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 to expand audio dynamic range.
3687 The filter accepts the following options:
3691 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3692 (unchanged sound) to 10.0 (maximum effect).
3695 Enable clipping. By default is enabled.
3698 @subsection Commands
3700 This filter supports the all above options as @ref{commands}.
3703 Apply a DC shift to the audio.
3705 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3706 in the recording chain) from the audio. The effect of a DC offset is reduced
3707 headroom and hence volume. The @ref{astats} filter can be used to determine if
3708 a signal has a DC offset.
3712 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3716 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3717 used to prevent clipping.
3722 Apply de-essing to the audio samples.
3726 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3730 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3734 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3738 Set the output mode.
3740 It accepts the following values:
3743 Pass input unchanged.
3746 Pass ess filtered out.
3751 Default value is @var{o}.
3757 Measure audio dynamic range.
3759 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3760 is found in transition material. And anything less that 8 have very poor dynamics
3761 and is very compressed.
3763 The filter accepts the following options:
3767 Set window length in seconds used to split audio into segments of equal length.
3768 Default is 3 seconds.
3772 Dynamic Audio Normalizer.
3774 This filter applies a certain amount of gain to the input audio in order
3775 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3776 contrast to more "simple" normalization algorithms, the Dynamic Audio
3777 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3778 This allows for applying extra gain to the "quiet" sections of the audio
3779 while avoiding distortions or clipping the "loud" sections. In other words:
3780 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3781 sections, in the sense that the volume of each section is brought to the
3782 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3783 this goal *without* applying "dynamic range compressing". It will retain 100%
3784 of the dynamic range *within* each section of the audio file.
3788 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3789 Default is 500 milliseconds.
3790 The Dynamic Audio Normalizer processes the input audio in small chunks,
3791 referred to as frames. This is required, because a peak magnitude has no
3792 meaning for just a single sample value. Instead, we need to determine the
3793 peak magnitude for a contiguous sequence of sample values. While a "standard"
3794 normalizer would simply use the peak magnitude of the complete file, the
3795 Dynamic Audio Normalizer determines the peak magnitude individually for each
3796 frame. The length of a frame is specified in milliseconds. By default, the
3797 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3798 been found to give good results with most files.
3799 Note that the exact frame length, in number of samples, will be determined
3800 automatically, based on the sampling rate of the individual input audio file.
3803 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3804 number. Default is 31.
3805 Probably the most important parameter of the Dynamic Audio Normalizer is the
3806 @code{window size} of the Gaussian smoothing filter. The filter's window size
3807 is specified in frames, centered around the current frame. For the sake of
3808 simplicity, this must be an odd number. Consequently, the default value of 31
3809 takes into account the current frame, as well as the 15 preceding frames and
3810 the 15 subsequent frames. Using a larger window results in a stronger
3811 smoothing effect and thus in less gain variation, i.e. slower gain
3812 adaptation. Conversely, using a smaller window results in a weaker smoothing
3813 effect and thus in more gain variation, i.e. faster gain adaptation.
3814 In other words, the more you increase this value, the more the Dynamic Audio
3815 Normalizer will behave like a "traditional" normalization filter. On the
3816 contrary, the more you decrease this value, the more the Dynamic Audio
3817 Normalizer will behave like a dynamic range compressor.
3820 Set the target peak value. This specifies the highest permissible magnitude
3821 level for the normalized audio input. This filter will try to approach the
3822 target peak magnitude as closely as possible, but at the same time it also
3823 makes sure that the normalized signal will never exceed the peak magnitude.
3824 A frame's maximum local gain factor is imposed directly by the target peak
3825 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3826 It is not recommended to go above this value.
3829 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3830 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3831 factor for each input frame, i.e. the maximum gain factor that does not
3832 result in clipping or distortion. The maximum gain factor is determined by
3833 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3834 additionally bounds the frame's maximum gain factor by a predetermined
3835 (global) maximum gain factor. This is done in order to avoid excessive gain
3836 factors in "silent" or almost silent frames. By default, the maximum gain
3837 factor is 10.0, For most inputs the default value should be sufficient and
3838 it usually is not recommended to increase this value. Though, for input
3839 with an extremely low overall volume level, it may be necessary to allow even
3840 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3841 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3842 Instead, a "sigmoid" threshold function will be applied. This way, the
3843 gain factors will smoothly approach the threshold value, but never exceed that
3847 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3848 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3849 This means that the maximum local gain factor for each frame is defined
3850 (only) by the frame's highest magnitude sample. This way, the samples can
3851 be amplified as much as possible without exceeding the maximum signal
3852 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3853 Normalizer can also take into account the frame's root mean square,
3854 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3855 determine the power of a time-varying signal. It is therefore considered
3856 that the RMS is a better approximation of the "perceived loudness" than
3857 just looking at the signal's peak magnitude. Consequently, by adjusting all
3858 frames to a constant RMS value, a uniform "perceived loudness" can be
3859 established. If a target RMS value has been specified, a frame's local gain
3860 factor is defined as the factor that would result in exactly that RMS value.
3861 Note, however, that the maximum local gain factor is still restricted by the
3862 frame's highest magnitude sample, in order to prevent clipping.
3865 Enable channels coupling. By default is enabled.
3866 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3867 amount. This means the same gain factor will be applied to all channels, i.e.
3868 the maximum possible gain factor is determined by the "loudest" channel.
3869 However, in some recordings, it may happen that the volume of the different
3870 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3871 In this case, this option can be used to disable the channel coupling. This way,
3872 the gain factor will be determined independently for each channel, depending
3873 only on the individual channel's highest magnitude sample. This allows for
3874 harmonizing the volume of the different channels.
3877 Enable DC bias correction. By default is disabled.
3878 An audio signal (in the time domain) is a sequence of sample values.
3879 In the Dynamic Audio Normalizer these sample values are represented in the
3880 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3881 audio signal, or "waveform", should be centered around the zero point.
3882 That means if we calculate the mean value of all samples in a file, or in a
3883 single frame, then the result should be 0.0 or at least very close to that
3884 value. If, however, there is a significant deviation of the mean value from
3885 0.0, in either positive or negative direction, this is referred to as a
3886 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3887 Audio Normalizer provides optional DC bias correction.
3888 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3889 the mean value, or "DC correction" offset, of each input frame and subtract
3890 that value from all of the frame's sample values which ensures those samples
3891 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3892 boundaries, the DC correction offset values will be interpolated smoothly
3893 between neighbouring frames.
3895 @item altboundary, b
3896 Enable alternative boundary mode. By default is disabled.
3897 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3898 around each frame. This includes the preceding frames as well as the
3899 subsequent frames. However, for the "boundary" frames, located at the very
3900 beginning and at the very end of the audio file, not all neighbouring
3901 frames are available. In particular, for the first few frames in the audio
3902 file, the preceding frames are not known. And, similarly, for the last few
3903 frames in the audio file, the subsequent frames are not known. Thus, the
3904 question arises which gain factors should be assumed for the missing frames
3905 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3906 to deal with this situation. The default boundary mode assumes a gain factor
3907 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3908 "fade out" at the beginning and at the end of the input, respectively.
3911 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3912 By default, the Dynamic Audio Normalizer does not apply "traditional"
3913 compression. This means that signal peaks will not be pruned and thus the
3914 full dynamic range will be retained within each local neighbourhood. However,
3915 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3916 normalization algorithm with a more "traditional" compression.
3917 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3918 (thresholding) function. If (and only if) the compression feature is enabled,
3919 all input frames will be processed by a soft knee thresholding function prior
3920 to the actual normalization process. Put simply, the thresholding function is
3921 going to prune all samples whose magnitude exceeds a certain threshold value.
3922 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3923 value. Instead, the threshold value will be adjusted for each individual
3925 In general, smaller parameters result in stronger compression, and vice versa.
3926 Values below 3.0 are not recommended, because audible distortion may appear.
3929 Set the target threshold value. This specifies the lowest permissible
3930 magnitude level for the audio input which will be normalized.
3931 If input frame volume is above this value frame will be normalized.
3932 Otherwise frame may not be normalized at all. The default value is set
3933 to 0, which means all input frames will be normalized.
3934 This option is mostly useful if digital noise is not wanted to be amplified.
3937 @subsection Commands
3939 This filter supports the all above options as @ref{commands}.
3943 Make audio easier to listen to on headphones.
3945 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3946 so that when listened to on headphones the stereo image is moved from
3947 inside your head (standard for headphones) to outside and in front of
3948 the listener (standard for speakers).
3954 Apply a two-pole peaking equalisation (EQ) filter. With this
3955 filter, the signal-level at and around a selected frequency can
3956 be increased or decreased, whilst (unlike bandpass and bandreject
3957 filters) that at all other frequencies is unchanged.
3959 In order to produce complex equalisation curves, this filter can
3960 be given several times, each with a different central frequency.
3962 The filter accepts the following options:
3966 Set the filter's central frequency in Hz.
3969 Set method to specify band-width of filter.
3984 Specify the band-width of a filter in width_type units.
3987 Set the required gain or attenuation in dB.
3988 Beware of clipping when using a positive gain.
3991 How much to use filtered signal in output. Default is 1.
3992 Range is between 0 and 1.
3995 Specify which channels to filter, by default all available are filtered.
3998 Normalize biquad coefficients, by default is disabled.
3999 Enabling it will normalize magnitude response at DC to 0dB.
4002 Set transform type of IIR filter.
4011 Set precison of filtering.
4014 Pick automatic sample format depending on surround filters.
4016 Always use signed 16-bit.
4018 Always use signed 32-bit.
4020 Always use float 32-bit.
4022 Always use float 64-bit.
4026 @subsection Examples
4029 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
4031 equalizer=f=1000:t=h:width=200:g=-10
4035 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
4037 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
4041 @subsection Commands
4043 This filter supports the following commands:
4046 Change equalizer frequency.
4047 Syntax for the command is : "@var{frequency}"
4050 Change equalizer width_type.
4051 Syntax for the command is : "@var{width_type}"
4054 Change equalizer width.
4055 Syntax for the command is : "@var{width}"
4058 Change equalizer gain.
4059 Syntax for the command is : "@var{gain}"
4062 Change equalizer mix.
4063 Syntax for the command is : "@var{mix}"
4066 @section extrastereo
4068 Linearly increases the difference between left and right channels which
4069 adds some sort of "live" effect to playback.
4071 The filter accepts the following options:
4075 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
4076 (average of both channels), with 1.0 sound will be unchanged, with
4077 -1.0 left and right channels will be swapped.
4080 Enable clipping. By default is enabled.
4083 @subsection Commands
4085 This filter supports the all above options as @ref{commands}.
4087 @section firequalizer
4088 Apply FIR Equalization using arbitrary frequency response.
4090 The filter accepts the following option:
4094 Set gain curve equation (in dB). The expression can contain variables:
4097 the evaluated frequency
4101 channel number, set to 0 when multichannels evaluation is disabled
4103 channel id, see libavutil/channel_layout.h, set to the first channel id when
4104 multichannels evaluation is disabled
4108 channel_layout, see libavutil/channel_layout.h
4113 @item gain_interpolate(f)
4114 interpolate gain on frequency f based on gain_entry
4115 @item cubic_interpolate(f)
4116 same as gain_interpolate, but smoother
4118 This option is also available as command. Default is @code{gain_interpolate(f)}.
4121 Set gain entry for gain_interpolate function. The expression can
4125 store gain entry at frequency f with value g
4127 This option is also available as command.
4130 Set filter delay in seconds. Higher value means more accurate.
4131 Default is @code{0.01}.
4134 Set filter accuracy in Hz. Lower value means more accurate.
4135 Default is @code{5}.
4138 Set window function. Acceptable values are:
4141 rectangular window, useful when gain curve is already smooth
4143 hann window (default)
4149 3-terms continuous 1st derivative nuttall window
4151 minimum 3-terms discontinuous nuttall window
4153 4-terms continuous 1st derivative nuttall window
4155 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4157 blackman-harris window
4163 If enabled, use fixed number of audio samples. This improves speed when
4164 filtering with large delay. Default is disabled.
4167 Enable multichannels evaluation on gain. Default is disabled.
4170 Enable zero phase mode by subtracting timestamp to compensate delay.
4171 Default is disabled.
4174 Set scale used by gain. Acceptable values are:
4177 linear frequency, linear gain
4179 linear frequency, logarithmic (in dB) gain (default)
4181 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
4183 logarithmic frequency, logarithmic gain
4187 Set file for dumping, suitable for gnuplot.
4190 Set scale for dumpfile. Acceptable values are same with scale option.
4194 Enable 2-channel convolution using complex FFT. This improves speed significantly.
4195 Default is disabled.
4198 Enable minimum phase impulse response. Default is disabled.
4201 @subsection Examples
4206 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4209 lowpass at 1000 Hz with gain_entry:
4211 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4214 custom equalization:
4216 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4219 higher delay with zero phase to compensate delay:
4221 firequalizer=delay=0.1:fixed=on:zero_phase=on
4224 lowpass on left channel, highpass on right channel:
4226 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4227 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4232 Apply a flanging effect to the audio.
4234 The filter accepts the following options:
4238 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4241 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4244 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4248 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4249 Default value is 71.
4252 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4255 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4256 Default value is @var{sinusoidal}.
4259 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4260 Default value is 25.
4263 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4264 Default is @var{linear}.
4268 Apply Haas effect to audio.
4270 Note that this makes most sense to apply on mono signals.
4271 With this filter applied to mono signals it give some directionality and
4272 stretches its stereo image.
4274 The filter accepts the following options:
4278 Set input level. By default is @var{1}, or 0dB
4281 Set output level. By default is @var{1}, or 0dB.
4284 Set gain applied to side part of signal. By default is @var{1}.
4287 Set kind of middle source. Can be one of the following:
4297 Pick middle part signal of stereo image.
4300 Pick side part signal of stereo image.
4304 Change middle phase. By default is disabled.
4307 Set left channel delay. By default is @var{2.05} milliseconds.
4310 Set left channel balance. By default is @var{-1}.
4313 Set left channel gain. By default is @var{1}.
4316 Change left phase. By default is disabled.
4319 Set right channel delay. By defaults is @var{2.12} milliseconds.
4322 Set right channel balance. By default is @var{1}.
4325 Set right channel gain. By default is @var{1}.
4328 Change right phase. By default is enabled.
4333 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4334 embedded HDCD codes is expanded into a 20-bit PCM stream.
4336 The filter supports the Peak Extend and Low-level Gain Adjustment features
4337 of HDCD, and detects the Transient Filter flag.
4340 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4343 When using the filter with wav, note the default encoding for wav is 16-bit,
4344 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4345 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4347 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4348 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4351 The filter accepts the following options:
4354 @item disable_autoconvert
4355 Disable any automatic format conversion or resampling in the filter graph.
4357 @item process_stereo
4358 Process the stereo channels together. If target_gain does not match between
4359 channels, consider it invalid and use the last valid target_gain.
4362 Set the code detect timer period in ms.
4365 Always extend peaks above -3dBFS even if PE isn't signaled.
4368 Replace audio with a solid tone and adjust the amplitude to signal some
4369 specific aspect of the decoding process. The output file can be loaded in
4370 an audio editor alongside the original to aid analysis.
4372 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4379 Gain adjustment level at each sample
4381 Samples where peak extend occurs
4383 Samples where the code detect timer is active
4385 Samples where the target gain does not match between channels
4391 Apply head-related transfer functions (HRTFs) to create virtual
4392 loudspeakers around the user for binaural listening via headphones.
4393 The HRIRs are provided via additional streams, for each channel
4394 one stereo input stream is needed.
4396 The filter accepts the following options:
4400 Set mapping of input streams for convolution.
4401 The argument is a '|'-separated list of channel names in order as they
4402 are given as additional stream inputs for filter.
4403 This also specify number of input streams. Number of input streams
4404 must be not less than number of channels in first stream plus one.
4407 Set gain applied to audio. Value is in dB. Default is 0.
4410 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4411 processing audio in time domain which is slow.
4412 @var{freq} is processing audio in frequency domain which is fast.
4413 Default is @var{freq}.
4416 Set custom gain for LFE channels. Value is in dB. Default is 0.
4419 Set size of frame in number of samples which will be processed at once.
4420 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4423 Set format of hrir stream.
4424 Default value is @var{stereo}. Alternative value is @var{multich}.
4425 If value is set to @var{stereo}, number of additional streams should
4426 be greater or equal to number of input channels in first input stream.
4427 Also each additional stream should have stereo number of channels.
4428 If value is set to @var{multich}, number of additional streams should
4429 be exactly one. Also number of input channels of additional stream
4430 should be equal or greater than twice number of channels of first input
4434 @subsection Examples
4438 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4439 each amovie filter use stereo file with IR coefficients as input.
4440 The files give coefficients for each position of virtual loudspeaker:
4443 -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"
4448 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4449 but now in @var{multich} @var{hrir} format.
4451 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"
4458 Apply a high-pass filter with 3dB point frequency.
4459 The filter can be either single-pole, or double-pole (the default).
4460 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4462 The filter accepts the following options:
4466 Set frequency in Hz. Default is 3000.
4469 Set number of poles. Default is 2.
4472 Set method to specify band-width of filter.
4487 Specify the band-width of a filter in width_type units.
4488 Applies only to double-pole filter.
4489 The default is 0.707q and gives a Butterworth response.
4492 How much to use filtered signal in output. Default is 1.
4493 Range is between 0 and 1.
4496 Specify which channels to filter, by default all available are filtered.
4499 Normalize biquad coefficients, by default is disabled.
4500 Enabling it will normalize magnitude response at DC to 0dB.
4503 Set transform type of IIR filter.
4512 Set precison of filtering.
4515 Pick automatic sample format depending on surround filters.
4517 Always use signed 16-bit.
4519 Always use signed 32-bit.
4521 Always use float 32-bit.
4523 Always use float 64-bit.
4527 @subsection Commands
4529 This filter supports the following commands:
4532 Change highpass frequency.
4533 Syntax for the command is : "@var{frequency}"
4536 Change highpass width_type.
4537 Syntax for the command is : "@var{width_type}"
4540 Change highpass width.
4541 Syntax for the command is : "@var{width}"
4544 Change highpass mix.
4545 Syntax for the command is : "@var{mix}"
4550 Join multiple input streams into one multi-channel stream.
4552 It accepts the following parameters:
4556 The number of input streams. It defaults to 2.
4558 @item channel_layout
4559 The desired output channel layout. It defaults to stereo.
4562 Map channels from inputs to output. The argument is a '|'-separated list of
4563 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4564 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4565 can be either the name of the input channel (e.g. FL for front left) or its
4566 index in the specified input stream. @var{out_channel} is the name of the output
4570 The filter will attempt to guess the mappings when they are not specified
4571 explicitly. It does so by first trying to find an unused matching input channel
4572 and if that fails it picks the first unused input channel.
4574 Join 3 inputs (with properly set channel layouts):
4576 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4579 Build a 5.1 output from 6 single-channel streams:
4581 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4582 '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'
4588 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4590 To enable compilation of this filter you need to configure FFmpeg with
4591 @code{--enable-ladspa}.
4595 Specifies the name of LADSPA plugin library to load. If the environment
4596 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4597 each one of the directories specified by the colon separated list in
4598 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4599 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4600 @file{/usr/lib/ladspa/}.
4603 Specifies the plugin within the library. Some libraries contain only
4604 one plugin, but others contain many of them. If this is not set filter
4605 will list all available plugins within the specified library.
4608 Set the '|' separated list of controls which are zero or more floating point
4609 values that determine the behavior of the loaded plugin (for example delay,
4611 Controls need to be defined using the following syntax:
4612 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4613 @var{valuei} is the value set on the @var{i}-th control.
4614 Alternatively they can be also defined using the following syntax:
4615 @var{value0}|@var{value1}|@var{value2}|..., where
4616 @var{valuei} is the value set on the @var{i}-th control.
4617 If @option{controls} is set to @code{help}, all available controls and
4618 their valid ranges are printed.
4620 @item sample_rate, s
4621 Specify the sample rate, default to 44100. Only used if plugin have
4625 Set the number of samples per channel per each output frame, default
4626 is 1024. Only used if plugin have zero inputs.
4629 Set the minimum duration of the sourced audio. See
4630 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4631 for the accepted syntax.
4632 Note that the resulting duration may be greater than the specified duration,
4633 as the generated audio is always cut at the end of a complete frame.
4634 If not specified, or the expressed duration is negative, the audio is
4635 supposed to be generated forever.
4636 Only used if plugin have zero inputs.
4639 Enable latency compensation, by default is disabled.
4640 Only used if plugin have inputs.
4643 @subsection Examples
4647 List all available plugins within amp (LADSPA example plugin) library:
4653 List all available controls and their valid ranges for @code{vcf_notch}
4654 plugin from @code{VCF} library:
4656 ladspa=f=vcf:p=vcf_notch:c=help
4660 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4663 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4667 Add reverberation to the audio using TAP-plugins
4668 (Tom's Audio Processing plugins):
4670 ladspa=file=tap_reverb:tap_reverb
4674 Generate white noise, with 0.2 amplitude:
4676 ladspa=file=cmt:noise_source_white:c=c0=.2
4680 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4681 @code{C* Audio Plugin Suite} (CAPS) library:
4683 ladspa=file=caps:Click:c=c1=20'
4687 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4689 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4693 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4694 @code{SWH Plugins} collection:
4696 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4700 Attenuate low frequencies using Multiband EQ from Steve Harris
4701 @code{SWH Plugins} collection:
4703 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4707 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4710 ladspa=caps:Narrower
4714 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4716 ladspa=caps:White:.2
4720 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4722 ladspa=caps:Fractal:c=c1=1
4726 Dynamic volume normalization using @code{VLevel} plugin:
4728 ladspa=vlevel-ladspa:vlevel_mono
4732 @subsection Commands
4734 This filter supports the following commands:
4737 Modify the @var{N}-th control value.
4739 If the specified value is not valid, it is ignored and prior one is kept.
4744 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4745 Support for both single pass (livestreams, files) and double pass (files) modes.
4746 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4747 detect true peaks, the audio stream will be upsampled to 192 kHz.
4748 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4750 The filter accepts the following options:
4754 Set integrated loudness target.
4755 Range is -70.0 - -5.0. Default value is -24.0.
4758 Set loudness range target.
4759 Range is 1.0 - 20.0. Default value is 7.0.
4762 Set maximum true peak.
4763 Range is -9.0 - +0.0. Default value is -2.0.
4765 @item measured_I, measured_i
4766 Measured IL of input file.
4767 Range is -99.0 - +0.0.
4769 @item measured_LRA, measured_lra
4770 Measured LRA of input file.
4771 Range is 0.0 - 99.0.
4773 @item measured_TP, measured_tp
4774 Measured true peak of input file.
4775 Range is -99.0 - +99.0.
4777 @item measured_thresh
4778 Measured threshold of input file.
4779 Range is -99.0 - +0.0.
4782 Set offset gain. Gain is applied before the true-peak limiter.
4783 Range is -99.0 - +99.0. Default is +0.0.
4786 Normalize by linearly scaling the source audio.
4787 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4788 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4789 be lower than source LRA and the change in integrated loudness shouldn't
4790 result in a true peak which exceeds the target TP. If any of these
4791 conditions aren't met, normalization mode will revert to @var{dynamic}.
4792 Options are @code{true} or @code{false}. Default is @code{true}.
4795 Treat mono input files as "dual-mono". If a mono file is intended for playback
4796 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4797 If set to @code{true}, this option will compensate for this effect.
4798 Multi-channel input files are not affected by this option.
4799 Options are true or false. Default is false.
4802 Set print format for stats. Options are summary, json, or none.
4803 Default value is none.
4808 Apply a low-pass filter with 3dB point frequency.
4809 The filter can be either single-pole or double-pole (the default).
4810 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4812 The filter accepts the following options:
4816 Set frequency in Hz. Default is 500.
4819 Set number of poles. Default is 2.
4822 Set method to specify band-width of filter.
4837 Specify the band-width of a filter in width_type units.
4838 Applies only to double-pole filter.
4839 The default is 0.707q and gives a Butterworth response.
4842 How much to use filtered signal in output. Default is 1.
4843 Range is between 0 and 1.
4846 Specify which channels to filter, by default all available are filtered.
4849 Normalize biquad coefficients, by default is disabled.
4850 Enabling it will normalize magnitude response at DC to 0dB.
4853 Set transform type of IIR filter.
4862 Set precison of filtering.
4865 Pick automatic sample format depending on surround filters.
4867 Always use signed 16-bit.
4869 Always use signed 32-bit.
4871 Always use float 32-bit.
4873 Always use float 64-bit.
4877 @subsection Examples
4880 Lowpass only LFE channel, it LFE is not present it does nothing:
4886 @subsection Commands
4888 This filter supports the following commands:
4891 Change lowpass frequency.
4892 Syntax for the command is : "@var{frequency}"
4895 Change lowpass width_type.
4896 Syntax for the command is : "@var{width_type}"
4899 Change lowpass width.
4900 Syntax for the command is : "@var{width}"
4904 Syntax for the command is : "@var{mix}"
4909 Load a LV2 (LADSPA Version 2) plugin.
4911 To enable compilation of this filter you need to configure FFmpeg with
4912 @code{--enable-lv2}.
4916 Specifies the plugin URI. You may need to escape ':'.
4919 Set the '|' separated list of controls which are zero or more floating point
4920 values that determine the behavior of the loaded plugin (for example delay,
4922 If @option{controls} is set to @code{help}, all available controls and
4923 their valid ranges are printed.
4925 @item sample_rate, s
4926 Specify the sample rate, default to 44100. Only used if plugin have
4930 Set the number of samples per channel per each output frame, default
4931 is 1024. Only used if plugin have zero inputs.
4934 Set the minimum duration of the sourced audio. See
4935 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4936 for the accepted syntax.
4937 Note that the resulting duration may be greater than the specified duration,
4938 as the generated audio is always cut at the end of a complete frame.
4939 If not specified, or the expressed duration is negative, the audio is
4940 supposed to be generated forever.
4941 Only used if plugin have zero inputs.
4944 @subsection Examples
4948 Apply bass enhancer plugin from Calf:
4950 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4954 Apply vinyl plugin from Calf:
4956 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4960 Apply bit crusher plugin from ArtyFX:
4962 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4967 Multiband Compress or expand the audio's dynamic range.
4969 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4970 This is akin to the crossover of a loudspeaker, and results in flat frequency
4971 response when absent compander action.
4973 It accepts the following parameters:
4977 This option syntax is:
4978 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4979 For explanation of each item refer to compand filter documentation.
4985 Mix channels with specific gain levels. The filter accepts the output
4986 channel layout followed by a set of channels definitions.
4988 This filter is also designed to efficiently remap the channels of an audio
4991 The filter accepts parameters of the form:
4992 "@var{l}|@var{outdef}|@var{outdef}|..."
4996 output channel layout or number of channels
4999 output channel specification, of the form:
5000 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
5003 output channel to define, either a channel name (FL, FR, etc.) or a channel
5004 number (c0, c1, etc.)
5007 multiplicative coefficient for the channel, 1 leaving the volume unchanged
5010 input channel to use, see out_name for details; it is not possible to mix
5011 named and numbered input channels
5014 If the `=' in a channel specification is replaced by `<', then the gains for
5015 that specification will be renormalized so that the total is 1, thus
5016 avoiding clipping noise.
5018 @subsection Mixing examples
5020 For example, if you want to down-mix from stereo to mono, but with a bigger
5021 factor for the left channel:
5023 pan=1c|c0=0.9*c0+0.1*c1
5026 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
5027 7-channels surround:
5029 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
5032 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
5033 that should be preferred (see "-ac" option) unless you have very specific
5036 @subsection Remapping examples
5038 The channel remapping will be effective if, and only if:
5041 @item gain coefficients are zeroes or ones,
5042 @item only one input per channel output,
5045 If all these conditions are satisfied, the filter will notify the user ("Pure
5046 channel mapping detected"), and use an optimized and lossless method to do the
5049 For example, if you have a 5.1 source and want a stereo audio stream by
5050 dropping the extra channels:
5052 pan="stereo| c0=FL | c1=FR"
5055 Given the same source, you can also switch front left and front right channels
5056 and keep the input channel layout:
5058 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
5061 If the input is a stereo audio stream, you can mute the front left channel (and
5062 still keep the stereo channel layout) with:
5067 Still with a stereo audio stream input, you can copy the right channel in both
5068 front left and right:
5070 pan="stereo| c0=FR | c1=FR"
5075 ReplayGain scanner filter. This filter takes an audio stream as an input and
5076 outputs it unchanged.
5077 At end of filtering it displays @code{track_gain} and @code{track_peak}.
5081 Convert the audio sample format, sample rate and channel layout. It is
5082 not meant to be used directly.
5085 Apply time-stretching and pitch-shifting with librubberband.
5087 To enable compilation of this filter, you need to configure FFmpeg with
5088 @code{--enable-librubberband}.
5090 The filter accepts the following options:
5094 Set tempo scale factor.
5097 Set pitch scale factor.
5100 Set transients detector.
5101 Possible values are:
5110 Possible values are:
5119 Possible values are:
5126 Set processing window size.
5127 Possible values are:
5136 Possible values are:
5143 Enable formant preservation when shift pitching.
5144 Possible values are:
5152 Possible values are:
5161 Possible values are:
5168 @subsection Commands
5170 This filter supports the following commands:
5173 Change filter tempo scale factor.
5174 Syntax for the command is : "@var{tempo}"
5177 Change filter pitch scale factor.
5178 Syntax for the command is : "@var{pitch}"
5181 @section sidechaincompress
5183 This filter acts like normal compressor but has the ability to compress
5184 detected signal using second input signal.
5185 It needs two input streams and returns one output stream.
5186 First input stream will be processed depending on second stream signal.
5187 The filtered signal then can be filtered with other filters in later stages of
5188 processing. See @ref{pan} and @ref{amerge} filter.
5190 The filter accepts the following options:
5194 Set input gain. Default is 1. Range is between 0.015625 and 64.
5197 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
5198 Default is @code{downward}.
5201 If a signal of second stream raises above this level it will affect the gain
5202 reduction of first stream.
5203 By default is 0.125. Range is between 0.00097563 and 1.
5206 Set a ratio about which the signal is reduced. 1:2 means that if the level
5207 raised 4dB above the threshold, it will be only 2dB above after the reduction.
5208 Default is 2. Range is between 1 and 20.
5211 Amount of milliseconds the signal has to rise above the threshold before gain
5212 reduction starts. Default is 20. Range is between 0.01 and 2000.
5215 Amount of milliseconds the signal has to fall below the threshold before
5216 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
5219 Set the amount by how much signal will be amplified after processing.
5220 Default is 1. Range is from 1 to 64.
5223 Curve the sharp knee around the threshold to enter gain reduction more softly.
5224 Default is 2.82843. Range is between 1 and 8.
5227 Choose if the @code{average} level between all channels of side-chain stream
5228 or the louder(@code{maximum}) channel of side-chain stream affects the
5229 reduction. Default is @code{average}.
5232 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5233 of @code{rms}. Default is @code{rms} which is mainly smoother.
