scale} is auto scaled to sum of weights.
19449 @subsection Examples
19453 Average 7 successive frames:
19455 tmix=frames=7:weights="1 1 1 1 1 1 1"
19459 Apply simple temporal convolution:
19461 tmix=frames=3:weights="-1 3 -1"
19465 Similar as above but only showing temporal differences:
19467 tmix=frames=3:weights="-1 2 -1":scale=1
19473 Tone map colors from different dynamic ranges.
19475 This filter expects data in single precision floating point, as it needs to
19476 operate on (and can output) out-of-range values. Another filter, such as
19477 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19479 The tonemapping algorithms implemented only work on linear light, so input
19480 data should be linearized beforehand (and possibly correctly tagged).
19483 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19486 @subsection Options
19487 The filter accepts the following options.
19491 Set the tone map algorithm to use.
19493 Possible values are:
19496 Do not apply any tone map, only desaturate overbright pixels.
19499 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19500 in-range values, while distorting out-of-range values.
19503 Stretch the entire reference gamut to a linear multiple of the display.
19506 Fit a logarithmic transfer between the tone curves.
19509 Preserve overall image brightness with a simple curve, using nonlinear
19510 contrast, which results in flattening details and degrading color accuracy.
19513 Preserve both dark and bright details better than @var{reinhard}, at the cost
19514 of slightly darkening everything. Use it when detail preservation is more
19515 important than color and brightness accuracy.
19518 Smoothly map out-of-range values, while retaining contrast and colors for
19519 in-range material as much as possible. Use it when color accuracy is more
19520 important than detail preservation.
19526 Tune the tone mapping algorithm.
19528 This affects the following algorithms:
19534 Specifies the scale factor to use while stretching.
19538 Specifies the exponent of the function.
19542 Specify an extra linear coefficient to multiply into the signal before clipping.
19546 Specify the local contrast coefficient at the display peak.
19547 Default to 0.5, which means that in-gamut values will be about half as bright
19554 Specify the transition point from linear to mobius transform. Every value
19555 below this point is guaranteed to be mapped 1:1. The higher the value, the
19556 more accurate the result will be, at the cost of losing bright details.
19557 Default to 0.3, which due to the steep initial slope still preserves in-range
19558 colors fairly accurately.
19562 Apply desaturation for highlights that exceed this level of brightness. The
19563 higher the parameter, the more color information will be preserved. This
19564 setting helps prevent unnaturally blown-out colors for super-highlights, by
19565 (smoothly) turning into white instead. This makes images feel more natural,
19566 at the cost of reducing information about out-of-range colors.
19568 The default of 2.0 is somewhat conservative and will mostly just apply to
19569 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19571 This option works only if the input frame has a supported color tag.
19574 Override signal/nominal/reference peak with this value. Useful when the
19575 embedded peak information in display metadata is not reliable or when tone
19576 mapping from a lower range to a higher range.
19581 Temporarily pad video frames.
19583 The filter accepts the following options:
19587 Specify number of delay frames before input video stream. Default is 0.
19590 Specify number of padding frames after input video stream.
19591 Set to -1 to pad indefinitely. Default is 0.
19594 Set kind of frames added to beginning of stream.
19595 Can be either @var{add} or @var{clone}.
19596 With @var{add} frames of solid-color are added.
19597 With @var{clone} frames are clones of first frame.
19598 Default is @var{add}.
19601 Set kind of frames added to end of stream.
19602 Can be either @var{add} or @var{clone}.
19603 With @var{add} frames of solid-color are added.
19604 With @var{clone} frames are clones of last frame.
19605 Default is @var{add}.
19607 @item start_duration, stop_duration
19608 Specify the duration of the start/stop delay. See
19609 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19610 for the accepted syntax.
19611 These options override @var{start} and @var{stop}. Default is 0.
19614 Specify the color of the padded area. For the syntax of this option,
19615 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19616 manual,ffmpeg-utils}.
19618 The default value of @var{color} is "black".
19624 Transpose rows with columns in the input video and optionally flip it.
19626 It accepts the following parameters:
19631 Specify the transposition direction.
19633 Can assume the following values:
19635 @item 0, 4, cclock_flip
19636 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19644 Rotate by 90 degrees clockwise, that is:
19652 Rotate by 90 degrees counterclockwise, that is:
19659 @item 3, 7, clock_flip
19660 Rotate by 90 degrees clockwise and vertically flip, that is:
19668 For values between 4-7, the transposition is only done if the input
19669 video geometry is portrait and not landscape. These values are
19670 deprecated, the @code{passthrough} option should be used instead.
19672 Numerical values are deprecated, and should be dropped in favor of
19673 symbolic constants.
19676 Do not apply the transposition if the input geometry matches the one
19677 specified by the specified value. It accepts the following values:
19680 Always apply transposition.
19682 Preserve portrait geometry (when @var{height} >= @var{width}).
19684 Preserve landscape geometry (when @var{width} >= @var{height}).
19687 Default value is @code{none}.
19690 For example to rotate by 90 degrees clockwise and preserve portrait
19693 transpose=dir=1:passthrough=portrait
19696 The command above can also be specified as:
19698 transpose=1:portrait
19701 @section transpose_npp
19703 Transpose rows with columns in the input video and optionally flip it.
19704 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19706 It accepts the following parameters:
19711 Specify the transposition direction.
19713 Can assume the following values:
19716 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19719 Rotate by 90 degrees clockwise.
19722 Rotate by 90 degrees counterclockwise.
19725 Rotate by 90 degrees clockwise and vertically flip.
19729 Do not apply the transposition if the input geometry matches the one
19730 specified by the specified value. It accepts the following values:
19733 Always apply transposition. (default)
19735 Preserve portrait geometry (when @var{height} >= @var{width}).
19737 Preserve landscape geometry (when @var{width} >= @var{height}).
19743 Trim the input so that the output contains one continuous subpart of the input.
19745 It accepts the following parameters:
19748 Specify the time of the start of the kept section, i.e. the frame with the
19749 timestamp @var{start} will be the first frame in the output.
19752 Specify the time of the first frame that will be dropped, i.e. the frame
19753 immediately preceding the one with the timestamp @var{end} will be the last
19754 frame in the output.
19757 This is the same as @var{start}, except this option sets the start timestamp
19758 in timebase units instead of seconds.
19761 This is the same as @var{end}, except this option sets the end timestamp
19762 in timebase units instead of seconds.
19765 The maximum duration of the output in seconds.
19768 The number of the first frame that should be passed to the output.
19771 The number of the first frame that should be dropped.
19774 @option{start}, @option{end}, and @option{duration} are expressed as time
19775 duration specifications; see
19776 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19777 for the accepted syntax.
19779 Note that the first two sets of the start/end options and the @option{duration}
19780 option look at the frame timestamp, while the _frame variants simply count the
19781 frames that pass through the filter. Also note that this filter does not modify
19782 the timestamps. If you wish for the output timestamps to start at zero, insert a
19783 setpts filter after the trim filter.
19785 If multiple start or end options are set, this filter tries to be greedy and
19786 keep all the frames that match at least one of the specified constraints. To keep
19787 only the part that matches all the constraints at once, chain multiple trim
19790 The defaults are such that all the input is kept. So it is possible to set e.g.
19791 just the end values to keep everything before the specified time.
19796 Drop everything except the second minute of input:
19798 ffmpeg -i INPUT -vf trim=60:120
19802 Keep only the first second:
19804 ffmpeg -i INPUT -vf trim=duration=1
19809 @section unpremultiply
19810 Apply alpha unpremultiply effect to input video stream using first plane
19811 of second stream as alpha.
19813 Both streams must have same dimensions and same pixel format.
19815 The filter accepts the following option:
19819 Set which planes will be processed, unprocessed planes will be copied.
19820 By default value 0xf, all planes will be processed.
19822 If the format has 1 or 2 components, then luma is bit 0.
19823 If the format has 3 or 4 components:
19824 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19825 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19826 If present, the alpha channel is always the last bit.
19829 Do not require 2nd input for processing, instead use alpha plane from input stream.
19835 Sharpen or blur the input video.
19837 It accepts the following parameters:
19840 @item luma_msize_x, lx
19841 Set the luma matrix horizontal size. It must be an odd integer between
19842 3 and 23. The default value is 5.
19844 @item luma_msize_y, ly
19845 Set the luma matrix vertical size. It must be an odd integer between 3
19846 and 23. The default value is 5.
19848 @item luma_amount, la
19849 Set the luma effect strength. It must be a floating point number, reasonable
19850 values lay between -1.5 and 1.5.
19852 Negative values will blur the input video, while positive values will
19853 sharpen it, a value of zero will disable the effect.
19855 Default value is 1.0.
19857 @item chroma_msize_x, cx
19858 Set the chroma matrix horizontal size. It must be an odd integer
19859 between 3 and 23. The default value is 5.
19861 @item chroma_msize_y, cy
19862 Set the chroma matrix vertical size. It must be an odd integer
19863 between 3 and 23. The default value is 5.
19865 @item chroma_amount, ca
19866 Set the chroma effect strength. It must be a floating point number, reasonable
19867 values lay between -1.5 and 1.5.
19869 Negative values will blur the input video, while positive values will
19870 sharpen it, a value of zero will disable the effect.
19872 Default value is 0.0.
19876 All parameters are optional and default to the equivalent of the
19877 string '5:5:1.0:5:5:0.0'.
19879 @subsection Examples
19883 Apply strong luma sharpen effect:
19885 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19889 Apply a strong blur of both luma and chroma parameters:
19891 unsharp=7:7:-2:7:7:-2
19898 Decompose a video made of tiled images into the individual images.
19900 The frame rate of the output video is the frame rate of the input video
19901 multiplied by the number of tiles.
19903 This filter does the reverse of @ref{tile}.
19905 The filter accepts the following options:
19910 Set the grid size (i.e. the number of lines and columns). For the syntax of
19911 this option, check the
19912 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19915 @subsection Examples
19919 Produce a 1-second video from a still image file made of 25 frames stacked
19920 vertically, like an analogic film reel:
19922 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19928 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19929 the image at several (or - in the case of @option{quality} level @code{8} - all)
19930 shifts and average the results.
19932 The way this differs from the behavior of spp is that uspp actually encodes &
19933 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19934 DCT similar to MJPEG.
19936 The filter accepts the following options:
19940 Set quality. This option defines the number of levels for averaging. It accepts
19941 an integer in the range 0-8. If set to @code{0}, the filter will have no
19942 effect. A value of @code{8} means the higher quality. For each increment of
19943 that value the speed drops by a factor of approximately 2. Default value is
19947 Force a constant quantization parameter. If not set, the filter will use the QP
19948 from the video stream (if available).
19953 Convert 360 videos between various formats.
19955 The filter accepts the following options:
19961 Set format of the input/output video.
19969 Equirectangular projection.
19974 Cubemap with 3x2/6x1/1x6 layout.
19976 Format specific options:
19981 Set padding proportion for the input/output cubemap. Values in decimals.
19988 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)
19991 Default value is @b{@samp{0}}.
19992 Maximum value is @b{@samp{0.1}}.
19996 Set fixed padding for the input/output cubemap. Values in pixels.
19998 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
20002 Set order of faces for the input/output cubemap. Choose one direction for each position.
20004 Designation of directions:
20020 Default value is @b{@samp{rludfb}}.
20024 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
20026 Designation of angles:
20029 0 degrees clockwise
20031 90 degrees clockwise
20033 180 degrees clockwise
20035 270 degrees clockwise
20038 Default value is @b{@samp{000000}}.
20042 Equi-Angular Cubemap.
20049 Format specific options:
20054 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20056 If diagonal field of view is set it overrides horizontal and vertical field of view.
20061 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20063 If diagonal field of view is set it overrides horizontal and vertical field of view.
20069 Format specific options:
20074 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20076 If diagonal field of view is set it overrides horizontal and vertical field of view.
20081 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20083 If diagonal field of view is set it overrides horizontal and vertical field of view.
20089 Facebook's 360 formats.
20092 Stereographic format.
20094 Format specific options:
20099 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20101 If diagonal field of view is set it overrides horizontal and vertical field of view.
20106 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20108 If diagonal field of view is set it overrides horizontal and vertical field of view.
20115 Ball format, gives significant distortion toward the back.
20118 Hammer-Aitoff map projection format.
20121 Sinusoidal map projection format.
20124 Fisheye projection.
20126 Format specific options:
20131 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20133 If diagonal field of view is set it overrides horizontal and vertical field of view.
20138 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20140 If diagonal field of view is set it overrides horizontal and vertical field of view.
20144 Pannini projection.
20146 Format specific options:
20149 Set output pannini parameter.
20152 Set input pannini parameter.
20156 Cylindrical projection.
20158 Format specific options:
20163 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20165 If diagonal field of view is set it overrides horizontal and vertical field of view.
20170 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20172 If diagonal field of view is set it overrides horizontal and vertical field of view.
20176 Perspective projection. @i{(output only)}
20178 Format specific options:
20181 Set perspective parameter.
20185 Tetrahedron projection.
20188 Truncated square pyramid projection.
20192 Half equirectangular projection.
20197 Format specific options:
20202 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20204 If diagonal field of view is set it overrides horizontal and vertical field of view.
20209 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20211 If diagonal field of view is set it overrides horizontal and vertical field of view.
20215 Orthographic format.
20217 Format specific options:
20222 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20224 If diagonal field of view is set it overrides horizontal and vertical field of view.
20229 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20231 If diagonal field of view is set it overrides horizontal and vertical field of view.
20235 Octahedron projection.
20239 Set interpolation method.@*
20240 @i{Note: more complex interpolation methods require much more memory to run.}
20250 Bilinear interpolation.
20252 Lagrange9 interpolation.
20255 Bicubic interpolation.
20258 Lanczos interpolation.
20261 Spline16 interpolation.
20264 Gaussian interpolation.
20266 Mitchell interpolation.
20269 Default value is @b{@samp{line}}.
20273 Set the output video resolution.
20275 Default resolution depends on formats.
20279 Set the input/output stereo format.
20290 Default value is @b{@samp{2d}} for input and output format.
20295 Set rotation for the output video. Values in degrees.
20298 Set rotation order for the output video. Choose one item for each position.
20309 Default value is @b{@samp{ypr}}.
20314 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
20318 Set if input video is flipped horizontally/vertically. Boolean values.
20321 Set if input video is transposed. Boolean value, by default disabled.
20324 Set if output video needs to be transposed. Boolean value, by default disabled.
20327 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
20330 @subsection Examples
20334 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
20336 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20339 Extract back view of Equi-Angular Cubemap:
20341 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20344 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
20346 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20350 @subsection Commands
20352 This filter supports subset of above options as @ref{commands}.
20354 @section vaguedenoiser
20356 Apply a wavelet based denoiser.
20358 It transforms each frame from the video input into the wavelet domain,
20359 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20360 the obtained coefficients. It does an inverse wavelet transform after.
20361 Due to wavelet properties, it should give a nice smoothed result, and
20362 reduced noise, without blurring picture features.
20364 This filter accepts the following options:
20368 The filtering strength. The higher, the more filtered the video will be.
20369 Hard thresholding can use a higher threshold than soft thresholding
20370 before the video looks overfiltered. Default value is 2.
20373 The filtering method the filter will use.
20375 It accepts the following values:
20378 All values under the threshold will be zeroed.
20381 All values under the threshold will be zeroed. All values above will be
20382 reduced by the threshold.
20385 Scales or nullifies coefficients - intermediary between (more) soft and
20386 (less) hard thresholding.