5236 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5239 How much to use compressed signal in output. Default is 1.
5240 Range is between 0 and 1.
5243 @subsection Commands
5245 This filter supports the all above options as @ref{commands}.
5247 @subsection Examples
5251 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5252 depending on the signal of 2nd input and later compressed signal to be
5253 merged with 2nd input:
5255 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5259 @section sidechaingate
5261 A sidechain gate acts like a normal (wideband) gate but has the ability to
5262 filter the detected signal before sending it to the gain reduction stage.
5263 Normally a gate uses the full range signal to detect a level above the
5265 For example: If you cut all lower frequencies from your sidechain signal
5266 the gate will decrease the volume of your track only if not enough highs
5267 appear. With this technique you are able to reduce the resonation of a
5268 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5270 It needs two input streams and returns one output stream.
5271 First input stream will be processed depending on second stream signal.
5273 The filter accepts the following options:
5277 Set input level before filtering.
5278 Default is 1. Allowed range is from 0.015625 to 64.
5281 Set the mode of operation. Can be @code{upward} or @code{downward}.
5282 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5283 will be amplified, expanding dynamic range in upward direction.
5284 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5287 Set the level of gain reduction when the signal is below the threshold.
5288 Default is 0.06125. Allowed range is from 0 to 1.
5289 Setting this to 0 disables reduction and then filter behaves like expander.
5292 If a signal rises above this level the gain reduction is released.
5293 Default is 0.125. Allowed range is from 0 to 1.
5296 Set a ratio about which the signal is reduced.
5297 Default is 2. Allowed range is from 1 to 9000.
5300 Amount of milliseconds the signal has to rise above the threshold before gain
5302 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5305 Amount of milliseconds the signal has to fall below the threshold before the
5306 reduction is increased again. Default is 250 milliseconds.
5307 Allowed range is from 0.01 to 9000.
5310 Set amount of amplification of signal after processing.
5311 Default is 1. Allowed range is from 1 to 64.
5314 Curve the sharp knee around the threshold to enter gain reduction more softly.
5315 Default is 2.828427125. Allowed range is from 1 to 8.
5318 Choose if exact signal should be taken for detection or an RMS like one.
5319 Default is rms. Can be peak or rms.
5322 Choose if the average level between all channels or the louder channel affects
5324 Default is average. Can be average or maximum.
5327 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5330 @subsection Commands
5332 This filter supports the all above options as @ref{commands}.
5334 @section silencedetect
5336 Detect silence in an audio stream.
5338 This filter logs a message when it detects that the input audio volume is less
5339 or equal to a noise tolerance value for a duration greater or equal to the
5340 minimum detected noise duration.
5342 The printed times and duration are expressed in seconds. The
5343 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5344 is set on the first frame whose timestamp equals or exceeds the detection
5345 duration and it contains the timestamp of the first frame of the silence.
5347 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5348 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5349 keys are set on the first frame after the silence. If @option{mono} is
5350 enabled, and each channel is evaluated separately, the @code{.X}
5351 suffixed keys are used, and @code{X} corresponds to the channel number.
5353 The filter accepts the following options:
5357 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5358 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5361 Set silence duration until notification (default is 2 seconds). See
5362 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5363 for the accepted syntax.
5366 Process each channel separately, instead of combined. By default is disabled.
5369 @subsection Examples
5373 Detect 5 seconds of silence with -50dB noise tolerance:
5375 silencedetect=n=-50dB:d=5
5379 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5380 tolerance in @file{silence.mp3}:
5382 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5386 @section silenceremove
5388 Remove silence from the beginning, middle or end of the audio.
5390 The filter accepts the following options:
5394 This value is used to indicate if audio should be trimmed at beginning of
5395 the audio. A value of zero indicates no silence should be trimmed from the
5396 beginning. When specifying a non-zero value, it trims audio up until it
5397 finds non-silence. Normally, when trimming silence from beginning of audio
5398 the @var{start_periods} will be @code{1} but it can be increased to higher
5399 values to trim all audio up to specific count of non-silence periods.
5400 Default value is @code{0}.
5402 @item start_duration
5403 Specify the amount of time that non-silence must be detected before it stops
5404 trimming audio. By increasing the duration, bursts of noises can be treated
5405 as silence and trimmed off. Default value is @code{0}.
5407 @item start_threshold
5408 This indicates what sample value should be treated as silence. For digital
5409 audio, a value of @code{0} may be fine but for audio recorded from analog,
5410 you may wish to increase the value to account for background noise.
5411 Can be specified in dB (in case "dB" is appended to the specified value)
5412 or amplitude ratio. Default value is @code{0}.
5415 Specify max duration of silence at beginning that will be kept after
5416 trimming. Default is 0, which is equal to trimming all samples detected
5420 Specify mode of detection of silence end in start of multi-channel audio.
5421 Can be @var{any} or @var{all}. Default is @var{any}.
5422 With @var{any}, any sample that is detected as non-silence will cause
5423 stopped trimming of silence.
5424 With @var{all}, only if all channels are detected as non-silence will cause
5425 stopped trimming of silence.
5428 Set the count for trimming silence from the end of audio.
5429 To remove silence from the middle of a file, specify a @var{stop_periods}
5430 that is negative. This value is then treated as a positive value and is
5431 used to indicate the effect should restart processing as specified by
5432 @var{start_periods}, making it suitable for removing periods of silence
5433 in the middle of the audio.
5434 Default value is @code{0}.
5437 Specify a duration of silence that must exist before audio is not copied any
5438 more. By specifying a higher duration, silence that is wanted can be left in
5440 Default value is @code{0}.
5442 @item stop_threshold
5443 This is the same as @option{start_threshold} but for trimming silence from
5445 Can be specified in dB (in case "dB" is appended to the specified value)
5446 or amplitude ratio. Default value is @code{0}.
5449 Specify max duration of silence at end that will be kept after
5450 trimming. Default is 0, which is equal to trimming all samples detected
5454 Specify mode of detection of silence start in end of multi-channel audio.
5455 Can be @var{any} or @var{all}. Default is @var{any}.
5456 With @var{any}, any sample that is detected as non-silence will cause
5457 stopped trimming of silence.
5458 With @var{all}, only if all channels are detected as non-silence will cause
5459 stopped trimming of silence.
5462 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5463 and works better with digital silence which is exactly 0.
5464 Default value is @code{rms}.
5467 Set duration in number of seconds used to calculate size of window in number
5468 of samples for detecting silence.
5469 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5472 @subsection Examples
5476 The following example shows how this filter can be used to start a recording
5477 that does not contain the delay at the start which usually occurs between
5478 pressing the record button and the start of the performance:
5480 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5484 Trim all silence encountered from beginning to end where there is more than 1
5485 second of silence in audio:
5487 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5491 Trim all digital silence samples, using peak detection, from beginning to end
5492 where there is more than 0 samples of digital silence in audio and digital
5493 silence is detected in all channels at same positions in stream:
5495 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5501 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5502 loudspeakers around the user for binaural listening via headphones (audio
5503 formats up to 9 channels supported).
5504 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5505 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5506 Austrian Academy of Sciences.
5508 To enable compilation of this filter you need to configure FFmpeg with
5509 @code{--enable-libmysofa}.
5511 The filter accepts the following options:
5515 Set the SOFA file used for rendering.
5518 Set gain applied to audio. Value is in dB. Default is 0.
5521 Set rotation of virtual loudspeakers in deg. Default is 0.
5524 Set elevation of virtual speakers in deg. Default is 0.
5527 Set distance in meters between loudspeakers and the listener with near-field
5528 HRTFs. Default is 1.
5531 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5532 processing audio in time domain which is slow.
5533 @var{freq} is processing audio in frequency domain which is fast.
5534 Default is @var{freq}.
5537 Set custom positions of virtual loudspeakers. Syntax for this option is:
5538 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5539 Each virtual loudspeaker is described with short channel name following with
5540 azimuth and elevation in degrees.
5541 Each virtual loudspeaker description is separated by '|'.
5542 For example to override front left and front right channel positions use:
5543 'speakers=FL 45 15|FR 345 15'.
5544 Descriptions with unrecognised channel names are ignored.
5547 Set custom gain for LFE channels. Value is in dB. Default is 0.
5550 Set custom frame size in number of samples. Default is 1024.
5551 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5552 is set to @var{freq}.
5555 Should all IRs be normalized upon importing SOFA file.
5556 By default is enabled.
5559 Should nearest IRs be interpolated with neighbor IRs if exact position
5560 does not match. By default is disabled.
5563 Minphase all IRs upon loading of SOFA file. By default is disabled.
5566 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5569 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5572 @subsection Examples
5576 Using ClubFritz6 sofa file:
5578 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5582 Using ClubFritz12 sofa file and bigger radius with small rotation:
5584 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5588 Similar as above but with custom speaker positions for front left, front right, back left and back right
5589 and also with custom gain:
5591 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5598 This filter expands or compresses each half-cycle of audio samples
5599 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5600 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5602 The filter accepts the following options:
5606 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5607 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5610 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5611 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5612 would be such that local peak value reaches target peak value but never to surpass it and that
5613 ratio between new and previous peak value does not surpass this option value.
5615 @item compression, c
5616 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5617 This option controls maximum local half-cycle of samples compression. This option is used
5618 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5619 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5620 that peak's half-cycle will be compressed by current compression factor.
5623 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5624 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5625 Any half-cycle samples with their local peak value below or same as this option value will be
5626 compressed by current compression factor, otherwise, if greater than threshold value they will be
5627 expanded with expansion factor so that it could reach peak target value but never surpass it.
5630 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5631 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5632 each new half-cycle until it reaches @option{expansion} value.
5633 Setting this options too high may lead to distortions.
5636 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5637 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5638 each new half-cycle until it reaches @option{compression} value.
5641 Specify which channels to filter, by default all available channels are filtered.
5644 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5645 option. When enabled any half-cycle of samples with their local peak value below or same as
5646 @option{threshold} option will be expanded otherwise it will be compressed.
5649 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5650 When disabled each filtered channel gain calculation is independent, otherwise when this option
5651 is enabled the minimum of all possible gains for each filtered channel is used.
5654 @subsection Commands
5656 This filter supports the all above options as @ref{commands}.
5658 @section stereotools
5660 This filter has some handy utilities to manage stereo signals, for converting
5661 M/S stereo recordings to L/R signal while having control over the parameters
5662 or spreading the stereo image of master track.
5664 The filter accepts the following options:
5668 Set input level before filtering for both channels. Defaults is 1.
5669 Allowed range is from 0.015625 to 64.
5672 Set output level after filtering for both channels. Defaults is 1.
5673 Allowed range is from 0.015625 to 64.
5676 Set input balance between both channels. Default is 0.
5677 Allowed range is from -1 to 1.
5680 Set output balance between both channels. Default is 0.
5681 Allowed range is from -1 to 1.
5684 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5685 clipping. Disabled by default.
5688 Mute the left channel. Disabled by default.
5691 Mute the right channel. Disabled by default.
5694 Change the phase of the left channel. Disabled by default.
5697 Change the phase of the right channel. Disabled by default.
5700 Set stereo mode. Available values are:
5704 Left/Right to Left/Right, this is default.
5707 Left/Right to Mid/Side.
5710 Mid/Side to Left/Right.
5713 Left/Right to Left/Left.
5716 Left/Right to Right/Right.
5719 Left/Right to Left + Right.
5722 Left/Right to Right/Left.
5725 Mid/Side to Left/Left.
5728 Mid/Side to Right/Right.
5731 Mid/Side to Right/Left.
5734 Left/Right to Left - Right.
5738 Set level of side signal. Default is 1.
5739 Allowed range is from 0.015625 to 64.
5742 Set balance of side signal. Default is 0.
5743 Allowed range is from -1 to 1.
5746 Set level of the middle signal. Default is 1.
5747 Allowed range is from 0.015625 to 64.
5750 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5753 Set stereo base between mono and inversed channels. Default is 0.
5754 Allowed range is from -1 to 1.
5757 Set delay in milliseconds how much to delay left from right channel and
5758 vice versa. Default is 0. Allowed range is from -20 to 20.
5761 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5764 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5766 @item bmode_in, bmode_out
5767 Set balance mode for balance_in/balance_out option.
5769 Can be one of the following:
5773 Classic balance mode. Attenuate one channel at time.
5774 Gain is raised up to 1.
5777 Similar as classic mode above but gain is raised up to 2.
5780 Equal power distribution, from -6dB to +6dB range.
5784 @subsection Commands
5786 This filter supports the all above options as @ref{commands}.
5788 @subsection Examples
5792 Apply karaoke like effect:
5794 stereotools=mlev=0.015625
5798 Convert M/S signal to L/R:
5800 "stereotools=mode=ms>lr"
5804 @section stereowiden
5806 This filter enhance the stereo effect by suppressing signal common to both
5807 channels and by delaying the signal of left into right and vice versa,
5808 thereby widening the stereo effect.
5810 The filter accepts the following options:
5814 Time in milliseconds of the delay of left signal into right and vice versa.
5815 Default is 20 milliseconds.
5818 Amount of gain in delayed signal into right and vice versa. Gives a delay
5819 effect of left signal in right output and vice versa which gives widening
5820 effect. Default is 0.3.
5823 Cross feed of left into right with inverted phase. This helps in suppressing
5824 the mono. If the value is 1 it will cancel all the signal common to both
5825 channels. Default is 0.3.
5828 Set level of input signal of original channel. Default is 0.8.
5831 @subsection Commands
5833 This filter supports the all above options except @code{delay} as @ref{commands}.
5835 @section superequalizer
5836 Apply 18 band equalizer.
5838 The filter accepts the following options:
5845 Set 131Hz band gain.
5847 Set 185Hz band gain.
5849 Set 262Hz band gain.
5851 Set 370Hz band gain.
5853 Set 523Hz band gain.
5855 Set 740Hz band gain.
5857 Set 1047Hz band gain.
5859 Set 1480Hz band gain.
5861 Set 2093Hz band gain.
5863 Set 2960Hz band gain.
5865 Set 4186Hz band gain.
5867 Set 5920Hz band gain.
5869 Set 8372Hz band gain.
5871 Set 11840Hz band gain.
5873 Set 16744Hz band gain.
5875 Set 20000Hz band gain.
5879 Apply audio surround upmix filter.
5881 This filter allows to produce multichannel output from audio stream.
5883 The filter accepts the following options:
5887 Set output channel layout. By default, this is @var{5.1}.
5889 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5890 for the required syntax.
5893 Set input channel layout. By default, this is @var{stereo}.
5895 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5896 for the required syntax.
5899 Set input volume level. By default, this is @var{1}.
5902 Set output volume level. By default, this is @var{1}.
5905 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5908 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5911 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5914 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5915 In @var{add} mode, LFE channel is created from input audio and added to output.
5916 In @var{sub} mode, LFE channel is created from input audio and added to output but
5917 also all non-LFE output channels are subtracted with output LFE channel.
5920 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5921 Default is @var{90}.
5924 Set front center input volume. By default, this is @var{1}.
5927 Set front center output volume. By default, this is @var{1}.
5930 Set front left input volume. By default, this is @var{1}.
5933 Set front left output volume. By default, this is @var{1}.
5936 Set front right input volume. By default, this is @var{1}.
5939 Set front right output volume. By default, this is @var{1}.
5942 Set side left input volume. By default, this is @var{1}.
5945 Set side left output volume. By default, this is @var{1}.
5948 Set side right input volume. By default, this is @var{1}.
5951 Set side right output volume. By default, this is @var{1}.
5954 Set back left input volume. By default, this is @var{1}.
5957 Set back left output volume. By default, this is @var{1}.
5960 Set back right input volume. By default, this is @var{1}.
5963 Set back right output volume. By default, this is @var{1}.
5966 Set back center input volume. By default, this is @var{1}.
5969 Set back center output volume. By default, this is @var{1}.
5972 Set LFE input volume. By default, this is @var{1}.
5975 Set LFE output volume. By default, this is @var{1}.
5978 Set spread usage of stereo image across X axis for all channels.
5981 Set spread usage of stereo image across Y axis for all channels.
5983 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5984 Set spread usage of stereo image across X axis for each channel.
5986 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5987 Set spread usage of stereo image across Y axis for each channel.
5990 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5993 Set window function.
5995 It accepts the following values:
6018 Default is @code{hann}.
6021 Set window overlap. If set to 1, the recommended overlap for selected
6022 window function will be picked. Default is @code{0.5}.
6025 @section treble, highshelf
6027 Boost or cut treble (upper) frequencies of the audio using a two-pole
6028 shelving filter with a response similar to that of a standard
6029 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
6031 The filter accepts the following options:
6035 Give the gain at whichever is the lower of ~22 kHz and the
6036 Nyquist frequency. Its useful range is about -20 (for a large cut)
6037 to +20 (for a large boost). Beware of clipping when using a positive gain.
6040 Set the filter's central frequency and so can be used
6041 to extend or reduce the frequency range to be boosted or cut.
6042 The default value is @code{3000} Hz.
6045 Set method to specify band-width of filter.
6060 Determine how steep is the filter's shelf transition.
6063 Set number of poles. Default is 2.
6066 How much to use filtered signal in output. Default is 1.
6067 Range is between 0 and 1.
6070 Specify which channels to filter, by default all available are filtered.
6073 Normalize biquad coefficients, by default is disabled.
6074 Enabling it will normalize magnitude response at DC to 0dB.
6077 Set transform type of IIR filter.
6086 Set precison of filtering.
6089 Pick automatic sample format depending on surround filters.
6091 Always use signed 16-bit.
6093 Always use signed 32-bit.
6095 Always use float 32-bit.
6097 Always use float 64-bit.
6101 @subsection Commands
6103 This filter supports the following commands:
6106 Change treble frequency.
6107 Syntax for the command is : "@var{frequency}"
6110 Change treble width_type.
6111 Syntax for the command is : "@var{width_type}"
6114 Change treble width.
6115 Syntax for the command is : "@var{width}"
6119 Syntax for the command is : "@var{gain}"
6123 Syntax for the command is : "@var{mix}"
6128 Sinusoidal amplitude modulation.
6130 The filter accepts the following options:
6134 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
6135 (20 Hz or lower) will result in a tremolo effect.
6136 This filter may also be used as a ring modulator by specifying
6137 a modulation frequency higher than 20 Hz.
6138 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6141 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6142 Default value is 0.5.
6147 Sinusoidal phase modulation.
6149 The filter accepts the following options:
6153 Modulation frequency in Hertz.
6154 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6157 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6158 Default value is 0.5.
6163 Adjust the input audio volume.
6165 It accepts the following parameters:
6169 Set audio volume expression.
6171 Output values are clipped to the maximum value.
6173 The output audio volume is given by the relation:
6175 @var{output_volume} = @var{volume} * @var{input_volume}
6178 The default value for @var{volume} is "1.0".
6181 This parameter represents the mathematical precision.
6183 It determines which input sample formats will be allowed, which affects the
6184 precision of the volume scaling.
6188 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
6190 32-bit floating-point; this limits input sample format to FLT. (default)
6192 64-bit floating-point; this limits input sample format to DBL.
6196 Choose the behaviour on encountering ReplayGain side data in input frames.
6200 Remove ReplayGain side data, ignoring its contents (the default).
6203 Ignore ReplayGain side data, but leave it in the frame.
6206 Prefer the track gain, if present.
6209 Prefer the album gain, if present.
6212 @item replaygain_preamp
6213 Pre-amplification gain in dB to apply to the selected replaygain gain.
6215 Default value for @var{replaygain_preamp} is 0.0.
6217 @item replaygain_noclip
6218 Prevent clipping by limiting the gain applied.
6220 Default value for @var{replaygain_noclip} is 1.
6223 Set when the volume expression is evaluated.
6225 It accepts the following values:
6228 only evaluate expression once during the filter initialization, or
6229 when the @samp{volume} command is sent
6232 evaluate expression for each incoming frame
6235 Default value is @samp{once}.
6238 The volume expression can contain the following parameters.
6242 frame number (starting at zero)
6245 @item nb_consumed_samples
6246 number of samples consumed by the filter
6248 number of samples in the current frame
6250 original frame position in the file
6256 PTS at start of stream
6258 time at start of stream
6264 last set volume value
6267 Note that when @option{eval} is set to @samp{once} only the
6268 @var{sample_rate} and @var{tb} variables are available, all other
6269 variables will evaluate to NAN.
6271 @subsection Commands
6273 This filter supports the following commands:
6276 Modify the volume expression.
6277 The command accepts the same syntax of the corresponding option.
6279 If the specified expression is not valid, it is kept at its current
6283 @subsection Examples
6287 Halve the input audio volume:
6291 volume=volume=-6.0206dB
6294 In all the above example the named key for @option{volume} can be
6295 omitted, for example like in:
6301 Increase input audio power by 6 decibels using fixed-point precision:
6303 volume=volume=6dB:precision=fixed
6307 Fade volume after time 10 with an annihilation period of 5 seconds:
6309 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6313 @section volumedetect
6315 Detect the volume of the input video.
6317 The filter has no parameters. The input is not modified. Statistics about
6318 the volume will be printed in the log when the input stream end is reached.
6320 In particular it will show the mean volume (root mean square), maximum
6321 volume (on a per-sample basis), and the beginning of a histogram of the
6322 registered volume values (from the maximum value to a cumulated 1/1000 of
6325 All volumes are in decibels relative to the maximum PCM value.
6327 @subsection Examples
6329 Here is an excerpt of the output:
6331 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6332 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6333 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6334 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6335 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6336 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6337 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6338 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6339 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6345 The mean square energy is approximately -27 dB, or 10^-2.7.
6347 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6349 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6352 In other words, raising the volume by +4 dB does not cause any clipping,
6353 raising it by +5 dB causes clipping for 6 samples, etc.
6355 @c man end AUDIO FILTERS
6357 @chapter Audio Sources
6358 @c man begin AUDIO SOURCES
6360 Below is a description of the currently available audio sources.
6364 Buffer audio frames, and make them available to the filter chain.
6366 This source is mainly intended for a programmatic use, in particular
6367 through the interface defined in @file{libavfilter/buffersrc.h}.
6369 It accepts the following parameters:
6373 The timebase which will be used for timestamps of submitted frames. It must be
6374 either a floating-point number or in @var{numerator}/@var{denominator} form.
6377 The sample rate of the incoming audio buffers.
6380 The sample format of the incoming audio buffers.
6381 Either a sample format name or its corresponding integer representation from
6382 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6384 @item channel_layout
6385 The channel layout of the incoming audio buffers.
6386 Either a channel layout name from channel_layout_map in
6387 @file{libavutil/channel_layout.c} or its corresponding integer representation
6388 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6391 The number of channels of the incoming audio buffers.
6392 If both @var{channels} and @var{channel_layout} are specified, then they
6397 @subsection Examples
6400 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6403 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6404 Since the sample format with name "s16p" corresponds to the number
6405 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6408 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6413 Generate an audio signal specified by an expression.
6415 This source accepts in input one or more expressions (one for each
6416 channel), which are evaluated and used to generate a corresponding
6419 This source accepts the following options:
6423 Set the '|'-separated expressions list for each separate channel. In case the
6424 @option{channel_layout} option is not specified, the selected channel layout
6425 depends on the number of provided expressions. Otherwise the last
6426 specified expression is applied to the remaining output channels.
6428 @item channel_layout, c
6429 Set the channel layout. The number of channels in the specified layout
6430 must be equal to the number of specified expressions.
6433 Set the minimum duration of the sourced audio. See
6434 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6435 for the accepted syntax.
6436 Note that the resulting duration may be greater than the specified
6437 duration, as the generated audio is always cut at the end of a
6440 If not specified, or the expressed duration is negative, the audio is
6441 supposed to be generated forever.
6444 Set the number of samples per channel per each output frame,
6447 @item sample_rate, s
6448 Specify the sample rate, default to 44100.
6451 Each expression in @var{exprs} can contain the following constants:
6455 number of the evaluated sample, starting from 0
6458 time of the evaluated sample expressed in seconds, starting from 0
6465 @subsection Examples
6475 Generate a sin signal with frequency of 440 Hz, set sample rate to
6478 aevalsrc="sin(440*2*PI*t):s=8000"
6482 Generate a two channels signal, specify the channel layout (Front
6483 Center + Back Center) explicitly:
6485 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6489 Generate white noise:
6491 aevalsrc="-2+random(0)"
6495 Generate an amplitude modulated signal:
6497 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6501 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6503 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6510 Generate a FIR coefficients using frequency sampling method.
6512 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6514 The filter accepts the following options:
6518 Set number of filter coefficents in output audio stream.
6519 Default value is 1025.
6522 Set frequency points from where magnitude and phase are set.
6523 This must be in non decreasing order, and first element must be 0, while last element
6524 must be 1. Elements are separated by white spaces.
6527 Set magnitude value for every frequency point set by @option{frequency}.
6528 Number of values must be same as number of frequency points.
6529 Values are separated by white spaces.
6532 Set phase value for every frequency point set by @option{frequency}.
6533 Number of values must be same as number of frequency points.
6534 Values are separated by white spaces.
6536 @item sample_rate, r
6537 Set sample rate, default is 44100.
6540 Set number of samples per each frame. Default is 1024.
6543 Set window function. Default is blackman.
6548 The null audio source, return unprocessed audio frames. It is mainly useful
6549 as a template and to be employed in analysis / debugging tools, or as
6550 the source for filters which ignore the input data (for example the sox
6553 This source accepts the following options:
6557 @item channel_layout, cl
6559 Specifies the channel layout, and can be either an integer or a string
6560 representing a channel layout. The default value of @var{channel_layout}
6563 Check the channel_layout_map definition in
6564 @file{libavutil/channel_layout.c} for the mapping between strings and
6565 channel layout values.
6567 @item sample_rate, r
6568 Specifies the sample rate, and defaults to 44100.
6571 Set the number of samples per requested frames.
6574 Set the duration of the sourced audio. See
6575 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6576 for the accepted syntax.
6578 If not specified, or the expressed duration is negative, the audio is
6579 supposed to be generated forever.
6582 @subsection Examples
6586 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6588 anullsrc=r=48000:cl=4
6592 Do the same operation with a more obvious syntax:
6594 anullsrc=r=48000:cl=mono
6598 All the parameters need to be explicitly defined.
6602 Synthesize a voice utterance using the libflite library.
6604 To enable compilation of this filter you need to configure FFmpeg with
6605 @code{--enable-libflite}.
6607 Note that versions of the flite library prior to 2.0 are not thread-safe.
6609 The filter accepts the following options:
6614 If set to 1, list the names of the available voices and exit
6615 immediately. Default value is 0.
6618 Set the maximum number of samples per frame. Default value is 512.
6621 Set the filename containing the text to speak.
6624 Set the text to speak.
6627 Set the voice to use for the speech synthesis. Default value is
6628 @code{kal}. See also the @var{list_voices} option.
6631 @subsection Examples
6635 Read from file @file{speech.txt}, and synthesize the text using the
6636 standard flite voice:
6638 flite=textfile=speech.txt
6642 Read the specified text selecting the @code{slt} voice:
6644 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6648 Input text to ffmpeg:
6650 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6654 Make @file{ffplay} speak the specified text, using @code{flite} and
6655 the @code{lavfi} device:
6657 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6661 For more information about libflite, check:
6662 @url{http://www.festvox.org/flite/}
6666 Generate a noise audio signal.
6668 The filter accepts the following options:
6671 @item sample_rate, r
6672 Specify the sample rate. Default value is 48000 Hz.
6675 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6679 Specify the duration of the generated audio stream. Not specifying this option
6680 results in noise with an infinite length.
6682 @item color, colour, c
6683 Specify the color of noise. Available noise colors are white, pink, brown,
6684 blue, violet and velvet. Default color is white.
6687 Specify a value used to seed the PRNG.
6690 Set the number of samples per each output frame, default is 1024.
6693 @subsection Examples
6698 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6700 anoisesrc=d=60:c=pink:r=44100:a=0.5
6706 Generate odd-tap Hilbert transform FIR coefficients.
6708 The resulting stream can be used with @ref{afir} filter for phase-shifting
6709 the signal by 90 degrees.
6711 This is used in many matrix coding schemes and for analytic signal generation.
6712 The process is often written as a multiplication by i (or j), the imaginary unit.
6714 The filter accepts the following options:
6718 @item sample_rate, s
6719 Set sample rate, default is 44100.
6722 Set length of FIR filter, default is 22051.
6725 Set number of samples per each frame.
6728 Set window function to be used when generating FIR coefficients.
6733 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6735 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6737 The filter accepts the following options:
6740 @item sample_rate, r
6741 Set sample rate, default is 44100.
6744 Set number of samples per each frame. Default is 1024.
6747 Set high-pass frequency. Default is 0.
6750 Set low-pass frequency. Default is 0.
6751 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6752 is higher than 0 then filter will create band-pass filter coefficients,
6753 otherwise band-reject filter coefficients.
6756 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6759 Set Kaiser window beta.
6762 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6765 Enable rounding, by default is disabled.
6768 Set number of taps for high-pass filter.
6771 Set number of taps for low-pass filter.
6776 Generate an audio signal made of a sine wave with amplitude 1/8.
6778 The audio signal is bit-exact.
6780 The filter accepts the following options:
6785 Set the carrier frequency. Default is 440 Hz.
6787 @item beep_factor, b
6788 Enable a periodic beep every second with frequency @var{beep_factor} times
6789 the carrier frequency. Default is 0, meaning the beep is disabled.
6791 @item sample_rate, r
6792 Specify the sample rate, default is 44100.
6795 Specify the duration of the generated audio stream.
6797 @item samples_per_frame
6798 Set the number of samples per output frame.
6800 The expression can contain the following constants:
6804 The (sequential) number of the output audio frame, starting from 0.
6807 The PTS (Presentation TimeStamp) of the output audio frame,
6808 expressed in @var{TB} units.
6811 The PTS of the output audio frame, expressed in seconds.
6814 The timebase of the output audio frames.
6817 Default is @code{1024}.
6820 @subsection Examples
6825 Generate a simple 440 Hz sine wave:
6831 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6835 sine=frequency=220:beep_factor=4:duration=5
6839 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6842 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6846 @c man end AUDIO SOURCES
6848 @chapter Audio Sinks
6849 @c man begin AUDIO SINKS
6851 Below is a description of the currently available audio sinks.
6853 @section abuffersink
6855 Buffer audio frames, and make them available to the end of filter chain.
6857 This sink is mainly intended for programmatic use, in particular
6858 through the interface defined in @file{libavfilter/buffersink.h}
6859 or the options system.
6861 It accepts a pointer to an AVABufferSinkContext structure, which
6862 defines the incoming buffers' formats, to be passed as the opaque
6863 parameter to @code{avfilter_init_filter} for initialization.
6866 Null audio sink; do absolutely nothing with the input audio. It is
6867 mainly useful as a template and for use in analysis / debugging
6870 @c man end AUDIO SINKS
6872 @chapter Video Filters
6873 @c man begin VIDEO FILTERS
6875 When you configure your FFmpeg build, you can disable any of the
6876 existing filters using @code{--disable-filters}.