20389 Default is garrote.
20392 Number of times, the wavelet will decompose the picture. Picture can't
20393 be decomposed beyond a particular point (typically, 8 for a 640x480
20394 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
20397 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
20400 A list of the planes to process. By default all planes are processed.
20403 The threshold type the filter will use.
20405 It accepts the following values:
20408 Threshold used is same for all decompositions.
20411 Threshold used depends also on each decomposition coefficients.
20414 Default is universal.
20417 @section vectorscope
20419 Display 2 color component values in the two dimensional graph (which is called
20422 This filter accepts the following options:
20426 Set vectorscope mode.
20428 It accepts the following values:
20432 Gray values are displayed on graph, higher brightness means more pixels have
20433 same component color value on location in graph. This is the default mode.
20436 Gray values are displayed on graph. Surrounding pixels values which are not
20437 present in video frame are drawn in gradient of 2 color components which are
20438 set by option @code{x} and @code{y}. The 3rd color component is static.
20441 Actual color components values present in video frame are displayed on graph.
20444 Similar as color2 but higher frequency of same values @code{x} and @code{y}
20445 on graph increases value of another color component, which is luminance by
20446 default values of @code{x} and @code{y}.
20449 Actual colors present in video frame are displayed on graph. If two different
20450 colors map to same position on graph then color with higher value of component
20451 not present in graph is picked.
20454 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
20455 component picked from radial gradient.
20459 Set which color component will be represented on X-axis. Default is @code{1}.
20462 Set which color component will be represented on Y-axis. Default is @code{2}.
20465 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20466 of color component which represents frequency of (X, Y) location in graph.
20471 No envelope, this is default.
20474 Instant envelope, even darkest single pixel will be clearly highlighted.
20477 Hold maximum and minimum values presented in graph over time. This way you
20478 can still spot out of range values without constantly looking at vectorscope.
20481 Peak and instant envelope combined together.
20485 Set what kind of graticule to draw.
20494 Set graticule opacity.
20497 Set graticule flags.
20501 Draw graticule for white point.
20504 Draw graticule for black point.
20507 Draw color points short names.
20511 Set background opacity.
20513 @item lthreshold, l
20514 Set low threshold for color component not represented on X or Y axis.
20515 Values lower than this value will be ignored. Default is 0.
20516 Note this value is multiplied with actual max possible value one pixel component
20517 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20520 @item hthreshold, h
20521 Set high threshold for color component not represented on X or Y axis.
20522 Values higher than this value will be ignored. Default is 1.
20523 Note this value is multiplied with actual max possible value one pixel component
20524 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20525 is 0.9 * 255 = 230.
20527 @item colorspace, c
20528 Set what kind of colorspace to use when drawing graticule.
20538 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20539 This means no tint, and output will remain gray.
20542 @anchor{vidstabdetect}
20543 @section vidstabdetect
20545 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20546 @ref{vidstabtransform} for pass 2.
20548 This filter generates a file with relative translation and rotation
20549 transform information about subsequent frames, which is then used by
20550 the @ref{vidstabtransform} filter.
20552 To enable compilation of this filter you need to configure FFmpeg with
20553 @code{--enable-libvidstab}.
20555 This filter accepts the following options:
20559 Set the path to the file used to write the transforms information.
20560 Default value is @file{transforms.trf}.
20563 Set how shaky the video is and how quick the camera is. It accepts an
20564 integer in the range 1-10, a value of 1 means little shakiness, a
20565 value of 10 means strong shakiness. Default value is 5.
20568 Set the accuracy of the detection process. It must be a value in the
20569 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20570 accuracy. Default value is 15.
20573 Set stepsize of the search process. The region around minimum is
20574 scanned with 1 pixel resolution. Default value is 6.
20577 Set minimum contrast. Below this value a local measurement field is
20578 discarded. Must be a floating point value in the range 0-1. Default
20582 Set reference frame number for tripod mode.
20584 If enabled, the motion of the frames is compared to a reference frame
20585 in the filtered stream, identified by the specified number. The idea
20586 is to compensate all movements in a more-or-less static scene and keep
20587 the camera view absolutely still.
20589 If set to 0, it is disabled. The frames are counted starting from 1.
20592 Show fields and transforms in the resulting frames. It accepts an
20593 integer in the range 0-2. Default value is 0, which disables any
20597 @subsection Examples
20601 Use default values:
20607 Analyze strongly shaky movie and put the results in file
20608 @file{mytransforms.trf}:
20610 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20614 Visualize the result of internal transformations in the resulting
20617 vidstabdetect=show=1
20621 Analyze a video with medium shakiness using @command{ffmpeg}:
20623 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20627 @anchor{vidstabtransform}
20628 @section vidstabtransform
20630 Video stabilization/deshaking: pass 2 of 2,
20631 see @ref{vidstabdetect} for pass 1.
20633 Read a file with transform information for each frame and
20634 apply/compensate them. Together with the @ref{vidstabdetect}
20635 filter this can be used to deshake videos. See also
20636 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20637 the @ref{unsharp} filter, see below.
20639 To enable compilation of this filter you need to configure FFmpeg with
20640 @code{--enable-libvidstab}.
20642 @subsection Options
20646 Set path to the file used to read the transforms. Default value is
20647 @file{transforms.trf}.
20650 Set the number of frames (value*2 + 1) used for lowpass filtering the
20651 camera movements. Default value is 10.
20653 For example a number of 10 means that 21 frames are used (10 in the
20654 past and 10 in the future) to smoothen the motion in the video. A
20655 larger value leads to a smoother video, but limits the acceleration of
20656 the camera (pan/tilt movements). 0 is a special case where a static
20657 camera is simulated.
20660 Set the camera path optimization algorithm.
20662 Accepted values are:
20665 gaussian kernel low-pass filter on camera motion (default)
20667 averaging on transformations
20671 Set maximal number of pixels to translate frames. Default value is -1,
20675 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20676 value is -1, meaning no limit.
20679 Specify how to deal with borders that may be visible due to movement
20682 Available values are:
20685 keep image information from previous frame (default)
20687 fill the border black
20691 Invert transforms if set to 1. Default value is 0.
20694 Consider transforms as relative to previous frame if set to 1,
20695 absolute if set to 0. Default value is 0.
20698 Set percentage to zoom. A positive value will result in a zoom-in
20699 effect, a negative value in a zoom-out effect. Default value is 0 (no
20703 Set optimal zooming to avoid borders.
20705 Accepted values are:
20710 optimal static zoom value is determined (only very strong movements
20711 will lead to visible borders) (default)
20713 optimal adaptive zoom value is determined (no borders will be
20714 visible), see @option{zoomspeed}
20717 Note that the value given at zoom is added to the one calculated here.
20720 Set percent to zoom maximally each frame (enabled when
20721 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20725 Specify type of interpolation.
20727 Available values are:
20732 linear only horizontal
20734 linear in both directions (default)
20736 cubic in both directions (slow)
20740 Enable virtual tripod mode if set to 1, which is equivalent to
20741 @code{relative=0:smoothing=0}. Default value is 0.
20743 Use also @code{tripod} option of @ref{vidstabdetect}.
20746 Increase log verbosity if set to 1. Also the detected global motions
20747 are written to the temporary file @file{global_motions.trf}. Default
20751 @subsection Examples
20755 Use @command{ffmpeg} for a typical stabilization with default values:
20757 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20760 Note the use of the @ref{unsharp} filter which is always recommended.
20763 Zoom in a bit more and load transform data from a given file:
20765 vidstabtransform=zoom=5:input="mytransforms.trf"
20769 Smoothen the video even more:
20771 vidstabtransform=smoothing=30
20777 Flip the input video vertically.
20779 For example, to vertically flip a video with @command{ffmpeg}:
20781 ffmpeg -i in.avi -vf "vflip" out.avi
20786 Detect variable frame rate video.
20788 This filter tries to detect if the input is variable or constant frame rate.
20790 At end it will output number of frames detected as having variable delta pts,
20791 and ones with constant delta pts.
20792 If there was frames with variable delta, than it will also show min, max and
20793 average delta encountered.
20797 Boost or alter saturation.
20799 The filter accepts the following options:
20802 Set strength of boost if positive value or strength of alter if negative value.
20803 Default is 0. Allowed range is from -2 to 2.
20806 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20809 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20812 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20815 Set the red luma coefficient.
20818 Set the green luma coefficient.
20821 Set the blue luma coefficient.
20824 If @code{intensity} is negative and this is set to 1, colors will change,
20825 otherwise colors will be less saturated, more towards gray.
20828 @subsection Commands
20830 This filter supports the all above options as @ref{commands}.
20835 Make or reverse a natural vignetting effect.
20837 The filter accepts the following options:
20841 Set lens angle expression as a number of radians.
20843 The value is clipped in the @code{[0,PI/2]} range.
20845 Default value: @code{"PI/5"}
20849 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20853 Set forward/backward mode.
20855 Available modes are:
20858 The larger the distance from the central point, the darker the image becomes.
20861 The larger the distance from the central point, the brighter the image becomes.
20862 This can be used to reverse a vignette effect, though there is no automatic
20863 detection to extract the lens @option{angle} and other settings (yet). It can
20864 also be used to create a burning effect.
20867 Default value is @samp{forward}.
20870 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20872 It accepts the following values:
20875 Evaluate expressions only once during the filter initialization.
20878 Evaluate expressions for each incoming frame. This is way slower than the
20879 @samp{init} mode since it requires all the scalers to be re-computed, but it
20880 allows advanced dynamic expressions.
20883 Default value is @samp{init}.
20886 Set dithering to reduce the circular banding effects. Default is @code{1}
20890 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20891 Setting this value to the SAR of the input will make a rectangular vignetting
20892 following the dimensions of the video.
20894 Default is @code{1/1}.
20897 @subsection Expressions
20899 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20900 following parameters.
20905 input width and height
20908 the number of input frame, starting from 0
20911 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20912 @var{TB} units, NAN if undefined
20915 frame rate of the input video, NAN if the input frame rate is unknown
20918 the PTS (Presentation TimeStamp) of the filtered video frame,
20919 expressed in seconds, NAN if undefined
20922 time base of the input video
20926 @subsection Examples
20930 Apply simple strong vignetting effect:
20936 Make a flickering vignetting:
20938 vignette='PI/4+random(1)*PI/50':eval=frame
20943 @section vmafmotion
20945 Obtain the average VMAF motion score of a video.
20946 It is one of the component metrics of VMAF.
20948 The obtained average motion score is printed through the logging system.
20950 The filter accepts the following options:
20954 If specified, the filter will use the named file to save the motion score of
20955 each frame with respect to the previous frame.
20956 When filename equals "-" the data is sent to standard output.
20961 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20965 Stack input videos vertically.
20967 All streams must be of same pixel format and of same width.
20969 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20970 to create same output.
20972 The filter accepts the following options:
20976 Set number of input streams. Default is 2.
20979 If set to 1, force the output to terminate when the shortest input
20980 terminates. Default value is 0.
20985 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20986 Deinterlacing Filter").
20988 Based on the process described by Martin Weston for BBC R&D, and
20989 implemented based on the de-interlace algorithm written by Jim
20990 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20991 uses filter coefficients calculated by BBC R&D.
20993 This filter uses field-dominance information in frame to decide which
20994 of each pair of fields to place first in the output.
20995 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20997 There are two sets of filter coefficients, so called "simple"
20998 and "complex". Which set of filter coefficients is used can
20999 be set by passing an optional parameter:
21003 Set the interlacing filter coefficients. Accepts one of the following values:
21007 Simple filter coefficient set.
21009 More-complex filter coefficient set.
21011 Default value is @samp{complex}.
21014 The interlacing mode to adopt. It accepts one of the following values:
21018 Output one frame for each frame.
21020 Output one frame for each field.
21023 The default value is @code{field}.
21026 The picture field parity assumed for the input interlaced video. It accepts one
21027 of the following values:
21031 Assume the top field is first.
21033 Assume the bottom field is first.
21035 Enable automatic detection of field parity.
21038 The default value is @code{auto}.
21039 If the interlacing is unknown or the decoder does not export this information,
21040 top field first will be assumed.
21043 Specify which frames to deinterlace. Accepts one of the following values:
21047 Deinterlace all frames,
21049 Only deinterlace frames marked as interlaced.
21052 Default value is @samp{all}.
21055 @subsection Commands
21056 This filter supports same @ref{commands} as options.
21059 Video waveform monitor.
21061 The waveform monitor plots color component intensity. By default luminance
21062 only. Each column of the waveform corresponds to a column of pixels in the
21065 It accepts the following options:
21069 Can be either @code{row}, or @code{column}. Default is @code{column}.
21070 In row mode, the graph on the left side represents color component value 0 and
21071 the right side represents value = 255. In column mode, the top side represents
21072 color component value = 0 and bottom side represents value = 255.
21075 Set intensity. Smaller values are useful to find out how many values of the same
21076 luminance are distributed across input rows/columns.
21077 Default value is @code{0.04}. Allowed range is [0, 1].
21080 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
21081 In mirrored mode, higher values will be represented on the left
21082 side for @code{row} mode and at the top for @code{column} mode. Default is
21083 @code{1} (mirrored).
21087 It accepts the following values:
21090 Presents information identical to that in the @code{parade}, except
21091 that the graphs representing color components are superimposed directly
21094 This display mode makes it easier to spot relative differences or similarities
21095 in overlapping areas of the color components that are supposed to be identical,
21096 such as neutral whites, grays, or blacks.
21099 Display separate graph for the color components side by side in
21100 @code{row} mode or one below the other in @code{column} mode.
21103 Display separate graph for the color components side by side in
21104 @code{column} mode or one below the other in @code{row} mode.
21106 Using this display mode makes it easy to spot color casts in the highlights
21107 and shadows of an image, by comparing the contours of the top and the bottom
21108 graphs of each waveform. Since whites, grays, and blacks are characterized
21109 by exactly equal amounts of red, green, and blue, neutral areas of the picture
21110 should display three waveforms of roughly equal width/height. If not, the
21111 correction is easy to perform by making level adjustments the three waveforms.
21113 Default is @code{stack}.
21115 @item components, c
21116 Set which color components to display. Default is 1, which means only luminance
21117 or red color component if input is in RGB colorspace. If is set for example to
21118 7 it will display all 3 (if) available color components.
21123 No envelope, this is default.
21126 Instant envelope, minimum and maximum values presented in graph will be easily
21127 visible even with small @code{step} value.
21130 Hold minimum and maximum values presented in graph across time. This way you
21131 can still spot out of range values without constantly looking at waveforms.
21134 Peak and instant envelope combined together.
21140 No filtering, this is default.
21143 Luma and chroma combined together.
21146 Similar as above, but shows difference between blue and red chroma.
21149 Similar as above, but use different colors.
21152 Similar as above, but again with different colors.
21155 Displays only chroma.
21158 Displays actual color value on waveform.
21161 Similar as above, but with luma showing frequency of chroma values.
21165 Set which graticule to display.
21169 Do not display graticule.
21172 Display green graticule showing legal broadcast ranges.
21175 Display orange graticule showing legal broadcast ranges.
21178 Display invert graticule showing legal broadcast ranges.
21182 Set graticule opacity.
21185 Set graticule flags.
21189 Draw numbers above lines. By default enabled.
21192 Draw dots instead of lines.
21196 Set scale used for displaying graticule.
21203 Default is digital.
21206 Set background opacity.
21210 Set tint for output.
21211 Only used with lowpass filter and when display is not overlay and input
21212 pixel formats are not RGB.
21215 @section weave, doubleweave
21217 The @code{weave} takes a field-based video input and join
21218 each two sequential fields into single frame, producing a new double
21219 height clip with half the frame rate and half the frame count.
21221 The @code{doubleweave} works same as @code{weave} but without
21222 halving frame rate and frame count.
21224 It accepts the following option:
21228 Set first field. Available values are:
21232 Set the frame as top-field-first.
21235 Set the frame as bottom-field-first.
21239 @subsection Examples
21243 Interlace video using @ref{select} and @ref{separatefields} filter:
21245 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21250 Apply the xBR high-quality magnification filter which is designed for pixel
21251 art. It follows a set of edge-detection rules, see
21252 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
21254 It accepts the following option:
21258 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
21259 @code{3xBR} and @code{4} for @code{4xBR}.
21260 Default is @code{3}.
21265 Apply cross fade from one input video stream to another input video stream.
21266 The cross fade is applied for specified duration.
21268 The filter accepts the following options:
21272 Set one of available transition effects:
21320 Default transition effect is fade.
21323 Set cross fade duration in seconds.
21324 Default duration is 1 second.
21327 Set cross fade start relative to first input stream in seconds.
21328 Default offset is 0.
21331 Set expression for custom transition effect.
21333 The expressions can use the following variables and functions:
21338 The coordinates of the current sample.
21342 The width and height of the image.
21345 Progress of transition effect.
21348 Currently processed plane.
21351 Return value of first input at current location and plane.
21354 Return value of second input at current location and plane.
21360 Return the value of the pixel at location (@var{x},@var{y}) of the
21361 first/second/third/fourth component of first input.
21367 Return the value of the pixel at location (@var{x},@var{y}) of the
21368 first/second/third/fourth component of second input.
21372 @subsection Examples
21376 Cross fade from one input video to another input video, with fade transition and duration of transition
21377 of 2 seconds starting at offset of 5 seconds:
21379 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21384 Pick median pixels from several input videos.
21386 The filter accepts the following options:
21390 Set number of inputs.
21391 Default is 3. Allowed range is from 3 to 255.
21392 If number of inputs is even number, than result will be mean value between two median values.
21395 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
21398 Set median percentile. Default value is @code{0.5}.
21399 Default value of @code{0.5} will pick always median values, while @code{0} will pick
21400 minimum values, and @code{1} maximum values.
21403 @subsection Commands
21405 This filter supports all above options as @ref{commands}, excluding option @code{inputs}.
21408 Stack video inputs into custom layout.
21410 All streams must be of same pixel format.
21412 The filter accepts the following options:
21416 Set number of input streams. Default is 2.
21419 Specify layout of inputs.
21420 This option requires the desired layout configuration to be explicitly set by the user.
21421 This sets position of each video input in output. Each input
21422 is separated by '|'.
21423 The first number represents the column, and the second number represents the row.
21424 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
21425 where X is video input from which to take width or height.
21426 Multiple values can be used when separated by '+'. In such
21427 case values are summed together.
21429 Note that if inputs are of different sizes gaps may appear, as not all of
21430 the output video frame will be filled. Similarly, videos can overlap each
21431 other if their position doesn't leave enough space for the full frame of
21434 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
21435 a layout must be set by the user.
21438 If set to 1, force the output to terminate when the shortest input
21439 terminates. Default value is 0.
21442 If set to valid color, all unused pixels will be filled with that color.
21443 By default fill is set to none, so it is disabled.
21446 @subsection Examples
21450 Display 4 inputs into 2x2 grid.
21454 input1(0, 0) | input3(w0, 0)
21455 input2(0, h0) | input4(w0, h0)
21459 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21462 Note that if inputs are of different sizes, gaps or overlaps may occur.
21465 Display 4 inputs into 1x4 grid.
21472 input4(0, h0+h1+h2)
21476 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21479 Note that if inputs are of different widths, unused space will appear.
21482 Display 9 inputs into 3x3 grid.
21486 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21487 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21488 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21492 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
21495 Note that if inputs are of different sizes, gaps or overlaps may occur.
21498 Display 16 inputs into 4x4 grid.
21502 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21503 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21504 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21505 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21509 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|
21510 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
21513 Note that if inputs are of different sizes, gaps or overlaps may occur.
21520 Deinterlace the input video ("yadif" means "yet another deinterlacing
21523 It accepts the following parameters:
21529 The interlacing mode to adopt. It accepts one of the following values:
21532 @item 0, send_frame
21533 Output one frame for each frame.
21534 @item 1, send_field
21535 Output one frame for each field.
21536 @item 2, send_frame_nospatial
21537 Like @code{send_frame}, but it skips the spatial interlacing check.
21538 @item 3, send_field_nospatial
21539 Like @code{send_field}, but it skips the spatial interlacing check.
21542 The default value is @code{send_frame}.
21545 The picture field parity assumed for the input interlaced video. It accepts one
21546 of the following values:
21550 Assume the top field is first.
21552 Assume the bottom field is first.
21554 Enable automatic detection of field parity.
21557 The default value is @code{auto}.
21558 If the interlacing is unknown or the decoder does not export this information,
21559 top field first will be assumed.
21562 Specify which frames to deinterlace. Accepts one of the following
21567 Deinterlace all frames.
21568 @item 1, interlaced
21569 Only deinterlace frames marked as interlaced.
21572 The default value is @code{all}.
21575 @section yadif_cuda
21577 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21578 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21581 It accepts the following parameters:
21587 The interlacing mode to adopt. It accepts one of the following values:
21590 @item 0, send_frame
21591 Output one frame for each frame.
21592 @item 1, send_field
21593 Output one frame for each field.
21594 @item 2, send_frame_nospatial
21595 Like @code{send_frame}, but it skips the spatial interlacing check.
21596 @item 3, send_field_nospatial
21597 Like @code{send_field}, but it skips the spatial interlacing check.
21600 The default value is @code{send_frame}.
21603 The picture field parity assumed for the input interlaced video. It accepts one
21604 of the following values:
21608 Assume the top field is first.
21610 Assume the bottom field is first.
21612 Enable automatic detection of field parity.
21615 The default value is @code{auto}.
21616 If the interlacing is unknown or the decoder does not export this information,
21617 top field first will be assumed.
21620 Specify which frames to deinterlace. Accepts one of the following
21625 Deinterlace all frames.
21626 @item 1, interlaced
21627 Only deinterlace frames marked as interlaced.
21630 The default value is @code{all}.
21635 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21636 The algorithm is described in
21637 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21639 It accepts the following parameters:
21643 Set the window radius. Default value is 3.
21646 Set which planes to filter. Default is only the first plane.
21649 Set blur strength. Default value is 128.
21652 @subsection Commands
21653 This filter supports same @ref{commands} as options.
21657 Apply Zoom & Pan effect.
21659 This filter accepts the following options:
21663 Set the zoom expression. Range is 1-10. Default is 1.
21667 Set the x and y expression. Default is 0.
21670 Set the duration expression in number of frames.
21671 This sets for how many number of frames effect will last for
21672 single input image.
21675 Set the output image size, default is 'hd720'.
21678 Set the output frame rate, default is '25'.
21681 Each expression can contain the following constants:
21700 Output frame count.
21703 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21705 @item out_time, time, ot
21706 The output timestamp expressed in seconds.
21710 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21711 for current input frame.
21715 'x' and 'y' of last output frame of previous input frame or 0 when there was
21716 not yet such frame (first input frame).
21719 Last calculated zoom from 'z' expression for current input frame.
21722 Last calculated zoom of last output frame of previous input frame.
21725 Number of output frames for current input frame. Calculated from 'd' expression
21726 for each input frame.
21729 number of output frames created for previous input frame
21732 Rational number: input width / input height
21735 sample aspect ratio
21738 display aspect ratio
21742 @subsection Examples
21746 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21748 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
21752 Zoom in up to 1.5x and pan always at center of picture:
21754 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21758 Same as above but without pausing:
21760 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21764 Zoom in 2x into center of picture only for the first second of the input video:
21766 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21773 Scale (resize) the input video, using the z.lib library:
21774 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21775 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21777 The zscale filter forces the output display aspect ratio to be the same
21778 as the input, by changing the output sample aspect ratio.
21780 If the input image format is different from the format requested by
21781 the next filter, the zscale filter will convert the input to the
21784 @subsection Options
21785 The filter accepts the following options.
21790 Set the output video dimension expression. Default value is the input
21793 If the @var{width} or @var{w} value is 0, the input width is used for
21794 the output. If the @var{height} or @var{h} value is 0, the input height
21795 is used for the output.
21797 If one and only one of the values is -n with n >= 1, the zscale filter
21798 will use a value that maintains the aspect ratio of the input image,
21799 calculated from the other specified dimension. After that it will,
21800 however, make sure that the calculated dimension is divisible by n and
21801 adjust the value if necessary.
21803 If both values are -n with n >= 1, the behavior will be identical to
21804 both values being set to 0 as previously detailed.
21806 See below for the list of accepted constants for use in the dimension
21810 Set the video size. For the syntax of this option, check the
21811 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21814 Set the dither type.
21816 Possible values are:
21821 @item error_diffusion
21827 Set the resize filter type.
21829 Possible values are:
21839 Default is bilinear.
21842 Set the color range.
21844 Possible values are:
21851 Default is same as input.
21854 Set the color primaries.
21856 Possible values are:
21866 Default is same as input.
21869 Set the transfer characteristics.
21871 Possible values are:
21885 Default is same as input.
21888 Set the colorspace matrix.
21890 Possible value are:
21901 Default is same as input.
21904 Set the input color range.
21906 Possible values are:
21913 Default is same as input.
21915 @item primariesin, pin
21916 Set the input color primaries.
21918 Possible values are:
21928 Default is same as input.
21930 @item transferin, tin
21931 Set the input transfer characteristics.
21933 Possible values are:
21944 Default is same as input.
21946 @item matrixin, min
21947 Set the input colorspace matrix.
21949 Possible value are:
21961 Set the output chroma location.
21963 Possible values are:
21974 @item chromalin, cin
21975 Set the input chroma location.
21977 Possible values are:
21989 Set the nominal peak luminance.
21992 The values of the @option{w} and @option{h} options are expressions
21993 containing the following constants:
21998 The input width and height
22002 These are the same as @var{in_w} and @var{in_h}.
22006 The output (scaled) width and height
22010 These are the same as @var{out_w} and @var{out_h}
22013 The same as @var{iw} / @var{ih}
22016 input sample aspect ratio
22019 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
22023 horizontal and vertical input chroma subsample values. For example for the
22024 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
22028 horizontal and vertical output chroma subsample values. For example for the
22029 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
22032 @subsection Commands
22034 This filter supports the following commands:
22038 Set the output video dimension expression.
22039 The command accepts the same syntax of the corresponding option.
22041 If the specified expression is not valid, it is kept at its current
22045 @c man end VIDEO FILTERS
22047 @chapter OpenCL Video Filters
22048 @c man begin OPENCL VIDEO FILTERS
22050 Below is a description of the currently available OpenCL video filters.
22052 To enable compilation of these filters you need to configure FFmpeg with
22053 @code{--enable-opencl}.
22055 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
22058 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
22059 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
22060 given device parameters.
22062 @item -filter_hw_device @var{name}
22063 Pass the hardware device called @var{name} to all filters in any filter graph.
22067 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
22071 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
22073 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
22077 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.
22079 @section avgblur_opencl
22081 Apply average blur filter.
22083 The filter accepts the following options:
22087 Set horizontal radius size.
22088 Range is @code{[1, 1024]} and default value is @code{1}.
22091 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22094 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
22097 @subsection Example
22101 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.
22103 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
22107 @section boxblur_opencl
22109 Apply a boxblur algorithm to the input video.
22111 It accepts the following parameters:
22115 @item luma_radius, lr
22116 @item luma_power, lp
22117 @item chroma_radius, cr
22118 @item chroma_power, cp
22119 @item alpha_radius, ar
22120 @item alpha_power, ap
22124 A description of the accepted options follows.
22127 @item luma_radius, lr
22128 @item chroma_radius, cr
22129 @item alpha_radius, ar
22130 Set an expression for the box radius in pixels used for blurring the
22131 corresponding input plane.
22133 The radius value must be a non-negative number, and must not be
22134 greater than the value of the expression @code{min(w,h)/2} for the
22135 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
22138 Default value for @option{luma_radius} is "2". If not specified,
22139 @option{chroma_radius} and @option{alpha_radius} default to the
22140 corresponding value set for @option{luma_radius}.
22142 The expressions can contain the following constants:
22146 The input width and height in pixels.
22150 The input chroma image width and height in pixels.
22154 The horizontal and vertical chroma subsample values. For example, for the
22155 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
22158 @item luma_power, lp
22159 @item chroma_power, cp
22160 @item alpha_power, ap
22161 Specify how many times the boxblur filter is applied to the
22162 corresponding plane.
22164 Default value for @option{luma_power} is 2. If not specified,
22165 @option{chroma_power} and @option{alpha_power} default to the
22166 corresponding value set for @option{luma_power}.
22168 A value of 0 will disable the effect.
22171 @subsection Examples
22173 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.
22177 Apply a boxblur filter with the luma, chroma, and alpha radius
22178 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.
22180 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
22181 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
22185 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.
22187 For the luma plane, a 2x2 box radius will be run once.
22189 For the chroma plane, a 4x4 box radius will be run 5 times.
22191 For the alpha plane, a 3x3 box radius will be run 7 times.
22193 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
22197 @section colorkey_opencl
22198 RGB colorspace color keying.
22200 The filter accepts the following options:
22204 The color which will be replaced with transparency.
22207 Similarity percentage with the key color.
22209 0.01 matches only the exact key color, while 1.0 matches everything.
22214 0.0 makes pixels either fully transparent, or not transparent at all.
22216 Higher values result in semi-transparent pixels, with a higher transparency
22217 the more similar the pixels color is to the key color.
22220 @subsection Examples
22224 Make every semi-green pixel in the input transparent with some slight blending:
22226 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22230 @section convolution_opencl
22232 Apply convolution of 3x3, 5x5, 7x7 matrix.
22234 The filter accepts the following options:
22241 Set matrix for each plane.
22242 Matrix is sequence of 9, 25 or 49 signed numbers.
22243 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
22249 Set multiplier for calculated value for each plane.
22250 If unset or 0, it will be sum of all matrix elements.
22251 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
22257 Set bias for each plane. This value is added to the result of the multiplication.
22258 Useful for making the overall image brighter or darker.
22259 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
22263 @subsection Examples
22269 -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
22275 -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
22279 Apply edge enhance:
22281 -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
22287 -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
22291 Apply laplacian edge detector which includes diagonals:
22293 -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
22299 -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
22303 @section erosion_opencl
22305 Apply erosion effect to the video.
22307 This filter replaces the pixel by the local(3x3) minimum.
22309 It accepts the following options:
22316 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22317 If @code{0}, plane will remain unchanged.
22320 Flag which specifies the pixel to refer to.
22321 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22323 Flags to local 3x3 coordinates region centered on @code{x}:
22332 @subsection Example
22336 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.
22338 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22342 @section deshake_opencl
22343 Feature-point based video stabilization filter.
22345 The filter accepts the following options:
22349 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
22352 Whether or not additional debug info should be displayed, both in the processed output and in the console.
22354 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
22356 Viewing point matches in the output video is only supported for RGB input.
22358 Defaults to @code{0}.
22360 @item adaptive_crop
22361 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
22363 Defaults to @code{1}.
22365 @item refine_features
22366 Whether or not feature points should be refined at a sub-pixel level.
22368 This can be turned off for a slight performance gain at the cost of precision.
22370 Defaults to @code{1}.
22372 @item smooth_strength
22373 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
22375 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
22377 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
22379 Defaults to @code{0.0}.
22381 @item smooth_window_multiplier
22382 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
22384 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
22386 Acceptable values range from @code{0.1} to @code{10.0}.
22388 Larger values increase the amount of motion data available for determining how to smooth the camera path,
22389 potentially improving smoothness, but also increase latency and memory usage.
22391 Defaults to @code{2.0}.
22395 @subsection Examples
22399 Stabilize a video with a fixed, medium smoothing strength:
22401 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22405 Stabilize a video with debugging (both in console and in rendered video):
22407 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22411 @section dilation_opencl
22413 Apply dilation effect to the video.
22415 This filter replaces the pixel by the local(3x3) maximum.
22417 It accepts the following options:
22424 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22425 If @code{0}, plane will remain unchanged.