6877 The configure output will show the video filters included in your
6880 Below is a description of the currently available video filters.
6884 Mark a region of interest in a video frame.
6886 The frame data is passed through unchanged, but metadata is attached
6887 to the frame indicating regions of interest which can affect the
6888 behaviour of later encoding. Multiple regions can be marked by
6889 applying the filter multiple times.
6893 Region distance in pixels from the left edge of the frame.
6895 Region distance in pixels from the top edge of the frame.
6897 Region width in pixels.
6899 Region height in pixels.
6901 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6902 and may contain the following variables:
6905 Width of the input frame.
6907 Height of the input frame.
6911 Quantisation offset to apply within the region.
6913 This must be a real value in the range -1 to +1. A value of zero
6914 indicates no quality change. A negative value asks for better quality
6915 (less quantisation), while a positive value asks for worse quality
6916 (greater quantisation).
6918 The range is calibrated so that the extreme values indicate the
6919 largest possible offset - if the rest of the frame is encoded with the
6920 worst possible quality, an offset of -1 indicates that this region
6921 should be encoded with the best possible quality anyway. Intermediate
6922 values are then interpolated in some codec-dependent way.
6924 For example, in 10-bit H.264 the quantisation parameter varies between
6925 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6926 this region should be encoded with a QP around one-tenth of the full
6927 range better than the rest of the frame. So, if most of the frame
6928 were to be encoded with a QP of around 30, this region would get a QP
6929 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6930 An extreme value of -1 would indicate that this region should be
6931 encoded with the best possible quality regardless of the treatment of
6932 the rest of the frame - that is, should be encoded at a QP of -12.
6934 If set to true, remove any existing regions of interest marked on the
6935 frame before adding the new one.
6938 @subsection Examples
6942 Mark the centre quarter of the frame as interesting.
6944 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6947 Mark the 100-pixel-wide region on the left edge of the frame as very
6948 uninteresting (to be encoded at much lower quality than the rest of
6951 addroi=0:0:100:ih:+1/5
6955 @section alphaextract
6957 Extract the alpha component from the input as a grayscale video. This
6958 is especially useful with the @var{alphamerge} filter.
6962 Add or replace the alpha component of the primary input with the
6963 grayscale value of a second input. This is intended for use with
6964 @var{alphaextract} to allow the transmission or storage of frame
6965 sequences that have alpha in a format that doesn't support an alpha
6968 For example, to reconstruct full frames from a normal YUV-encoded video
6969 and a separate video created with @var{alphaextract}, you might use:
6971 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6976 Amplify differences between current pixel and pixels of adjacent frames in
6977 same pixel location.
6979 This filter accepts the following options:
6983 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6984 For example radius of 3 will instruct filter to calculate average of 7 frames.
6987 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6990 Set threshold for difference amplification. Any difference greater or equal to
6991 this value will not alter source pixel. Default is 10.
6992 Allowed range is from 0 to 65535.
6995 Set tolerance for difference amplification. Any difference lower to
6996 this value will not alter source pixel. Default is 0.
6997 Allowed range is from 0 to 65535.
7000 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7001 This option controls maximum possible value that will decrease source pixel value.
7004 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7005 This option controls maximum possible value that will increase source pixel value.
7008 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
7011 @subsection Commands
7013 This filter supports the following @ref{commands} that corresponds to option of same name:
7025 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
7026 and libavformat to work. On the other hand, it is limited to ASS (Advanced
7027 Substation Alpha) subtitles files.
7029 This filter accepts the following option in addition to the common options from
7030 the @ref{subtitles} filter:
7034 Set the shaping engine
7036 Available values are:
7039 The default libass shaping engine, which is the best available.
7041 Fast, font-agnostic shaper that can do only substitutions
7043 Slower shaper using OpenType for substitutions and positioning
7046 The default is @code{auto}.
7050 Apply an Adaptive Temporal Averaging Denoiser to the video input.
7052 The filter accepts the following options:
7056 Set threshold A for 1st plane. Default is 0.02.
7057 Valid range is 0 to 0.3.
7060 Set threshold B for 1st plane. Default is 0.04.
7061 Valid range is 0 to 5.
7064 Set threshold A for 2nd plane. Default is 0.02.
7065 Valid range is 0 to 0.3.
7068 Set threshold B for 2nd plane. Default is 0.04.
7069 Valid range is 0 to 5.
7072 Set threshold A for 3rd plane. Default is 0.02.
7073 Valid range is 0 to 0.3.
7076 Set threshold B for 3rd plane. Default is 0.04.
7077 Valid range is 0 to 5.
7079 Threshold A is designed to react on abrupt changes in the input signal and
7080 threshold B is designed to react on continuous changes in the input signal.
7083 Set number of frames filter will use for averaging. Default is 9. Must be odd
7084 number in range [5, 129].
7087 Set what planes of frame filter will use for averaging. Default is all.
7090 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
7091 Alternatively can be set to @code{s} serial.
7093 Parallel can be faster then serial, while other way around is never true.
7094 Parallel will abort early on first change being greater then thresholds, while serial
7095 will continue processing other side of frames if they are equal or below thresholds.
7098 @subsection Commands
7099 This filter supports same @ref{commands} as options except option @code{s}.
7100 The command accepts the same syntax of the corresponding option.
7104 Apply average blur filter.
7106 The filter accepts the following options:
7110 Set horizontal radius size.
7113 Set which planes to filter. By default all planes are filtered.
7116 Set vertical radius size, if zero it will be same as @code{sizeX}.
7117 Default is @code{0}.
7120 @subsection Commands
7121 This filter supports same commands as options.
7122 The command accepts the same syntax of the corresponding option.
7124 If the specified expression is not valid, it is kept at its current
7129 Compute the bounding box for the non-black pixels in the input frame
7132 This filter computes the bounding box containing all the pixels with a
7133 luminance value greater than the minimum allowed value.
7134 The parameters describing the bounding box are printed on the filter
7137 The filter accepts the following option:
7141 Set the minimal luminance value. Default is @code{16}.
7145 Apply bilateral filter, spatial smoothing while preserving edges.
7147 The filter accepts the following options:
7150 Set sigma of gaussian function to calculate spatial weight.
7151 Allowed range is 0 to 512. Default is 0.1.
7154 Set sigma of gaussian function to calculate range weight.
7155 Allowed range is 0 to 1. Default is 0.1.
7158 Set planes to filter. Default is first only.
7161 @section bitplanenoise
7163 Show and measure bit plane noise.
7165 The filter accepts the following options:
7169 Set which plane to analyze. Default is @code{1}.
7172 Filter out noisy pixels from @code{bitplane} set above.
7173 Default is disabled.
7176 @section blackdetect
7178 Detect video intervals that are (almost) completely black. Can be
7179 useful to detect chapter transitions, commercials, or invalid
7182 The filter outputs its detection analysis to both the log as well as
7183 frame metadata. If a black segment of at least the specified minimum
7184 duration is found, a line with the start and end timestamps as well
7185 as duration is printed to the log with level @code{info}. In addition,
7186 a log line with level @code{debug} is printed per frame showing the
7187 black amount detected for that frame.
7189 The filter also attaches metadata to the first frame of a black
7190 segment with key @code{lavfi.black_start} and to the first frame
7191 after the black segment ends with key @code{lavfi.black_end}. The
7192 value is the frame's timestamp. This metadata is added regardless
7193 of the minimum duration specified.
7195 The filter accepts the following options:
7198 @item black_min_duration, d
7199 Set the minimum detected black duration expressed in seconds. It must
7200 be a non-negative floating point number.
7202 Default value is 2.0.
7204 @item picture_black_ratio_th, pic_th
7205 Set the threshold for considering a picture "black".
7206 Express the minimum value for the ratio:
7208 @var{nb_black_pixels} / @var{nb_pixels}
7211 for which a picture is considered black.
7212 Default value is 0.98.
7214 @item pixel_black_th, pix_th
7215 Set the threshold for considering a pixel "black".
7217 The threshold expresses the maximum pixel luminance value for which a
7218 pixel is considered "black". The provided value is scaled according to
7219 the following equation:
7221 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
7224 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
7225 the input video format, the range is [0-255] for YUV full-range
7226 formats and [16-235] for YUV non full-range formats.
7228 Default value is 0.10.
7231 The following example sets the maximum pixel threshold to the minimum
7232 value, and detects only black intervals of 2 or more seconds:
7234 blackdetect=d=2:pix_th=0.00
7239 Detect frames that are (almost) completely black. Can be useful to
7240 detect chapter transitions or commercials. Output lines consist of
7241 the frame number of the detected frame, the percentage of blackness,
7242 the position in the file if known or -1 and the timestamp in seconds.
7244 In order to display the output lines, you need to set the loglevel at
7245 least to the AV_LOG_INFO value.
7247 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
7248 The value represents the percentage of pixels in the picture that
7249 are below the threshold value.
7251 It accepts the following parameters:
7256 The percentage of the pixels that have to be below the threshold; it defaults to
7259 @item threshold, thresh
7260 The threshold below which a pixel value is considered black; it defaults to
7268 Blend two video frames into each other.
7270 The @code{blend} filter takes two input streams and outputs one
7271 stream, the first input is the "top" layer and second input is
7272 "bottom" layer. By default, the output terminates when the longest input terminates.
7274 The @code{tblend} (time blend) filter takes two consecutive frames
7275 from one single stream, and outputs the result obtained by blending
7276 the new frame on top of the old frame.
7278 A description of the accepted options follows.
7286 Set blend mode for specific pixel component or all pixel components in case
7287 of @var{all_mode}. Default value is @code{normal}.
7289 Available values for component modes are:
7331 Set blend opacity for specific pixel component or all pixel components in case
7332 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7339 Set blend expression for specific pixel component or all pixel components in case
7340 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7342 The expressions can use the following variables:
7346 The sequential number of the filtered frame, starting from @code{0}.
7350 the coordinates of the current sample
7354 the width and height of currently filtered plane
7358 Width and height scale for the plane being filtered. It is the
7359 ratio between the dimensions of the current plane to the luma plane,
7360 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7361 the luma plane and @code{0.5,0.5} for the chroma planes.
7364 Time of the current frame, expressed in seconds.
7367 Value of pixel component at current location for first video frame (top layer).
7370 Value of pixel component at current location for second video frame (bottom layer).
7374 The @code{blend} filter also supports the @ref{framesync} options.
7376 @subsection Examples
7380 Apply transition from bottom layer to top layer in first 10 seconds:
7382 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7386 Apply linear horizontal transition from top layer to bottom layer:
7388 blend=all_expr='A*(X/W)+B*(1-X/W)'
7392 Apply 1x1 checkerboard effect:
7394 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7398 Apply uncover left effect:
7400 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7404 Apply uncover down effect:
7406 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7410 Apply uncover up-left effect:
7412 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7416 Split diagonally video and shows top and bottom layer on each side:
7418 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7422 Display differences between the current and the previous frame:
7424 tblend=all_mode=grainextract
7430 Denoise frames using Block-Matching 3D algorithm.
7432 The filter accepts the following options.
7436 Set denoising strength. Default value is 1.
7437 Allowed range is from 0 to 999.9.
7438 The denoising algorithm is very sensitive to sigma, so adjust it
7439 according to the source.
7442 Set local patch size. This sets dimensions in 2D.
7445 Set sliding step for processing blocks. Default value is 4.
7446 Allowed range is from 1 to 64.
7447 Smaller values allows processing more reference blocks and is slower.
7450 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7451 When set to 1, no block matching is done. Larger values allows more blocks
7453 Allowed range is from 1 to 256.
7456 Set radius for search block matching. Default is 9.
7457 Allowed range is from 1 to INT32_MAX.
7460 Set step between two search locations for block matching. Default is 1.
7461 Allowed range is from 1 to 64. Smaller is slower.
7464 Set threshold of mean square error for block matching. Valid range is 0 to
7468 Set thresholding parameter for hard thresholding in 3D transformed domain.
7469 Larger values results in stronger hard-thresholding filtering in frequency
7473 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7474 Default is @code{basic}.
7477 If enabled, filter will use 2nd stream for block matching.
7478 Default is disabled for @code{basic} value of @var{estim} option,
7479 and always enabled if value of @var{estim} is @code{final}.
7482 Set planes to filter. Default is all available except alpha.
7485 @subsection Examples
7489 Basic filtering with bm3d:
7491 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7495 Same as above, but filtering only luma:
7497 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7501 Same as above, but with both estimation modes:
7503 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
7507 Same as above, but prefilter with @ref{nlmeans} filter instead:
7509 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
7515 Apply a boxblur algorithm to the input video.
7517 It accepts the following parameters:
7521 @item luma_radius, lr
7522 @item luma_power, lp
7523 @item chroma_radius, cr
7524 @item chroma_power, cp
7525 @item alpha_radius, ar
7526 @item alpha_power, ap
7530 A description of the accepted options follows.
7533 @item luma_radius, lr
7534 @item chroma_radius, cr
7535 @item alpha_radius, ar
7536 Set an expression for the box radius in pixels used for blurring the
7537 corresponding input plane.
7539 The radius value must be a non-negative number, and must not be
7540 greater than the value of the expression @code{min(w,h)/2} for the
7541 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7544 Default value for @option{luma_radius} is "2". If not specified,
7545 @option{chroma_radius} and @option{alpha_radius} default to the
7546 corresponding value set for @option{luma_radius}.
7548 The expressions can contain the following constants:
7552 The input width and height in pixels.
7556 The input chroma image width and height in pixels.
7560 The horizontal and vertical chroma subsample values. For example, for the
7561 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7564 @item luma_power, lp
7565 @item chroma_power, cp
7566 @item alpha_power, ap
7567 Specify how many times the boxblur filter is applied to the
7568 corresponding plane.
7570 Default value for @option{luma_power} is 2. If not specified,
7571 @option{chroma_power} and @option{alpha_power} default to the
7572 corresponding value set for @option{luma_power}.
7574 A value of 0 will disable the effect.
7577 @subsection Examples
7581 Apply a boxblur filter with the luma, chroma, and alpha radii
7584 boxblur=luma_radius=2:luma_power=1
7589 Set the luma radius to 2, and alpha and chroma radius to 0:
7591 boxblur=2:1:cr=0:ar=0
7595 Set the luma and chroma radii to a fraction of the video dimension:
7597 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7603 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7604 Deinterlacing Filter").
7606 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7607 interpolation algorithms.
7608 It accepts the following parameters:
7612 The interlacing mode to adopt. It accepts one of the following values:
7616 Output one frame for each frame.
7618 Output one frame for each field.
7621 The default value is @code{send_field}.
7624 The picture field parity assumed for the input interlaced video. It accepts one
7625 of the following values:
7629 Assume the top field is first.
7631 Assume the bottom field is first.
7633 Enable automatic detection of field parity.
7636 The default value is @code{auto}.
7637 If the interlacing is unknown or the decoder does not export this information,
7638 top field first will be assumed.
7641 Specify which frames to deinterlace. Accepts one of the following
7646 Deinterlace all frames.
7648 Only deinterlace frames marked as interlaced.
7651 The default value is @code{all}.
7656 Apply Contrast Adaptive Sharpen filter to video stream.
7658 The filter accepts the following options:
7662 Set the sharpening strength. Default value is 0.
7665 Set planes to filter. Default value is to filter all
7666 planes except alpha plane.
7670 Remove all color information for all colors except for certain one.
7672 The filter accepts the following options:
7676 The color which will not be replaced with neutral chroma.
7679 Similarity percentage with the above color.
7680 0.01 matches only the exact key color, while 1.0 matches everything.
7684 0.0 makes pixels either fully gray, or not gray at all.
7685 Higher values result in more preserved color.
7688 Signals that the color passed is already in YUV instead of RGB.
7690 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7691 This can be used to pass exact YUV values as hexadecimal numbers.
7694 @subsection Commands
7695 This filter supports same @ref{commands} as options.
7696 The command accepts the same syntax of the corresponding option.
7698 If the specified expression is not valid, it is kept at its current
7702 YUV colorspace color/chroma keying.
7704 The filter accepts the following options:
7708 The color which will be replaced with transparency.
7711 Similarity percentage with the key color.
7713 0.01 matches only the exact key color, while 1.0 matches everything.
7718 0.0 makes pixels either fully transparent, or not transparent at all.
7720 Higher values result in semi-transparent pixels, with a higher transparency
7721 the more similar the pixels color is to the key color.
7724 Signals that the color passed is already in YUV instead of RGB.
7726 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7727 This can be used to pass exact YUV values as hexadecimal numbers.
7730 @subsection Commands
7731 This filter supports same @ref{commands} as options.
7732 The command accepts the same syntax of the corresponding option.
7734 If the specified expression is not valid, it is kept at its current
7737 @subsection Examples
7741 Make every green pixel in the input image transparent:
7743 ffmpeg -i input.png -vf chromakey=green out.png
7747 Overlay a greenscreen-video on top of a static black background.
7749 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
7754 Reduce chrominance noise.
7756 The filter accepts the following options:
7760 Set threshold for averaging chrominance values.
7761 Sum of absolute difference of Y, U and V pixel components of current
7762 pixel and neighbour pixels lower than this threshold will be used in
7763 averaging. Luma component is left unchanged and is copied to output.
7764 Default value is 30. Allowed range is from 1 to 200.
7767 Set horizontal radius of rectangle used for averaging.
7768 Allowed range is from 1 to 100. Default value is 5.
7771 Set vertical radius of rectangle used for averaging.
7772 Allowed range is from 1 to 100. Default value is 5.
7775 Set horizontal step when averaging. Default value is 1.
7776 Allowed range is from 1 to 50.
7777 Mostly useful to speed-up filtering.
7780 Set vertical step when averaging. Default value is 1.
7781 Allowed range is from 1 to 50.
7782 Mostly useful to speed-up filtering.
7785 Set Y threshold for averaging chrominance values.
7786 Set finer control for max allowed difference between Y components
7787 of current pixel and neigbour pixels.
7788 Default value is 200. Allowed range is from 1 to 200.
7791 Set U threshold for averaging chrominance values.
7792 Set finer control for max allowed difference between U components
7793 of current pixel and neigbour pixels.
7794 Default value is 200. Allowed range is from 1 to 200.
7797 Set V threshold for averaging chrominance values.
7798 Set finer control for max allowed difference between V components
7799 of current pixel and neigbour pixels.
7800 Default value is 200. Allowed range is from 1 to 200.
7803 @subsection Commands
7804 This filter supports same @ref{commands} as options.
7805 The command accepts the same syntax of the corresponding option.
7807 @section chromashift
7808 Shift chroma pixels horizontally and/or vertically.
7810 The filter accepts the following options:
7813 Set amount to shift chroma-blue horizontally.
7815 Set amount to shift chroma-blue vertically.
7817 Set amount to shift chroma-red horizontally.
7819 Set amount to shift chroma-red vertically.
7821 Set edge mode, can be @var{smear}, default, or @var{warp}.
7824 @subsection Commands
7826 This filter supports the all above options as @ref{commands}.
7830 Display CIE color diagram with pixels overlaid onto it.
7832 The filter accepts the following options:
7847 @item uhdtv, rec2020
7861 Set what gamuts to draw.
7863 See @code{system} option for available values.
7866 Set ciescope size, by default set to 512.
7869 Set intensity used to map input pixel values to CIE diagram.
7872 Set contrast used to draw tongue colors that are out of active color system gamut.
7875 Correct gamma displayed on scope, by default enabled.
7878 Show white point on CIE diagram, by default disabled.
7881 Set input gamma. Used only with XYZ input color space.
7886 Visualize information exported by some codecs.
7888 Some codecs can export information through frames using side-data or other
7889 means. For example, some MPEG based codecs export motion vectors through the
7890 @var{export_mvs} flag in the codec @option{flags2} option.
7892 The filter accepts the following option:
7896 Set motion vectors to visualize.
7898 Available flags for @var{mv} are:
7902 forward predicted MVs of P-frames
7904 forward predicted MVs of B-frames
7906 backward predicted MVs of B-frames
7910 Display quantization parameters using the chroma planes.
7913 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7915 Available flags for @var{mv_type} are:
7919 forward predicted MVs
7921 backward predicted MVs
7924 @item frame_type, ft
7925 Set frame type to visualize motion vectors of.
7927 Available flags for @var{frame_type} are:
7931 intra-coded frames (I-frames)
7933 predicted frames (P-frames)
7935 bi-directionally predicted frames (B-frames)
7939 @subsection Examples
7943 Visualize forward predicted MVs of all frames using @command{ffplay}:
7945 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7949 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7951 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7955 @section colorbalance
7956 Modify intensity of primary colors (red, green and blue) of input frames.
7958 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7959 regions for the red-cyan, green-magenta or blue-yellow balance.
7961 A positive adjustment value shifts the balance towards the primary color, a negative
7962 value towards the complementary color.
7964 The filter accepts the following options:
7970 Adjust red, green and blue shadows (darkest pixels).
7975 Adjust red, green and blue midtones (medium pixels).
7980 Adjust red, green and blue highlights (brightest pixels).
7982 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7985 Preserve lightness when changing color balance. Default is disabled.
7988 @subsection Examples
7992 Add red color cast to shadows:
7998 @subsection Commands
8000 This filter supports the all above options as @ref{commands}.
8002 @section colorchannelmixer
8004 Adjust video input frames by re-mixing color channels.
8006 This filter modifies a color channel by adding the values associated to
8007 the other channels of the same pixels. For example if the value to
8008 modify is red, the output value will be:
8010 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
8013 The filter accepts the following options:
8020 Adjust contribution of input red, green, blue and alpha channels for output red channel.
8021 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
8027 Adjust contribution of input red, green, blue and alpha channels for output green channel.
8028 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
8034 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
8035 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
8041 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
8042 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
8044 Allowed ranges for options are @code{[-2.0, 2.0]}.
8047 @subsection Examples
8051 Convert source to grayscale:
8053 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
8056 Simulate sepia tones:
8058 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
8062 @subsection Commands
8064 This filter supports the all above options as @ref{commands}.
8067 RGB colorspace color keying.
8069 The filter accepts the following options:
8073 The color which will be replaced with transparency.
8076 Similarity percentage with the key color.
8078 0.01 matches only the exact key color, while 1.0 matches everything.
8083 0.0 makes pixels either fully transparent, or not transparent at all.
8085 Higher values result in semi-transparent pixels, with a higher transparency
8086 the more similar the pixels color is to the key color.
8089 @subsection Examples
8093 Make every green pixel in the input image transparent:
8095 ffmpeg -i input.png -vf colorkey=green out.png
8099 Overlay a greenscreen-video on top of a static background image.
8101 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
8105 @subsection Commands
8106 This filter supports same @ref{commands} as options.
8107 The command accepts the same syntax of the corresponding option.
8109 If the specified expression is not valid, it is kept at its current
8113 Remove all color information for all RGB colors except for certain one.
8115 The filter accepts the following options:
8119 The color which will not be replaced with neutral gray.
8122 Similarity percentage with the above color.
8123 0.01 matches only the exact key color, while 1.0 matches everything.
8126 Blend percentage. 0.0 makes pixels fully gray.
8127 Higher values result in more preserved color.
8130 @subsection Commands
8131 This filter supports same @ref{commands} as options.
8132 The command accepts the same syntax of the corresponding option.
8134 If the specified expression is not valid, it is kept at its current
8137 @section colorlevels
8139 Adjust video input frames using levels.
8141 The filter accepts the following options:
8148 Adjust red, green, blue and alpha input black point.
8149 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
8155 Adjust red, green, blue and alpha input white point.
8156 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
8158 Input levels are used to lighten highlights (bright tones), darken shadows
8159 (dark tones), change the balance of bright and dark tones.
8165 Adjust red, green, blue and alpha output black point.
8166 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
8172 Adjust red, green, blue and alpha output white point.
8173 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
8175 Output levels allows manual selection of a constrained output level range.
8178 @subsection Examples
8182 Make video output darker:
8184 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
8190 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
8194 Make video output lighter:
8196 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
8200 Increase brightness:
8202 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
8206 @subsection Commands
8208 This filter supports the all above options as @ref{commands}.
8210 @section colormatrix
8212 Convert color matrix.
8214 The filter accepts the following options:
8219 Specify the source and destination color matrix. Both values must be
8222 The accepted values are:
8250 For example to convert from BT.601 to SMPTE-240M, use the command:
8252 colormatrix=bt601:smpte240m
8257 Convert colorspace, transfer characteristics or color primaries.
8258 Input video needs to have an even size.
8260 The filter accepts the following options:
8265 Specify all color properties at once.
8267 The accepted values are:
8297 Specify output colorspace.
8299 The accepted values are:
8308 BT.470BG or BT.601-6 625
8311 SMPTE-170M or BT.601-6 525
8320 BT.2020 with non-constant luminance
8326 Specify output transfer characteristics.
8328 The accepted values are:
8340 Constant gamma of 2.2
8343 Constant gamma of 2.8
8346 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8364 BT.2020 for 10-bits content
8367 BT.2020 for 12-bits content
8373 Specify output color primaries.
8375 The accepted values are:
8384 BT.470BG or BT.601-6 625
8387 SMPTE-170M or BT.601-6 525
8411 Specify output color range.
8413 The accepted values are:
8416 TV (restricted) range
8419 MPEG (restricted) range
8430 Specify output color format.
8432 The accepted values are:
8435 YUV 4:2:0 planar 8-bits
8438 YUV 4:2:0 planar 10-bits
8441 YUV 4:2:0 planar 12-bits
8444 YUV 4:2:2 planar 8-bits
8447 YUV 4:2:2 planar 10-bits
8450 YUV 4:2:2 planar 12-bits
8453 YUV 4:4:4 planar 8-bits
8456 YUV 4:4:4 planar 10-bits
8459 YUV 4:4:4 planar 12-bits
8464 Do a fast conversion, which skips gamma/primary correction. This will take
8465 significantly less CPU, but will be mathematically incorrect. To get output
8466 compatible with that produced by the colormatrix filter, use fast=1.
8469 Specify dithering mode.
8471 The accepted values are:
8477 Floyd-Steinberg dithering
8481 Whitepoint adaptation mode.
8483 The accepted values are:
8486 Bradford whitepoint adaptation
8489 von Kries whitepoint adaptation
8492 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8496 Override all input properties at once. Same accepted values as @ref{all}.
8499 Override input colorspace. Same accepted values as @ref{space}.
8502 Override input color primaries. Same accepted values as @ref{primaries}.
8505 Override input transfer characteristics. Same accepted values as @ref{trc}.
8508 Override input color range. Same accepted values as @ref{range}.
8512 The filter converts the transfer characteristics, color space and color
8513 primaries to the specified user values. The output value, if not specified,
8514 is set to a default value based on the "all" property. If that property is
8515 also not specified, the filter will log an error. The output color range and
8516 format default to the same value as the input color range and format. The
8517 input transfer characteristics, color space, color primaries and color range
8518 should be set on the input data. If any of these are missing, the filter will
8519 log an error and no conversion will take place.
8521 For example to convert the input to SMPTE-240M, use the command:
8523 colorspace=smpte240m
8526 @section convolution
8528 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8530 The filter accepts the following options:
8537 Set matrix for each plane.
8538 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8539 and from 1 to 49 odd number of signed integers in @var{row} mode.
8545 Set multiplier for calculated value for each plane.
8546 If unset or 0, it will be sum of all matrix elements.
8552 Set bias for each plane. This value is added to the result of the multiplication.
8553 Useful for making the overall image brighter or darker. Default is 0.0.
8559 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8560 Default is @var{square}.
8563 @subsection Commands
8565 This filter supports the all above options as @ref{commands}.
8567 @subsection Examples
8573 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"
8579 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"
8585 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"
8591 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"
8595 Apply laplacian edge detector which includes diagonals:
8597 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"
8603 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"
8609 Apply 2D convolution of video stream in frequency domain using second stream
8612 The filter accepts the following options:
8616 Set which planes to process.
8619 Set which impulse video frames will be processed, can be @var{first}
8620 or @var{all}. Default is @var{all}.
8623 The @code{convolve} filter also supports the @ref{framesync} options.
8627 Copy the input video source unchanged to the output. This is mainly useful for
8632 Video filtering on GPU using Apple's CoreImage API on OSX.
8634 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8635 processed by video hardware. However, software-based OpenGL implementations
8636 exist which means there is no guarantee for hardware processing. It depends on
8639 There are many filters and image generators provided by Apple that come with a
8640 large variety of options. The filter has to be referenced by its name along
8643 The coreimage filter accepts the following options:
8646 List all available filters and generators along with all their respective
8647 options as well as possible minimum and maximum values along with the default
8654 Specify all filters by their respective name and options.
8655 Use @var{list_filters} to determine all valid filter names and options.
8656 Numerical options are specified by a float value and are automatically clamped
8657 to their respective value range. Vector and color options have to be specified
8658 by a list of space separated float values. Character escaping has to be done.
8659 A special option name @code{default} is available to use default options for a
8662 It is required to specify either @code{default} or at least one of the filter options.
8663 All omitted options are used with their default values.
8664 The syntax of the filter string is as follows:
8666 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8670 Specify a rectangle where the output of the filter chain is copied into the
8671 input image. It is given by a list of space separated float values:
8673 output_rect=x\ y\ width\ height
8675 If not given, the output rectangle equals the dimensions of the input image.
8676 The output rectangle is automatically cropped at the borders of the input
8677 image. Negative values are valid for each component.
8679 output_rect=25\ 25\ 100\ 100
8683 Several filters can be chained for successive processing without GPU-HOST
8684 transfers allowing for fast processing of complex filter chains.
8685 Currently, only filters with zero (generators) or exactly one (filters) input
8686 image and one output image are supported. Also, transition filters are not yet
8689 Some filters generate output images with additional padding depending on the
8690 respective filter kernel. The padding is automatically removed to ensure the
8691 filter output has the same size as the input image.
8693 For image generators, the size of the output image is determined by the
8694 previous output image of the filter chain or the input image of the whole
8695 filterchain, respectively. The generators do not use the pixel information of
8696 this image to generate their output. However, the generated output is
8697 blended onto this image, resulting in partial or complete coverage of the
8700 The @ref{coreimagesrc} video source can be used for generating input images
8701 which are directly fed into the filter chain. By using it, providing input
8702 images by another video source or an input video is not required.
8704 @subsection Examples
8709 List all filters available:
8711 coreimage=list_filters=true
8715 Use the CIBoxBlur filter with default options to blur an image:
8717 coreimage=filter=CIBoxBlur@@default
8721 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8722 its center at 100x100 and a radius of 50 pixels:
8724 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8728 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8729 given as complete and escaped command-line for Apple's standard bash shell:
8731 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8737 Cover a rectangular object
8739 It accepts the following options:
8743 Filepath of the optional cover image, needs to be in yuv420.
8748 It accepts the following values:
8751 cover it by the supplied image
8753 cover it by interpolating the surrounding pixels
8756 Default value is @var{blur}.
8759 @subsection Examples
8763 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8765 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8771 Crop the input video to given dimensions.
8773 It accepts the following parameters:
8777 The width of the output video. It defaults to @code{iw}.
8778 This expression is evaluated only once during the filter
8779 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8782 The height of the output video. It defaults to @code{ih}.
8783 This expression is evaluated only once during the filter
8784 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8787 The horizontal position, in the input video, of the left edge of the output
8788 video. It defaults to @code{(in_w-out_w)/2}.
8789 This expression is evaluated per-frame.
8792 The vertical position, in the input video, of the top edge of the output video.
8793 It defaults to @code{(in_h-out_h)/2}.
8794 This expression is evaluated per-frame.
8797 If set to 1 will force the output display aspect ratio
8798 to be the same of the input, by changing the output sample aspect
8799 ratio. It defaults to 0.
8802 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8803 width/height/x/y as specified and will not be rounded to nearest smaller value.
8807 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8808 expressions containing the following constants:
8813 The computed values for @var{x} and @var{y}. They are evaluated for
8818 The input width and height.
8822 These are the same as @var{in_w} and @var{in_h}.
8826 The output (cropped) width and height.
8830 These are the same as @var{out_w} and @var{out_h}.
8833 same as @var{iw} / @var{ih}
8836 input sample aspect ratio
8839 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8843 horizontal and vertical chroma subsample values. For example for the
8844 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8847 The number of the input frame, starting from 0.
8850 the position in the file of the input frame, NAN if unknown
8853 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8857 The expression for @var{out_w} may depend on the value of @var{out_h},
8858 and the expression for @var{out_h} may depend on @var{out_w}, but they
8859 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8860 evaluated after @var{out_w} and @var{out_h}.
8862 The @var{x} and @var{y} parameters specify the expressions for the
8863 position of the top-left corner of the output (non-cropped) area. They
8864 are evaluated for each frame. If the evaluated value is not valid, it
8865 is approximated to the nearest valid value.
8867 The expression for @var{x} may depend on @var{y}, and the expression
8868 for @var{y} may depend on @var{x}.
8870 @subsection Examples
8874 Crop area with size 100x100 at position (12,34).
8879 Using named options, the example above becomes:
8881 crop=w=100:h=100:x=12:y=34
8885 Crop the central input area with size 100x100:
8891 Crop the central input area with size 2/3 of the input video:
8893 crop=2/3*in_w:2/3*in_h
8897 Crop the input video central square:
8904 Delimit the rectangle with the top-left corner placed at position
8905 100:100 and the right-bottom corner corresponding to the right-bottom
8906 corner of the input image.
8908 crop=in_w-100:in_h-100:100:100
8912 Crop 10 pixels from the left and right borders, and 20 pixels from
8913 the top and bottom borders
8915 crop=in_w-2*10:in_h-2*20
8919 Keep only the bottom right quarter of the input image:
8921 crop=in_w/2:in_h/2:in_w/2:in_h/2
8925 Crop height for getting Greek harmony:
8927 crop=in_w:1/PHI*in_w
8931 Apply trembling effect:
8933 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)
8937 Apply erratic camera effect depending on timestamp:
8939 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)"
8943 Set x depending on the value of y:
8945 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8949 @subsection Commands
8951 This filter supports the following commands:
8957 Set width/height of the output video and the horizontal/vertical position
8959 The command accepts the same syntax of the corresponding option.
8961 If the specified expression is not valid, it is kept at its current
8967 Auto-detect the crop size.
8969 It calculates the necessary cropping parameters and prints the
8970 recommended parameters via the logging system. The detected dimensions
8971 correspond to the non-black area of the input video.
8973 It accepts the following parameters:
8978 Set higher black value threshold, which can be optionally specified
8979 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8980 value greater to the set value is considered non-black. It defaults to 24.
8981 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8982 on the bitdepth of the pixel format.
8985 The value which the width/height should be divisible by. It defaults to
8986 16. The offset is automatically adjusted to center the video. Use 2 to
8987 get only even dimensions (needed for 4:2:2 video). 16 is best when
8988 encoding to most video codecs.
8991 Set the number of initial frames for which evaluation is skipped.
8992 Default is 2. Range is 0 to INT_MAX.
8994 @item reset_count, reset
8995 Set the counter that determines after how many frames cropdetect will
8996 reset the previously detected largest video area and start over to
8997 detect the current optimal crop area. Default value is 0.
8999 This can be useful when channel logos distort the video area. 0
9000 indicates 'never reset', and returns the largest area encountered during
9007 Delay video filtering until a given wallclock timestamp. The filter first
9008 passes on @option{preroll} amount of frames, then it buffers at most
9009 @option{buffer} amount of frames and waits for the cue. After reaching the cue
9010 it forwards the buffered frames and also any subsequent frames coming in its
9013 The filter can be used synchronize the output of multiple ffmpeg processes for
9014 realtime output devices like decklink. By putting the delay in the filtering
9015 chain and pre-buffering frames the process can pass on data to output almost
9016 immediately after the target wallclock timestamp is reached.
9018 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
9024 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
9027 The duration of content to pass on as preroll expressed in seconds. Default is 0.
9030 The maximum duration of content to buffer before waiting for the cue expressed
9031 in seconds. Default is 0.
9038 Apply color adjustments using curves.
9040 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
9041 component (red, green and blue) has its values defined by @var{N} key points
9042 tied from each other using a smooth curve. The x-axis represents the pixel
9043 values from the input frame, and the y-axis the new pixel values to be set for
9046 By default, a component curve is defined by the two points @var{(0;0)} and
9047 @var{(1;1)}. This creates a straight line where each original pixel value is
9048 "adjusted" to its own value, which means no change to the image.
9050 The filter allows you to redefine these two points and add some more. A new
9051 curve (using a natural cubic spline interpolation) will be define to pass
9052 smoothly through all these new coordinates. The new defined points needs to be
9053 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
9054 be in the @var{[0;1]} interval. If the computed curves happened to go outside
9055 the vector spaces, the values will be clipped accordingly.
9057 The filter accepts the following options:
9061 Select one of the available color presets. This option can be used in addition
9062 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
9063 options takes priority on the preset values.
9064 Available presets are:
9067 @item color_negative
9070 @item increase_contrast
9072 @item linear_contrast
9073 @item medium_contrast
9075 @item strong_contrast
9078 Default is @code{none}.
9080 Set the master key points. These points will define a second pass mapping. It
9081 is sometimes called a "luminance" or "value" mapping. It can be used with
9082 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
9083 post-processing LUT.
9085 Set the key points for the red component.
9087 Set the key points for the green component.
9089 Set the key points for the blue component.
9091 Set the key points for all components (not including master).
9092 Can be used in addition to the other key points component
9093 options. In this case, the unset component(s) will fallback on this
9094 @option{all} setting.
9096 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
9098 Save Gnuplot script of the curves in specified file.
9101 To avoid some filtergraph syntax conflicts, each key points list need to be
9102 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
9104 @subsection Examples
9108 Increase slightly the middle level of blue:
9110 curves=blue='0/0 0.5/0.58 1/1'
9116 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'
9118 Here we obtain the following coordinates for each components:
9121 @code{(0;0.11) (0.42;0.51) (1;0.95)}
9123 @code{(0;0) (0.50;0.48) (1;1)}
9125 @code{(0;0.22) (0.49;0.44) (1;0.80)}
9129 The previous example can also be achieved with the associated built-in preset:
9131 curves=preset=vintage
9141 Use a Photoshop preset and redefine the points of the green component:
9143 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
9147 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
9148 and @command{gnuplot}:
9150 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
9151 gnuplot -p /tmp/curves.plt
9157 Video data analysis filter.
9159 This filter shows hexadecimal pixel values of part of video.
9161 The filter accepts the following options:
9165 Set output video size.
9168 Set x offset from where to pick pixels.
9171 Set y offset from where to pick pixels.
9174 Set scope mode, can be one of the following:
9177 Draw hexadecimal pixel values with white color on black background.
9180 Draw hexadecimal pixel values with input video pixel color on black
9184 Draw hexadecimal pixel values on color background picked from input video,
9185 the text color is picked in such way so its always visible.
9189 Draw rows and columns numbers on left and top of video.
9192 Set background opacity.
9195 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
9199 Apply Directional blur filter.
9201 The filter accepts the following options:
9205 Set angle of directional blur. Default is @code{45}.
9208 Set radius of directional blur. Default is @code{5}.
9211 Set which planes to filter. By default all planes are filtered.
9214 @subsection Commands
9215 This filter supports same @ref{commands} as options.
9216 The command accepts the same syntax of the corresponding option.
9218 If the specified expression is not valid, it is kept at its current
9223 Denoise frames using 2D DCT (frequency domain filtering).
9225 This filter is not designed for real time.
9227 The filter accepts the following options:
9231 Set the noise sigma constant.
9233 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
9234 coefficient (absolute value) below this threshold with be dropped.
9236 If you need a more advanced filtering, see @option{expr}.
9238 Default is @code{0}.
9241 Set number overlapping pixels for each block. Since the filter can be slow, you
9242 may want to reduce this value, at the cost of a less effective filter and the
9243 risk of various artefacts.
9245 If the overlapping value doesn't permit processing the whole input width or
9246 height, a warning will be displayed and according borders won't be denoised.
9248 Default value is @var{blocksize}-1, which is the best possible setting.
9251 Set the coefficient factor expression.
9253 For each coefficient of a DCT block, this expression will be evaluated as a
9254 multiplier value for the coefficient.
9256 If this is option is set, the @option{sigma} option will be ignored.
9258 The absolute value of the coefficient can be accessed through the @var{c}
9262 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
9263 @var{blocksize}, which is the width and height of the processed blocks.
9265 The default value is @var{3} (8x8) and can be raised to @var{4} for a
9266 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
9267 on the speed processing. Also, a larger block size does not necessarily means a
9271 @subsection Examples
9273 Apply a denoise with a @option{sigma} of @code{4.5}:
9278 The same operation can be achieved using the expression system:
9280 dctdnoiz=e='gte(c, 4.5*3)'
9283 Violent denoise using a block size of @code{16x16}:
9290 Remove banding artifacts from input video.
9291 It works by replacing banded pixels with average value of referenced pixels.
9293 The filter accepts the following options:
9300 Set banding detection threshold for each plane. Default is 0.02.
9301 Valid range is 0.00003 to 0.5.
9302 If difference between current pixel and reference pixel is less than threshold,
9303 it will be considered as banded.
9306 Banding detection range in pixels. Default is 16. If positive, random number
9307 in range 0 to set value will be used. If negative, exact absolute value
9309 The range defines square of four pixels around current pixel.
9312 Set direction in radians from which four pixel will be compared. If positive,
9313 random direction from 0 to set direction will be picked. If negative, exact of
9314 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9315 will pick only pixels on same row and -PI/2 will pick only pixels on same
9319 If enabled, current pixel is compared with average value of all four
9320 surrounding pixels. The default is enabled. If disabled current pixel is
9321 compared with all four surrounding pixels. The pixel is considered banded
9322 if only all four differences with surrounding pixels are less than threshold.
9325 If enabled, current pixel is changed if and only if all pixel components are banded,
9326 e.g. banding detection threshold is triggered for all color components.
9327 The default is disabled.
9332 Remove blocking artifacts from input video.
9334 The filter accepts the following options:
9338 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9339 This controls what kind of deblocking is applied.
9342 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9348 Set blocking detection thresholds. Allowed range is 0 to 1.
9349 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9350 Using higher threshold gives more deblocking strength.
9351 Setting @var{alpha} controls threshold detection at exact edge of block.
9352 Remaining options controls threshold detection near the edge. Each one for
9353 below/above or left/right. Setting any of those to @var{0} disables
9357 Set planes to filter. Default is to filter all available planes.
9360 @subsection Examples
9364 Deblock using weak filter and block size of 4 pixels.
9366 deblock=filter=weak:block=4
9370 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9371 deblocking more edges.
9373 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9377 Similar as above, but filter only first plane.
9379 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9383 Similar as above, but filter only second and third plane.
9385 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9392 Drop duplicated frames at regular intervals.
9394 The filter accepts the following options:
9398 Set the number of frames from which one will be dropped. Setting this to
9399 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9400 Default is @code{5}.
9403 Set the threshold for duplicate detection. If the difference metric for a frame
9404 is less than or equal to this value, then it is declared as duplicate. Default
9408 Set scene change threshold. Default is @code{15}.
9412 Set the size of the x and y-axis blocks used during metric calculations.
9413 Larger blocks give better noise suppression, but also give worse detection of
9414 small movements. Must be a power of two. Default is @code{32}.
9417 Mark main input as a pre-processed input and activate clean source input
9418 stream. This allows the input to be pre-processed with various filters to help
9419 the metrics calculation while keeping the frame selection lossless. When set to
9420 @code{1}, the first stream is for the pre-processed input, and the second
9421 stream is the clean source from where the kept frames are chosen. Default is
9425 Set whether or not chroma is considered in the metric calculations. Default is
9431 Apply 2D deconvolution of video stream in frequency domain using second stream
9434 The filter accepts the following options:
9438 Set which planes to process.
9441 Set which impulse video frames will be processed, can be @var{first}
9442 or @var{all}. Default is @var{all}.
9445 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9446 and height are not same and not power of 2 or if stream prior to convolving
9450 The @code{deconvolve} filter also supports the @ref{framesync} options.
9454 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9456 It accepts the following options:
9460 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9461 @var{rainbows} for cross-color reduction.
9464 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9467 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9470 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9473 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9478 Apply deflate effect to the video.
9480 This filter replaces the pixel by the local(3x3) average by taking into account
9481 only values lower than the pixel.
9483 It accepts the following options:
9490 Limit the maximum change for each plane, default is 65535.
9491 If 0, plane will remain unchanged.
9494 @subsection Commands
9496 This filter supports the all above options as @ref{commands}.
9500 Remove temporal frame luminance variations.
9502 It accepts the following options:
9506 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9509 Set averaging mode to smooth temporal luminance variations.
9511 Available values are:
9536 Do not actually modify frame. Useful when one only wants metadata.
9541 Remove judder produced by partially interlaced telecined content.
9543 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9544 source was partially telecined content then the output of @code{pullup,dejudder}
9545 will have a variable frame rate. May change the recorded frame rate of the
9546 container. Aside from that change, this filter will not affect constant frame
9549 The option available in this filter is:
9553 Specify the length of the window over which the judder repeats.
9555 Accepts any integer greater than 1. Useful values are:
9559 If the original was telecined from 24 to 30 fps (Film to NTSC).
9562 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9565 If a mixture of the two.
9568 The default is @samp{4}.
9573 Suppress a TV station logo by a simple interpolation of the surrounding
9574 pixels. Just set a rectangle covering the logo and watch it disappear
9575 (and sometimes something even uglier appear - your mileage may vary).
9577 It accepts the following parameters:
9582 Specify the top left corner coordinates of the logo. They must be
9587 Specify the width and height of the logo to clear. They must be
9591 Specify the thickness of the fuzzy edge of the rectangle (added to
9592 @var{w} and @var{h}). The default value is 1. This option is
9593 deprecated, setting higher values should no longer be necessary and
9597 When set to 1, a green rectangle is drawn on the screen to simplify
9598 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9599 The default value is 0.
9601 The rectangle is drawn on the outermost pixels which will be (partly)
9602 replaced with interpolated values. The values of the next pixels
9603 immediately outside this rectangle in each direction will be used to
9604 compute the interpolated pixel values inside the rectangle.
9608 @subsection Examples
9612 Set a rectangle covering the area with top left corner coordinates 0,0
9613 and size 100x77, and a band of size 10:
9615 delogo=x=0:y=0:w=100:h=77:band=10
9623 Remove the rain in the input image/video by applying the derain methods based on
9624 convolutional neural networks. Supported models:
9628 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9629 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9632 Training as well as model generation scripts are provided in
9633 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9635 Native model files (.model) can be generated from TensorFlow model
9636 files (.pb) by using tools/python/convert.py
9638 The filter accepts the following options:
9642 Specify which filter to use. This option accepts the following values:
9646 Derain filter. To conduct derain filter, you need to use a derain model.
9649 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9651 Default value is @samp{derain}.
9654 Specify which DNN backend to use for model loading and execution. This option accepts
9655 the following values:
9659 Native implementation of DNN loading and execution.
9662 TensorFlow backend. To enable this backend you
9663 need to install the TensorFlow for C library (see
9664 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9665 @code{--enable-libtensorflow}
9667 Default value is @samp{native}.
9670 Set path to model file specifying network architecture and its parameters.
9671 Note that different backends use different file formats. TensorFlow and native
9672 backend can load files for only its format.
9675 It can also be finished with @ref{dnn_processing} filter.
9679 Attempt to fix small changes in horizontal and/or vertical shift. This
9680 filter helps remove camera shake from hand-holding a camera, bumping a
9681 tripod, moving on a vehicle, etc.
9683 The filter accepts the following options:
9691 Specify a rectangular area where to limit the search for motion
9693 If desired the search for motion vectors can be limited to a
9694 rectangular area of the frame defined by its top left corner, width
9695 and height. These parameters have the same meaning as the drawbox
9696 filter which can be used to visualise the position of the bounding
9699 This is useful when simultaneous movement of subjects within the frame
9700 might be confused for camera motion by the motion vector search.
9702 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9703 then the full frame is used. This allows later options to be set
9704 without specifying the bounding box for the motion vector search.
9706 Default - search the whole frame.
9710 Specify the maximum extent of movement in x and y directions in the
9711 range 0-64 pixels. Default 16.
9714 Specify how to generate pixels to fill blanks at the edge of the
9715 frame. Available values are:
9718 Fill zeroes at blank locations
9720 Original image at blank locations
9722 Extruded edge value at blank locations
9724 Mirrored edge at blank locations
9726 Default value is @samp{mirror}.
9729 Specify the blocksize to use for motion search. Range 4-128 pixels,
9733 Specify the contrast threshold for blocks. Only blocks with more than
9734 the specified contrast (difference between darkest and lightest
9735 pixels) will be considered. Range 1-255, default 125.
9738 Specify the search strategy. Available values are:
9741 Set exhaustive search
9743 Set less exhaustive search.
9745 Default value is @samp{exhaustive}.
9748 If set then a detailed log of the motion search is written to the
9755 Remove unwanted contamination of foreground colors, caused by reflected color of
9756 greenscreen or bluescreen.
9758 This filter accepts the following options:
9762 Set what type of despill to use.
9765 Set how spillmap will be generated.
9768 Set how much to get rid of still remaining spill.
9771 Controls amount of red in spill area.
9774 Controls amount of green in spill area.
9775 Should be -1 for greenscreen.
9778 Controls amount of blue in spill area.
9779 Should be -1 for bluescreen.
9782 Controls brightness of spill area, preserving colors.
9785 Modify alpha from generated spillmap.
9788 @subsection Commands
9790 This filter supports the all above options as @ref{commands}.
9794 Apply an exact inverse of the telecine operation. It requires a predefined
9795 pattern specified using the pattern option which must be the same as that passed
9796 to the telecine filter.
9798 This filter accepts the following options:
9807 The default value is @code{top}.
9811 A string of numbers representing the pulldown pattern you wish to apply.
9812 The default value is @code{23}.
9815 A number representing position of the first frame with respect to the telecine
9816 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9821 Apply dilation effect to the video.
9823 This filter replaces the pixel by the local(3x3) maximum.
9825 It accepts the following options:
9832 Limit the maximum change for each plane, default is 65535.
9833 If 0, plane will remain unchanged.
9836 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9839 Flags to local 3x3 coordinates maps like this:
9846 @subsection Commands
9848 This filter supports the all above options as @ref{commands}.
9852 Displace pixels as indicated by second and third input stream.
9854 It takes three input streams and outputs one stream, the first input is the
9855 source, and second and third input are displacement maps.
9857 The second input specifies how much to displace pixels along the
9858 x-axis, while the third input specifies how much to displace pixels
9860 If one of displacement map streams terminates, last frame from that
9861 displacement map will be used.
9863 Note that once generated, displacements maps can be reused over and over again.
9865 A description of the accepted options follows.
9869 Set displace behavior for pixels that are out of range.
9871 Available values are:
9874 Missing pixels are replaced by black pixels.
9877 Adjacent pixels will spread out to replace missing pixels.
9880 Out of range pixels are wrapped so they point to pixels of other side.
9883 Out of range pixels will be replaced with mirrored pixels.
9885 Default is @samp{smear}.
9889 @subsection Examples
9893 Add ripple effect to rgb input of video size hd720:
9895 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
9899 Add wave effect to rgb input of video size hd720:
9901 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
9905 @anchor{dnn_processing}
9906 @section dnn_processing
9908 Do image processing with deep neural networks. It works together with another filter
9909 which converts the pixel format of the Frame to what the dnn network requires.
9911 The filter accepts the following options:
9915 Specify which DNN backend to use for model loading and execution. This option accepts
9916 the following values:
9920 Native implementation of DNN loading and execution.
9923 TensorFlow backend. To enable this backend you
9924 need to install the TensorFlow for C library (see
9925 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9926 @code{--enable-libtensorflow}
9929 OpenVINO backend. To enable this backend you
9930 need to build and install the OpenVINO for C library (see
9931 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9932 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9933 be needed if the header files and libraries are not installed into system path)
9937 Default value is @samp{native}.
9940 Set path to model file specifying network architecture and its parameters.
9941 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9942 backend can load files for only its format.
9944 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9947 Set the input name of the dnn network.
9950 Set the output name of the dnn network.
9954 @subsection Examples
9958 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9960 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9964 Halve the pixel value of the frame with format gray32f:
9966 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
9970 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9972 ./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
9976 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9978 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9985 Draw a colored box on the input image.
9987 It accepts the following parameters:
9992 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9996 The expressions which specify the width and height of the box; if 0 they are interpreted as
9997 the input width and height. It defaults to 0.
10000 Specify the color of the box to write. For the general syntax of this option,
10001 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10002 value @code{invert} is used, the box edge color is the same as the
10003 video with inverted luma.
10006 The expression which sets the thickness of the box edge.
10007 A value of @code{fill} will create a filled box. Default value is @code{3}.
10009 See below for the list of accepted constants.
10012 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
10013 will overwrite the video's color and alpha pixels.
10014 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
10017 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10018 following constants:
10022 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10026 horizontal and vertical chroma subsample values. For example for the
10027 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10031 The input width and height.
10034 The input sample aspect ratio.
10038 The x and y offset coordinates where the box is drawn.
10042 The width and height of the drawn box.
10045 The thickness of the drawn box.
10047 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10048 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10052 @subsection Examples
10056 Draw a black box around the edge of the input image:
10062 Draw a box with color red and an opacity of 50%:
10064 drawbox=10:20:200:60:red@@0.5
10067 The previous example can be specified as:
10069 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
10073 Fill the box with pink color:
10075 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
10079 Draw a 2-pixel red 2.40:1 mask:
10081 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
10085 @subsection Commands
10086 This filter supports same commands as options.
10087 The command accepts the same syntax of the corresponding option.
10089 If the specified expression is not valid, it is kept at its current
10094 Draw a graph using input video metadata.
10096 It accepts the following parameters:
10100 Set 1st frame metadata key from which metadata values will be used to draw a graph.
10103 Set 1st foreground color expression.
10106 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
10109 Set 2nd foreground color expression.
10112 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
10115 Set 3rd foreground color expression.
10118 Set 4th frame metadata key from which metadata values will be used to draw a graph.
10121 Set 4th foreground color expression.
10124 Set minimal value of metadata value.
10127 Set maximal value of metadata value.
10130 Set graph background color. Default is white.
10135 Available values for mode is:
10142 Default is @code{line}.
10147 Available values for slide is:
10150 Draw new frame when right border is reached.
10153 Replace old columns with new ones.
10156 Scroll from right to left.
10159 Scroll from left to right.
10162 Draw single picture.
10165 Default is @code{frame}.
10168 Set size of graph video. For the syntax of this option, check the
10169 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
10170 The default value is @code{900x256}.
10173 Set the output frame rate. Default value is @code{25}.
10175 The foreground color expressions can use the following variables:
10178 Minimal value of metadata value.
10181 Maximal value of metadata value.
10184 Current metadata key value.
10187 The color is defined as 0xAABBGGRR.
10190 Example using metadata from @ref{signalstats} filter:
10192 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
10195 Example using metadata from @ref{ebur128} filter:
10197 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
10202 Draw a grid on the input image.
10204 It accepts the following parameters:
10209 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
10213 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
10214 input width and height, respectively, minus @code{thickness}, so image gets
10215 framed. Default to 0.
10218 Specify the color of the grid. For the general syntax of this option,
10219 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10220 value @code{invert} is used, the grid color is the same as the
10221 video with inverted luma.
10224 The expression which sets the thickness of the grid line. Default value is @code{1}.
10226 See below for the list of accepted constants.
10229 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
10230 will overwrite the video's color and alpha pixels.
10231 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
10234 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10235 following constants:
10239 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10243 horizontal and vertical chroma subsample values. For example for the
10244 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10248 The input grid cell width and height.
10251 The input sample aspect ratio.
10255 The x and y coordinates of some point of grid intersection (meant to configure offset).
10259 The width and height of the drawn cell.
10262 The thickness of the drawn cell.
10264 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10265 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10269 @subsection Examples
10273 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
10275 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
10279 Draw a white 3x3 grid with an opacity of 50%:
10281 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10285 @subsection Commands
10286 This filter supports same commands as options.
10287 The command accepts the same syntax of the corresponding option.
10289 If the specified expression is not valid, it is kept at its current
10295 Draw a text string or text from a specified file on top of a video, using the
10296 libfreetype library.
10298 To enable compilation of this filter, you need to configure FFmpeg with
10299 @code{--enable-libfreetype}.
10300 To enable default font fallback and the @var{font} option you need to
10301 configure FFmpeg with @code{--enable-libfontconfig}.
10302 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10303 @code{--enable-libfribidi}.
10307 It accepts the following parameters:
10312 Used to draw a box around text using the background color.
10313 The value must be either 1 (enable) or 0 (disable).
10314 The default value of @var{box} is 0.
10317 Set the width of the border to be drawn around the box using @var{boxcolor}.
10318 The default value of @var{boxborderw} is 0.
10321 The color to be used for drawing box around text. For the syntax of this
10322 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10324 The default value of @var{boxcolor} is "white".
10327 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10328 The default value of @var{line_spacing} is 0.
10331 Set the width of the border to be drawn around the text using @var{bordercolor}.
10332 The default value of @var{borderw} is 0.
10335 Set the color to be used for drawing border around text. For the syntax of this
10336 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10338 The default value of @var{bordercolor} is "black".
10341 Select how the @var{text} is expanded. Can be either @code{none},
10342 @code{strftime} (deprecated) or
10343 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10347 Set a start time for the count. Value is in microseconds. Only applied
10348 in the deprecated strftime expansion mode. To emulate in normal expansion
10349 mode use the @code{pts} function, supplying the start time (in seconds)
10350 as the second argument.
10353 If true, check and fix text coords to avoid clipping.
10356 The color to be used for drawing fonts. For the syntax of this option, check
10357 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10359 The default value of @var{fontcolor} is "black".
10361 @item fontcolor_expr
10362 String which is expanded the same way as @var{text} to obtain dynamic
10363 @var{fontcolor} value. By default this option has empty value and is not
10364 processed. When this option is set, it overrides @var{fontcolor} option.
10367 The font family to be used for drawing text. By default Sans.
10370 The font file to be used for drawing text. The path must be included.
10371 This parameter is mandatory if the fontconfig support is disabled.
10374 Draw the text applying alpha blending. The value can
10375 be a number between 0.0 and 1.0.
10376 The expression accepts the same variables @var{x, y} as well.
10377 The default value is 1.
10378 Please see @var{fontcolor_expr}.
10381 The font size to be used for drawing text.
10382 The default value of @var{fontsize} is 16.
10385 If set to 1, attempt to shape the text (for example, reverse the order of
10386 right-to-left text and join Arabic characters) before drawing it.
10387 Otherwise, just draw the text exactly as given.
10388 By default 1 (if supported).
10390 @item ft_load_flags
10391 The flags to be used for loading the fonts.
10393 The flags map the corresponding flags supported by libfreetype, and are
10394 a combination of the following values:
10401 @item vertical_layout
10402 @item force_autohint
10405 @item ignore_global_advance_width
10407 @item ignore_transform
10409 @item linear_design
10413 Default value is "default".
10415 For more information consult the documentation for the FT_LOAD_*
10419 The color to be used for drawing a shadow behind the drawn text. For the
10420 syntax of this option, check the @ref{color syntax,,"Color" section in the
10421 ffmpeg-utils manual,ffmpeg-utils}.
10423 The default value of @var{shadowcolor} is "black".
10427 The x and y offsets for the text shadow position with respect to the
10428 position of the text. They can be either positive or negative
10429 values. The default value for both is "0".
10432 The starting frame number for the n/frame_num variable. The default value
10436 The size in number of spaces to use for rendering the tab.
10437 Default value is 4.
10440 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10441 format. It can be used with or without text parameter. @var{timecode_rate}
10442 option must be specified.
10444 @item timecode_rate, rate, r
10445 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10446 integer. Minimum value is "1".
10447 Drop-frame timecode is supported for frame rates 30 & 60.
10450 If set to 1, the output of the timecode option will wrap around at 24 hours.
10451 Default is 0 (disabled).
10454 The text string to be drawn. The text must be a sequence of UTF-8
10455 encoded characters.
10456 This parameter is mandatory if no file is specified with the parameter
10460 A text file containing text to be drawn. The text must be a sequence
10461 of UTF-8 encoded characters.
10463 This parameter is mandatory if no text string is specified with the
10464 parameter @var{text}.
10466 If both @var{text} and @var{textfile} are specified, an error is thrown.
10469 If set to 1, the @var{textfile} will be reloaded before each frame.
10470 Be sure to update it atomically, or it may be read partially, or even fail.
10474 The expressions which specify the offsets where text will be drawn
10475 within the video frame. They are relative to the top/left border of the
10478 The default value of @var{x} and @var{y} is "0".
10480 See below for the list of accepted constants and functions.
10483 The parameters for @var{x} and @var{y} are expressions containing the
10484 following constants and functions:
10488 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10492 horizontal and vertical chroma subsample values. For example for the
10493 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10496 the height of each text line
10504 @item max_glyph_a, ascent
10505 the maximum distance from the baseline to the highest/upper grid
10506 coordinate used to place a glyph outline point, for all the rendered
10508 It is a positive value, due to the grid's orientation with the Y axis
10511 @item max_glyph_d, descent
10512 the maximum distance from the baseline to the lowest grid coordinate
10513 used to place a glyph outline point, for all the rendered glyphs.
10514 This is a negative value, due to the grid's orientation, with the Y axis
10518 maximum glyph height, that is the maximum height for all the glyphs
10519 contained in the rendered text, it is equivalent to @var{ascent} -
10523 maximum glyph width, that is the maximum width for all the glyphs
10524 contained in the rendered text
10527 the number of input frame, starting from 0
10529 @item rand(min, max)
10530 return a random number included between @var{min} and @var{max}
10533 The input sample aspect ratio.
10536 timestamp expressed in seconds, NAN if the input timestamp is unknown
10539 the height of the rendered text
10542 the width of the rendered text
10546 the x and y offset coordinates where the text is drawn.