22428 Flag which specifies the pixel to refer to.
22429 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22431 Flags to local 3x3 coordinates region centered on @code{x}:
22440 @subsection Example
22444 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.
22446 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22450 @section nlmeans_opencl
22452 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
22454 @section overlay_opencl
22456 Overlay one video on top of another.
22458 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
22459 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
22461 The filter accepts the following options:
22466 Set the x coordinate of the overlaid video on the main video.
22467 Default value is @code{0}.
22470 Set the y coordinate of the overlaid video on the main video.
22471 Default value is @code{0}.
22475 @subsection Examples
22479 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
22481 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22484 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
22486 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22491 @section pad_opencl
22493 Add paddings to the input image, and place the original input at the
22494 provided @var{x}, @var{y} coordinates.
22496 It accepts the following options:
22501 Specify an expression for the size of the output image with the
22502 paddings added. If the value for @var{width} or @var{height} is 0, the
22503 corresponding input size is used for the output.
22505 The @var{width} expression can reference the value set by the
22506 @var{height} expression, and vice versa.
22508 The default value of @var{width} and @var{height} is 0.
22512 Specify the offsets to place the input image at within the padded area,
22513 with respect to the top/left border of the output image.
22515 The @var{x} expression can reference the value set by the @var{y}
22516 expression, and vice versa.
22518 The default value of @var{x} and @var{y} is 0.
22520 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22521 so the input image is centered on the padded area.
22524 Specify the color of the padded area. For the syntax of this option,
22525 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22526 manual,ffmpeg-utils}.
22529 Pad to an aspect instead to a resolution.
22532 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22533 options are expressions containing the following constants:
22538 The input video width and height.
22542 These are the same as @var{in_w} and @var{in_h}.
22546 The output width and height (the size of the padded area), as
22547 specified by the @var{width} and @var{height} expressions.
22551 These are the same as @var{out_w} and @var{out_h}.
22555 The x and y offsets as specified by the @var{x} and @var{y}
22556 expressions, or NAN if not yet specified.
22559 same as @var{iw} / @var{ih}
22562 input sample aspect ratio
22565 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22568 @section prewitt_opencl
22570 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22572 The filter accepts the following option:
22576 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22579 Set value which will be multiplied with filtered result.
22580 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22583 Set value which will be added to filtered result.
22584 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22587 @subsection Example
22591 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22593 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22597 @anchor{program_opencl}
22598 @section program_opencl
22600 Filter video using an OpenCL program.
22605 OpenCL program source file.
22608 Kernel name in program.
22611 Number of inputs to the filter. Defaults to 1.
22614 Size of output frames. Defaults to the same as the first input.
22618 The @code{program_opencl} filter also supports the @ref{framesync} options.
22620 The program source file must contain a kernel function with the given name,
22621 which will be run once for each plane of the output. Each run on a plane
22622 gets enqueued as a separate 2D global NDRange with one work-item for each
22623 pixel to be generated. The global ID offset for each work-item is therefore
22624 the coordinates of a pixel in the destination image.
22626 The kernel function needs to take the following arguments:
22629 Destination image, @var{__write_only image2d_t}.
22631 This image will become the output; the kernel should write all of it.
22633 Frame index, @var{unsigned int}.
22635 This is a counter starting from zero and increasing by one for each frame.
22637 Source images, @var{__read_only image2d_t}.
22639 These are the most recent images on each input. The kernel may read from
22640 them to generate the output, but they can't be written to.
22647 Copy the input to the output (output must be the same size as the input).
22649 __kernel void copy(__write_only image2d_t destination,
22650 unsigned int index,
22651 __read_only image2d_t source)
22653 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22655 int2 location = (int2)(get_global_id(0), get_global_id(1));
22657 float4 value = read_imagef(source, sampler, location);
22659 write_imagef(destination, location, value);
22664 Apply a simple transformation, rotating the input by an amount increasing
22665 with the index counter. Pixel values are linearly interpolated by the
22666 sampler, and the output need not have the same dimensions as the input.
22668 __kernel void rotate_image(__write_only image2d_t dst,
22669 unsigned int index,
22670 __read_only image2d_t src)
22672 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22673 CLK_FILTER_LINEAR);
22675 float angle = (float)index / 100.0f;
22677 float2 dst_dim = convert_float2(get_image_dim(dst));
22678 float2 src_dim = convert_float2(get_image_dim(src));
22680 float2 dst_cen = dst_dim / 2.0f;
22681 float2 src_cen = src_dim / 2.0f;
22683 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22685 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22687 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22688 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22690 src_pos = src_pos * src_dim / dst_dim;
22692 float2 src_loc = src_pos + src_cen;
22694 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22695 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22696 write_imagef(dst, dst_loc, 0.5f);
22698 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22703 Blend two inputs together, with the amount of each input used varying
22704 with the index counter.
22706 __kernel void blend_images(__write_only image2d_t dst,
22707 unsigned int index,
22708 __read_only image2d_t src1,
22709 __read_only image2d_t src2)
22711 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22712 CLK_FILTER_LINEAR);
22714 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22716 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22717 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22718 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22720 float4 val1 = read_imagef(src1, sampler, src1_loc);
22721 float4 val2 = read_imagef(src2, sampler, src2_loc);
22723 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22729 @section roberts_opencl
22730 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22732 The filter accepts the following option:
22736 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22739 Set value which will be multiplied with filtered result.
22740 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22743 Set value which will be added to filtered result.
22744 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22747 @subsection Example
22751 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22753 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22757 @section sobel_opencl
22759 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22761 The filter accepts the following option:
22765 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22768 Set value which will be multiplied with filtered result.
22769 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22772 Set value which will be added to filtered result.
22773 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22776 @subsection Example
22780 Apply sobel operator with scale set to 2 and delta set to 10
22782 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22786 @section tonemap_opencl
22788 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22790 It accepts the following parameters:
22794 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22797 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22800 Apply desaturation for highlights that exceed this level of brightness. The
22801 higher the parameter, the more color information will be preserved. This
22802 setting helps prevent unnaturally blown-out colors for super-highlights, by
22803 (smoothly) turning into white instead. This makes images feel more natural,
22804 at the cost of reducing information about out-of-range colors.
22806 The default value is 0.5, and the algorithm here is a little different from
22807 the cpu version tonemap currently. A setting of 0.0 disables this option.
22810 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22811 is used to detect whether the scene has changed or not. If the distance between
22812 the current frame average brightness and the current running average exceeds
22813 a threshold value, we would re-calculate scene average and peak brightness.
22814 The default value is 0.2.
22817 Specify the output pixel format.
22819 Currently supported formats are:
22826 Set the output color range.
22828 Possible values are:
22834 Default is same as input.
22837 Set the output color primaries.
22839 Possible values are:
22845 Default is same as input.
22848 Set the output transfer characteristics.
22850 Possible values are:
22859 Set the output colorspace matrix.
22861 Possible value are:
22867 Default is same as input.
22871 @subsection Example
22875 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22877 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22881 @section unsharp_opencl
22883 Sharpen or blur the input video.
22885 It accepts the following parameters:
22888 @item luma_msize_x, lx
22889 Set the luma matrix horizontal size.
22890 Range is @code{[1, 23]} and default value is @code{5}.
22892 @item luma_msize_y, ly
22893 Set the luma matrix vertical size.
22894 Range is @code{[1, 23]} and default value is @code{5}.
22896 @item luma_amount, la
22897 Set the luma effect strength.
22898 Range is @code{[-10, 10]} and default value is @code{1.0}.
22900 Negative values will blur the input video, while positive values will
22901 sharpen it, a value of zero will disable the effect.
22903 @item chroma_msize_x, cx
22904 Set the chroma matrix horizontal size.
22905 Range is @code{[1, 23]} and default value is @code{5}.
22907 @item chroma_msize_y, cy
22908 Set the chroma matrix vertical size.
22909 Range is @code{[1, 23]} and default value is @code{5}.
22911 @item chroma_amount, ca
22912 Set the chroma effect strength.
22913 Range is @code{[-10, 10]} and default value is @code{0.0}.
22915 Negative values will blur the input video, while positive values will
22916 sharpen it, a value of zero will disable the effect.
22920 All parameters are optional and default to the equivalent of the
22921 string '5:5:1.0:5:5:0.0'.
22923 @subsection Examples
22927 Apply strong luma sharpen effect:
22929 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22933 Apply a strong blur of both luma and chroma parameters:
22935 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22939 @section xfade_opencl
22941 Cross fade two videos with custom transition effect by using OpenCL.
22943 It accepts the following options:
22947 Set one of possible transition effects.
22951 Select custom transition effect, the actual transition description
22952 will be picked from source and kernel options.
22964 Default transition is fade.
22968 OpenCL program source file for custom transition.
22971 Set name of kernel to use for custom transition from program source file.
22974 Set duration of video transition.
22977 Set time of start of transition relative to first video.
22980 The program source file must contain a kernel function with the given name,
22981 which will be run once for each plane of the output. Each run on a plane
22982 gets enqueued as a separate 2D global NDRange with one work-item for each
22983 pixel to be generated. The global ID offset for each work-item is therefore
22984 the coordinates of a pixel in the destination image.
22986 The kernel function needs to take the following arguments:
22989 Destination image, @var{__write_only image2d_t}.
22991 This image will become the output; the kernel should write all of it.
22994 First Source image, @var{__read_only image2d_t}.
22995 Second Source image, @var{__read_only image2d_t}.
22997 These are the most recent images on each input. The kernel may read from
22998 them to generate the output, but they can't be written to.
23001 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
23008 Apply dots curtain transition effect:
23010 __kernel void blend_images(__write_only image2d_t dst,
23011 __read_only image2d_t src1,
23012 __read_only image2d_t src2,
23015 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
23016 CLK_FILTER_LINEAR);
23017 int2 p = (int2)(get_global_id(0), get_global_id(1));
23018 float2 rp = (float2)(get_global_id(0), get_global_id(1));
23019 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
23022 float2 dots = (float2)(20.0, 20.0);
23023 float2 center = (float2)(0,0);
23026 float4 val1 = read_imagef(src1, sampler, p);
23027 float4 val2 = read_imagef(src2, sampler, p);
23028 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
23030 write_imagef(dst, p, next ? val1 : val2);
23036 @c man end OPENCL VIDEO FILTERS
23038 @chapter VAAPI Video Filters
23039 @c man begin VAAPI VIDEO FILTERS
23041 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
23043 To enable compilation of these filters you need to configure FFmpeg with
23044 @code{--enable-vaapi}.
23046 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}
23048 @section tonemap_vaapi
23050 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
23051 It maps the dynamic range of HDR10 content to the SDR content.
23052 It currently only accepts HDR10 as input.
23054 It accepts the following parameters:
23058 Specify the output pixel format.
23060 Currently supported formats are:
23069 Set the output color primaries.
23071 Default is same as input.
23074 Set the output transfer characteristics.
23079 Set the output colorspace matrix.
23081 Default is same as input.
23085 @subsection Example
23089 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
23091 tonemap_vaapi=format=p010:t=bt2020-10
23095 @c man end VAAPI VIDEO FILTERS
23097 @chapter Video Sources
23098 @c man begin VIDEO SOURCES
23100 Below is a description of the currently available video sources.
23104 Buffer video frames, and make them available to the filter chain.
23106 This source is mainly intended for a programmatic use, in particular
23107 through the interface defined in @file{libavfilter/buffersrc.h}.
23109 It accepts the following parameters:
23114 Specify the size (width and height) of the buffered video frames. For the
23115 syntax of this option, check the
23116 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23119 The input video width.
23122 The input video height.
23125 A string representing the pixel format of the buffered video frames.
23126 It may be a number corresponding to a pixel format, or a pixel format
23130 Specify the timebase assumed by the timestamps of the buffered frames.
23133 Specify the frame rate expected for the video stream.
23135 @item pixel_aspect, sar
23136 The sample (pixel) aspect ratio of the input video.
23139 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
23140 to the filtergraph description to specify swscale flags for automatically
23141 inserted scalers. See @ref{Filtergraph syntax}.
23143 @item hw_frames_ctx
23144 When using a hardware pixel format, this should be a reference to an
23145 AVHWFramesContext describing input frames.
23150 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
23153 will instruct the source to accept video frames with size 320x240 and
23154 with format "yuv410p", assuming 1/24 as the timestamps timebase and
23155 square pixels (1:1 sample aspect ratio).
23156 Since the pixel format with name "yuv410p" corresponds to the number 6
23157 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
23158 this example corresponds to:
23160 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
23163 Alternatively, the options can be specified as a flat string, but this
23164 syntax is deprecated:
23166 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
23170 Create a pattern generated by an elementary cellular automaton.
23172 The initial state of the cellular automaton can be defined through the
23173 @option{filename} and @option{pattern} options. If such options are
23174 not specified an initial state is created randomly.
23176 At each new frame a new row in the video is filled with the result of
23177 the cellular automaton next generation. The behavior when the whole
23178 frame is filled is defined by the @option{scroll} option.
23180 This source accepts the following options:
23184 Read the initial cellular automaton state, i.e. the starting row, from
23185 the specified file.
23186 In the file, each non-whitespace character is considered an alive
23187 cell, a newline will terminate the row, and further characters in the
23188 file will be ignored.
23191 Read the initial cellular automaton state, i.e. the starting row, from
23192 the specified string.
23194 Each non-whitespace character in the string is considered an alive
23195 cell, a newline will terminate the row, and further characters in the
23196 string will be ignored.
23199 Set the video rate, that is the number of frames generated per second.
23202 @item random_fill_ratio, ratio
23203 Set the random fill ratio for the initial cellular automaton row. It
23204 is a floating point number value ranging from 0 to 1, defaults to
23207 This option is ignored when a file or a pattern is specified.
23209 @item random_seed, seed
23210 Set the seed for filling randomly the initial row, must be an integer
23211 included between 0 and UINT32_MAX. If not specified, or if explicitly
23212 set to -1, the filter will try to use a good random seed on a best
23216 Set the cellular automaton rule, it is a number ranging from 0 to 255.
23217 Default value is 110.
23220 Set the size of the output video. For the syntax of this option, check the
23221 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23223 If @option{filename} or @option{pattern} is specified, the size is set
23224 by default to the width of the specified initial state row, and the
23225 height is set to @var{width} * PHI.
23227 If @option{size} is set, it must contain the width of the specified
23228 pattern string, and the specified pattern will be centered in the
23231 If a filename or a pattern string is not specified, the size value
23232 defaults to "320x518" (used for a randomly generated initial state).
23235 If set to 1, scroll the output upward when all the rows in the output
23236 have been already filled. If set to 0, the new generated row will be
23237 written over the top row just after the bottom row is filled.
23240 @item start_full, full
23241 If set to 1, completely fill the output with generated rows before
23242 outputting the first frame.
23243 This is the default behavior, for disabling set the value to 0.
23246 If set to 1, stitch the left and right row edges together.
23247 This is the default behavior, for disabling set the value to 0.
23250 @subsection Examples
23254 Read the initial state from @file{pattern}, and specify an output of
23257 cellauto=f=pattern:s=200x400
23261 Generate a random initial row with a width of 200 cells, with a fill
23264 cellauto=ratio=2/3:s=200x200
23268 Create a pattern generated by rule 18 starting by a single alive cell
23269 centered on an initial row with width 100:
23271 cellauto=p=@@:s=100x400:full=0:rule=18
23275 Specify a more elaborated initial pattern:
23277 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
23282 @anchor{coreimagesrc}
23283 @section coreimagesrc
23284 Video source generated on GPU using Apple's CoreImage API on OSX.
23286 This video source is a specialized version of the @ref{coreimage} video filter.
23287 Use a core image generator at the beginning of the applied filterchain to
23288 generate the content.