10548 These parameters allow the @var{x} and @var{y} expressions to refer
10549 to each other, so you can for example specify @code{y=x/dar}.
10552 A one character description of the current frame's picture type.
10555 The current packet's position in the input file or stream
10556 (in bytes, from the start of the input). A value of -1 indicates
10557 this info is not available.
10560 The current packet's duration, in seconds.
10563 The current packet's size (in bytes).
10566 @anchor{drawtext_expansion}
10567 @subsection Text expansion
10569 If @option{expansion} is set to @code{strftime},
10570 the filter recognizes strftime() sequences in the provided text and
10571 expands them accordingly. Check the documentation of strftime(). This
10572 feature is deprecated.
10574 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10576 If @option{expansion} is set to @code{normal} (which is the default),
10577 the following expansion mechanism is used.
10579 The backslash character @samp{\}, followed by any character, always expands to
10580 the second character.
10582 Sequences of the form @code{%@{...@}} are expanded. The text between the
10583 braces is a function name, possibly followed by arguments separated by ':'.
10584 If the arguments contain special characters or delimiters (':' or '@}'),
10585 they should be escaped.
10587 Note that they probably must also be escaped as the value for the
10588 @option{text} option in the filter argument string and as the filter
10589 argument in the filtergraph description, and possibly also for the shell,
10590 that makes up to four levels of escaping; using a text file avoids these
10593 The following functions are available:
10598 The expression evaluation result.
10600 It must take one argument specifying the expression to be evaluated,
10601 which accepts the same constants and functions as the @var{x} and
10602 @var{y} values. Note that not all constants should be used, for
10603 example the text size is not known when evaluating the expression, so
10604 the constants @var{text_w} and @var{text_h} will have an undefined
10607 @item expr_int_format, eif
10608 Evaluate the expression's value and output as formatted integer.
10610 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10611 The second argument specifies the output format. Allowed values are @samp{x},
10612 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10613 @code{printf} function.
10614 The third parameter is optional and sets the number of positions taken by the output.
10615 It can be used to add padding with zeros from the left.
10618 The time at which the filter is running, expressed in UTC.
10619 It can accept an argument: a strftime() format string.
10622 The time at which the filter is running, expressed in the local time zone.
10623 It can accept an argument: a strftime() format string.
10626 Frame metadata. Takes one or two arguments.
10628 The first argument is mandatory and specifies the metadata key.
10630 The second argument is optional and specifies a default value, used when the
10631 metadata key is not found or empty.
10633 Available metadata can be identified by inspecting entries
10634 starting with TAG included within each frame section
10635 printed by running @code{ffprobe -show_frames}.
10637 String metadata generated in filters leading to
10638 the drawtext filter are also available.
10641 The frame number, starting from 0.
10644 A one character description of the current picture type.
10647 The timestamp of the current frame.
10648 It can take up to three arguments.
10650 The first argument is the format of the timestamp; it defaults to @code{flt}
10651 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10652 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10653 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10654 @code{localtime} stands for the timestamp of the frame formatted as
10655 local time zone time.
10657 The second argument is an offset added to the timestamp.
10659 If the format is set to @code{hms}, a third argument @code{24HH} may be
10660 supplied to present the hour part of the formatted timestamp in 24h format
10663 If the format is set to @code{localtime} or @code{gmtime},
10664 a third argument may be supplied: a strftime() format string.
10665 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10668 @subsection Commands
10670 This filter supports altering parameters via commands:
10673 Alter existing filter parameters.
10675 Syntax for the argument is the same as for filter invocation, e.g.
10678 fontsize=56:fontcolor=green:text='Hello World'
10681 Full filter invocation with sendcmd would look like this:
10684 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10688 If the entire argument can't be parsed or applied as valid values then the filter will
10689 continue with its existing parameters.
10691 @subsection Examples
10695 Draw "Test Text" with font FreeSerif, using the default values for the
10696 optional parameters.
10699 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10703 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10704 and y=50 (counting from the top-left corner of the screen), text is
10705 yellow with a red box around it. Both the text and the box have an
10709 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10710 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10713 Note that the double quotes are not necessary if spaces are not used
10714 within the parameter list.
10717 Show the text at the center of the video frame:
10719 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10723 Show the text at a random position, switching to a new position every 30 seconds:
10725 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)"
10729 Show a text line sliding from right to left in the last row of the video
10730 frame. The file @file{LONG_LINE} is assumed to contain a single line
10733 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10737 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10739 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10743 Draw a single green letter "g", at the center of the input video.
10744 The glyph baseline is placed at half screen height.
10746 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10750 Show text for 1 second every 3 seconds:
10752 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10756 Use fontconfig to set the font. Note that the colons need to be escaped.
10758 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10762 Draw "Test Text" with font size dependent on height of the video.
10764 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10768 Print the date of a real-time encoding (see strftime(3)):
10770 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10774 Show text fading in and out (appearing/disappearing):
10777 DS=1.0 # display start
10778 DE=10.0 # display end
10779 FID=1.5 # fade in duration
10780 FOD=5 # fade out duration
10781 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 @}"
10785 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10786 and the @option{fontsize} value are included in the @option{y} offset.
10788 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10789 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10793 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10794 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10795 must have option @option{-export_path_metadata 1} for the special metadata fields
10796 to be available for filters.
10798 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10803 For more information about libfreetype, check:
10804 @url{http://www.freetype.org/}.
10806 For more information about fontconfig, check:
10807 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10809 For more information about libfribidi, check:
10810 @url{http://fribidi.org/}.
10812 @section edgedetect
10814 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10816 The filter accepts the following options:
10821 Set low and high threshold values used by the Canny thresholding
10824 The high threshold selects the "strong" edge pixels, which are then
10825 connected through 8-connectivity with the "weak" edge pixels selected
10826 by the low threshold.
10828 @var{low} and @var{high} threshold values must be chosen in the range
10829 [0,1], and @var{low} should be lesser or equal to @var{high}.
10831 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10835 Define the drawing mode.
10839 Draw white/gray wires on black background.
10842 Mix the colors to create a paint/cartoon effect.
10845 Apply Canny edge detector on all selected planes.
10847 Default value is @var{wires}.
10850 Select planes for filtering. By default all available planes are filtered.
10853 @subsection Examples
10857 Standard edge detection with custom values for the hysteresis thresholding:
10859 edgedetect=low=0.1:high=0.4
10863 Painting effect without thresholding:
10865 edgedetect=mode=colormix:high=0
10871 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10873 For each input image, the filter will compute the optimal mapping from
10874 the input to the output given the codebook length, that is the number
10875 of distinct output colors.
10877 This filter accepts the following options.
10880 @item codebook_length, l
10881 Set codebook length. The value must be a positive integer, and
10882 represents the number of distinct output colors. Default value is 256.
10885 Set the maximum number of iterations to apply for computing the optimal
10886 mapping. The higher the value the better the result and the higher the
10887 computation time. Default value is 1.
10890 Set a random seed, must be an integer included between 0 and
10891 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10892 will try to use a good random seed on a best effort basis.
10895 Set pal8 output pixel format. This option does not work with codebook
10896 length greater than 256.
10901 Measure graylevel entropy in histogram of color channels of video frames.
10903 It accepts the following parameters:
10907 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10909 @var{diff} mode measures entropy of histogram delta values, absolute differences
10910 between neighbour histogram values.
10914 Set brightness, contrast, saturation and approximate gamma adjustment.
10916 The filter accepts the following options:
10920 Set the contrast expression. The value must be a float value in range
10921 @code{-1000.0} to @code{1000.0}. The default value is "1".
10924 Set the brightness expression. The value must be a float value in
10925 range @code{-1.0} to @code{1.0}. The default value is "0".
10928 Set the saturation expression. The value must be a float in
10929 range @code{0.0} to @code{3.0}. The default value is "1".
10932 Set the gamma expression. The value must be a float in range
10933 @code{0.1} to @code{10.0}. The default value is "1".
10936 Set the gamma expression for red. The value must be a float in
10937 range @code{0.1} to @code{10.0}. The default value is "1".
10940 Set the gamma expression for green. The value must be a float in range
10941 @code{0.1} to @code{10.0}. The default value is "1".
10944 Set the gamma expression for blue. The value must be a float in range
10945 @code{0.1} to @code{10.0}. The default value is "1".
10948 Set the gamma weight expression. It can be used to reduce the effect
10949 of a high gamma value on bright image areas, e.g. keep them from
10950 getting overamplified and just plain white. The value must be a float
10951 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10952 gamma correction all the way down while @code{1.0} leaves it at its
10953 full strength. Default is "1".
10956 Set when the expressions for brightness, contrast, saturation and
10957 gamma expressions are evaluated.
10959 It accepts the following values:
10962 only evaluate expressions once during the filter initialization or
10963 when a command is processed
10966 evaluate expressions for each incoming frame
10969 Default value is @samp{init}.
10972 The expressions accept the following parameters:
10975 frame count of the input frame starting from 0
10978 byte position of the corresponding packet in the input file, NAN if
10982 frame rate of the input video, NAN if the input frame rate is unknown
10985 timestamp expressed in seconds, NAN if the input timestamp is unknown
10988 @subsection Commands
10989 The filter supports the following commands:
10993 Set the contrast expression.
10996 Set the brightness expression.
10999 Set the saturation expression.
11002 Set the gamma expression.
11005 Set the gamma_r expression.
11008 Set gamma_g expression.
11011 Set gamma_b expression.
11014 Set gamma_weight expression.
11016 The command accepts the same syntax of the corresponding option.
11018 If the specified expression is not valid, it is kept at its current
11025 Apply erosion effect to the video.
11027 This filter replaces the pixel by the local(3x3) minimum.
11029 It accepts the following options:
11036 Limit the maximum change for each plane, default is 65535.
11037 If 0, plane will remain unchanged.
11040 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
11043 Flags to local 3x3 coordinates maps like this:
11050 @subsection Commands
11052 This filter supports the all above options as @ref{commands}.
11054 @section extractplanes
11056 Extract color channel components from input video stream into
11057 separate grayscale video streams.
11059 The filter accepts the following option:
11063 Set plane(s) to extract.
11065 Available values for planes are:
11076 Choosing planes not available in the input will result in an error.
11077 That means you cannot select @code{r}, @code{g}, @code{b} planes
11078 with @code{y}, @code{u}, @code{v} planes at same time.
11081 @subsection Examples
11085 Extract luma, u and v color channel component from input video frame
11086 into 3 grayscale outputs:
11088 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
11094 Apply a fade-in/out effect to the input video.
11096 It accepts the following parameters:
11100 The effect type can be either "in" for a fade-in, or "out" for a fade-out
11102 Default is @code{in}.
11104 @item start_frame, s
11105 Specify the number of the frame to start applying the fade
11106 effect at. Default is 0.
11109 The number of frames that the fade effect lasts. At the end of the
11110 fade-in effect, the output video will have the same intensity as the input video.
11111 At the end of the fade-out transition, the output video will be filled with the
11112 selected @option{color}.
11116 If set to 1, fade only alpha channel, if one exists on the input.
11117 Default value is 0.
11119 @item start_time, st
11120 Specify the timestamp (in seconds) of the frame to start to apply the fade
11121 effect. If both start_frame and start_time are specified, the fade will start at
11122 whichever comes last. Default is 0.
11125 The number of seconds for which the fade effect has to last. At the end of the
11126 fade-in effect the output video will have the same intensity as the input video,
11127 at the end of the fade-out transition the output video will be filled with the
11128 selected @option{color}.
11129 If both duration and nb_frames are specified, duration is used. Default is 0
11130 (nb_frames is used by default).
11133 Specify the color of the fade. Default is "black".
11136 @subsection Examples
11140 Fade in the first 30 frames of video:
11145 The command above is equivalent to:
11151 Fade out the last 45 frames of a 200-frame video:
11154 fade=type=out:start_frame=155:nb_frames=45
11158 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
11160 fade=in:0:25, fade=out:975:25
11164 Make the first 5 frames yellow, then fade in from frame 5-24:
11166 fade=in:5:20:color=yellow
11170 Fade in alpha over first 25 frames of video:
11172 fade=in:0:25:alpha=1
11176 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
11178 fade=t=in:st=5.5:d=0.5
11184 Denoise frames using 3D FFT (frequency domain filtering).
11186 The filter accepts the following options:
11190 Set the noise sigma constant. This sets denoising strength.
11191 Default value is 1. Allowed range is from 0 to 30.
11192 Using very high sigma with low overlap may give blocking artifacts.
11195 Set amount of denoising. By default all detected noise is reduced.
11196 Default value is 1. Allowed range is from 0 to 1.
11199 Set size of block, Default is 4, can be 3, 4, 5 or 6.
11200 Actual size of block in pixels is 2 to power of @var{block}, so by default
11201 block size in pixels is 2^4 which is 16.
11204 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
11207 Set number of previous frames to use for denoising. By default is set to 0.
11210 Set number of next frames to to use for denoising. By default is set to 0.
11213 Set planes which will be filtered, by default are all available filtered
11218 Apply arbitrary expressions to samples in frequency domain
11222 Adjust the dc value (gain) of the luma plane of the image. The filter
11223 accepts an integer value in range @code{0} to @code{1000}. The default
11224 value is set to @code{0}.
11227 Adjust the dc value (gain) of the 1st chroma plane of the image. The
11228 filter accepts an integer value in range @code{0} to @code{1000}. The
11229 default value is set to @code{0}.
11232 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
11233 filter accepts an integer value in range @code{0} to @code{1000}. The
11234 default value is set to @code{0}.
11237 Set the frequency domain weight expression for the luma plane.
11240 Set the frequency domain weight expression for the 1st chroma plane.
11243 Set the frequency domain weight expression for the 2nd chroma plane.
11246 Set when the expressions are evaluated.
11248 It accepts the following values:
11251 Only evaluate expressions once during the filter initialization.
11254 Evaluate expressions for each incoming frame.
11257 Default value is @samp{init}.
11259 The filter accepts the following variables:
11262 The coordinates of the current sample.
11266 The width and height of the image.
11269 The number of input frame, starting from 0.
11272 @subsection Examples
11278 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11284 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11290 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11296 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11303 Extract a single field from an interlaced image using stride
11304 arithmetic to avoid wasting CPU time. The output frames are marked as
11307 The filter accepts the following options:
11311 Specify whether to extract the top (if the value is @code{0} or
11312 @code{top}) or the bottom field (if the value is @code{1} or
11318 Create new frames by copying the top and bottom fields from surrounding frames
11319 supplied as numbers by the hint file.
11323 Set file containing hints: absolute/relative frame numbers.
11325 There must be one line for each frame in a clip. Each line must contain two
11326 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11327 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11328 is current frame number for @code{absolute} mode or out of [-1, 1] range
11329 for @code{relative} mode. First number tells from which frame to pick up top
11330 field and second number tells from which frame to pick up bottom field.
11332 If optionally followed by @code{+} output frame will be marked as interlaced,
11333 else if followed by @code{-} output frame will be marked as progressive, else
11334 it will be marked same as input frame.
11335 If optionally followed by @code{t} output frame will use only top field, or in
11336 case of @code{b} it will use only bottom field.
11337 If line starts with @code{#} or @code{;} that line is skipped.
11340 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11343 Example of first several lines of @code{hint} file for @code{relative} mode:
11345 0,0 - # first frame
11346 1,0 - # second frame, use third's frame top field and second's frame bottom field
11347 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11362 @section fieldmatch
11364 Field matching filter for inverse telecine. It is meant to reconstruct the
11365 progressive frames from a telecined stream. The filter does not drop duplicated
11366 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11367 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11369 The separation of the field matching and the decimation is notably motivated by
11370 the possibility of inserting a de-interlacing filter fallback between the two.
11371 If the source has mixed telecined and real interlaced content,
11372 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11373 But these remaining combed frames will be marked as interlaced, and thus can be
11374 de-interlaced by a later filter such as @ref{yadif} before decimation.
11376 In addition to the various configuration options, @code{fieldmatch} can take an
11377 optional second stream, activated through the @option{ppsrc} option. If
11378 enabled, the frames reconstruction will be based on the fields and frames from
11379 this second stream. This allows the first input to be pre-processed in order to
11380 help the various algorithms of the filter, while keeping the output lossless
11381 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11382 or brightness/contrast adjustments can help.
11384 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11385 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11386 which @code{fieldmatch} is based on. While the semantic and usage are very
11387 close, some behaviour and options names can differ.
11389 The @ref{decimate} filter currently only works for constant frame rate input.
11390 If your input has mixed telecined (30fps) and progressive content with a lower
11391 framerate like 24fps use the following filterchain to produce the necessary cfr
11392 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11394 The filter accepts the following options:
11398 Specify the assumed field order of the input stream. Available values are:
11402 Auto detect parity (use FFmpeg's internal parity value).
11404 Assume bottom field first.
11406 Assume top field first.
11409 Note that it is sometimes recommended not to trust the parity announced by the
11412 Default value is @var{auto}.
11415 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11416 sense that it won't risk creating jerkiness due to duplicate frames when
11417 possible, but if there are bad edits or blended fields it will end up
11418 outputting combed frames when a good match might actually exist. On the other
11419 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11420 but will almost always find a good frame if there is one. The other values are
11421 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11422 jerkiness and creating duplicate frames versus finding good matches in sections
11423 with bad edits, orphaned fields, blended fields, etc.
11425 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11427 Available values are:
11431 2-way matching (p/c)
11433 2-way matching, and trying 3rd match if still combed (p/c + n)
11435 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11437 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11438 still combed (p/c + n + u/b)
11440 3-way matching (p/c/n)
11442 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11443 detected as combed (p/c/n + u/b)
11446 The parenthesis at the end indicate the matches that would be used for that
11447 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11450 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11453 Default value is @var{pc_n}.
11456 Mark the main input stream as a pre-processed input, and enable the secondary
11457 input stream as the clean source to pick the fields from. See the filter
11458 introduction for more details. It is similar to the @option{clip2} feature from
11461 Default value is @code{0} (disabled).
11464 Set the field to match from. It is recommended to set this to the same value as
11465 @option{order} unless you experience matching failures with that setting. In
11466 certain circumstances changing the field that is used to match from can have a
11467 large impact on matching performance. Available values are:
11471 Automatic (same value as @option{order}).
11473 Match from the bottom field.
11475 Match from the top field.
11478 Default value is @var{auto}.
11481 Set whether or not chroma is included during the match comparisons. In most
11482 cases it is recommended to leave this enabled. You should set this to @code{0}
11483 only if your clip has bad chroma problems such as heavy rainbowing or other
11484 artifacts. Setting this to @code{0} could also be used to speed things up at
11485 the cost of some accuracy.
11487 Default value is @code{1}.
11491 These define an exclusion band which excludes the lines between @option{y0} and
11492 @option{y1} from being included in the field matching decision. An exclusion
11493 band can be used to ignore subtitles, a logo, or other things that may
11494 interfere with the matching. @option{y0} sets the starting scan line and
11495 @option{y1} sets the ending line; all lines in between @option{y0} and
11496 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11497 @option{y0} and @option{y1} to the same value will disable the feature.
11498 @option{y0} and @option{y1} defaults to @code{0}.
11501 Set the scene change detection threshold as a percentage of maximum change on
11502 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11503 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11504 @option{scthresh} is @code{[0.0, 100.0]}.
11506 Default value is @code{12.0}.
11509 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11510 account the combed scores of matches when deciding what match to use as the
11511 final match. Available values are:
11515 No final matching based on combed scores.
11517 Combed scores are only used when a scene change is detected.
11519 Use combed scores all the time.
11522 Default is @var{sc}.
11525 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11526 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11527 Available values are:
11531 No forced calculation.
11533 Force p/c/n calculations.
11535 Force p/c/n/u/b calculations.
11538 Default value is @var{none}.
11541 This is the area combing threshold used for combed frame detection. This
11542 essentially controls how "strong" or "visible" combing must be to be detected.
11543 Larger values mean combing must be more visible and smaller values mean combing
11544 can be less visible or strong and still be detected. Valid settings are from
11545 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11546 be detected as combed). This is basically a pixel difference value. A good
11547 range is @code{[8, 12]}.
11549 Default value is @code{9}.
11552 Sets whether or not chroma is considered in the combed frame decision. Only
11553 disable this if your source has chroma problems (rainbowing, etc.) that are
11554 causing problems for the combed frame detection with chroma enabled. Actually,
11555 using @option{chroma}=@var{0} is usually more reliable, except for the case
11556 where there is chroma only combing in the source.
11558 Default value is @code{0}.
11562 Respectively set the x-axis and y-axis size of the window used during combed
11563 frame detection. This has to do with the size of the area in which
11564 @option{combpel} pixels are required to be detected as combed for a frame to be
11565 declared combed. See the @option{combpel} parameter description for more info.
11566 Possible values are any number that is a power of 2 starting at 4 and going up
11569 Default value is @code{16}.
11572 The number of combed pixels inside any of the @option{blocky} by
11573 @option{blockx} size blocks on the frame for the frame to be detected as
11574 combed. While @option{cthresh} controls how "visible" the combing must be, this
11575 setting controls "how much" combing there must be in any localized area (a
11576 window defined by the @option{blockx} and @option{blocky} settings) on the
11577 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11578 which point no frames will ever be detected as combed). This setting is known
11579 as @option{MI} in TFM/VFM vocabulary.
11581 Default value is @code{80}.
11584 @anchor{p/c/n/u/b meaning}
11585 @subsection p/c/n/u/b meaning
11587 @subsubsection p/c/n
11589 We assume the following telecined stream:
11592 Top fields: 1 2 2 3 4
11593 Bottom fields: 1 2 3 4 4
11596 The numbers correspond to the progressive frame the fields relate to. Here, the
11597 first two frames are progressive, the 3rd and 4th are combed, and so on.
11599 When @code{fieldmatch} is configured to run a matching from bottom
11600 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11605 B 1 2 3 4 4 <-- matching reference
11614 As a result of the field matching, we can see that some frames get duplicated.
11615 To perform a complete inverse telecine, you need to rely on a decimation filter
11616 after this operation. See for instance the @ref{decimate} filter.
11618 The same operation now matching from top fields (@option{field}=@var{top})
11623 T 1 2 2 3 4 <-- matching reference
11633 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11634 basically, they refer to the frame and field of the opposite parity:
11637 @item @var{p} matches the field of the opposite parity in the previous frame
11638 @item @var{c} matches the field of the opposite parity in the current frame
11639 @item @var{n} matches the field of the opposite parity in the next frame
11644 The @var{u} and @var{b} matching are a bit special in the sense that they match
11645 from the opposite parity flag. In the following examples, we assume that we are
11646 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11647 'x' is placed above and below each matched fields.
11649 With bottom matching (@option{field}=@var{bottom}):
11654 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11655 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11663 With top matching (@option{field}=@var{top}):
11668 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11669 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11677 @subsection Examples
11679 Simple IVTC of a top field first telecined stream:
11681 fieldmatch=order=tff:combmatch=none, decimate
11684 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11686 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11689 @section fieldorder
11691 Transform the field order of the input video.
11693 It accepts the following parameters:
11698 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11699 for bottom field first.
11702 The default value is @samp{tff}.
11704 The transformation is done by shifting the picture content up or down
11705 by one line, and filling the remaining line with appropriate picture content.
11706 This method is consistent with most broadcast field order converters.
11708 If the input video is not flagged as being interlaced, or it is already
11709 flagged as being of the required output field order, then this filter does
11710 not alter the incoming video.
11712 It is very useful when converting to or from PAL DV material,
11713 which is bottom field first.
11717 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11720 @section fifo, afifo
11722 Buffer input images and send them when they are requested.
11724 It is mainly useful when auto-inserted by the libavfilter
11727 It does not take parameters.
11729 @section fillborders
11731 Fill borders of the input video, without changing video stream dimensions.
11732 Sometimes video can have garbage at the four edges and you may not want to
11733 crop video input to keep size multiple of some number.
11735 This filter accepts the following options:
11739 Number of pixels to fill from left border.
11742 Number of pixels to fill from right border.
11745 Number of pixels to fill from top border.
11748 Number of pixels to fill from bottom border.
11753 It accepts the following values:
11756 fill pixels using outermost pixels
11759 fill pixels using mirroring (half sample symmetric)
11762 fill pixels with constant value
11765 fill pixels using reflecting (whole sample symmetric)
11768 fill pixels using wrapping
11771 Default is @var{smear}.
11774 Set color for pixels in fixed mode. Default is @var{black}.
11777 @subsection Commands
11778 This filter supports same @ref{commands} as options.
11779 The command accepts the same syntax of the corresponding option.
11781 If the specified expression is not valid, it is kept at its current
11786 Find a rectangular object
11788 It accepts the following options:
11792 Filepath of the object image, needs to be in gray8.
11795 Detection threshold, default is 0.5.
11798 Number of mipmaps, default is 3.
11800 @item xmin, ymin, xmax, ymax
11801 Specifies the rectangle in which to search.
11804 @subsection Examples
11808 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11810 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11816 Flood area with values of same pixel components with another values.
11818 It accepts the following options:
11821 Set pixel x coordinate.
11824 Set pixel y coordinate.
11827 Set source #0 component value.
11830 Set source #1 component value.
11833 Set source #2 component value.
11836 Set source #3 component value.
11839 Set destination #0 component value.
11842 Set destination #1 component value.
11845 Set destination #2 component value.
11848 Set destination #3 component value.
11854 Convert the input video to one of the specified pixel formats.
11855 Libavfilter will try to pick one that is suitable as input to
11858 It accepts the following parameters:
11862 A '|'-separated list of pixel format names, such as
11863 "pix_fmts=yuv420p|monow|rgb24".
11867 @subsection Examples
11871 Convert the input video to the @var{yuv420p} format
11873 format=pix_fmts=yuv420p
11876 Convert the input video to any of the formats in the list
11878 format=pix_fmts=yuv420p|yuv444p|yuv410p
11885 Convert the video to specified constant frame rate by duplicating or dropping
11886 frames as necessary.
11888 It accepts the following parameters:
11892 The desired output frame rate. The default is @code{25}.
11895 Assume the first PTS should be the given value, in seconds. This allows for
11896 padding/trimming at the start of stream. By default, no assumption is made
11897 about the first frame's expected PTS, so no padding or trimming is done.
11898 For example, this could be set to 0 to pad the beginning with duplicates of
11899 the first frame if a video stream starts after the audio stream or to trim any
11900 frames with a negative PTS.
11903 Timestamp (PTS) rounding method.
11905 Possible values are:
11912 round towards -infinity
11914 round towards +infinity
11918 The default is @code{near}.
11921 Action performed when reading the last frame.
11923 Possible values are:
11926 Use same timestamp rounding method as used for other frames.
11928 Pass through last frame if input duration has not been reached yet.
11930 The default is @code{round}.
11934 Alternatively, the options can be specified as a flat string:
11935 @var{fps}[:@var{start_time}[:@var{round}]].
11937 See also the @ref{setpts} filter.
11939 @subsection Examples
11943 A typical usage in order to set the fps to 25:
11949 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11951 fps=fps=film:round=near
11957 Pack two different video streams into a stereoscopic video, setting proper
11958 metadata on supported codecs. The two views should have the same size and
11959 framerate and processing will stop when the shorter video ends. Please note
11960 that you may conveniently adjust view properties with the @ref{scale} and
11963 It accepts the following parameters:
11967 The desired packing format. Supported values are:
11972 The views are next to each other (default).
11975 The views are on top of each other.
11978 The views are packed by line.
11981 The views are packed by column.
11984 The views are temporally interleaved.
11993 # Convert left and right views into a frame-sequential video
11994 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11996 # Convert views into a side-by-side video with the same output resolution as the input
11997 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
12002 Change the frame rate by interpolating new video output frames from the source
12005 This filter is not designed to function correctly with interlaced media. If
12006 you wish to change the frame rate of interlaced media then you are required
12007 to deinterlace before this filter and re-interlace after this filter.
12009 A description of the accepted options follows.
12013 Specify the output frames per second. This option can also be specified
12014 as a value alone. The default is @code{50}.
12017 Specify the start of a range where the output frame will be created as a
12018 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12019 the default is @code{15}.
12022 Specify the end of a range where the output frame will be created as a
12023 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12024 the default is @code{240}.
12027 Specify the level at which a scene change is detected as a value between
12028 0 and 100 to indicate a new scene; a low value reflects a low
12029 probability for the current frame to introduce a new scene, while a higher
12030 value means the current frame is more likely to be one.
12031 The default is @code{8.2}.
12034 Specify flags influencing the filter process.
12036 Available value for @var{flags} is:
12039 @item scene_change_detect, scd
12040 Enable scene change detection using the value of the option @var{scene}.
12041 This flag is enabled by default.
12047 Select one frame every N-th frame.
12049 This filter accepts the following option:
12052 Select frame after every @code{step} frames.
12053 Allowed values are positive integers higher than 0. Default value is @code{1}.
12056 @section freezedetect
12058 Detect frozen video.
12060 This filter logs a message and sets frame metadata when it detects that the
12061 input video has no significant change in content during a specified duration.
12062 Video freeze detection calculates the mean average absolute difference of all
12063 the components of video frames and compares it to a noise floor.
12065 The printed times and duration are expressed in seconds. The
12066 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
12067 whose timestamp equals or exceeds the detection duration and it contains the
12068 timestamp of the first frame of the freeze. The
12069 @code{lavfi.freezedetect.freeze_duration} and
12070 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
12073 The filter accepts the following options:
12077 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
12078 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
12082 Set freeze duration until notification (default is 2 seconds).
12085 @section freezeframes
12087 Freeze video frames.
12089 This filter freezes video frames using frame from 2nd input.
12091 The filter accepts the following options:
12095 Set number of first frame from which to start freeze.
12098 Set number of last frame from which to end freeze.
12101 Set number of frame from 2nd input which will be used instead of replaced frames.
12107 Apply a frei0r effect to the input video.
12109 To enable the compilation of this filter, you need to install the frei0r
12110 header and configure FFmpeg with @code{--enable-frei0r}.
12112 It accepts the following parameters:
12117 The name of the frei0r effect to load. If the environment variable
12118 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
12119 directories specified by the colon-separated list in @env{FREI0R_PATH}.
12120 Otherwise, the standard frei0r paths are searched, in this order:
12121 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
12122 @file{/usr/lib/frei0r-1/}.
12124 @item filter_params
12125 A '|'-separated list of parameters to pass to the frei0r effect.
12129 A frei0r effect parameter can be a boolean (its value is either
12130 "y" or "n"), a double, a color (specified as
12131 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
12132 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
12133 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
12134 a position (specified as @var{X}/@var{Y}, where
12135 @var{X} and @var{Y} are floating point numbers) and/or a string.
12137 The number and types of parameters depend on the loaded effect. If an
12138 effect parameter is not specified, the default value is set.