23290 The coreimagesrc video source accepts the following options:
23292 @item list_generators
23293 List all available generators along with all their respective options as well as
23294 possible minimum and maximum values along with the default values.
23296 list_generators=true
23300 Specify the size of the sourced video. For the syntax of this option, check the
23301 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23302 The default value is @code{320x240}.
23305 Specify the frame rate of the sourced video, as the number of frames
23306 generated per second. It has to be a string in the format
23307 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23308 number or a valid video frame rate abbreviation. The default value is
23312 Set the sample aspect ratio of the sourced video.
23315 Set the duration of the sourced video. See
23316 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23317 for the accepted syntax.
23319 If not specified, or the expressed duration is negative, the video is
23320 supposed to be generated forever.
23323 Additionally, all options of the @ref{coreimage} video filter are accepted.
23324 A complete filterchain can be used for further processing of the
23325 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
23326 and examples for details.
23328 @subsection Examples
23333 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23334 given as complete and escaped command-line for Apple's standard bash shell:
23336 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
23338 This example is equivalent to the QRCode example of @ref{coreimage} without the
23339 need for a nullsrc video source.
23344 Generate several gradients.
23348 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23349 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23352 Set frame rate, expressed as number of frames per second. Default
23355 @item c0, c1, c2, c3, c4, c5, c6, c7
23356 Set 8 colors. Default values for colors is to pick random one.
23358 @item x0, y0, y0, y1
23359 Set gradient line source and destination points. If negative or out of range, random ones
23363 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
23366 Set seed for picking gradient line points.
23369 Set the duration of the sourced video. See
23370 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23371 for the accepted syntax.
23373 If not specified, or the expressed duration is negative, the video is
23374 supposed to be generated forever.
23377 Set speed of gradients rotation.
23381 @section mandelbrot
23383 Generate a Mandelbrot set fractal, and progressively zoom towards the
23384 point specified with @var{start_x} and @var{start_y}.
23386 This source accepts the following options:
23391 Set the terminal pts value. Default value is 400.
23394 Set the terminal scale value.
23395 Must be a floating point value. Default value is 0.3.
23398 Set the inner coloring mode, that is the algorithm used to draw the
23399 Mandelbrot fractal internal region.
23401 It shall assume one of the following values:
23406 Show time until convergence.
23408 Set color based on point closest to the origin of the iterations.
23413 Default value is @var{mincol}.
23416 Set the bailout value. Default value is 10.0.
23419 Set the maximum of iterations performed by the rendering
23420 algorithm. Default value is 7189.
23423 Set outer coloring mode.
23424 It shall assume one of following values:
23426 @item iteration_count
23427 Set iteration count mode.
23428 @item normalized_iteration_count
23429 set normalized iteration count mode.
23431 Default value is @var{normalized_iteration_count}.
23434 Set frame rate, expressed as number of frames per second. Default
23438 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23439 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23442 Set the initial scale value. Default value is 3.0.
23445 Set the initial x position. Must be a floating point value between
23446 -100 and 100. Default value is -0.743643887037158704752191506114774.
23449 Set the initial y position. Must be a floating point value between
23450 -100 and 100. Default value is -0.131825904205311970493132056385139.
23455 Generate various test patterns, as generated by the MPlayer test filter.
23457 The size of the generated video is fixed, and is 256x256.
23458 This source is useful in particular for testing encoding features.
23460 This source accepts the following options:
23465 Specify the frame rate of the sourced video, as the number of frames
23466 generated per second. It has to be a string in the format
23467 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23468 number or a valid video frame rate abbreviation. The default value is
23472 Set the duration of the sourced video. See
23473 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23474 for the accepted syntax.
23476 If not specified, or the expressed duration is negative, the video is
23477 supposed to be generated forever.
23481 Set the number or the name of the test to perform. Supported tests are:
23495 @item max_frames, m
23496 Set the maximum number of frames generated for each test, default value is 30.
23500 Default value is "all", which will cycle through the list of all tests.
23505 mptestsrc=t=dc_luma
23508 will generate a "dc_luma" test pattern.
23510 @section frei0r_src
23512 Provide a frei0r source.
23514 To enable compilation of this filter you need to install the frei0r
23515 header and configure FFmpeg with @code{--enable-frei0r}.
23517 This source accepts the following parameters:
23522 The size of the video to generate. For the syntax of this option, check the
23523 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23526 The framerate of the generated video. It may be a string of the form
23527 @var{num}/@var{den} or a frame rate abbreviation.
23530 The name to the frei0r source to load. For more information regarding frei0r and
23531 how to set the parameters, read the @ref{frei0r} section in the video filters
23534 @item filter_params
23535 A '|'-separated list of parameters to pass to the frei0r source.
23539 For example, to generate a frei0r partik0l source with size 200x200
23540 and frame rate 10 which is overlaid on the overlay filter main input:
23542 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23547 Generate a life pattern.
23549 This source is based on a generalization of John Conway's life game.
23551 The sourced input represents a life grid, each pixel represents a cell
23552 which can be in one of two possible states, alive or dead. Every cell
23553 interacts with its eight neighbours, which are the cells that are
23554 horizontally, vertically, or diagonally adjacent.
23556 At each interaction the grid evolves according to the adopted rule,
23557 which specifies the number of neighbor alive cells which will make a
23558 cell stay alive or born. The @option{rule} option allows one to specify
23561 This source accepts the following options:
23565 Set the file from which to read the initial grid state. In the file,
23566 each non-whitespace character is considered an alive cell, and newline
23567 is used to delimit the end of each row.
23569 If this option is not specified, the initial grid is generated
23573 Set the video rate, that is the number of frames generated per second.
23576 @item random_fill_ratio, ratio
23577 Set the random fill ratio for the initial random grid. It is a
23578 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23579 It is ignored when a file is specified.
23581 @item random_seed, seed
23582 Set the seed for filling the initial random grid, must be an integer
23583 included between 0 and UINT32_MAX. If not specified, or if explicitly
23584 set to -1, the filter will try to use a good random seed on a best
23590 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23591 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23592 @var{NS} specifies the number of alive neighbor cells which make a
23593 live cell stay alive, and @var{NB} the number of alive neighbor cells
23594 which make a dead cell to become alive (i.e. to "born").
23595 "s" and "b" can be used in place of "S" and "B", respectively.
23597 Alternatively a rule can be specified by an 18-bits integer. The 9
23598 high order bits are used to encode the next cell state if it is alive
23599 for each number of neighbor alive cells, the low order bits specify
23600 the rule for "borning" new cells. Higher order bits encode for an
23601 higher number of neighbor cells.
23602 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23603 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23605 Default value is "S23/B3", which is the original Conway's game of life
23606 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23607 cells, and will born a new cell if there are three alive cells around
23611 Set the size of the output video. For the syntax of this option, check the
23612 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23614 If @option{filename} is specified, the size is set by default to the
23615 same size of the input file. If @option{size} is set, it must contain
23616 the size specified in the input file, and the initial grid defined in
23617 that file is centered in the larger resulting area.
23619 If a filename is not specified, the size value defaults to "320x240"
23620 (used for a randomly generated initial grid).
23623 If set to 1, stitch the left and right grid edges together, and the
23624 top and bottom edges also. Defaults to 1.
23627 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23628 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23629 value from 0 to 255.
23632 Set the color of living (or new born) cells.
23635 Set the color of dead cells. If @option{mold} is set, this is the first color
23636 used to represent a dead cell.
23639 Set mold color, for definitely dead and moldy cells.
23641 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23642 ffmpeg-utils manual,ffmpeg-utils}.
23645 @subsection Examples
23649 Read a grid from @file{pattern}, and center it on a grid of size
23652 life=f=pattern:s=300x300
23656 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23658 life=ratio=2/3:s=200x200
23662 Specify a custom rule for evolving a randomly generated grid:
23668 Full example with slow death effect (mold) using @command{ffplay}:
23670 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23677 @anchor{haldclutsrc}
23680 @anchor{pal100bars}
23681 @anchor{rgbtestsrc}
23683 @anchor{smptehdbars}
23686 @anchor{yuvtestsrc}
23687 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23689 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23691 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23693 The @code{color} source provides an uniformly colored input.
23695 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23696 @ref{haldclut} filter.
23698 The @code{nullsrc} source returns unprocessed video frames. It is
23699 mainly useful to be employed in analysis / debugging tools, or as the
23700 source for filters which ignore the input data.
23702 The @code{pal75bars} source generates a color bars pattern, based on
23703 EBU PAL recommendations with 75% color levels.
23705 The @code{pal100bars} source generates a color bars pattern, based on
23706 EBU PAL recommendations with 100% color levels.
23708 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23709 detecting RGB vs BGR issues. You should see a red, green and blue
23710 stripe from top to bottom.
23712 The @code{smptebars} source generates a color bars pattern, based on
23713 the SMPTE Engineering Guideline EG 1-1990.
23715 The @code{smptehdbars} source generates a color bars pattern, based on
23716 the SMPTE RP 219-2002.
23718 The @code{testsrc} source generates a test video pattern, showing a
23719 color pattern, a scrolling gradient and a timestamp. This is mainly
23720 intended for testing purposes.
23722 The @code{testsrc2} source is similar to testsrc, but supports more
23723 pixel formats instead of just @code{rgb24}. This allows using it as an
23724 input for other tests without requiring a format conversion.
23726 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23727 see a y, cb and cr stripe from top to bottom.
23729 The sources accept the following parameters:
23734 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23735 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23736 pixels to be used as identity matrix for 3D lookup tables. Each component is
23737 coded on a @code{1/(N*N)} scale.
23740 Specify the color of the source, only available in the @code{color}
23741 source. For the syntax of this option, check the
23742 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23745 Specify the size of the sourced video. For the syntax of this option, check the
23746 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23747 The default value is @code{320x240}.
23749 This option is not available with the @code{allrgb}, @code{allyuv}, and
23750 @code{haldclutsrc} filters.
23753 Specify the frame rate of the sourced video, as the number of frames
23754 generated per second. It has to be a string in the format
23755 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23756 number or a valid video frame rate abbreviation. The default value is
23760 Set the duration of the sourced video. See
23761 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23762 for the accepted syntax.
23764 If not specified, or the expressed duration is negative, the video is
23765 supposed to be generated forever.
23767 Since the frame rate is used as time base, all frames including the last one
23768 will have their full duration. If the specified duration is not a multiple
23769 of the frame duration, it will be rounded up.
23772 Set the sample aspect ratio of the sourced video.
23775 Specify the alpha (opacity) of the background, only available in the
23776 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23777 255 (fully opaque, the default).
23780 Set the number of decimals to show in the timestamp, only available in the
23781 @code{testsrc} source.
23783 The displayed timestamp value will correspond to the original
23784 timestamp value multiplied by the power of 10 of the specified
23785 value. Default value is 0.
23788 @subsection Examples
23792 Generate a video with a duration of 5.3 seconds, with size
23793 176x144 and a frame rate of 10 frames per second:
23795 testsrc=duration=5.3:size=qcif:rate=10
23799 The following graph description will generate a red source
23800 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23803 color=c=red@@0.2:s=qcif:r=10
23807 If the input content is to be ignored, @code{nullsrc} can be used. The
23808 following command generates noise in the luminance plane by employing
23809 the @code{geq} filter:
23811 nullsrc=s=256x256, geq=random(1)*255:128:128
23815 @subsection Commands
23817 The @code{color} source supports the following commands:
23821 Set the color of the created image. Accepts the same syntax of the
23822 corresponding @option{color} option.
23827 Generate video using an OpenCL program.
23832 OpenCL program source file.
23835 Kernel name in program.
23838 Size of frames to generate. This must be set.
23841 Pixel format to use for the generated frames. This must be set.
23844 Number of frames generated every second. Default value is '25'.
23848 For details of how the program loading works, see the @ref{program_opencl}
23855 Generate a colour ramp by setting pixel values from the position of the pixel
23856 in the output image. (Note that this will work with all pixel formats, but
23857 the generated output will not be the same.)
23859 __kernel void ramp(__write_only image2d_t dst,
23860 unsigned int index)
23862 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23865 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23867 write_imagef(dst, loc, val);
23872 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23874 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23875 unsigned int index)
23877 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23879 float4 value = 0.0f;
23880 int x = loc.x + index;
23881 int y = loc.y + index;
23882 while (x > 0 || y > 0) {
23883 if (x % 3 == 1 && y % 3 == 1) {
23891 write_imagef(dst, loc, value);
23897 @section sierpinski
23899 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23901 This source accepts the following options:
23905 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23906 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23909 Set frame rate, expressed as number of frames per second. Default
23913 Set seed which is used for random panning.
23916 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23919 Set fractal type, can be default @code{carpet} or @code{triangle}.
23922 @c man end VIDEO SOURCES
23924 @chapter Video Sinks
23925 @c man begin VIDEO SINKS
23927 Below is a description of the currently available video sinks.
23929 @section buffersink
23931 Buffer video frames, and make them available to the end of the filter
23934 This sink is mainly intended for programmatic use, in particular
23935 through the interface defined in @file{libavfilter/buffersink.h}
23936 or the options system.
23938 It accepts a pointer to an AVBufferSinkContext structure, which
23939 defines the incoming buffers' formats, to be passed as the opaque
23940 parameter to @code{avfilter_init_filter} for initialization.
23944 Null video sink: do absolutely nothing with the input video. It is
23945 mainly useful as a template and for use in analysis / debugging
23948 @c man end VIDEO SINKS
23950 @chapter Multimedia Filters
23951 @c man begin MULTIMEDIA FILTERS
23953 Below is a description of the currently available multimedia filters.
23957 Convert input audio to a video output, displaying the audio bit scope.
23959 The filter accepts the following options:
23963 Set frame rate, expressed as number of frames per second. Default
23967 Specify the video size for the output. For the syntax of this option, check the
23968 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23969 Default value is @code{1024x256}.
23972 Specify list of colors separated by space or by '|' which will be used to
23973 draw channels. Unrecognized or missing colors will be replaced
23977 @section adrawgraph
23978 Draw a graph using input audio metadata.
23980 See @ref{drawgraph}
23982 @section agraphmonitor
23984 See @ref{graphmonitor}.
23986 @section ahistogram
23988 Convert input audio to a video output, displaying the volume histogram.
23990 The filter accepts the following options:
23994 Specify how histogram is calculated.
23996 It accepts the following values:
23999 Use single histogram for all channels.
24001 Use separate histogram for each channel.
24003 Default is @code{single}.
24006 Set frame rate, expressed as number of frames per second. Default
24010 Specify the video size for the output. For the syntax of this option, check the
24011 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24012 Default value is @code{hd720}.
24017 It accepts the following values:
24028 reverse logarithmic
24030 Default is @code{log}.
24033 Set amplitude scale.
24035 It accepts the following values:
24042 Default is @code{log}.
24045 Set how much frames to accumulate in histogram.
24046 Default is 1. Setting this to -1 accumulates all frames.
24049 Set histogram ratio of window height.
24052 Set sonogram sliding.
24054 It accepts the following values:
24057 replace old rows with new ones.
24059 scroll from top to bottom.
24061 Default is @code{replace}.
24064 @section aphasemeter
24066 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
24067 representing mean phase of current audio frame. A video output can also be produced and is
24068 enabled by default. The audio is passed through as first output.
24070 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
24071 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
24072 and @code{1} means channels are in phase.
24074 The filter accepts the following options, all related to its video output:
24078 Set the output frame rate. Default value is @code{25}.
24081 Set the video size for the output. For the syntax of this option, check the
24082 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24083 Default value is @code{800x400}.
24088 Specify the red, green, blue contrast. Default values are @code{2},
24089 @code{7} and @code{1}.
24090 Allowed range is @code{[0, 255]}.
24093 Set color which will be used for drawing median phase. If color is
24094 @code{none} which is default, no median phase value will be drawn.
24097 Enable video output. Default is enabled.
24100 @subsection phasing detection
24102 The filter also detects out of phase and mono sequences in stereo streams.
24103 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
24105 The filter accepts the following options for this detection:
24109 Enable mono and out of phase detection. Default is disabled.
24112 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
24113 Allowed range is @code{[0, 1]}.
24116 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
24117 Allowed range is @code{[90, 180]}.
24120 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
24123 @subsection Examples
24127 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
24129 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
24133 @section avectorscope
24135 Convert input audio to a video output, representing the audio vector
24138 The filter is used to measure the difference between channels of stereo
24139 audio stream. A monaural signal, consisting of identical left and right
24140 signal, results in straight vertical line. Any stereo separation is visible
24141 as a deviation from this line, creating a Lissajous figure.
24142 If the straight (or deviation from it) but horizontal line appears this
24143 indicates that the left and right channels are out of phase.
24145 The filter accepts the following options:
24149 Set the vectorscope mode.
24151 Available values are:
24154 Lissajous rotated by 45 degrees.
24157 Same as above but not rotated.
24160 Shape resembling half of circle.
24163 Default value is @samp{lissajous}.
24166 Set the video size for the output. For the syntax of this option, check the
24167 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24168 Default value is @code{400x400}.
24171 Set the output frame rate. Default value is @code{25}.
24177 Specify the red, green, blue and alpha contrast. Default values are @code{40},
24178 @code{160}, @code{80} and @code{255}.
24179 Allowed range is @code{[0, 255]}.
24185 Specify the red, green, blue and alpha fade. Default values are @code{15},
24186 @code{10}, @code{5} and @code{5}.
24187 Allowed range is @code{[0, 255]}.
24190 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
24191 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
24194 Set the vectorscope drawing mode.
24196 Available values are:
24199 Draw dot for each sample.
24202 Draw line between previous and current sample.
24205 Default value is @samp{dot}.
24208 Specify amplitude scale of audio samples.
24210 Available values are:
24226 Swap left channel axis with right channel axis.
24236 Mirror only x axis.
24239 Mirror only y axis.
24247 @subsection Examples
24251 Complete example using @command{ffplay}:
24253 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24254 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24258 @section bench, abench
24260 Benchmark part of a filtergraph.
24262 The filter accepts the following options:
24266 Start or stop a timer.
24268 Available values are:
24271 Get the current time, set it as frame metadata (using the key
24272 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
24275 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
24276 the input frame metadata to get the time difference. Time difference, average,
24277 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
24278 @code{min}) are then printed. The timestamps are expressed in seconds.
24282 @subsection Examples
24286 Benchmark @ref{selectivecolor} filter:
24288 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24294 Concatenate audio and video streams, joining them together one after the
24297 The filter works on segments of synchronized video and audio streams. All
24298 segments must have the same number of streams of each type, and that will
24299 also be the number of streams at output.
24301 The filter accepts the following options:
24306 Set the number of segments. Default is 2.
24309 Set the number of output video streams, that is also the number of video
24310 streams in each segment. Default is 1.
24313 Set the number of output audio streams, that is also the number of audio
24314 streams in each segment. Default is 0.
24317 Activate unsafe mode: do not fail if segments have a different format.
24321 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
24322 @var{a} audio outputs.
24324 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
24325 segment, in the same order as the outputs, then the inputs for the second
24328 Related streams do not always have exactly the same duration, for various
24329 reasons including codec frame size or sloppy authoring. For that reason,
24330 related synchronized streams (e.g. a video and its audio track) should be
24331 concatenated at once. The concat filter will use the duration of the longest
24332 stream in each segment (except the last one), and if necessary pad shorter
24333 audio streams with silence.
24335 For this filter to work correctly, all segments must start at timestamp 0.
24337 All corresponding streams must have the same parameters in all segments; the
24338 filtering system will automatically select a common pixel format for video
24339 streams, and a common sample format, sample rate and channel layout for
24340 audio streams, but other settings, such as resolution, must be converted
24341 explicitly by the user.
24343 Different frame rates are acceptable but will result in variable frame rate
24344 at output; be sure to configure the output file to handle it.
24346 @subsection Examples
24350 Concatenate an opening, an episode and an ending, all in bilingual version
24351 (video in stream 0, audio in streams 1 and 2):
24353 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24354 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24355 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24356 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24360 Concatenate two parts, handling audio and video separately, using the
24361 (a)movie sources, and adjusting the resolution:
24363 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24364 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24365 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24367 Note that a desync will happen at the stitch if the audio and video streams
24368 do not have exactly the same duration in the first file.
24372 @subsection Commands
24374 This filter supports the following commands:
24377 Close the current segment and step to the next one
24383 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
24384 level. By default, it logs a message at a frequency of 10Hz with the
24385 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
24386 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
24388 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
24389 sample format is double-precision floating point. The input stream will be converted to
24390 this specification, if needed. Users may need to insert aformat and/or aresample filters
24391 after this filter to obtain the original parameters.
24393 The filter also has a video output (see the @var{video} option) with a real
24394 time graph to observe the loudness evolution. The graphic contains the logged
24395 message mentioned above, so it is not printed anymore when this option is set,
24396 unless the verbose logging is set. The main graphing area contains the
24397 short-term loudness (3 seconds of analysis), and the gauge on the right is for
24398 the momentary loudness (400 milliseconds), but can optionally be configured
24399 to instead display short-term loudness (see @var{gauge}).
24401 The green area marks a +/- 1LU target range around the target loudness
24402 (-23LUFS by default, unless modified through @var{target}).
24404 More information about the Loudness Recommendation EBU R128 on
24405 @url{http://tech.ebu.ch/loudness}.
24407 The filter accepts the following options:
24412 Activate the video output. The audio stream is passed unchanged whether this
24413 option is set or no. The video stream will be the first output stream if
24414 activated. Default is @code{0}.
24417 Set the video size. This option is for video only. For the syntax of this
24419 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24420 Default and minimum resolution is @code{640x480}.
24423 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
24424 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
24425 other integer value between this range is allowed.
24428 Set metadata injection. If set to @code{1}, the audio input will be segmented
24429 into 100ms output frames, each of them containing various loudness information
24430 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
24432 Default is @code{0}.
24435 Force the frame logging level.
24437 Available values are:
24440 information logging level
24442 verbose logging level
24445 By default, the logging level is set to @var{info}. If the @option{video} or
24446 the @option{metadata} options are set, it switches to @var{verbose}.
24451 Available modes can be cumulated (the option is a @code{flag} type). Possible
24455 Disable any peak mode (default).
24457 Enable sample-peak mode.
24459 Simple peak mode looking for the higher sample value. It logs a message
24460 for sample-peak (identified by @code{SPK}).
24462 Enable true-peak mode.
24464 If enabled, the peak lookup is done on an over-sampled version of the input
24465 stream for better peak accuracy. It logs a message for true-peak.
24466 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
24467 This mode requires a build with @code{libswresample}.
24471 Treat mono input files as "dual mono". If a mono file is intended for playback
24472 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
24473 If set to @code{true}, this option will compensate for this effect.
24474 Multi-channel input files are not affected by this option.
24477 Set a specific pan law to be used for the measurement of dual mono files.
24478 This parameter is optional, and has a default value of -3.01dB.
24481 Set a specific target level (in LUFS) used as relative zero in the visualization.
24482 This parameter is optional and has a default value of -23LUFS as specified
24483 by EBU R128. However, material published online may prefer a level of -16LUFS
24484 (e.g. for use with podcasts or video platforms).
24487 Set the value displayed by the gauge. Valid values are @code{momentary} and s
24488 @code{shortterm}. By default the momentary value will be used, but in certain
24489 scenarios it may be more useful to observe the short term value instead (e.g.
24493 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24494 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24495 video output, not the summary or continuous log output.
24498 @subsection Examples
24502 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24504 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24508 Run an analysis with @command{ffmpeg}:
24510 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24514 @section interleave, ainterleave
24516 Temporally interleave frames from several inputs.
24518 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24520 These filters read frames from several inputs and send the oldest
24521 queued frame to the output.
24523 Input streams must have well defined, monotonically increasing frame
24526 In order to submit one frame to output, these filters need to enqueue
24527 at least one frame for each input, so they cannot work in case one
24528 input is not yet terminated and will not receive incoming frames.
24530 For example consider the case when one input is a @code{select} filter
24531 which always drops input frames. The @code{interleave} filter will keep
24532 reading from that input, but it will never be able to send new frames
24533 to output until the input sends an end-of-stream signal.
24535 Also, depending on inputs synchronization, the filters will drop
24536 frames in case one input receives more frames than the other ones, and
24537 the queue is already filled.
24539 These filters accept the following options:
24543 Set the number of different inputs, it is 2 by default.
24546 How to determine the end-of-stream.
24550 The duration of the longest input. (default)
24553 The duration of the shortest input.
24556 The duration of the first input.
24561 @subsection Examples
24565 Interleave frames belonging to different streams using @command{ffmpeg}:
24567 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24571 Add flickering blur effect:
24573 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24577 @section metadata, ametadata
24579 Manipulate frame metadata.
24581 This filter accepts the following options:
24585 Set mode of operation of the filter.
24587 Can be one of the following:
24591 If both @code{value} and @code{key} is set, select frames
24592 which have such metadata. If only @code{key} is set, select
24593 every frame that has such key in metadata.
24596 Add new metadata @code{key} and @code{value}. If key is already available
24600 Modify value of already present key.
24603 If @code{value} is set, delete only keys that have such value.
24604 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24608 Print key and its value if metadata was found. If @code{key} is not set print all
24609 metadata values available in frame.
24613 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24616 Set metadata value which will be used. This option is mandatory for
24617 @code{modify} and @code{add} mode.
24620 Which function to use when comparing metadata value and @code{value}.
24622 Can be one of following:
24626 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24629 Values are interpreted as strings, returns true if metadata value starts with
24630 the @code{value} option string.
24633 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24636 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24639 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24642 Values are interpreted as floats, returns true if expression from option @code{expr}
24646 Values are interpreted as strings, returns true if metadata value ends with
24647 the @code{value} option string.
24651 Set expression which is used when @code{function} is set to @code{expr}.
24652 The expression is evaluated through the eval API and can contain the following
24657 Float representation of @code{value} from metadata key.
24660 Float representation of @code{value} as supplied by user in @code{value} option.
24664 If specified in @code{print} mode, output is written to the named file. Instead of
24665 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24666 for standard output. If @code{file} option is not set, output is written to the log
24667 with AV_LOG_INFO loglevel.
24670 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24674 @subsection Examples
24678 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24681 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24684 Print silencedetect output to file @file{metadata.txt}.
24686 silencedetect,ametadata=mode=print:file=metadata.txt
24689 Direct all metadata to a pipe with file descriptor 4.
24691 metadata=mode=print:file='pipe\:4'
24695 @section perms, aperms
24697 Set read/write permissions for the output frames.
24699 These filters are mainly aimed at developers to test direct path in the
24700 following filter in the filtergraph.
24702 The filters accept the following options:
24706 Select the permissions mode.
24708 It accepts the following values:
24711 Do nothing. This is the default.
24713 Set all the output frames read-only.
24715 Set all the output frames directly writable.
24717 Make the frame read-only if writable, and writable if read-only.
24719 Set each output frame read-only or writable randomly.
24723 Set the seed for the @var{random} mode, must be an integer included between
24724 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24725 @code{-1}, the filter will try to use a good random seed on a best effort
24729 Note: in case of auto-inserted filter between the permission filter and the
24730 following one, the permission might not be received as expected in that
24731 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24732 perms/aperms filter can avoid this problem.
24734 @section realtime, arealtime
24736 Slow down filtering to match real time approximately.
24738 These filters will pause the filtering for a variable amount of time to
24739 match the output rate with the input timestamps.
24740 They are similar to the @option{re} option to @code{ffmpeg}.
24742 They accept the following options:
24746 Time limit for the pauses. Any pause longer than that will be considered
24747 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24749 Speed factor for processing. The value must be a float larger than zero.
24750 Values larger than 1.0 will result in faster than realtime processing,
24751 smaller will slow processing down. The @var{limit} is automatically adapted
24752 accordingly. Default is 1.0.
24754 A processing speed faster than what is possible without these filters cannot
24759 @section select, aselect
24761 Select frames to pass in output.
24763 This filter accepts the following options:
24768 Set expression, which is evaluated for each input frame.
24770 If the expression is evaluated to zero, the frame is discarded.
24772 If the evaluation result is negative or NaN, the frame is sent to the
24773 first output; otherwise it is sent to the output with index
24774 @code{ceil(val)-1}, assuming that the input index starts from 0.
24776 For example a value of @code{1.2} corresponds to the output with index
24777 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24780 Set the number of outputs. The output to which to send the selected
24781 frame is based on the result of the evaluation. Default value is 1.
24784 The expression can contain the following constants:
24788 The (sequential) number of the filtered frame, starting from 0.
24791 The (sequential) number of the selected frame, starting from 0.
24793 @item prev_selected_n
24794 The sequential number of the last selected frame. It's NAN if undefined.
24797 The timebase of the input timestamps.
24800 The PTS (Presentation TimeStamp) of the filtered video frame,
24801 expressed in @var{TB} units. It's NAN if undefined.
24804 The PTS of the filtered video frame,
24805 expressed in seconds. It's NAN if undefined.
24808 The PTS of the previously filtered video frame. It's NAN if undefined.
24810 @item prev_selected_pts
24811 The PTS of the last previously filtered video frame. It's NAN if undefined.
24813 @item prev_selected_t
24814 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24817 The PTS of the first video frame in the video. It's NAN if undefined.
24820 The time of the first video frame in the video. It's NAN if undefined.
24822 @item pict_type @emph{(video only)}
24823 The type of the filtered frame. It can assume one of the following
24835 @item interlace_type @emph{(video only)}
24836 The frame interlace type. It can assume one of the following values:
24839 The frame is progressive (not interlaced).
24841 The frame is top-field-first.
24843 The frame is bottom-field-first.
24846 @item consumed_sample_n @emph{(audio only)}
24847 the number of selected samples before the current frame
24849 @item samples_n @emph{(audio only)}
24850 the number of samples in the current frame
24852 @item sample_rate @emph{(audio only)}
24853 the input sample rate
24856 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24859 the position in the file of the filtered frame, -1 if the information
24860 is not available (e.g. for synthetic video)
24862 @item scene @emph{(video only)}
24863 value between 0 and 1 to indicate a new scene; a low value reflects a low
24864 probability for the current frame to introduce a new scene, while a higher
24865 value means the current frame is more likely to be one (see the example below)
24867 @item concatdec_select
24868 The concat demuxer can select only part of a concat input file by setting an
24869 inpoint and an outpoint, but the output packets may not be entirely contained
24870 in the selected interval. By using this variable, it is possible to skip frames
24871 generated by the concat demuxer which are not exactly contained in the selected
24874 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24875 and the @var{lavf.concat.duration} packet metadata values which are also
24876 present in the decoded frames.
24878 The @var{concatdec_select} variable is -1 if the frame pts is at least
24879 start_time and either the duration metadata is missing or the frame pts is less
24880 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24883 That basically means that an input frame is selected if its pts is within the
24884 interval set by the concat demuxer.
24888 The default value of the select expression is "1".
24890 @subsection Examples
24894 Select all frames in input:
24899 The example above is the same as:
24911 Select only I-frames:
24913 select='eq(pict_type\,I)'
24917 Select one frame every 100:
24919 select='not(mod(n\,100))'
24923 Select only frames contained in the 10-20 time interval:
24925 select=between(t\,10\,20)
24929 Select only I-frames contained in the 10-20 time interval:
24931 select=between(t\,10\,20)*eq(pict_type\,I)
24935 Select frames with a minimum distance of 10 seconds:
24937 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24941 Use aselect to select only audio frames with samples number > 100:
24943 aselect='gt(samples_n\,100)'
24947 Create a mosaic of the first scenes:
24949 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24952 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24956 Send even and odd frames to separate outputs, and compose them:
24958 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24962 Select useful frames from an ffconcat file which is using inpoints and
24963 outpoints but where the source files are not intra frame only.