12140 @subsection Examples
12144 Apply the distort0r effect, setting the first two double parameters:
12146 frei0r=filter_name=distort0r:filter_params=0.5|0.01
12150 Apply the colordistance effect, taking a color as the first parameter:
12152 frei0r=colordistance:0.2/0.3/0.4
12153 frei0r=colordistance:violet
12154 frei0r=colordistance:0x112233
12158 Apply the perspective effect, specifying the top left and top right image
12161 frei0r=perspective:0.2/0.2|0.8/0.2
12165 For more information, see
12166 @url{http://frei0r.dyne.org}
12168 @subsection Commands
12170 This filter supports the @option{filter_params} option as @ref{commands}.
12174 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
12176 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
12177 processing filter, one of them is performed once per block, not per pixel.
12178 This allows for much higher speed.
12180 The filter accepts the following options:
12184 Set quality. This option defines the number of levels for averaging. It accepts
12185 an integer in the range 4-5. Default value is @code{4}.
12188 Force a constant quantization parameter. It accepts an integer in range 0-63.
12189 If not set, the filter will use the QP from the video stream (if available).
12192 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
12193 more details but also more artifacts, while higher values make the image smoother
12194 but also blurrier. Default value is @code{0} − PSNR optimal.
12196 @item use_bframe_qp
12197 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
12198 option may cause flicker since the B-Frames have often larger QP. Default is
12199 @code{0} (not enabled).
12205 Apply Gaussian blur filter.
12207 The filter accepts the following options:
12211 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
12214 Set number of steps for Gaussian approximation. Default is @code{1}.
12217 Set which planes to filter. By default all planes are filtered.
12220 Set vertical sigma, if negative it will be same as @code{sigma}.
12221 Default is @code{-1}.
12224 @subsection Commands
12225 This filter supports same commands as options.
12226 The command accepts the same syntax of the corresponding option.
12228 If the specified expression is not valid, it is kept at its current
12233 Apply generic equation to each pixel.
12235 The filter accepts the following options:
12238 @item lum_expr, lum
12239 Set the luminance expression.
12241 Set the chrominance blue expression.
12243 Set the chrominance red expression.
12244 @item alpha_expr, a
12245 Set the alpha expression.
12247 Set the red expression.
12248 @item green_expr, g
12249 Set the green expression.
12251 Set the blue expression.
12254 The colorspace is selected according to the specified options. If one
12255 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
12256 options is specified, the filter will automatically select a YCbCr
12257 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
12258 @option{blue_expr} options is specified, it will select an RGB
12261 If one of the chrominance expression is not defined, it falls back on the other
12262 one. If no alpha expression is specified it will evaluate to opaque value.
12263 If none of chrominance expressions are specified, they will evaluate
12264 to the luminance expression.
12266 The expressions can use the following variables and functions:
12270 The sequential number of the filtered frame, starting from @code{0}.
12274 The coordinates of the current sample.
12278 The width and height of the image.
12282 Width and height scale depending on the currently filtered plane. It is the
12283 ratio between the corresponding luma plane number of pixels and the current
12284 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
12285 @code{0.5,0.5} for chroma planes.
12288 Time of the current frame, expressed in seconds.
12291 Return the value of the pixel at location (@var{x},@var{y}) of the current
12295 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12299 Return the value of the pixel at location (@var{x},@var{y}) of the
12300 blue-difference chroma plane. Return 0 if there is no such plane.
12303 Return the value of the pixel at location (@var{x},@var{y}) of the
12304 red-difference chroma plane. Return 0 if there is no such plane.
12309 Return the value of the pixel at location (@var{x},@var{y}) of the
12310 red/green/blue component. Return 0 if there is no such component.
12313 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12314 plane. Return 0 if there is no such plane.
12316 @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)
12317 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12318 sums of samples within a rectangle. See the functions without the sum postfix.
12320 @item interpolation
12321 Set one of interpolation methods:
12326 Default is bilinear.
12329 For functions, if @var{x} and @var{y} are outside the area, the value will be
12330 automatically clipped to the closer edge.
12332 Please note that this filter can use multiple threads in which case each slice
12333 will have its own expression state. If you want to use only a single expression
12334 state because your expressions depend on previous state then you should limit
12335 the number of filter threads to 1.
12337 @subsection Examples
12341 Flip the image horizontally:
12347 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12348 wavelength of 100 pixels:
12350 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12354 Generate a fancy enigmatic moving light:
12356 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
12360 Generate a quick emboss effect:
12362 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12366 Modify RGB components depending on pixel position:
12368 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12372 Create a radial gradient that is the same size as the input (also see
12373 the @ref{vignette} filter):
12375 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12381 Fix the banding artifacts that are sometimes introduced into nearly flat
12382 regions by truncation to 8-bit color depth.
12383 Interpolate the gradients that should go where the bands are, and
12386 It is designed for playback only. Do not use it prior to
12387 lossy compression, because compression tends to lose the dither and
12388 bring back the bands.
12390 It accepts the following parameters:
12395 The maximum amount by which the filter will change any one pixel. This is also
12396 the threshold for detecting nearly flat regions. Acceptable values range from
12397 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12401 The neighborhood to fit the gradient to. A larger radius makes for smoother
12402 gradients, but also prevents the filter from modifying the pixels near detailed
12403 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12404 values will be clipped to the valid range.
12408 Alternatively, the options can be specified as a flat string:
12409 @var{strength}[:@var{radius}]
12411 @subsection Examples
12415 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12421 Specify radius, omitting the strength (which will fall-back to the default
12429 @anchor{graphmonitor}
12430 @section graphmonitor
12431 Show various filtergraph stats.
12433 With this filter one can debug complete filtergraph.
12434 Especially issues with links filling with queued frames.
12436 The filter accepts the following options:
12440 Set video output size. Default is @var{hd720}.
12443 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12446 Set output mode, can be @var{fulll} or @var{compact}.
12447 In @var{compact} mode only filters with some queued frames have displayed stats.
12450 Set flags which enable which stats are shown in video.
12452 Available values for flags are:
12455 Display number of queued frames in each link.
12457 @item frame_count_in
12458 Display number of frames taken from filter.
12460 @item frame_count_out
12461 Display number of frames given out from filter.
12464 Display current filtered frame pts.
12467 Display current filtered frame time.
12470 Display time base for filter link.
12473 Display used format for filter link.
12476 Display video size or number of audio channels in case of audio used by filter link.
12479 Display video frame rate or sample rate in case of audio used by filter link.
12482 Display link output status.
12486 Set upper limit for video rate of output stream, Default value is @var{25}.
12487 This guarantee that output video frame rate will not be higher than this value.
12491 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12492 and corrects the scene colors accordingly.
12494 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12496 The filter accepts the following options:
12500 The order of differentiation to be applied on the scene. Must be chosen in the range
12501 [0,2] and default value is 1.
12504 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12505 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12506 max value instead of calculating Minkowski distance.
12509 The standard deviation of Gaussian blur to be applied on the scene. Must be
12510 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12511 can't be equal to 0 if @var{difford} is greater than 0.
12514 @subsection Examples
12520 greyedge=difford=1:minknorm=5:sigma=2
12526 greyedge=difford=1:minknorm=0:sigma=2
12534 Apply a Hald CLUT to a video stream.
12536 First input is the video stream to process, and second one is the Hald CLUT.
12537 The Hald CLUT input can be a simple picture or a complete video stream.
12539 The filter accepts the following options:
12543 Force termination when the shortest input terminates. Default is @code{0}.
12545 Continue applying the last CLUT after the end of the stream. A value of
12546 @code{0} disable the filter after the last frame of the CLUT is reached.
12547 Default is @code{1}.
12550 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12551 filters share the same internals).
12553 This filter also supports the @ref{framesync} options.
12555 More information about the Hald CLUT can be found on Eskil Steenberg's website
12556 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12558 @subsection Workflow examples
12560 @subsubsection Hald CLUT video stream
12562 Generate an identity Hald CLUT stream altered with various effects:
12564 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
12567 Note: make sure you use a lossless codec.
12569 Then use it with @code{haldclut} to apply it on some random stream:
12571 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12574 The Hald CLUT will be applied to the 10 first seconds (duration of
12575 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12576 to the remaining frames of the @code{mandelbrot} stream.
12578 @subsubsection Hald CLUT with preview
12580 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12581 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12582 biggest possible square starting at the top left of the picture. The remaining
12583 padding pixels (bottom or right) will be ignored. This area can be used to add
12584 a preview of the Hald CLUT.
12586 Typically, the following generated Hald CLUT will be supported by the
12587 @code{haldclut} filter:
12590 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12591 pad=iw+320 [padded_clut];
12592 smptebars=s=320x256, split [a][b];
12593 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12594 [main][b] overlay=W-320" -frames:v 1 clut.png
12597 It contains the original and a preview of the effect of the CLUT: SMPTE color
12598 bars are displayed on the right-top, and below the same color bars processed by
12601 Then, the effect of this Hald CLUT can be visualized with:
12603 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12608 Flip the input video horizontally.
12610 For example, to horizontally flip the input video with @command{ffmpeg}:
12612 ffmpeg -i in.avi -vf "hflip" out.avi
12616 This filter applies a global color histogram equalization on a
12619 It can be used to correct video that has a compressed range of pixel
12620 intensities. The filter redistributes the pixel intensities to
12621 equalize their distribution across the intensity range. It may be
12622 viewed as an "automatically adjusting contrast filter". This filter is
12623 useful only for correcting degraded or poorly captured source
12626 The filter accepts the following options:
12630 Determine the amount of equalization to be applied. As the strength
12631 is reduced, the distribution of pixel intensities more-and-more
12632 approaches that of the input frame. The value must be a float number
12633 in the range [0,1] and defaults to 0.200.
12636 Set the maximum intensity that can generated and scale the output
12637 values appropriately. The strength should be set as desired and then
12638 the intensity can be limited if needed to avoid washing-out. The value
12639 must be a float number in the range [0,1] and defaults to 0.210.
12642 Set the antibanding level. If enabled the filter will randomly vary
12643 the luminance of output pixels by a small amount to avoid banding of
12644 the histogram. Possible values are @code{none}, @code{weak} or
12645 @code{strong}. It defaults to @code{none}.
12651 Compute and draw a color distribution histogram for the input video.
12653 The computed histogram is a representation of the color component
12654 distribution in an image.
12656 Standard histogram displays the color components distribution in an image.
12657 Displays color graph for each color component. Shows distribution of
12658 the Y, U, V, A or R, G, B components, depending on input format, in the
12659 current frame. Below each graph a color component scale meter is shown.
12661 The filter accepts the following options:
12665 Set height of level. Default value is @code{200}.
12666 Allowed range is [50, 2048].
12669 Set height of color scale. Default value is @code{12}.
12670 Allowed range is [0, 40].
12674 It accepts the following values:
12677 Per color component graphs are placed below each other.
12680 Per color component graphs are placed side by side.
12683 Presents information identical to that in the @code{parade}, except
12684 that the graphs representing color components are superimposed directly
12687 Default is @code{stack}.
12690 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12691 Default is @code{linear}.
12694 Set what color components to display.
12695 Default is @code{7}.
12698 Set foreground opacity. Default is @code{0.7}.
12701 Set background opacity. Default is @code{0.5}.
12704 @subsection Examples
12709 Calculate and draw histogram:
12711 ffplay -i input -vf histogram
12719 This is a high precision/quality 3d denoise filter. It aims to reduce
12720 image noise, producing smooth images and making still images really
12721 still. It should enhance compressibility.
12723 It accepts the following optional parameters:
12727 A non-negative floating point number which specifies spatial luma strength.
12728 It defaults to 4.0.
12730 @item chroma_spatial
12731 A non-negative floating point number which specifies spatial chroma strength.
12732 It defaults to 3.0*@var{luma_spatial}/4.0.
12735 A floating point number which specifies luma temporal strength. It defaults to
12736 6.0*@var{luma_spatial}/4.0.
12739 A floating point number which specifies chroma temporal strength. It defaults to
12740 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12743 @subsection Commands
12744 This filter supports same @ref{commands} as options.
12745 The command accepts the same syntax of the corresponding option.
12747 If the specified expression is not valid, it is kept at its current
12750 @anchor{hwdownload}
12751 @section hwdownload
12753 Download hardware frames to system memory.
12755 The input must be in hardware frames, and the output a non-hardware format.
12756 Not all formats will be supported on the output - it may be necessary to insert
12757 an additional @option{format} filter immediately following in the graph to get
12758 the output in a supported format.
12762 Map hardware frames to system memory or to another device.
12764 This filter has several different modes of operation; which one is used depends
12765 on the input and output formats:
12768 Hardware frame input, normal frame output
12770 Map the input frames to system memory and pass them to the output. If the
12771 original hardware frame is later required (for example, after overlaying
12772 something else on part of it), the @option{hwmap} filter can be used again
12773 in the next mode to retrieve it.
12775 Normal frame input, hardware frame output
12777 If the input is actually a software-mapped hardware frame, then unmap it -
12778 that is, return the original hardware frame.
12780 Otherwise, a device must be provided. Create new hardware surfaces on that
12781 device for the output, then map them back to the software format at the input
12782 and give those frames to the preceding filter. This will then act like the
12783 @option{hwupload} filter, but may be able to avoid an additional copy when
12784 the input is already in a compatible format.
12786 Hardware frame input and output
12788 A device must be supplied for the output, either directly or with the
12789 @option{derive_device} option. The input and output devices must be of
12790 different types and compatible - the exact meaning of this is
12791 system-dependent, but typically it means that they must refer to the same
12792 underlying hardware context (for example, refer to the same graphics card).
12794 If the input frames were originally created on the output device, then unmap
12795 to retrieve the original frames.
12797 Otherwise, map the frames to the output device - create new hardware frames
12798 on the output corresponding to the frames on the input.
12801 The following additional parameters are accepted:
12805 Set the frame mapping mode. Some combination of:
12808 The mapped frame should be readable.
12810 The mapped frame should be writeable.
12812 The mapping will always overwrite the entire frame.
12814 This may improve performance in some cases, as the original contents of the
12815 frame need not be loaded.
12817 The mapping must not involve any copying.
12819 Indirect mappings to copies of frames are created in some cases where either
12820 direct mapping is not possible or it would have unexpected properties.
12821 Setting this flag ensures that the mapping is direct and will fail if that is
12824 Defaults to @var{read+write} if not specified.
12826 @item derive_device @var{type}
12827 Rather than using the device supplied at initialisation, instead derive a new
12828 device of type @var{type} from the device the input frames exist on.
12831 In a hardware to hardware mapping, map in reverse - create frames in the sink
12832 and map them back to the source. This may be necessary in some cases where
12833 a mapping in one direction is required but only the opposite direction is
12834 supported by the devices being used.
12836 This option is dangerous - it may break the preceding filter in undefined
12837 ways if there are any additional constraints on that filter's output.
12838 Do not use it without fully understanding the implications of its use.
12844 Upload system memory frames to hardware surfaces.
12846 The device to upload to must be supplied when the filter is initialised. If
12847 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12848 option or with the @option{derive_device} option. The input and output devices
12849 must be of different types and compatible - the exact meaning of this is
12850 system-dependent, but typically it means that they must refer to the same
12851 underlying hardware context (for example, refer to the same graphics card).
12853 The following additional parameters are accepted:
12856 @item derive_device @var{type}
12857 Rather than using the device supplied at initialisation, instead derive a new
12858 device of type @var{type} from the device the input frames exist on.
12861 @anchor{hwupload_cuda}
12862 @section hwupload_cuda
12864 Upload system memory frames to a CUDA device.
12866 It accepts the following optional parameters:
12870 The number of the CUDA device to use
12875 Apply a high-quality magnification filter designed for pixel art. This filter
12876 was originally created by Maxim Stepin.
12878 It accepts the following option:
12882 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12883 @code{hq3x} and @code{4} for @code{hq4x}.
12884 Default is @code{3}.
12888 Stack input videos horizontally.
12890 All streams must be of same pixel format and of same height.
12892 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12893 to create same output.
12895 The filter accepts the following option:
12899 Set number of input streams. Default is 2.
12902 If set to 1, force the output to terminate when the shortest input
12903 terminates. Default value is 0.
12908 Modify the hue and/or the saturation of the input.
12910 It accepts the following parameters:
12914 Specify the hue angle as a number of degrees. It accepts an expression,
12915 and defaults to "0".
12918 Specify the saturation in the [-10,10] range. It accepts an expression and
12922 Specify the hue angle as a number of radians. It accepts an
12923 expression, and defaults to "0".
12926 Specify the brightness in the [-10,10] range. It accepts an expression and
12930 @option{h} and @option{H} are mutually exclusive, and can't be
12931 specified at the same time.
12933 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12934 expressions containing the following constants:
12938 frame count of the input frame starting from 0
12941 presentation timestamp of the input frame expressed in time base units
12944 frame rate of the input video, NAN if the input frame rate is unknown
12947 timestamp expressed in seconds, NAN if the input timestamp is unknown
12950 time base of the input video
12953 @subsection Examples
12957 Set the hue to 90 degrees and the saturation to 1.0:
12963 Same command but expressing the hue in radians:
12969 Rotate hue and make the saturation swing between 0
12970 and 2 over a period of 1 second:
12972 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12976 Apply a 3 seconds saturation fade-in effect starting at 0:
12978 hue="s=min(t/3\,1)"
12981 The general fade-in expression can be written as:
12983 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12987 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12989 hue="s=max(0\, min(1\, (8-t)/3))"
12992 The general fade-out expression can be written as:
12994 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12999 @subsection Commands
13001 This filter supports the following commands:
13007 Modify the hue and/or the saturation and/or brightness of the input video.
13008 The command accepts the same syntax of the corresponding option.
13010 If the specified expression is not valid, it is kept at its current
13014 @section hysteresis
13016 Grow first stream into second stream by connecting components.
13017 This makes it possible to build more robust edge masks.
13019 This filter accepts the following options:
13023 Set which planes will be processed as bitmap, unprocessed planes will be
13024 copied from first stream.
13025 By default value 0xf, all planes will be processed.
13028 Set threshold which is used in filtering. If pixel component value is higher than
13029 this value filter algorithm for connecting components is activated.
13030 By default value is 0.
13033 The @code{hysteresis} filter also supports the @ref{framesync} options.
13037 Detect video interlacing type.
13039 This filter tries to detect if the input frames are interlaced, progressive,
13040 top or bottom field first. It will also try to detect fields that are
13041 repeated between adjacent frames (a sign of telecine).
13043 Single frame detection considers only immediately adjacent frames when classifying each frame.
13044 Multiple frame detection incorporates the classification history of previous frames.
13046 The filter will log these metadata values:
13049 @item single.current_frame
13050 Detected type of current frame using single-frame detection. One of:
13051 ``tff'' (top field first), ``bff'' (bottom field first),
13052 ``progressive'', or ``undetermined''
13055 Cumulative number of frames detected as top field first using single-frame detection.
13058 Cumulative number of frames detected as top field first using multiple-frame detection.
13061 Cumulative number of frames detected as bottom field first using single-frame detection.
13063 @item multiple.current_frame
13064 Detected type of current frame using multiple-frame detection. One of:
13065 ``tff'' (top field first), ``bff'' (bottom field first),
13066 ``progressive'', or ``undetermined''
13069 Cumulative number of frames detected as bottom field first using multiple-frame detection.
13071 @item single.progressive
13072 Cumulative number of frames detected as progressive using single-frame detection.
13074 @item multiple.progressive
13075 Cumulative number of frames detected as progressive using multiple-frame detection.
13077 @item single.undetermined
13078 Cumulative number of frames that could not be classified using single-frame detection.
13080 @item multiple.undetermined
13081 Cumulative number of frames that could not be classified using multiple-frame detection.
13083 @item repeated.current_frame
13084 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
13086 @item repeated.neither
13087 Cumulative number of frames with no repeated field.
13090 Cumulative number of frames with the top field repeated from the previous frame's top field.
13092 @item repeated.bottom
13093 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
13096 The filter accepts the following options:
13100 Set interlacing threshold.
13102 Set progressive threshold.
13104 Threshold for repeated field detection.
13106 Number of frames after which a given frame's contribution to the
13107 statistics is halved (i.e., it contributes only 0.5 to its
13108 classification). The default of 0 means that all frames seen are given
13109 full weight of 1.0 forever.
13110 @item analyze_interlaced_flag
13111 When this is not 0 then idet will use the specified number of frames to determine
13112 if the interlaced flag is accurate, it will not count undetermined frames.
13113 If the flag is found to be accurate it will be used without any further
13114 computations, if it is found to be inaccurate it will be cleared without any
13115 further computations. This allows inserting the idet filter as a low computational
13116 method to clean up the interlaced flag
13121 Deinterleave or interleave fields.
13123 This filter allows one to process interlaced images fields without
13124 deinterlacing them. Deinterleaving splits the input frame into 2
13125 fields (so called half pictures). Odd lines are moved to the top
13126 half of the output image, even lines to the bottom half.
13127 You can process (filter) them independently and then re-interleave them.
13129 The filter accepts the following options:
13133 @item chroma_mode, c
13134 @item alpha_mode, a
13135 Available values for @var{luma_mode}, @var{chroma_mode} and
13136 @var{alpha_mode} are:
13142 @item deinterleave, d
13143 Deinterleave fields, placing one above the other.
13145 @item interleave, i
13146 Interleave fields. Reverse the effect of deinterleaving.
13148 Default value is @code{none}.
13150 @item luma_swap, ls
13151 @item chroma_swap, cs
13152 @item alpha_swap, as
13153 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
13156 @subsection Commands
13158 This filter supports the all above options as @ref{commands}.
13162 Apply inflate effect to the video.
13164 This filter replaces the pixel by the local(3x3) average by taking into account
13165 only values higher than the pixel.
13167 It accepts the following options:
13174 Limit the maximum change for each plane, default is 65535.
13175 If 0, plane will remain unchanged.
13178 @subsection Commands
13180 This filter supports the all above options as @ref{commands}.
13184 Simple interlacing filter from progressive contents. This interleaves upper (or
13185 lower) lines from odd frames with lower (or upper) lines from even frames,
13186 halving the frame rate and preserving image height.
13189 Original Original New Frame
13190 Frame 'j' Frame 'j+1' (tff)
13191 ========== =========== ==================
13192 Line 0 --------------------> Frame 'j' Line 0
13193 Line 1 Line 1 ----> Frame 'j+1' Line 1
13194 Line 2 ---------------------> Frame 'j' Line 2
13195 Line 3 Line 3 ----> Frame 'j+1' Line 3
13197 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
13200 It accepts the following optional parameters:
13204 This determines whether the interlaced frame is taken from the even
13205 (tff - default) or odd (bff) lines of the progressive frame.
13208 Vertical lowpass filter to avoid twitter interlacing and
13209 reduce moire patterns.
13213 Disable vertical lowpass filter
13216 Enable linear filter (default)
13219 Enable complex filter. This will slightly less reduce twitter and moire
13220 but better retain detail and subjective sharpness impression.
13227 Deinterlace input video by applying Donald Graft's adaptive kernel
13228 deinterling. Work on interlaced parts of a video to produce
13229 progressive frames.
13231 The description of the accepted parameters follows.
13235 Set the threshold which affects the filter's tolerance when
13236 determining if a pixel line must be processed. It must be an integer
13237 in the range [0,255] and defaults to 10. A value of 0 will result in
13238 applying the process on every pixels.
13241 Paint pixels exceeding the threshold value to white if set to 1.
13245 Set the fields order. Swap fields if set to 1, leave fields alone if
13249 Enable additional sharpening if set to 1. Default is 0.
13252 Enable twoway sharpening if set to 1. Default is 0.
13255 @subsection Examples
13259 Apply default values:
13261 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
13265 Enable additional sharpening:
13271 Paint processed pixels in white:
13279 Slowly update darker pixels.
13281 This filter makes short flashes of light appear longer.
13282 This filter accepts the following options:
13286 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13289 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13292 @section lenscorrection
13294 Correct radial lens distortion
13296 This filter can be used to correct for radial distortion as can result from the use
13297 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13298 one can use tools available for example as part of opencv or simply trial-and-error.
13299 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13300 and extract the k1 and k2 coefficients from the resulting matrix.
13302 Note that effectively the same filter is available in the open-source tools Krita and
13303 Digikam from the KDE project.
13305 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13306 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13307 brightness distribution, so you may want to use both filters together in certain
13308 cases, though you will have to take care of ordering, i.e. whether vignetting should
13309 be applied before or after lens correction.
13311 @subsection Options
13313 The filter accepts the following options:
13317 Relative x-coordinate of the focal point of the image, and thereby the center of the
13318 distortion. This value has a range [0,1] and is expressed as fractions of the image
13319 width. Default is 0.5.
13321 Relative y-coordinate of the focal point of the image, and thereby the center of the
13322 distortion. This value has a range [0,1] and is expressed as fractions of the image
13323 height. Default is 0.5.
13325 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13326 no correction. Default is 0.
13328 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13329 0 means no correction. Default is 0.
13332 The formula that generates the correction is:
13334 @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)
13336 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13337 distances from the focal point in the source and target images, respectively.
13341 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13343 The @code{lensfun} filter requires the camera make, camera model, and lens model
13344 to apply the lens correction. The filter will load the lensfun database and
13345 query it to find the corresponding camera and lens entries in the database. As
13346 long as these entries can be found with the given options, the filter can
13347 perform corrections on frames. Note that incomplete strings will result in the
13348 filter choosing the best match with the given options, and the filter will
13349 output the chosen camera and lens models (logged with level "info"). You must
13350 provide the make, camera model, and lens model as they are required.
13352 The filter accepts the following options:
13356 The make of the camera (for example, "Canon"). This option is required.
13359 The model of the camera (for example, "Canon EOS 100D"). This option is
13363 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13364 option is required.
13367 The type of correction to apply. The following values are valid options:
13371 Enables fixing lens vignetting.
13374 Enables fixing lens geometry. This is the default.
13377 Enables fixing chromatic aberrations.
13380 Enables fixing lens vignetting and lens geometry.
13383 Enables fixing lens vignetting and chromatic aberrations.
13386 Enables fixing both lens geometry and chromatic aberrations.
13389 Enables all possible corrections.
13393 The focal length of the image/video (zoom; expected constant for video). For
13394 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13395 range should be chosen when using that lens. Default 18.
13398 The aperture of the image/video (expected constant for video). Note that
13399 aperture is only used for vignetting correction. Default 3.5.
13401 @item focus_distance
13402 The focus distance of the image/video (expected constant for video). Note that
13403 focus distance is only used for vignetting and only slightly affects the
13404 vignetting correction process. If unknown, leave it at the default value (which
13408 The scale factor which is applied after transformation. After correction the
13409 video is no longer necessarily rectangular. This parameter controls how much of
13410 the resulting image is visible. The value 0 means that a value will be chosen
13411 automatically such that there is little or no unmapped area in the output
13412 image. 1.0 means that no additional scaling is done. Lower values may result
13413 in more of the corrected image being visible, while higher values may avoid
13414 unmapped areas in the output.
13416 @item target_geometry
13417 The target geometry of the output image/video. The following values are valid
13421 @item rectilinear (default)
13424 @item equirectangular
13425 @item fisheye_orthographic
13426 @item fisheye_stereographic
13427 @item fisheye_equisolid
13428 @item fisheye_thoby
13431 Apply the reverse of image correction (instead of correcting distortion, apply
13434 @item interpolation
13435 The type of interpolation used when correcting distortion. The following values
13440 @item linear (default)
13445 @subsection Examples
13449 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13450 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13454 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
13458 Apply the same as before, but only for the first 5 seconds of video.
13461 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
13468 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13469 score between two input videos.
13471 The obtained VMAF score is printed through the logging system.
13473 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13474 After installing the library it can be enabled using:
13475 @code{./configure --enable-libvmaf}.
13476 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13478 The filter has following options:
13482 Set the model path which is to be used for SVM.
13483 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13486 Set the file path to be used to store logs.
13489 Set the format of the log file (csv, json or xml).
13491 @item enable_transform
13492 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13493 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13494 Default value: @code{false}
13497 Invokes the phone model which will generate VMAF scores higher than in the
13498 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13499 Default value: @code{false}
13502 Enables computing psnr along with vmaf.
13503 Default value: @code{false}
13506 Enables computing ssim along with vmaf.
13507 Default value: @code{false}
13510 Enables computing ms_ssim along with vmaf.
13511 Default value: @code{false}
13514 Set the pool method to be used for computing vmaf.
13515 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13518 Set number of threads to be used when computing vmaf.
13519 Default value: @code{0}, which makes use of all available logical processors.
13522 Set interval for frame subsampling used when computing vmaf.
13523 Default value: @code{1}
13525 @item enable_conf_interval
13526 Enables confidence interval.
13527 Default value: @code{false}
13530 This filter also supports the @ref{framesync} options.
13532 @subsection Examples
13535 On the below examples the input file @file{main.mpg} being processed is
13536 compared with the reference file @file{ref.mpg}.
13539 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13543 Example with options:
13545 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13549 Example with options and different containers:
13551 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 -
13557 Limits the pixel components values to the specified range [min, max].
13559 The filter accepts the following options:
13563 Lower bound. Defaults to the lowest allowed value for the input.
13566 Upper bound. Defaults to the highest allowed value for the input.
13569 Specify which planes will be processed. Defaults to all available.
13572 @subsection Commands
13574 This filter supports the all above options as @ref{commands}.
13580 The filter accepts the following options:
13584 Set the number of loops. Setting this value to -1 will result in infinite loops.
13588 Set maximal size in number of frames. Default is 0.
13591 Set first frame of loop. Default is 0.
13594 @subsection Examples
13598 Loop single first frame infinitely:
13600 loop=loop=-1:size=1:start=0
13604 Loop single first frame 10 times:
13606 loop=loop=10:size=1:start=0
13610 Loop 10 first frames 5 times:
13612 loop=loop=5:size=10:start=0
13618 Apply a 1D LUT to an input video.
13620 The filter accepts the following options:
13624 Set the 1D LUT file name.
13626 Currently supported formats:
13635 Select interpolation mode.
13637 Available values are:
13641 Use values from the nearest defined point.
13643 Interpolate values using the linear interpolation.
13645 Interpolate values using the cosine interpolation.
13647 Interpolate values using the cubic interpolation.
13649 Interpolate values using the spline interpolation.
13656 Apply a 3D LUT to an input video.
13658 The filter accepts the following options:
13662 Set the 3D LUT file name.
13664 Currently supported formats:
13678 Select interpolation mode.
13680 Available values are:
13684 Use values from the nearest defined point.
13686 Interpolate values using the 8 points defining a cube.
13688 Interpolate values using a tetrahedron.
13694 Turn certain luma values into transparency.
13696 The filter accepts the following options:
13700 Set the luma which will be used as base for transparency.
13701 Default value is @code{0}.
13704 Set the range of luma values to be keyed out.
13705 Default value is @code{0.01}.
13708 Set the range of softness. Default value is @code{0}.
13709 Use this to control gradual transition from zero to full transparency.
13712 @subsection Commands
13713 This filter supports same @ref{commands} as options.
13714 The command accepts the same syntax of the corresponding option.
13716 If the specified expression is not valid, it is kept at its current
13719 @section lut, lutrgb, lutyuv
13721 Compute a look-up table for binding each pixel component input value
13722 to an output value, and apply it to the input video.
13724 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13725 to an RGB input video.
13727 These filters accept the following parameters:
13730 set first pixel component expression
13732 set second pixel component expression
13734 set third pixel component expression
13736 set fourth pixel component expression, corresponds to the alpha component
13739 set red component expression
13741 set green component expression
13743 set blue component expression
13745 alpha component expression
13748 set Y/luminance component expression
13750 set U/Cb component expression
13752 set V/Cr component expression
13755 Each of them specifies the expression to use for computing the lookup table for
13756 the corresponding pixel component values.
13758 The exact component associated to each of the @var{c*} options depends on the
13761 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13762 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13764 The expressions can contain the following constants and functions:
13769 The input width and height.
13772 The input value for the pixel component.
13775 The input value, clipped to the @var{minval}-@var{maxval} range.
13778 The maximum value for the pixel component.
13781 The minimum value for the pixel component.
13784 The negated value for the pixel component value, clipped to the
13785 @var{minval}-@var{maxval} range; it corresponds to the expression
13786 "maxval-clipval+minval".
13789 The computed value in @var{val}, clipped to the
13790 @var{minval}-@var{maxval} range.
13792 @item gammaval(gamma)
13793 The computed gamma correction value of the pixel component value,
13794 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13796 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13800 All expressions default to "val".
13802 @subsection Examples
13806 Negate input video:
13808 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13809 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13812 The above is the same as:
13814 lutrgb="r=negval:g=negval:b=negval"
13815 lutyuv="y=negval:u=negval:v=negval"
13825 Remove chroma components, turning the video into a graytone image:
13827 lutyuv="u=128:v=128"
13831 Apply a luma burning effect:
13837 Remove green and blue components:
13843 Set a constant alpha channel value on input:
13845 format=rgba,lutrgb=a="maxval-minval/2"
13849 Correct luminance gamma by a factor of 0.5:
13851 lutyuv=y=gammaval(0.5)
13855 Discard least significant bits of luma:
13857 lutyuv=y='bitand(val, 128+64+32)'
13861 Technicolor like effect:
13863 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13867 @section lut2, tlut2
13869 The @code{lut2} filter takes two input streams and outputs one
13872 The @code{tlut2} (time lut2) filter takes two consecutive frames
13873 from one single stream.
13875 This filter accepts the following parameters:
13878 set first pixel component expression
13880 set second pixel component expression
13882 set third pixel component expression
13884 set fourth pixel component expression, corresponds to the alpha component
13887 set output bit depth, only available for @code{lut2} filter. By default is 0,
13888 which means bit depth is automatically picked from first input format.
13891 The @code{lut2} filter also supports the @ref{framesync} options.
13893 Each of them specifies the expression to use for computing the lookup table for
13894 the corresponding pixel component values.
13896 The exact component associated to each of the @var{c*} options depends on the
13899 The expressions can contain the following constants:
13904 The input width and height.
13907 The first input value for the pixel component.
13910 The second input value for the pixel component.
13913 The first input video bit depth.
13916 The second input video bit depth.
13919 All expressions default to "x".
13921 @subsection Examples
13925 Highlight differences between two RGB video streams:
13927 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)'
13931 Highlight differences between two YUV video streams:
13933 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)'
13937 Show max difference between two video streams:
13939 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)))'
13943 @section maskedclamp
13945 Clamp the first input stream with the second input and third input stream.
13947 Returns the value of first stream to be between second input
13948 stream - @code{undershoot} and third input stream + @code{overshoot}.
13950 This filter accepts the following options:
13953 Default value is @code{0}.
13956 Default value is @code{0}.
13959 Set which planes will be processed as bitmap, unprocessed planes will be
13960 copied from first stream.
13961 By default value 0xf, all planes will be processed.
13966 Merge the second and third input stream into output stream using absolute differences
13967 between second input stream and first input stream and absolute difference between
13968 third input stream and first input stream. The picked value will be from second input
13969 stream if second absolute difference is greater than first one or from third input stream
13972 This filter accepts the following options:
13975 Set which planes will be processed as bitmap, unprocessed planes will be
13976 copied from first stream.
13977 By default value 0xf, all planes will be processed.
13980 @section maskedmerge
13982 Merge the first input stream with the second input stream using per pixel
13983 weights in the third input stream.
13985 A value of 0 in the third stream pixel component means that pixel component
13986 from first stream is returned unchanged, while maximum value (eg. 255 for
13987 8-bit videos) means that pixel component from second stream is returned
13988 unchanged. Intermediate values define the amount of merging between both
13989 input stream's pixel components.
13991 This filter accepts the following options:
13994 Set which planes will be processed as bitmap, unprocessed planes will be
13995 copied from first stream.
13996 By default value 0xf, all planes will be processed.
14001 Merge the second and third input stream into output stream using absolute differences
14002 between second input stream and first input stream and absolute difference between
14003 third input stream and first input stream. The picked value will be from second input
14004 stream if second absolute difference is less than first one or from third input stream
14007 This filter accepts the following options:
14010 Set which planes will be processed as bitmap, unprocessed planes will be
14011 copied from first stream.
14012 By default value 0xf, all planes will be processed.
14015 @section maskedthreshold
14016 Pick pixels comparing absolute difference of two video streams with fixed
14019 If absolute difference between pixel component of first and second video
14020 stream is equal or lower than user supplied threshold than pixel component
14021 from first video stream is picked, otherwise pixel component from second
14022 video stream is picked.
14024 This filter accepts the following options:
14027 Set threshold used when picking pixels from absolute difference from two input
14031 Set which planes will be processed as bitmap, unprocessed planes will be
14032 copied from second stream.
14033 By default value 0xf, all planes will be processed.
14037 Create mask from input video.
14039 For example it is useful to create motion masks after @code{tblend} filter.
14041 This filter accepts the following options:
14045 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
14048 Set high threshold. Any pixel component higher than this value will be set to max value
14049 allowed for current pixel format.
14052 Set planes to filter, by default all available planes are filtered.
14055 Fill all frame pixels with this value.
14058 Set max average pixel value for frame. If sum of all pixel components is higher that this
14059 average, output frame will be completely filled with value set by @var{fill} option.
14060 Typically useful for scene changes when used in combination with @code{tblend} filter.
14065 Apply motion-compensation deinterlacing.
14067 It needs one field per frame as input and must thus be used together
14068 with yadif=1/3 or equivalent.
14070 This filter accepts the following options:
14073 Set the deinterlacing mode.
14075 It accepts one of the following values:
14080 use iterative motion estimation
14082 like @samp{slow}, but use multiple reference frames.
14084 Default value is @samp{fast}.
14087 Set the picture field parity assumed for the input video. It must be
14088 one of the following values:
14092 assume top field first
14094 assume bottom field first
14097 Default value is @samp{bff}.
14100 Set per-block quantization parameter (QP) used by the internal
14103 Higher values should result in a smoother motion vector field but less
14104 optimal individual vectors. Default value is 1.
14109 Pick median pixel from certain rectangle defined by radius.
14111 This filter accepts the following options:
14115 Set horizontal radius size. Default value is @code{1}.
14116 Allowed range is integer from 1 to 127.
14119 Set which planes to process. Default is @code{15}, which is all available planes.
14122 Set vertical radius size. Default value is @code{0}.
14123 Allowed range is integer from 0 to 127.
14124 If it is 0, value will be picked from horizontal @code{radius} option.
14127 Set median percentile. Default value is @code{0.5}.
14128 Default value of @code{0.5} will pick always median values, while @code{0} will pick
14129 minimum values, and @code{1} maximum values.
14132 @subsection Commands
14133 This filter supports same @ref{commands} as options.
14134 The command accepts the same syntax of the corresponding option.
14136 If the specified expression is not valid, it is kept at its current
14139 @section mergeplanes
14141 Merge color channel components from several video streams.
14143 The filter accepts up to 4 input streams, and merge selected input
14144 planes to the output video.
14146 This filter accepts the following options:
14149 Set input to output plane mapping. Default is @code{0}.
14151 The mappings is specified as a bitmap. It should be specified as a
14152 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
14153 mapping for the first plane of the output stream. 'A' sets the number of
14154 the input stream to use (from 0 to 3), and 'a' the plane number of the
14155 corresponding input to use (from 0 to 3). The rest of the mappings is
14156 similar, 'Bb' describes the mapping for the output stream second
14157 plane, 'Cc' describes the mapping for the output stream third plane and
14158 'Dd' describes the mapping for the output stream fourth plane.
14161 Set output pixel format. Default is @code{yuva444p}.
14164 @subsection Examples
14168 Merge three gray video streams of same width and height into single video stream:
14170 [a0][a1][a2]mergeplanes=0x001020:yuv444p
14174 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
14176 [a0][a1]mergeplanes=0x00010210:yuva444p
14180 Swap Y and A plane in yuva444p stream:
14182 format=yuva444p,mergeplanes=0x03010200:yuva444p
14186 Swap U and V plane in yuv420p stream:
14188 format=yuv420p,mergeplanes=0x000201:yuv420p
14192 Cast a rgb24 clip to yuv444p:
14194 format=rgb24,mergeplanes=0x000102:yuv444p
14200 Estimate and export motion vectors using block matching algorithms.
14201 Motion vectors are stored in frame side data to be used by other filters.
14203 This filter accepts the following options:
14206 Specify the motion estimation method. Accepts one of the following values:
14210 Exhaustive search algorithm.
14212 Three step search algorithm.
14214 Two dimensional logarithmic search algorithm.
14216 New three step search algorithm.
14218 Four step search algorithm.
14220 Diamond search algorithm.
14222 Hexagon-based search algorithm.
14224 Enhanced predictive zonal search algorithm.
14226 Uneven multi-hexagon search algorithm.
14228 Default value is @samp{esa}.
14231 Macroblock size. Default @code{16}.
14234 Search parameter. Default @code{7}.
14237 @section midequalizer
14239 Apply Midway Image Equalization effect using two video streams.
14241 Midway Image Equalization adjusts a pair of images to have the same
14242 histogram, while maintaining their dynamics as much as possible. It's
14243 useful for e.g. matching exposures from a pair of stereo cameras.
14245 This filter has two inputs and one output, which must be of same pixel format, but
14246 may be of different sizes. The output of filter is first input adjusted with
14247 midway histogram of both inputs.
14249 This filter accepts the following option:
14253 Set which planes to process. Default is @code{15}, which is all available planes.
14256 @section minterpolate
14258 Convert the video to specified frame rate using motion interpolation.
14260 This filter accepts the following options:
14263 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}.
14266 Motion interpolation mode. Following values are accepted:
14269 Duplicate previous or next frame for interpolating new ones.
14271 Blend source frames. Interpolated frame is mean of previous and next frames.
14273 Motion compensated interpolation. Following options are effective when this mode is selected:
14277 Motion compensation mode. Following values are accepted:
14280 Overlapped block motion compensation.
14282 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
14284 Default mode is @samp{obmc}.
14287 Motion estimation mode. Following values are accepted:
14290 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14292 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14294 Default mode is @samp{bilat}.
14297 The algorithm to be used for motion estimation. Following values are accepted:
14300 Exhaustive search algorithm.
14302 Three step search algorithm.
14304 Two dimensional logarithmic search algorithm.
14306 New three step search algorithm.
14308 Four step search algorithm.
14310 Diamond search algorithm.
14312 Hexagon-based search algorithm.
14314 Enhanced predictive zonal search algorithm.
14316 Uneven multi-hexagon search algorithm.
14318 Default algorithm is @samp{epzs}.
14321 Macroblock size. Default @code{16}.
14324 Motion estimation search parameter. Default @code{32}.
14327 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).
14332 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:
14335 Disable scene change detection.
14337 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14339 Default method is @samp{fdiff}.
14341 @item scd_threshold
14342 Scene change detection threshold. Default is @code{10.}.
14347 Mix several video input streams into one video stream.
14349 A description of the accepted options follows.
14353 The number of inputs. If unspecified, it defaults to 2.
14356 Specify weight of each input video stream as sequence.
14357 Each weight is separated by space. If number of weights
14358 is smaller than number of @var{frames} last specified
14359 weight will be used for all remaining unset weights.
14362 Specify scale, if it is set it will be multiplied with sum
14363 of each weight multiplied with pixel values to give final destination
14364 pixel value. By default @var{scale} is auto scaled to sum of weights.
14367 Specify how end of stream is determined.
14370 The duration of the longest input. (default)
14373 The duration of the shortest input.
14376 The duration of the first input.
14380 @section mpdecimate
14382 Drop frames that do not differ greatly from the previous frame in
14383 order to reduce frame rate.
14385 The main use of this filter is for very-low-bitrate encoding
14386 (e.g. streaming over dialup modem), but it could in theory be used for
14387 fixing movies that were inverse-telecined incorrectly.
14389 A description of the accepted options follows.
14393 Set the maximum number of consecutive frames which can be dropped (if
14394 positive), or the minimum interval between dropped frames (if
14395 negative). If the value is 0, the frame is dropped disregarding the
14396 number of previous sequentially dropped frames.
14398 Default value is 0.
14403 Set the dropping threshold values.
14405 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14406 represent actual pixel value differences, so a threshold of 64
14407 corresponds to 1 unit of difference for each pixel, or the same spread
14408 out differently over the block.
14410 A frame is a candidate for dropping if no 8x8 blocks differ by more
14411 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14412 meaning the whole image) differ by more than a threshold of @option{lo}.
14414 Default value for @option{hi} is 64*12, default value for @option{lo} is
14415 64*5, and default value for @option{frac} is 0.33.
14421 Negate (invert) the input video.
14423 It accepts the following option:
14428 With value 1, it negates the alpha component, if present. Default value is 0.
14434 Denoise frames using Non-Local Means algorithm.
14436 Each pixel is adjusted by looking for other pixels with similar contexts. This
14437 context similarity is defined by comparing their surrounding patches of size
14438 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14441 Note that the research area defines centers for patches, which means some
14442 patches will be made of pixels outside that research area.
14444 The filter accepts the following options.
14448 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14451 Set patch size. Default is 7. Must be odd number in range [0, 99].
14454 Same as @option{p} but for chroma planes.
14456 The default value is @var{0} and means automatic.
14459 Set research size. Default is 15. Must be odd number in range [0, 99].
14462 Same as @option{r} but for chroma planes.
14464 The default value is @var{0} and means automatic.
14469 Deinterlace video using neural network edge directed interpolation.
14471 This filter accepts the following options:
14475 Mandatory option, without binary file filter can not work.
14476 Currently file can be found here:
14477 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14480 Set which frames to deinterlace, by default it is @code{all}.
14481 Can be @code{all} or @code{interlaced}.
14484 Set mode of operation.
14486 Can be one of the following:
14490 Use frame flags, both fields.
14492 Use frame flags, single field.
14494 Use top field only.
14496 Use bottom field only.
14498 Use both fields, top first.
14500 Use both fields, bottom first.
14504 Set which planes to process, by default filter process all frames.
14507 Set size of local neighborhood around each pixel, used by the predictor neural
14510 Can be one of the following:
14523 Set the number of neurons in predictor neural network.
14524 Can be one of the following:
14535 Controls the number of different neural network predictions that are blended
14536 together to compute the final output value. Can be @code{fast}, default or
14540 Set which set of weights to use in the predictor.
14541 Can be one of the following:
14545 weights trained to minimize absolute error
14547 weights trained to minimize squared error
14551 Controls whether or not the prescreener neural network is used to decide
14552 which pixels should be processed by the predictor neural network and which
14553 can be handled by simple cubic interpolation.
14554 The prescreener is trained to know whether cubic interpolation will be
14555 sufficient for a pixel or whether it should be predicted by the predictor nn.
14556 The computational complexity of the prescreener nn is much less than that of
14557 the predictor nn. Since most pixels can be handled by cubic interpolation,
14558 using the prescreener generally results in much faster processing.
14559 The prescreener is pretty accurate, so the difference between using it and not
14560 using it is almost always unnoticeable.
14562 Can be one of the following:
14570 Default is @code{new}.
14573 Set various debugging flags.
14578 Force libavfilter not to use any of the specified pixel formats for the
14579 input to the next filter.
14581 It accepts the following parameters:
14585 A '|'-separated list of pixel format names, such as
14586 pix_fmts=yuv420p|monow|rgb24".
14590 @subsection Examples
14594 Force libavfilter to use a format different from @var{yuv420p} for the
14595 input to the vflip filter:
14597 noformat=pix_fmts=yuv420p,vflip
14601 Convert the input video to any of the formats not contained in the list:
14603 noformat=yuv420p|yuv444p|yuv410p
14609 Add noise on video input frame.
14611 The filter accepts the following options:
14619 Set noise seed for specific pixel component or all pixel components in case
14620 of @var{all_seed}. Default value is @code{123457}.
14622 @item all_strength, alls
14623 @item c0_strength, c0s
14624 @item c1_strength, c1s
14625 @item c2_strength, c2s
14626 @item c3_strength, c3s
14627 Set noise strength for specific pixel component or all pixel components in case
14628 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14630 @item all_flags, allf
14631 @item c0_flags, c0f
14632 @item c1_flags, c1f
14633 @item c2_flags, c2f
14634 @item c3_flags, c3f
14635 Set pixel component flags or set flags for all components if @var{all_flags}.
14636 Available values for component flags are:
14639 averaged temporal noise (smoother)
14641 mix random noise with a (semi)regular pattern
14643 temporal noise (noise pattern changes between frames)
14645 uniform noise (gaussian otherwise)
14649 @subsection Examples
14651 Add temporal and uniform noise to input video:
14653 noise=alls=20:allf=t+u
14658 Normalize RGB video (aka histogram stretching, contrast stretching).
14659 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14661 For each channel of each frame, the filter computes the input range and maps
14662 it linearly to the user-specified output range. The output range defaults
14663 to the full dynamic range from pure black to pure white.
14665 Temporal smoothing can be used on the input range to reduce flickering (rapid
14666 changes in brightness) caused when small dark or bright objects enter or leave
14667 the scene. This is similar to the auto-exposure (automatic gain control) on a
14668 video camera, and, like a video camera, it may cause a period of over- or
14669 under-exposure of the video.
14671 The R,G,B channels can be normalized independently, which may cause some
14672 color shifting, or linked together as a single channel, which prevents
14673 color shifting. Linked normalization preserves hue. Independent normalization
14674 does not, so it can be used to remove some color casts. Independent and linked
14675 normalization can be combined in any ratio.
14677 The normalize filter accepts the following options:
14682 Colors which define the output range. The minimum input value is mapped to
14683 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14684 The defaults are black and white respectively. Specifying white for
14685 @var{blackpt} and black for @var{whitept} will give color-inverted,
14686 normalized video. Shades of grey can be used to reduce the dynamic range
14687 (contrast). Specifying saturated colors here can create some interesting
14691 The number of previous frames to use for temporal smoothing. The input range
14692 of each channel is smoothed using a rolling average over the current frame
14693 and the @var{smoothing} previous frames. The default is 0 (no temporal
14697 Controls the ratio of independent (color shifting) channel normalization to
14698 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14699 independent. Defaults to 1.0 (fully independent).
14702 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14703 expensive no-op. Defaults to 1.0 (full strength).
14707 @subsection Commands
14708 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14709 The command accepts the same syntax of the corresponding option.
14711 If the specified expression is not valid, it is kept at its current
14714 @subsection Examples
14716 Stretch video contrast to use the full dynamic range, with no temporal
14717 smoothing; may flicker depending on the source content:
14719 normalize=blackpt=black:whitept=white:smoothing=0
14722 As above, but with 50 frames of temporal smoothing; flicker should be
14723 reduced, depending on the source content:
14725 normalize=blackpt=black:whitept=white:smoothing=50
14728 As above, but with hue-preserving linked channel normalization:
14730 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14733 As above, but with half strength:
14735 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14738 Map the darkest input color to red, the brightest input color to cyan:
14740 normalize=blackpt=red:whitept=cyan
14745 Pass the video source unchanged to the output.
14748 Optical Character Recognition
14750 This filter uses Tesseract for optical character recognition. To enable
14751 compilation of this filter, you need to configure FFmpeg with
14752 @code{--enable-libtesseract}.
14754 It accepts the following options:
14758 Set datapath to tesseract data. Default is to use whatever was
14759 set at installation.
14762 Set language, default is "eng".
14765 Set character whitelist.
14768 Set character blacklist.
14771 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14772 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14776 Apply a video transform using libopencv.
14778 To enable this filter, install the libopencv library and headers and
14779 configure FFmpeg with @code{--enable-libopencv}.
14781 It accepts the following parameters:
14786 The name of the libopencv filter to apply.
14788 @item filter_params
14789 The parameters to pass to the libopencv filter. If not specified, the default
14790 values are assumed.
14794 Refer to the official libopencv documentation for more precise
14796 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14798 Several libopencv filters are supported; see the following subsections.
14803 Dilate an image by using a specific structuring element.
14804 It corresponds to the libopencv function @code{cvDilate}.
14806 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14808 @var{struct_el} represents a structuring element, and has the syntax:
14809 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14811 @var{cols} and @var{rows} represent the number of columns and rows of
14812 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14813 point, and @var{shape} the shape for the structuring element. @var{shape}
14814 must be "rect", "cross", "ellipse", or "custom".
14816 If the value for @var{shape} is "custom", it must be followed by a
14817 string of the form "=@var{filename}". The file with name
14818 @var{filename} is assumed to represent a binary image, with each
14819 printable character corresponding to a bright pixel. When a custom
14820 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14821 or columns and rows of the read file are assumed instead.
14823 The default value for @var{struct_el} is "3x3+0x0/rect".
14825 @var{nb_iterations} specifies the number of times the transform is
14826 applied to the image, and defaults to 1.
14830 # Use the default values
14833 # Dilate using a structuring element with a 5x5 cross, iterating two times
14834 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14836 # Read the shape from the file diamond.shape, iterating two times.
14837 # The file diamond.shape may contain a pattern of characters like this
14843 # The specified columns and rows are ignored
14844 # but the anchor point coordinates are not
14845 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14850 Erode an image by using a specific structuring element.
14851 It corresponds to the libopencv function @code{cvErode}.
14853 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14854 with the same syntax and semantics as the @ref{dilate} filter.
14858 Smooth the input video.
14860 The filter takes the following parameters:
14861 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14863 @var{type} is the type of smooth filter to apply, and must be one of
14864 the following values: "blur", "blur_no_scale", "median", "gaussian",
14865 or "bilateral". The default value is "gaussian".
14867 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14868 depends on the smooth type. @var{param1} and
14869 @var{param2} accept integer positive values or 0. @var{param3} and
14870 @var{param4} accept floating point values.
14872 The default value for @var{param1} is 3. The default value for the
14873 other parameters is 0.
14875 These parameters correspond to the parameters assigned to the
14876 libopencv function @code{cvSmooth}.
14878 @section oscilloscope
14880 2D Video Oscilloscope.
14882 Useful to measure spatial impulse, step responses, chroma delays, etc.
14884 It accepts the following parameters:
14888 Set scope center x position.
14891 Set scope center y position.
14894 Set scope size, relative to frame diagonal.
14897 Set scope tilt/rotation.
14903 Set trace center x position.
14906 Set trace center y position.
14909 Set trace width, relative to width of frame.
14912 Set trace height, relative to height of frame.
14915 Set which components to trace. By default it traces first three components.
14918 Draw trace grid. By default is enabled.
14921 Draw some statistics. By default is enabled.
14924 Draw scope. By default is enabled.
14927 @subsection Commands
14928 This filter supports same @ref{commands} as options.
14929 The command accepts the same syntax of the corresponding option.
14931 If the specified expression is not valid, it is kept at its current
14934 @subsection Examples
14938 Inspect full first row of video frame.
14940 oscilloscope=x=0.5:y=0:s=1
14944 Inspect full last row of video frame.
14946 oscilloscope=x=0.5:y=1:s=1
14950 Inspect full 5th line of video frame of height 1080.
14952 oscilloscope=x=0.5:y=5/1080:s=1
14956 Inspect full last column of video frame.
14958 oscilloscope=x=1:y=0.5:s=1:t=1
14966 Overlay one video on top of another.
14968 It takes two inputs and has one output. The first input is the "main"
14969 video on which the second input is overlaid.
14971 It accepts the following parameters:
14973 A description of the accepted options follows.
14978 Set the expression for the x and y coordinates of the overlaid video
14979 on the main video. Default value is "0" for both expressions. In case
14980 the expression is invalid, it is set to a huge value (meaning that the
14981 overlay will not be displayed within the output visible area).
14984 See @ref{framesync}.
14987 Set when the expressions for @option{x}, and @option{y} are evaluated.
14989 It accepts the following values:
14992 only evaluate expressions once during the filter initialization or
14993 when a command is processed
14996 evaluate expressions for each incoming frame
14999 Default value is @samp{frame}.
15002 See @ref{framesync}.
15005 Set the format for the output video.
15007 It accepts the following values:
15010 force YUV420 output
15013 force YUV420p10 output
15016 force YUV422 output
15019 force YUV422p10 output
15022 force YUV444 output
15025 force packed RGB output
15028 force planar RGB output
15031 automatically pick format
15034 Default value is @samp{yuv420}.
15037 See @ref{framesync}.
15040 Set format of alpha of the overlaid video, it can be @var{straight} or
15041 @var{premultiplied}. Default is @var{straight}.
15044 The @option{x}, and @option{y} expressions can contain the following
15050 The main input width and height.
15054 The overlay input width and height.
15058 The computed values for @var{x} and @var{y}. They are evaluated for
15063 horizontal and vertical chroma subsample values of the output
15064 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
15068 the number of input frame, starting from 0
15071 the position in the file of the input frame, NAN if unknown
15074 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
15078 This filter also supports the @ref{framesync} options.
15080 Note that the @var{n}, @var{pos}, @var{t} variables are available only
15081 when evaluation is done @emph{per frame}, and will evaluate to NAN
15082 when @option{eval} is set to @samp{init}.
15084 Be aware that frames are taken from each input video in timestamp
15085 order, hence, if their initial timestamps differ, it is a good idea
15086 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
15087 have them begin in the same zero timestamp, as the example for
15088 the @var{movie} filter does.
15090 You can chain together more overlays but you should test the
15091 efficiency of such approach.
15093 @subsection Commands
15095 This filter supports the following commands:
15099 Modify the x and y of the overlay input.
15100 The command accepts the same syntax of the corresponding option.
15102 If the specified expression is not valid, it is kept at its current
15106 @subsection Examples
15110 Draw the overlay at 10 pixels from the bottom right corner of the main
15113 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15116 Using named options the example above becomes:
15118 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15122 Insert a transparent PNG logo in the bottom left corner of the input,
15123 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
15125 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15129 Insert 2 different transparent PNG logos (second logo on bottom
15130 right corner) using the @command{ffmpeg} tool:
15132 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
15136 Add a transparent color layer on top of the main video; @code{WxH}
15137 must specify the size of the main input to the overlay filter:
15139 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
15143 Play an original video and a filtered version (here with the deshake
15144 filter) side by side using the @command{ffplay} tool:
15146 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15149 The above command is the same as:
15151 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15155 Make a sliding overlay appearing from the left to the right top part of the
15156 screen starting since time 2:
15158 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15162 Compose output by putting two input videos side to side:
15164 ffmpeg -i left.avi -i right.avi -filter_complex "
15165 nullsrc=size=200x100 [background];
15166 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15167 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15168 [background][left] overlay=shortest=1 [background+left];
15169 [background+left][right] overlay=shortest=1:x=100 [left+right]
15174 Mask 10-20 seconds of a video by applying the delogo filter to a section
15176 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15177 -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]'
15182 Chain several overlays in cascade:
15184 nullsrc=s=200x200 [bg];
15185 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15186 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15187 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15188 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15189 [in3] null, [mid2] overlay=100:100 [out0]
15194 @anchor{overlay_cuda}
15195 @section overlay_cuda
15197 Overlay one video on top of another.
15199 This is the CUDA variant of the @ref{overlay} filter.
15200 It only accepts CUDA frames. The underlying input pixel formats have to match.
15202 It takes two inputs and has one output. The first input is the "main"
15203 video on which the second input is overlaid.
15205 It accepts the following parameters:
15210 Set the x and y coordinates of the overlaid video on the main video.
15211 Default value is "0" for both expressions.
15214 See @ref{framesync}.
15217 See @ref{framesync}.
15220 See @ref{framesync}.
15224 This filter also supports the @ref{framesync} options.
15228 Apply Overcomplete Wavelet denoiser.
15230 The filter accepts the following options:
15236 Larger depth values will denoise lower frequency components more, but
15237 slow down filtering.
15239 Must be an int in the range 8-16, default is @code{8}.
15241 @item luma_strength, ls
15244 Must be a double value in the range 0-1000, default is @code{1.0}.
15246 @item chroma_strength, cs
15247 Set chroma strength.
15249 Must be a double value in the range 0-1000, default is @code{1.0}.
15255 Add paddings to the input image, and place the original input at the
15256 provided @var{x}, @var{y} coordinates.
15258 It accepts the following parameters:
15263 Specify an expression for the size of the output image with the
15264 paddings added. If the value for @var{width} or @var{height} is 0, the
15265 corresponding input size is used for the output.
15267 The @var{width} expression can reference the value set by the
15268 @var{height} expression, and vice versa.
15270 The default value of @var{width} and @var{height} is 0.
15274 Specify the offsets to place the input image at within the padded area,
15275 with respect to the top/left border of the output image.
15277 The @var{x} expression can reference the value set by the @var{y}
15278 expression, and vice versa.
15280 The default value of @var{x} and @var{y} is 0.
15282 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15283 so the input image is centered on the padded area.
15286 Specify the color of the padded area. For the syntax of this option,
15287 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15288 manual,ffmpeg-utils}.
15290 The default value of @var{color} is "black".
15293 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15295 It accepts the following values:
15299 Only evaluate expressions once during the filter initialization or when
15300 a command is processed.
15303 Evaluate expressions for each incoming frame.
15307 Default value is @samp{init}.
15310 Pad to aspect instead to a resolution.
15314 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15315 options are expressions containing the following constants:
15320 The input video width and height.
15324 These are the same as @var{in_w} and @var{in_h}.
15328 The output width and height (the size of the padded area), as
15329 specified by the @var{width} and @var{height} expressions.
15333 These are the same as @var{out_w} and @var{out_h}.
15337 The x and y offsets as specified by the @var{x} and @var{y}
15338 expressions, or NAN if not yet specified.