24965 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24969 @section sendcmd, asendcmd
24971 Send commands to filters in the filtergraph.
24973 These filters read commands to be sent to other filters in the
24976 @code{sendcmd} must be inserted between two video filters,
24977 @code{asendcmd} must be inserted between two audio filters, but apart
24978 from that they act the same way.
24980 The specification of commands can be provided in the filter arguments
24981 with the @var{commands} option, or in a file specified by the
24982 @var{filename} option.
24984 These filters accept the following options:
24987 Set the commands to be read and sent to the other filters.
24989 Set the filename of the commands to be read and sent to the other
24993 @subsection Commands syntax
24995 A commands description consists of a sequence of interval
24996 specifications, comprising a list of commands to be executed when a
24997 particular event related to that interval occurs. The occurring event
24998 is typically the current frame time entering or leaving a given time
25001 An interval is specified by the following syntax:
25003 @var{START}[-@var{END}] @var{COMMANDS};
25006 The time interval is specified by the @var{START} and @var{END} times.
25007 @var{END} is optional and defaults to the maximum time.
25009 The current frame time is considered within the specified interval if
25010 it is included in the interval [@var{START}, @var{END}), that is when
25011 the time is greater or equal to @var{START} and is lesser than
25014 @var{COMMANDS} consists of a sequence of one or more command
25015 specifications, separated by ",", relating to that interval. The
25016 syntax of a command specification is given by:
25018 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
25021 @var{FLAGS} is optional and specifies the type of events relating to
25022 the time interval which enable sending the specified command, and must
25023 be a non-null sequence of identifier flags separated by "+" or "|" and
25024 enclosed between "[" and "]".
25026 The following flags are recognized:
25029 The command is sent when the current frame timestamp enters the
25030 specified interval. In other words, the command is sent when the
25031 previous frame timestamp was not in the given interval, and the
25035 The command is sent when the current frame timestamp leaves the
25036 specified interval. In other words, the command is sent when the
25037 previous frame timestamp was in the given interval, and the
25041 The command @var{ARG} is interpreted as expression and result of
25042 expression is passed as @var{ARG}.
25044 The expression is evaluated through the eval API and can contain the following
25049 Original position in the file of the frame, or undefined if undefined
25050 for the current frame.
25053 The presentation timestamp in input.
25056 The count of the input frame for video or audio, starting from 0.
25059 The time in seconds of the current frame.
25062 The start time in seconds of the current command interval.
25065 The end time in seconds of the current command interval.
25068 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
25073 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
25076 @var{TARGET} specifies the target of the command, usually the name of
25077 the filter class or a specific filter instance name.
25079 @var{COMMAND} specifies the name of the command for the target filter.
25081 @var{ARG} is optional and specifies the optional list of argument for
25082 the given @var{COMMAND}.
25084 Between one interval specification and another, whitespaces, or
25085 sequences of characters starting with @code{#} until the end of line,
25086 are ignored and can be used to annotate comments.
25088 A simplified BNF description of the commands specification syntax
25091 @var{COMMAND_FLAG} ::= "enter" | "leave"
25092 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
25093 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
25094 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
25095 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
25096 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
25099 @subsection Examples
25103 Specify audio tempo change at second 4:
25105 asendcmd=c='4.0 atempo tempo 1.5',atempo
25109 Target a specific filter instance:
25111 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
25115 Specify a list of drawtext and hue commands in a file.
25117 # show text in the interval 5-10
25118 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
25119 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
25121 # desaturate the image in the interval 15-20
25122 15.0-20.0 [enter] hue s 0,
25123 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
25125 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
25127 # apply an exponential saturation fade-out effect, starting from time 25
25128 25 [enter] hue s exp(25-t)
25131 A filtergraph allowing to read and process the above command list
25132 stored in a file @file{test.cmd}, can be specified with:
25134 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
25139 @section setpts, asetpts
25141 Change the PTS (presentation timestamp) of the input frames.
25143 @code{setpts} works on video frames, @code{asetpts} on audio frames.
25145 This filter accepts the following options:
25150 The expression which is evaluated for each frame to construct its timestamp.
25154 The expression is evaluated through the eval API and can contain the following
25158 @item FRAME_RATE, FR
25159 frame rate, only defined for constant frame-rate video
25162 The presentation timestamp in input
25165 The count of the input frame for video or the number of consumed samples,
25166 not including the current frame for audio, starting from 0.
25168 @item NB_CONSUMED_SAMPLES
25169 The number of consumed samples, not including the current frame (only
25172 @item NB_SAMPLES, S
25173 The number of samples in the current frame (only audio)
25175 @item SAMPLE_RATE, SR
25176 The audio sample rate.
25179 The PTS of the first frame.
25182 the time in seconds of the first frame
25185 State whether the current frame is interlaced.
25188 the time in seconds of the current frame
25191 original position in the file of the frame, or undefined if undefined
25192 for the current frame
25195 The previous input PTS.
25198 previous input time in seconds
25201 The previous output PTS.
25204 previous output time in seconds
25207 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
25211 The wallclock (RTC) time at the start of the movie in microseconds.
25214 The timebase of the input timestamps.
25218 @subsection Examples
25222 Start counting PTS from zero
25224 setpts=PTS-STARTPTS
25228 Apply fast motion effect:
25234 Apply slow motion effect:
25240 Set fixed rate of 25 frames per second:
25246 Set fixed rate 25 fps with some jitter:
25248 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25252 Apply an offset of 10 seconds to the input PTS:
25258 Generate timestamps from a "live source" and rebase onto the current timebase:
25260 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25264 Generate timestamps by counting samples:
25273 Force color range for the output video frame.
25275 The @code{setrange} filter marks the color range property for the
25276 output frames. It does not change the input frame, but only sets the
25277 corresponding property, which affects how the frame is treated by
25280 The filter accepts the following options:
25285 Available values are:
25289 Keep the same color range property.
25291 @item unspecified, unknown
25292 Set the color range as unspecified.
25294 @item limited, tv, mpeg
25295 Set the color range as limited.
25297 @item full, pc, jpeg
25298 Set the color range as full.
25302 @section settb, asettb
25304 Set the timebase to use for the output frames timestamps.
25305 It is mainly useful for testing timebase configuration.
25307 It accepts the following parameters:
25312 The expression which is evaluated into the output timebase.
25316 The value for @option{tb} is an arithmetic expression representing a
25317 rational. The expression can contain the constants "AVTB" (the default
25318 timebase), "intb" (the input timebase) and "sr" (the sample rate,
25319 audio only). Default value is "intb".
25321 @subsection Examples
25325 Set the timebase to 1/25:
25331 Set the timebase to 1/10:
25337 Set the timebase to 1001/1000:
25343 Set the timebase to 2*intb:
25349 Set the default timebase value:
25356 Convert input audio to a video output representing frequency spectrum
25357 logarithmically using Brown-Puckette constant Q transform algorithm with
25358 direct frequency domain coefficient calculation (but the transform itself
25359 is not really constant Q, instead the Q factor is actually variable/clamped),
25360 with musical tone scale, from E0 to D#10.
25362 The filter accepts the following options:
25366 Specify the video size for the output. It must be even. For the syntax of this option,
25367 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25368 Default value is @code{1920x1080}.
25371 Set the output frame rate. Default value is @code{25}.
25374 Set the bargraph height. It must be even. Default value is @code{-1} which
25375 computes the bargraph height automatically.
25378 Set the axis height. It must be even. Default value is @code{-1} which computes
25379 the axis height automatically.
25382 Set the sonogram height. It must be even. Default value is @code{-1} which
25383 computes the sonogram height automatically.
25386 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
25387 instead. Default value is @code{1}.
25389 @item sono_v, volume
25390 Specify the sonogram volume expression. It can contain variables:
25393 the @var{bar_v} evaluated expression
25394 @item frequency, freq, f
25395 the frequency where it is evaluated
25396 @item timeclamp, tc
25397 the value of @var{timeclamp} option
25401 @item a_weighting(f)
25402 A-weighting of equal loudness
25403 @item b_weighting(f)
25404 B-weighting of equal loudness
25405 @item c_weighting(f)
25406 C-weighting of equal loudness.
25408 Default value is @code{16}.
25410 @item bar_v, volume2
25411 Specify the bargraph volume expression. It can contain variables:
25414 the @var{sono_v} evaluated expression
25415 @item frequency, freq, f
25416 the frequency where it is evaluated
25417 @item timeclamp, tc
25418 the value of @var{timeclamp} option
25422 @item a_weighting(f)
25423 A-weighting of equal loudness
25424 @item b_weighting(f)
25425 B-weighting of equal loudness
25426 @item c_weighting(f)
25427 C-weighting of equal loudness.
25429 Default value is @code{sono_v}.
25431 @item sono_g, gamma
25432 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
25433 higher gamma makes the spectrum having more range. Default value is @code{3}.
25434 Acceptable range is @code{[1, 7]}.
25436 @item bar_g, gamma2
25437 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
25441 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
25442 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
25444 @item timeclamp, tc
25445 Specify the transform timeclamp. At low frequency, there is trade-off between
25446 accuracy in time domain and frequency domain. If timeclamp is lower,
25447 event in time domain is represented more accurately (such as fast bass drum),
25448 otherwise event in frequency domain is represented more accurately
25449 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
25452 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
25453 limits future samples by applying asymmetric windowing in time domain, useful
25454 when low latency is required. Accepted range is @code{[0, 1]}.
25457 Specify the transform base frequency. Default value is @code{20.01523126408007475},
25458 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
25461 Specify the transform end frequency. Default value is @code{20495.59681441799654},
25462 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
25465 This option is deprecated and ignored.
25468 Specify the transform length in time domain. Use this option to control accuracy
25469 trade-off between time domain and frequency domain at every frequency sample.
25470 It can contain variables:
25472 @item frequency, freq, f
25473 the frequency where it is evaluated
25474 @item timeclamp, tc
25475 the value of @var{timeclamp} option.
25477 Default value is @code{384*tc/(384+tc*f)}.
25480 Specify the transform count for every video frame. Default value is @code{6}.
25481 Acceptable range is @code{[1, 30]}.
25484 Specify the transform count for every single pixel. Default value is @code{0},
25485 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
25488 Specify font file for use with freetype to draw the axis. If not specified,
25489 use embedded font. Note that drawing with font file or embedded font is not
25490 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25494 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25495 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25499 Specify font color expression. This is arithmetic expression that should return
25500 integer value 0xRRGGBB. It can contain variables:
25502 @item frequency, freq, f
25503 the frequency where it is evaluated
25504 @item timeclamp, tc
25505 the value of @var{timeclamp} option
25510 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25511 @item r(x), g(x), b(x)
25512 red, green, and blue value of intensity x.
25514 Default value is @code{st(0, (midi(f)-59.5)/12);
25515 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25516 r(1-ld(1)) + b(ld(1))}.
25519 Specify image file to draw the axis. This option override @var{fontfile} and
25520 @var{fontcolor} option.
25523 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25524 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25525 Default value is @code{1}.
25528 Set colorspace. The accepted values are:
25531 Unspecified (default)
25540 BT.470BG or BT.601-6 625
25543 SMPTE-170M or BT.601-6 525
25549 BT.2020 with non-constant luminance
25554 Set spectrogram color scheme. This is list of floating point values with format
25555 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25556 The default is @code{1|0.5|0|0|0.5|1}.
25560 @subsection Examples
25564 Playing audio while showing the spectrum:
25566 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25570 Same as above, but with frame rate 30 fps:
25572 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25576 Playing at 1280x720:
25578 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25582 Disable sonogram display:
25588 A1 and its harmonics: A1, A2, (near)E3, A3:
25590 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),
25591 asplit[a][out1]; [a] showcqt [out0]'
25595 Same as above, but with more accuracy in frequency domain:
25597 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),
25598 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25604 bar_v=10:sono_v=bar_v*a_weighting(f)
25608 Custom gamma, now spectrum is linear to the amplitude.
25614 Custom tlength equation:
25616 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)))'
25620 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25622 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25626 Custom font using fontconfig:
25628 font='Courier New,Monospace,mono|bold'
25632 Custom frequency range with custom axis using image file:
25634 axisfile=myaxis.png:basefreq=40:endfreq=10000
25640 Convert input audio to video output representing the audio power spectrum.
25641 Audio amplitude is on Y-axis while frequency is on X-axis.
25643 The filter accepts the following options:
25647 Specify size of video. For the syntax of this option, check the
25648 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25649 Default is @code{1024x512}.
25653 This set how each frequency bin will be represented.
25655 It accepts the following values:
25661 Default is @code{bar}.
25664 Set amplitude scale.
25666 It accepts the following values:
25680 Default is @code{log}.
25683 Set frequency scale.
25685 It accepts the following values:
25694 Reverse logarithmic scale.
25696 Default is @code{lin}.
25699 Set window size. Allowed range is from 16 to 65536.
25701 Default is @code{2048}
25704 Set windowing function.
25706 It accepts the following values:
25729 Default is @code{hanning}.
25732 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25733 which means optimal overlap for selected window function will be picked.
25736 Set time averaging. Setting this to 0 will display current maximal peaks.
25737 Default is @code{1}, which means time averaging is disabled.
25740 Specify list of colors separated by space or by '|' which will be used to
25741 draw channel frequencies. Unrecognized or missing colors will be replaced
25745 Set channel display mode.
25747 It accepts the following values:
25752 Default is @code{combined}.
25755 Set minimum amplitude used in @code{log} amplitude scaler.
25758 Set data display mode.
25760 It accepts the following values:
25766 Default is @code{magnitude}.
25769 @section showspatial
25771 Convert stereo input audio to a video output, representing the spatial relationship
25772 between two channels.
25774 The filter accepts the following options:
25778 Specify the video size for the output. For the syntax of this option, check the
25779 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25780 Default value is @code{512x512}.
25783 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25786 Set window function.
25788 It accepts the following values:
25813 Default value is @code{hann}.
25816 Set ratio of overlap window. Default value is @code{0.5}.
25817 When value is @code{1} overlap is set to recommended size for specific
25818 window function currently used.
25821 @anchor{showspectrum}
25822 @section showspectrum
25824 Convert input audio to a video output, representing the audio frequency
25827 The filter accepts the following options:
25831 Specify the video size for the output. For the syntax of this option, check the
25832 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25833 Default value is @code{640x512}.
25836 Specify how the spectrum should slide along the window.
25838 It accepts the following values:
25841 the samples start again on the left when they reach the right
25843 the samples scroll from right to left
25845 frames are only produced when the samples reach the right
25847 the samples scroll from left to right
25850 Default value is @code{replace}.
25853 Specify display mode.
25855 It accepts the following values:
25858 all channels are displayed in the same row
25860 all channels are displayed in separate rows
25863 Default value is @samp{combined}.
25866 Specify display color mode.
25868 It accepts the following values:
25871 each channel is displayed in a separate color
25873 each channel is displayed using the same color scheme
25875 each channel is displayed using the rainbow color scheme
25877 each channel is displayed using the moreland color scheme
25879 each channel is displayed using the nebulae color scheme
25881 each channel is displayed using the fire color scheme
25883 each channel is displayed using the fiery color scheme
25885 each channel is displayed using the fruit color scheme
25887 each channel is displayed using the cool color scheme
25889 each channel is displayed using the magma color scheme
25891 each channel is displayed using the green color scheme
25893 each channel is displayed using the viridis color scheme
25895 each channel is displayed using the plasma color scheme
25897 each channel is displayed using the cividis color scheme
25899 each channel is displayed using the terrain color scheme
25902 Default value is @samp{channel}.
25905 Specify scale used for calculating intensity color values.
25907 It accepts the following values:
25912 square root, default
25923 Default value is @samp{sqrt}.
25926 Specify frequency scale.
25928 It accepts the following values:
25936 Default value is @samp{lin}.
25939 Set saturation modifier for displayed colors. Negative values provide
25940 alternative color scheme. @code{0} is no saturation at all.
25941 Saturation must be in [-10.0, 10.0] range.
25942 Default value is @code{1}.
25945 Set window function.
25947 It accepts the following values:
25972 Default value is @code{hann}.
25975 Set orientation of time vs frequency axis. Can be @code{vertical} or
25976 @code{horizontal}. Default is @code{vertical}.
25979 Set ratio of overlap window. Default value is @code{0}.
25980 When value is @code{1} overlap is set to recommended size for specific
25981 window function currently used.
25984 Set scale gain for calculating intensity color values.
25985 Default value is @code{1}.
25988 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25991 Set color rotation, must be in [-1.0, 1.0] range.
25992 Default value is @code{0}.
25995 Set start frequency from which to display spectrogram. Default is @code{0}.
25998 Set stop frequency to which to display spectrogram. Default is @code{0}.
26001 Set upper frame rate limit. Default is @code{auto}, unlimited.
26004 Draw time and frequency axes and legends. Default is disabled.
26007 The usage is very similar to the showwaves filter; see the examples in that
26010 @subsection Examples
26014 Large window with logarithmic color scaling:
26016 showspectrum=s=1280x480:scale=log
26020 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
26022 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
26023 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
26027 @section showspectrumpic
26029 Convert input audio to a single video frame, representing the audio frequency
26032 The filter accepts the following options:
26036 Specify the video size for the output. For the syntax of this option, check the
26037 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26038 Default value is @code{4096x2048}.
26041 Specify display mode.
26043 It accepts the following values:
26046 all channels are displayed in the same row
26048 all channels are displayed in separate rows
26050 Default value is @samp{combined}.
26053 Specify display color mode.
26055 It accepts the following values:
26058 each channel is displayed in a separate color
26060 each channel is displayed using the same color scheme
26062 each channel is displayed using the rainbow color scheme
26064 each channel is displayed using the moreland color scheme
26066 each channel is displayed using the nebulae color scheme
26068 each channel is displayed using the fire color scheme
26070 each channel is displayed using the fiery color scheme
26072 each channel is displayed using the fruit color scheme
26074 each channel is displayed using the cool color scheme
26076 each channel is displayed using the magma color scheme
26078 each channel is displayed using the green color scheme
26080 each channel is displayed using the viridis color scheme
26082 each channel is displayed using the plasma color scheme
26084 each channel is displayed using the cividis color scheme
26086 each channel is displayed using the terrain color scheme
26088 Default value is @samp{intensity}.
26091 Specify scale used for calculating intensity color values.
26093 It accepts the following values:
26098 square root, default
26108 Default value is @samp{log}.
26111 Specify frequency scale.
26113 It accepts the following values:
26121 Default value is @samp{lin}.
26124 Set saturation modifier for displayed colors. Negative values provide
26125 alternative color scheme. @code{0} is no saturation at all.
26126 Saturation must be in [-10.0, 10.0] range.
26127 Default value is @code{1}.
26130 Set window function.
26132 It accepts the following values:
26156 Default value is @code{hann}.
26159 Set orientation of time vs frequency axis. Can be @code{vertical} or
26160 @code{horizontal}. Default is @code{vertical}.
26163 Set scale gain for calculating intensity color values.
26164 Default value is @code{1}.
26167 Draw time and frequency axes and legends. Default is enabled.
26170 Set color rotation, must be in [-1.0, 1.0] range.
26171 Default value is @code{0}.
26174 Set start frequency from which to display spectrogram. Default is @code{0}.
26177 Set stop frequency to which to display spectrogram. Default is @code{0}.
26180 @subsection Examples
26184 Extract an audio spectrogram of a whole audio track
26185 in a 1024x1024 picture using @command{ffmpeg}:
26187 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
26191 @section showvolume
26193 Convert input audio volume to a video output.
26195 The filter accepts the following options:
26202 Set border width, allowed range is [0, 5]. Default is 1.
26205 Set channel width, allowed range is [80, 8192]. Default is 400.
26208 Set channel height, allowed range is [1, 900]. Default is 20.
26211 Set fade, allowed range is [0, 1]. Default is 0.95.
26214 Set volume color expression.
26216 The expression can use the following variables:
26220 Current max volume of channel in dB.
26226 Current channel number, starting from 0.
26230 If set, displays channel names. Default is enabled.
26233 If set, displays volume values. Default is enabled.
26236 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
26237 default is @code{h}.
26240 Set step size, allowed range is [0, 5]. Default is 0, which means
26244 Set background opacity, allowed range is [0, 1]. Default is 0.
26247 Set metering mode, can be peak: @code{p} or rms: @code{r},
26248 default is @code{p}.
26251 Set display scale, can be linear: @code{lin} or log: @code{log},
26252 default is @code{lin}.
26256 If set to > 0., display a line for the max level
26257 in the previous seconds.
26258 default is disabled: @code{0.}
26261 The color of the max line. Use when @code{dm} option is set to > 0.
26262 default is: @code{orange}
26267 Convert input audio to a video output, representing the samples waves.
26269 The filter accepts the following options:
26273 Specify the video size for the output. For the syntax of this option, check the
26274 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26275 Default value is @code{600x240}.
26280 Available values are:
26283 Draw a point for each sample.
26286 Draw a vertical line for each sample.
26289 Draw a point for each sample and a line between them.
26292 Draw a centered vertical line for each sample.
26295 Default value is @code{point}.
26298 Set the number of samples which are printed on the same column. A
26299 larger value will decrease the frame rate. Must be a positive
26300 integer. This option can be set only if the value for @var{rate}
26301 is not explicitly specified.
26304 Set the (approximate) output frame rate. This is done by setting the
26305 option @var{n}. Default value is "25".
26307 @item split_channels
26308 Set if channels should be drawn separately or overlap. Default value is 0.
26311 Set colors separated by '|' which are going to be used for drawing of each channel.
26314 Set amplitude scale.
26316 Available values are:
26334 Set the draw mode. This is mostly useful to set for high @var{n}.
26336 Available values are:
26339 Scale pixel values for each drawn sample.
26342 Draw every sample directly.
26345 Default value is @code{scale}.
26348 @subsection Examples
26352 Output the input file audio and the corresponding video representation
26355 amovie=a.mp3,asplit[out0],showwaves[out1]
26359 Create a synthetic signal and show it with showwaves, forcing a
26360 frame rate of 30 frames per second:
26362 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26366 @section showwavespic
26368 Convert input audio to a single video frame, representing the samples waves.
26370 The filter accepts the following options:
26374 Specify the video size for the output. For the syntax of this option, check the
26375 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26376 Default value is @code{600x240}.
26378 @item split_channels
26379 Set if channels should be drawn separately or overlap. Default value is 0.
26382 Set colors separated by '|' which are going to be used for drawing of each channel.
26385 Set amplitude scale.
26387 Available values are:
26407 Available values are:
26410 Scale pixel values for each drawn sample.
26413 Draw every sample directly.
26416 Default value is @code{scale}.
26419 Set the filter mode.
26421 Available values are:
26424 Use average samples values for each drawn sample.
26427 Use peak samples values for each drawn sample.
26430 Default value is @code{average}.
26433 @subsection Examples
26437 Extract a channel split representation of the wave form of a whole audio track
26438 in a 1024x800 picture using @command{ffmpeg}:
26440 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26444 @section sidedata, asidedata
26446 Delete frame side data, or select frames based on it.
26448 This filter accepts the following options:
26452 Set mode of operation of the filter.
26454 Can be one of the following:
26458 Select every frame with side data of @code{type}.
26461 Delete side data of @code{type}. If @code{type} is not set, delete all side
26467 Set side data type used with all modes. Must be set for @code{select} mode. For
26468 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
26469 in @file{libavutil/frame.h}. For example, to choose
26470 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
26474 @section spectrumsynth
26476 Synthesize audio from 2 input video spectrums, first input stream represents
26477 magnitude across time and second represents phase across time.
26478 The filter will transform from frequency domain as displayed in videos back
26479 to time domain as presented in audio output.
26481 This filter is primarily created for reversing processed @ref{showspectrum}
26482 filter outputs, but can synthesize sound from other spectrograms too.
26483 But in such case results are going to be poor if the phase data is not
26484 available, because in such cases phase data need to be recreated, usually
26485 it's just recreated from random noise.
26486 For best results use gray only output (@code{channel} color mode in
26487 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
26488 @code{lin} scale for phase video. To produce phase, for 2nd video, use
26489 @code{data} option. Inputs videos should generally use @code{fullframe}
26490 slide mode as that saves resources needed for decoding video.
26492 The filter accepts the following options:
26496 Specify sample rate of output audio, the sample rate of audio from which
26497 spectrum was generated may differ.
26500 Set number of channels represented in input video spectrums.
26503 Set scale which was used when generating magnitude input spectrum.
26504 Can be @code{lin} or @code{log}. Default is @code{log}.
26507 Set slide which was used when generating inputs spectrums.
26508 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26509 Default is @code{fullframe}.
26512 Set window function used for resynthesis.
26515 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26516 which means optimal overlap for selected window function will be picked.
26519 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26520 Default is @code{vertical}.
26523 @subsection Examples
26527 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26528 then resynthesize videos back to audio with spectrumsynth:
26530 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
26531 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
26532 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26536 @section split, asplit
26538 Split input into several identical outputs.
26540 @code{asplit} works with audio input, @code{split} with video.
26542 The filter accepts a single parameter which specifies the number of outputs. If
26543 unspecified, it defaults to 2.
26545 @subsection Examples
26549 Create two separate outputs from the same input:
26551 [in] split [out0][out1]
26555 To create 3 or more outputs, you need to specify the number of
26558 [in] asplit=3 [out0][out1][out2]
26562 Create two separate outputs from the same input, one cropped and
26565 [in] split [splitout1][splitout2];
26566 [splitout1] crop=100:100:0:0 [cropout];
26567 [splitout2] pad=200:200:100:100 [padout];
26571 Create 5 copies of the input audio with @command{ffmpeg}:
26573 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26579 Receive commands sent through a libzmq client, and forward them to
26580 filters in the filtergraph.
26582 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26583 must be inserted between two video filters, @code{azmq} between two
26584 audio filters. Both are capable to send messages to any filter type.
26586 To enable these filters you need to install the libzmq library and
26587 headers and configure FFmpeg with @code{--enable-libzmq}.
26589 For more information about libzmq see:
26590 @url{http://www.zeromq.org/}
26592 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26593 receives messages sent through a network interface defined by the
26594 @option{bind_address} (or the abbreviation "@option{b}") option.
26595 Default value of this option is @file{tcp://localhost:5555}. You may
26596 want to alter this value to your needs, but do not forget to escape any
26597 ':' signs (see @ref{filtergraph escaping}).
26599 The received message must be in the form:
26601 @var{TARGET} @var{COMMAND} [@var{ARG}]
26604 @var{TARGET} specifies the target of the command, usually the name of
26605 the filter class or a specific filter instance name. The default
26606 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26607 but you can override this by using the @samp{filter_name@@id} syntax
26608 (see @ref{Filtergraph syntax}).
26610 @var{COMMAND} specifies the name of the command for the target filter.
26612 @var{ARG} is optional and specifies the optional argument list for the
26613 given @var{COMMAND}.
26615 Upon reception, the message is processed and the corresponding command
26616 is injected into the filtergraph. Depending on the result, the filter
26617 will send a reply to the client, adopting the format:
26619 @var{ERROR_CODE} @var{ERROR_REASON}
26623 @var{MESSAGE} is optional.
26625 @subsection Examples
26627 Look at @file{tools/zmqsend} for an example of a zmq client which can
26628 be used to send commands processed by these filters.
26630 Consider the following filtergraph generated by @command{ffplay}.
26631 In this example the last overlay filter has an instance name. All other
26632 filters will have default instance names.
26635 ffplay -dumpgraph 1 -f lavfi "
26636 color=s=100x100:c=red [l];
26637 color=s=100x100:c=blue [r];
26638 nullsrc=s=200x100, zmq [bg];
26639 [bg][l] overlay [bg+l];
26640 [bg+l][r] overlay@@my=x=100 "
26643 To change the color of the left side of the video, the following
26644 command can be used:
26646 echo Parsed_color_0 c yellow | tools/zmqsend
26649 To change the right side:
26651 echo Parsed_color_1 c pink | tools/zmqsend
26654 To change the position of the right side:
26656 echo overlay@@my x 150 | tools/zmqsend
26660 @c man end MULTIMEDIA FILTERS
26662 @chapter Multimedia Sources
26663 @c man begin MULTIMEDIA SOURCES
26665 Below is a description of the currently available multimedia sources.
26669 This is the same as @ref{movie} source, except it selects an audio
26675 Read audio and/or video stream(s) from a movie container.
26677 It accepts the following parameters:
26681 The name of the resource to read (not necessarily a file; it can also be a
26682 device or a stream accessed through some protocol).
26684 @item format_name, f
26685 Specifies the format assumed for the movie to read, and can be either
26686 the name of a container or an input device. If not specified, the
26687 format is guessed from @var{movie_name} or by probing.
26689 @item seek_point, sp
26690 Specifies the seek point in seconds. The frames will be output
26691 starting from this seek point. The parameter is evaluated with
26692 @code{av_strtod}, so the numerical value may be suffixed by an IS
26693 postfix. The default value is "0".
26696 Specifies the streams to read. Several streams can be specified,
26697 separated by "+". The source will then have as many outputs, in the
26698 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26699 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26700 respectively the default (best suited) video and audio stream. Default
26701 is "dv", or "da" if the filter is called as "amovie".
26703 @item stream_index, si
26704 Specifies the index of the video stream to read. If the value is -1,
26705 the most suitable video stream will be automatically selected. The default
26706 value is "-1". Deprecated. If the filter is called "amovie", it will select
26707 audio instead of video.
26710 Specifies how many times to read the stream in sequence.
26711 If the value is 0, the stream will be looped infinitely.
26712 Default value is "1".
26714 Note that when the movie is looped the source timestamps are not
26715 changed, so it will generate non monotonically increasing timestamps.
26717 @item discontinuity
26718 Specifies the time difference between frames above which the point is
26719 considered a timestamp discontinuity which is removed by adjusting the later
26723 It allows overlaying a second video on top of the main input of
26724 a filtergraph, as shown in this graph:
26726 input -----------> deltapts0 --> overlay --> output
26729 movie --> scale--> deltapts1 -------+
26731 @subsection Examples
26735 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26736 on top of the input labelled "in":
26738 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26739 [in] setpts=PTS-STARTPTS [main];
26740 [main][over] overlay=16:16 [out]
26744 Read from a video4linux2 device, and overlay it on top of the input
26747 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26748 [in] setpts=PTS-STARTPTS [main];
26749 [main][over] overlay=16:16 [out]
26753 Read the first video stream and the audio stream with id 0x81 from
26754 dvd.vob; the video is connected to the pad named "video" and the audio is
26755 connected to the pad named "audio":
26757 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26761 @subsection Commands
26763 Both movie and amovie support the following commands:
26766 Perform seek using "av_seek_frame".
26767 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26770 @var{stream_index}: If stream_index is -1, a default
26771 stream is selected, and @var{timestamp} is automatically converted
26772 from AV_TIME_BASE units to the stream specific time_base.
26774 @var{timestamp}: Timestamp in AVStream.time_base units
26775 or, if no stream is specified, in AV_TIME_BASE units.
26777 @var{flags}: Flags which select direction and seeking mode.
26781 Get movie duration in AV_TIME_BASE units.
26785 @c man end MULTIMEDIA SOURCES