15341 same as @var{iw} / @var{ih}
15344 input sample aspect ratio
15347 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15351 The horizontal and vertical chroma subsample values. For example for the
15352 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15355 @subsection Examples
15359 Add paddings with the color "violet" to the input video. The output video
15360 size is 640x480, and the top-left corner of the input video is placed at
15363 pad=640:480:0:40:violet
15366 The example above is equivalent to the following command:
15368 pad=width=640:height=480:x=0:y=40:color=violet
15372 Pad the input to get an output with dimensions increased by 3/2,
15373 and put the input video at the center of the padded area:
15375 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15379 Pad the input to get a squared output with size equal to the maximum
15380 value between the input width and height, and put the input video at
15381 the center of the padded area:
15383 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15387 Pad the input to get a final w/h ratio of 16:9:
15389 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15393 In case of anamorphic video, in order to set the output display aspect
15394 correctly, it is necessary to use @var{sar} in the expression,
15395 according to the relation:
15397 (ih * X / ih) * sar = output_dar
15398 X = output_dar / sar
15401 Thus the previous example needs to be modified to:
15403 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15407 Double the output size and put the input video in the bottom-right
15408 corner of the output padded area:
15410 pad="2*iw:2*ih:ow-iw:oh-ih"
15414 @anchor{palettegen}
15415 @section palettegen
15417 Generate one palette for a whole video stream.
15419 It accepts the following options:
15423 Set the maximum number of colors to quantize in the palette.
15424 Note: the palette will still contain 256 colors; the unused palette entries
15427 @item reserve_transparent
15428 Create a palette of 255 colors maximum and reserve the last one for
15429 transparency. Reserving the transparency color is useful for GIF optimization.
15430 If not set, the maximum of colors in the palette will be 256. You probably want
15431 to disable this option for a standalone image.
15434 @item transparency_color
15435 Set the color that will be used as background for transparency.
15438 Set statistics mode.
15440 It accepts the following values:
15443 Compute full frame histograms.
15445 Compute histograms only for the part that differs from previous frame. This
15446 might be relevant to give more importance to the moving part of your input if
15447 the background is static.
15449 Compute new histogram for each frame.
15452 Default value is @var{full}.
15455 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15456 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15457 color quantization of the palette. This information is also visible at
15458 @var{info} logging level.
15460 @subsection Examples
15464 Generate a representative palette of a given video using @command{ffmpeg}:
15466 ffmpeg -i input.mkv -vf palettegen palette.png
15470 @section paletteuse
15472 Use a palette to downsample an input video stream.
15474 The filter takes two inputs: one video stream and a palette. The palette must
15475 be a 256 pixels image.
15477 It accepts the following options:
15481 Select dithering mode. Available algorithms are:
15484 Ordered 8x8 bayer dithering (deterministic)
15486 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15487 Note: this dithering is sometimes considered "wrong" and is included as a
15489 @item floyd_steinberg
15490 Floyd and Steingberg dithering (error diffusion)
15492 Frankie Sierra dithering v2 (error diffusion)
15494 Frankie Sierra dithering v2 "Lite" (error diffusion)
15497 Default is @var{sierra2_4a}.
15500 When @var{bayer} dithering is selected, this option defines the scale of the
15501 pattern (how much the crosshatch pattern is visible). A low value means more
15502 visible pattern for less banding, and higher value means less visible pattern
15503 at the cost of more banding.
15505 The option must be an integer value in the range [0,5]. Default is @var{2}.
15508 If set, define the zone to process
15512 Only the changing rectangle will be reprocessed. This is similar to GIF
15513 cropping/offsetting compression mechanism. This option can be useful for speed
15514 if only a part of the image is changing, and has use cases such as limiting the
15515 scope of the error diffusal @option{dither} to the rectangle that bounds the
15516 moving scene (it leads to more deterministic output if the scene doesn't change
15517 much, and as a result less moving noise and better GIF compression).
15520 Default is @var{none}.
15523 Take new palette for each output frame.
15525 @item alpha_threshold
15526 Sets the alpha threshold for transparency. Alpha values above this threshold
15527 will be treated as completely opaque, and values below this threshold will be
15528 treated as completely transparent.
15530 The option must be an integer value in the range [0,255]. Default is @var{128}.
15533 @subsection Examples
15537 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15538 using @command{ffmpeg}:
15540 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15544 @section perspective
15546 Correct perspective of video not recorded perpendicular to the screen.
15548 A description of the accepted parameters follows.
15559 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15560 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15561 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15562 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15563 then the corners of the source will be sent to the specified coordinates.
15565 The expressions can use the following variables:
15570 the width and height of video frame.
15574 Output frame count.
15577 @item interpolation
15578 Set interpolation for perspective correction.
15580 It accepts the following values:
15586 Default value is @samp{linear}.
15589 Set interpretation of coordinate options.
15591 It accepts the following values:
15595 Send point in the source specified by the given coordinates to
15596 the corners of the destination.
15598 @item 1, destination
15600 Send the corners of the source to the point in the destination specified
15601 by the given coordinates.
15603 Default value is @samp{source}.
15607 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15609 It accepts the following values:
15612 only evaluate expressions once during the filter initialization or
15613 when a command is processed
15616 evaluate expressions for each incoming frame
15619 Default value is @samp{init}.
15624 Delay interlaced video by one field time so that the field order changes.
15626 The intended use is to fix PAL movies that have been captured with the
15627 opposite field order to the film-to-video transfer.
15629 A description of the accepted parameters follows.
15635 It accepts the following values:
15638 Capture field order top-first, transfer bottom-first.
15639 Filter will delay the bottom field.
15642 Capture field order bottom-first, transfer top-first.
15643 Filter will delay the top field.
15646 Capture and transfer with the same field order. This mode only exists
15647 for the documentation of the other options to refer to, but if you
15648 actually select it, the filter will faithfully do nothing.
15651 Capture field order determined automatically by field flags, transfer
15653 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15654 basis using field flags. If no field information is available,
15655 then this works just like @samp{u}.
15658 Capture unknown or varying, transfer opposite.
15659 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15660 analyzing the images and selecting the alternative that produces best
15661 match between the fields.
15664 Capture top-first, transfer unknown or varying.
15665 Filter selects among @samp{t} and @samp{p} using image analysis.
15668 Capture bottom-first, transfer unknown or varying.
15669 Filter selects among @samp{b} and @samp{p} using image analysis.
15672 Capture determined by field flags, transfer unknown or varying.
15673 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15674 image analysis. If no field information is available, then this works just
15675 like @samp{U}. This is the default mode.
15678 Both capture and transfer unknown or varying.
15679 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15683 @section photosensitivity
15684 Reduce various flashes in video, so to help users with epilepsy.
15686 It accepts the following options:
15689 Set how many frames to use when filtering. Default is 30.
15692 Set detection threshold factor. Default is 1.
15696 Set how many pixels to skip when sampling frames. Default is 1.
15697 Allowed range is from 1 to 1024.
15700 Leave frames unchanged. Default is disabled.
15703 @section pixdesctest
15705 Pixel format descriptor test filter, mainly useful for internal
15706 testing. The output video should be equal to the input video.
15710 format=monow, pixdesctest
15713 can be used to test the monowhite pixel format descriptor definition.
15717 Display sample values of color channels. Mainly useful for checking color
15718 and levels. Minimum supported resolution is 640x480.
15720 The filters accept the following options:
15724 Set scope X position, relative offset on X axis.
15727 Set scope Y position, relative offset on Y axis.
15736 Set window opacity. This window also holds statistics about pixel area.
15739 Set window X position, relative offset on X axis.
15742 Set window Y position, relative offset on Y axis.
15747 Enable the specified chain of postprocessing subfilters using libpostproc. This
15748 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15749 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15750 Each subfilter and some options have a short and a long name that can be used
15751 interchangeably, i.e. dr/dering are the same.
15753 The filters accept the following options:
15757 Set postprocessing subfilters string.
15760 All subfilters share common options to determine their scope:
15764 Honor the quality commands for this subfilter.
15767 Do chrominance filtering, too (default).
15770 Do luminance filtering only (no chrominance).
15773 Do chrominance filtering only (no luminance).
15776 These options can be appended after the subfilter name, separated by a '|'.
15778 Available subfilters are:
15781 @item hb/hdeblock[|difference[|flatness]]
15782 Horizontal deblocking filter
15785 Difference factor where higher values mean more deblocking (default: @code{32}).
15787 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15790 @item vb/vdeblock[|difference[|flatness]]
15791 Vertical deblocking filter
15794 Difference factor where higher values mean more deblocking (default: @code{32}).
15796 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15799 @item ha/hadeblock[|difference[|flatness]]
15800 Accurate horizontal deblocking filter
15803 Difference factor where higher values mean more deblocking (default: @code{32}).
15805 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15808 @item va/vadeblock[|difference[|flatness]]
15809 Accurate vertical deblocking filter
15812 Difference factor where higher values mean more deblocking (default: @code{32}).
15814 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15818 The horizontal and vertical deblocking filters share the difference and
15819 flatness values so you cannot set different horizontal and vertical
15823 @item h1/x1hdeblock
15824 Experimental horizontal deblocking filter
15826 @item v1/x1vdeblock
15827 Experimental vertical deblocking filter
15832 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15835 larger -> stronger filtering
15837 larger -> stronger filtering
15839 larger -> stronger filtering
15842 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15845 Stretch luminance to @code{0-255}.
15848 @item lb/linblenddeint
15849 Linear blend deinterlacing filter that deinterlaces the given block by
15850 filtering all lines with a @code{(1 2 1)} filter.
15852 @item li/linipoldeint
15853 Linear interpolating deinterlacing filter that deinterlaces the given block by
15854 linearly interpolating every second line.
15856 @item ci/cubicipoldeint
15857 Cubic interpolating deinterlacing filter deinterlaces the given block by
15858 cubically interpolating every second line.
15860 @item md/mediandeint
15861 Median deinterlacing filter that deinterlaces the given block by applying a
15862 median filter to every second line.
15864 @item fd/ffmpegdeint
15865 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15866 second line with a @code{(-1 4 2 4 -1)} filter.
15869 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15870 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15872 @item fq/forceQuant[|quantizer]
15873 Overrides the quantizer table from the input with the constant quantizer you
15881 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15884 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15887 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15890 @subsection Examples
15894 Apply horizontal and vertical deblocking, deringing and automatic
15895 brightness/contrast:
15901 Apply default filters without brightness/contrast correction:
15907 Apply default filters and temporal denoiser:
15909 pp=default/tmpnoise|1|2|3
15913 Apply deblocking on luminance only, and switch vertical deblocking on or off
15914 automatically depending on available CPU time:
15921 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15922 similar to spp = 6 with 7 point DCT, where only the center sample is
15925 The filter accepts the following options:
15929 Force a constant quantization parameter. It accepts an integer in range
15930 0 to 63. If not set, the filter will use the QP from the video stream
15934 Set thresholding mode. Available modes are:
15938 Set hard thresholding.
15940 Set soft thresholding (better de-ringing effect, but likely blurrier).
15942 Set medium thresholding (good results, default).
15946 @section premultiply
15947 Apply alpha premultiply effect to input video stream using first plane
15948 of second stream as alpha.
15950 Both streams must have same dimensions and same pixel format.
15952 The filter accepts the following option:
15956 Set which planes will be processed, unprocessed planes will be copied.
15957 By default value 0xf, all planes will be processed.
15960 Do not require 2nd input for processing, instead use alpha plane from input stream.
15964 Apply prewitt operator to input video stream.
15966 The filter accepts the following option:
15970 Set which planes will be processed, unprocessed planes will be copied.
15971 By default value 0xf, all planes will be processed.
15974 Set value which will be multiplied with filtered result.
15977 Set value which will be added to filtered result.
15980 @subsection Commands
15982 This filter supports the all above options as @ref{commands}.
15984 @section pseudocolor
15986 Alter frame colors in video with pseudocolors.
15988 This filter accepts the following options:
15992 set pixel first component expression
15995 set pixel second component expression
15998 set pixel third component expression
16001 set pixel fourth component expression, corresponds to the alpha component
16004 set component to use as base for altering colors
16007 Each of them specifies the expression to use for computing the lookup table for
16008 the corresponding pixel component values.
16010 The expressions can contain the following constants and functions:
16015 The input width and height.
16018 The input value for the pixel component.
16020 @item ymin, umin, vmin, amin
16021 The minimum allowed component value.
16023 @item ymax, umax, vmax, amax
16024 The maximum allowed component value.
16027 All expressions default to "val".
16029 @subsection Examples
16033 Change too high luma values to gradient:
16035 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'"
16041 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16042 Ratio) between two input videos.
16044 This filter takes in input two input videos, the first input is
16045 considered the "main" source and is passed unchanged to the
16046 output. The second input is used as a "reference" video for computing
16049 Both video inputs must have the same resolution and pixel format for
16050 this filter to work correctly. Also it assumes that both inputs
16051 have the same number of frames, which are compared one by one.
16053 The obtained average PSNR is printed through the logging system.
16055 The filter stores the accumulated MSE (mean squared error) of each
16056 frame, and at the end of the processing it is averaged across all frames
16057 equally, and the following formula is applied to obtain the PSNR:
16060 PSNR = 10*log10(MAX^2/MSE)
16063 Where MAX is the average of the maximum values of each component of the
16066 The description of the accepted parameters follows.
16069 @item stats_file, f
16070 If specified the filter will use the named file to save the PSNR of
16071 each individual frame. When filename equals "-" the data is sent to
16074 @item stats_version
16075 Specifies which version of the stats file format to use. Details of
16076 each format are written below.
16077 Default value is 1.
16079 @item stats_add_max
16080 Determines whether the max value is output to the stats log.
16081 Default value is 0.
16082 Requires stats_version >= 2. If this is set and stats_version < 2,
16083 the filter will return an error.
16086 This filter also supports the @ref{framesync} options.
16088 The file printed if @var{stats_file} is selected, contains a sequence of
16089 key/value pairs of the form @var{key}:@var{value} for each compared
16092 If a @var{stats_version} greater than 1 is specified, a header line precedes
16093 the list of per-frame-pair stats, with key value pairs following the frame
16094 format with the following parameters:
16097 @item psnr_log_version
16098 The version of the log file format. Will match @var{stats_version}.
16101 A comma separated list of the per-frame-pair parameters included in
16105 A description of each shown per-frame-pair parameter follows:
16109 sequential number of the input frame, starting from 1
16112 Mean Square Error pixel-by-pixel average difference of the compared
16113 frames, averaged over all the image components.
16115 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16116 Mean Square Error pixel-by-pixel average difference of the compared
16117 frames for the component specified by the suffix.
16119 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16120 Peak Signal to Noise ratio of the compared frames for the component
16121 specified by the suffix.
16123 @item max_avg, max_y, max_u, max_v
16124 Maximum allowed value for each channel, and average over all
16128 @subsection Examples
16133 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16134 [main][ref] psnr="stats_file=stats.log" [out]
16137 On this example the input file being processed is compared with the
16138 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
16139 is stored in @file{stats.log}.
16142 Another example with different containers:
16144 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 -
16151 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16152 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
16155 The pullup filter is designed to take advantage of future context in making
16156 its decisions. This filter is stateless in the sense that it does not lock
16157 onto a pattern to follow, but it instead looks forward to the following
16158 fields in order to identify matches and rebuild progressive frames.
16160 To produce content with an even framerate, insert the fps filter after
16161 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
16162 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
16164 The filter accepts the following options:
16171 These options set the amount of "junk" to ignore at the left, right, top, and
16172 bottom of the image, respectively. Left and right are in units of 8 pixels,
16173 while top and bottom are in units of 2 lines.
16174 The default is 8 pixels on each side.
16177 Set the strict breaks. Setting this option to 1 will reduce the chances of
16178 filter generating an occasional mismatched frame, but it may also cause an
16179 excessive number of frames to be dropped during high motion sequences.
16180 Conversely, setting it to -1 will make filter match fields more easily.
16181 This may help processing of video where there is slight blurring between
16182 the fields, but may also cause there to be interlaced frames in the output.
16183 Default value is @code{0}.
16186 Set the metric plane to use. It accepts the following values:
16192 Use chroma blue plane.
16195 Use chroma red plane.
16198 This option may be set to use chroma plane instead of the default luma plane
16199 for doing filter's computations. This may improve accuracy on very clean
16200 source material, but more likely will decrease accuracy, especially if there
16201 is chroma noise (rainbow effect) or any grayscale video.
16202 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
16203 load and make pullup usable in realtime on slow machines.
16206 For best results (without duplicated frames in the output file) it is
16207 necessary to change the output frame rate. For example, to inverse
16208 telecine NTSC input:
16210 ffmpeg -i input -vf pullup -r 24000/1001 ...
16215 Change video quantization parameters (QP).
16217 The filter accepts the following option:
16221 Set expression for quantization parameter.
16224 The expression is evaluated through the eval API and can contain, among others,
16225 the following constants:
16229 1 if index is not 129, 0 otherwise.
16232 Sequential index starting from -129 to 128.
16235 @subsection Examples
16239 Some equation like:
16247 Flush video frames from internal cache of frames into a random order.
16248 No frame is discarded.
16249 Inspired by @ref{frei0r} nervous filter.
16253 Set size in number of frames of internal cache, in range from @code{2} to
16254 @code{512}. Default is @code{30}.
16257 Set seed for random number generator, must be an integer included between
16258 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
16259 less than @code{0}, the filter will try to use a good random seed on a
16263 @section readeia608
16265 Read closed captioning (EIA-608) information from the top lines of a video frame.
16267 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
16268 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
16269 with EIA-608 data (starting from 0). A description of each metadata value follows:
16272 @item lavfi.readeia608.X.cc
16273 The two bytes stored as EIA-608 data (printed in hexadecimal).
16275 @item lavfi.readeia608.X.line
16276 The number of the line on which the EIA-608 data was identified and read.
16279 This filter accepts the following options:
16283 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16286 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16289 Set the ratio of width reserved for sync code detection.
16290 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16293 Enable checking the parity bit. In the event of a parity error, the filter will output
16294 @code{0x00} for that character. Default is false.
16297 Lowpass lines prior to further processing. Default is enabled.
16300 @subsection Commands
16302 This filter supports the all above options as @ref{commands}.
16304 @subsection Examples
16308 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16310 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
16316 Read vertical interval timecode (VITC) information from the top lines of a
16319 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16320 timecode value, if a valid timecode has been detected. Further metadata key
16321 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16322 timecode data has been found or not.
16324 This filter accepts the following options:
16328 Set the maximum number of lines to scan for VITC data. If the value is set to
16329 @code{-1} the full video frame is scanned. Default is @code{45}.
16332 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16333 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16336 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16337 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16340 @subsection Examples
16344 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16345 draw @code{--:--:--:--} as a placeholder:
16347 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16353 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16355 Destination pixel at position (X, Y) will be picked from source (x, y) position
16356 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16357 value for pixel will be used for destination pixel.
16359 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16360 will have Xmap/Ymap video stream dimensions.
16361 Xmap and Ymap input video streams are 16bit depth, single channel.
16365 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16366 Default is @code{color}.
16369 Specify the color of the unmapped pixels. For the syntax of this option,
16370 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16371 manual,ffmpeg-utils}. Default color is @code{black}.
16374 @section removegrain
16376 The removegrain filter is a spatial denoiser for progressive video.
16380 Set mode for the first plane.
16383 Set mode for the second plane.
16386 Set mode for the third plane.
16389 Set mode for the fourth plane.
16392 Range of mode is from 0 to 24. Description of each mode follows:
16396 Leave input plane unchanged. Default.
16399 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16402 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16405 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16408 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16409 This is equivalent to a median filter.
16412 Line-sensitive clipping giving the minimal change.
16415 Line-sensitive clipping, intermediate.
16418 Line-sensitive clipping, intermediate.
16421 Line-sensitive clipping, intermediate.
16424 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16427 Replaces the target pixel with the closest neighbour.
16430 [1 2 1] horizontal and vertical kernel blur.
16436 Bob mode, interpolates top field from the line where the neighbours
16437 pixels are the closest.
16440 Bob mode, interpolates bottom field from the line where the neighbours
16441 pixels are the closest.
16444 Bob mode, interpolates top field. Same as 13 but with a more complicated
16445 interpolation formula.
16448 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16449 interpolation formula.
16452 Clips the pixel with the minimum and maximum of respectively the maximum and
16453 minimum of each pair of opposite neighbour pixels.
16456 Line-sensitive clipping using opposite neighbours whose greatest distance from
16457 the current pixel is minimal.
16460 Replaces the pixel with the average of its 8 neighbours.
16463 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16466 Clips pixels using the averages of opposite neighbour.
16469 Same as mode 21 but simpler and faster.
16472 Small edge and halo removal, but reputed useless.
16478 @section removelogo
16480 Suppress a TV station logo, using an image file to determine which
16481 pixels comprise the logo. It works by filling in the pixels that
16482 comprise the logo with neighboring pixels.
16484 The filter accepts the following options:
16488 Set the filter bitmap file, which can be any image format supported by
16489 libavformat. The width and height of the image file must match those of the
16490 video stream being processed.
16493 Pixels in the provided bitmap image with a value of zero are not
16494 considered part of the logo, non-zero pixels are considered part of
16495 the logo. If you use white (255) for the logo and black (0) for the
16496 rest, you will be safe. For making the filter bitmap, it is
16497 recommended to take a screen capture of a black frame with the logo
16498 visible, and then using a threshold filter followed by the erode
16499 filter once or twice.
16501 If needed, little splotches can be fixed manually. Remember that if
16502 logo pixels are not covered, the filter quality will be much
16503 reduced. Marking too many pixels as part of the logo does not hurt as
16504 much, but it will increase the amount of blurring needed to cover over
16505 the image and will destroy more information than necessary, and extra
16506 pixels will slow things down on a large logo.
16508 @section repeatfields
16510 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16511 fields based on its value.
16515 Reverse a video clip.
16517 Warning: This filter requires memory to buffer the entire clip, so trimming
16520 @subsection Examples
16524 Take the first 5 seconds of a clip, and reverse it.
16531 Shift R/G/B/A pixels horizontally and/or vertically.
16533 The filter accepts the following options:
16536 Set amount to shift red horizontally.
16538 Set amount to shift red vertically.
16540 Set amount to shift green horizontally.
16542 Set amount to shift green vertically.
16544 Set amount to shift blue horizontally.
16546 Set amount to shift blue vertically.
16548 Set amount to shift alpha horizontally.
16550 Set amount to shift alpha vertically.
16552 Set edge mode, can be @var{smear}, default, or @var{warp}.
16555 @subsection Commands
16557 This filter supports the all above options as @ref{commands}.
16560 Apply roberts cross operator to input video stream.
16562 The filter accepts the following option:
16566 Set which planes will be processed, unprocessed planes will be copied.
16567 By default value 0xf, all planes will be processed.
16570 Set value which will be multiplied with filtered result.
16573 Set value which will be added to filtered result.
16576 @subsection Commands
16578 This filter supports the all above options as @ref{commands}.
16582 Rotate video by an arbitrary angle expressed in radians.
16584 The filter accepts the following options:
16586 A description of the optional parameters follows.
16589 Set an expression for the angle by which to rotate the input video
16590 clockwise, expressed as a number of radians. A negative value will
16591 result in a counter-clockwise rotation. By default it is set to "0".
16593 This expression is evaluated for each frame.
16596 Set the output width expression, default value is "iw".
16597 This expression is evaluated just once during configuration.
16600 Set the output height expression, default value is "ih".
16601 This expression is evaluated just once during configuration.
16604 Enable bilinear interpolation if set to 1, a value of 0 disables
16605 it. Default value is 1.
16608 Set the color used to fill the output area not covered by the rotated
16609 image. For the general syntax of this option, check the
16610 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16611 If the special value "none" is selected then no
16612 background is printed (useful for example if the background is never shown).
16614 Default value is "black".
16617 The expressions for the angle and the output size can contain the
16618 following constants and functions:
16622 sequential number of the input frame, starting from 0. It is always NAN
16623 before the first frame is filtered.
16626 time in seconds of the input frame, it is set to 0 when the filter is
16627 configured. It is always NAN before the first frame is filtered.
16631 horizontal and vertical chroma subsample values. For example for the
16632 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16636 the input video width and height
16640 the output width and height, that is the size of the padded area as
16641 specified by the @var{width} and @var{height} expressions
16645 the minimal width/height required for completely containing the input
16646 video rotated by @var{a} radians.
16648 These are only available when computing the @option{out_w} and
16649 @option{out_h} expressions.
16652 @subsection Examples
16656 Rotate the input by PI/6 radians clockwise:
16662 Rotate the input by PI/6 radians counter-clockwise:
16668 Rotate the input by 45 degrees clockwise:
16674 Apply a constant rotation with period T, starting from an angle of PI/3:
16676 rotate=PI/3+2*PI*t/T
16680 Make the input video rotation oscillating with a period of T
16681 seconds and an amplitude of A radians:
16683 rotate=A*sin(2*PI/T*t)
16687 Rotate the video, output size is chosen so that the whole rotating
16688 input video is always completely contained in the output:
16690 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16694 Rotate the video, reduce the output size so that no background is ever
16697 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16701 @subsection Commands
16703 The filter supports the following commands:
16707 Set the angle expression.
16708 The command accepts the same syntax of the corresponding option.
16710 If the specified expression is not valid, it is kept at its current
16716 Apply Shape Adaptive Blur.
16718 The filter accepts the following options:
16721 @item luma_radius, lr
16722 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16723 value is 1.0. A greater value will result in a more blurred image, and
16724 in slower processing.
16726 @item luma_pre_filter_radius, lpfr
16727 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16730 @item luma_strength, ls
16731 Set luma maximum difference between pixels to still be considered, must
16732 be a value in the 0.1-100.0 range, default value is 1.0.
16734 @item chroma_radius, cr
16735 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16736 greater value will result in a more blurred image, and in slower
16739 @item chroma_pre_filter_radius, cpfr
16740 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16742 @item chroma_strength, cs
16743 Set chroma maximum difference between pixels to still be considered,
16744 must be a value in the -0.9-100.0 range.
16747 Each chroma option value, if not explicitly specified, is set to the
16748 corresponding luma option value.
16753 Scale (resize) the input video, using the libswscale library.
16755 The scale filter forces the output display aspect ratio to be the same
16756 of the input, by changing the output sample aspect ratio.
16758 If the input image format is different from the format requested by
16759 the next filter, the scale filter will convert the input to the
16762 @subsection Options
16763 The filter accepts the following options, or any of the options
16764 supported by the libswscale scaler.
16766 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16767 the complete list of scaler options.
16772 Set the output video dimension expression. Default value is the input
16775 If the @var{width} or @var{w} value is 0, the input width is used for
16776 the output. If the @var{height} or @var{h} value is 0, the input height
16777 is used for the output.
16779 If one and only one of the values is -n with n >= 1, the scale filter
16780 will use a value that maintains the aspect ratio of the input image,
16781 calculated from the other specified dimension. After that it will,
16782 however, make sure that the calculated dimension is divisible by n and
16783 adjust the value if necessary.
16785 If both values are -n with n >= 1, the behavior will be identical to
16786 both values being set to 0 as previously detailed.
16788 See below for the list of accepted constants for use in the dimension
16792 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16796 Only evaluate expressions once during the filter initialization or when a command is processed.
16799 Evaluate expressions for each incoming frame.
16803 Default value is @samp{init}.
16807 Set the interlacing mode. It accepts the following values:
16811 Force interlaced aware scaling.
16814 Do not apply interlaced scaling.
16817 Select interlaced aware scaling depending on whether the source frames
16818 are flagged as interlaced or not.
16821 Default value is @samp{0}.
16824 Set libswscale scaling flags. See
16825 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16826 complete list of values. If not explicitly specified the filter applies
16830 @item param0, param1
16831 Set libswscale input parameters for scaling algorithms that need them. See
16832 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16833 complete documentation. If not explicitly specified the filter applies
16839 Set the video size. For the syntax of this option, check the
16840 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16842 @item in_color_matrix
16843 @item out_color_matrix
16844 Set in/output YCbCr color space type.
16846 This allows the autodetected value to be overridden as well as allows forcing
16847 a specific value used for the output and encoder.
16849 If not specified, the color space type depends on the pixel format.
16855 Choose automatically.
16858 Format conforming to International Telecommunication Union (ITU)
16859 Recommendation BT.709.
16862 Set color space conforming to the United States Federal Communications
16863 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16868 Set color space conforming to:
16872 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16875 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16878 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16883 Set color space conforming to SMPTE ST 240:1999.
16886 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16891 Set in/output YCbCr sample range.
16893 This allows the autodetected value to be overridden as well as allows forcing
16894 a specific value used for the output and encoder. If not specified, the
16895 range depends on the pixel format. Possible values:
16899 Choose automatically.
16902 Set full range (0-255 in case of 8-bit luma).
16904 @item mpeg/limited/tv
16905 Set "MPEG" range (16-235 in case of 8-bit luma).
16908 @item force_original_aspect_ratio
16909 Enable decreasing or increasing output video width or height if necessary to
16910 keep the original aspect ratio. Possible values:
16914 Scale the video as specified and disable this feature.
16917 The output video dimensions will automatically be decreased if needed.
16920 The output video dimensions will automatically be increased if needed.
16924 One useful instance of this option is that when you know a specific device's
16925 maximum allowed resolution, you can use this to limit the output video to
16926 that, while retaining the aspect ratio. For example, device A allows
16927 1280x720 playback, and your video is 1920x800. Using this option (set it to
16928 decrease) and specifying 1280x720 to the command line makes the output
16931 Please note that this is a different thing than specifying -1 for @option{w}
16932 or @option{h}, you still need to specify the output resolution for this option
16935 @item force_divisible_by
16936 Ensures that both the output dimensions, width and height, are divisible by the
16937 given integer when used together with @option{force_original_aspect_ratio}. This
16938 works similar to using @code{-n} in the @option{w} and @option{h} options.
16940 This option respects the value set for @option{force_original_aspect_ratio},
16941 increasing or decreasing the resolution accordingly. The video's aspect ratio
16942 may be slightly modified.
16944 This option can be handy if you need to have a video fit within or exceed
16945 a defined resolution using @option{force_original_aspect_ratio} but also have
16946 encoder restrictions on width or height divisibility.
16950 The values of the @option{w} and @option{h} options are expressions
16951 containing the following constants:
16956 The input width and height
16960 These are the same as @var{in_w} and @var{in_h}.
16964 The output (scaled) width and height
16968 These are the same as @var{out_w} and @var{out_h}
16971 The same as @var{iw} / @var{ih}
16974 input sample aspect ratio
16977 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16981 horizontal and vertical input chroma subsample values. For example for the
16982 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16986 horizontal and vertical output chroma subsample values. For example for the
16987 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16990 The (sequential) number of the input frame, starting from 0.
16991 Only available with @code{eval=frame}.
16994 The presentation timestamp of the input frame, expressed as a number of
16995 seconds. Only available with @code{eval=frame}.
16998 The position (byte offset) of the frame in the input stream, or NaN if
16999 this information is unavailable and/or meaningless (for example in case of synthetic video).
17000 Only available with @code{eval=frame}.
17003 @subsection Examples
17007 Scale the input video to a size of 200x100
17012 This is equivalent to:
17023 Specify a size abbreviation for the output size:
17028 which can also be written as:
17034 Scale the input to 2x:
17036 scale=w=2*iw:h=2*ih
17040 The above is the same as:
17042 scale=2*in_w:2*in_h
17046 Scale the input to 2x with forced interlaced scaling:
17048 scale=2*iw:2*ih:interl=1
17052 Scale the input to half size:
17054 scale=w=iw/2:h=ih/2
17058 Increase the width, and set the height to the same size:
17064 Seek Greek harmony:
17071 Increase the height, and set the width to 3/2 of the height:
17073 scale=w=3/2*oh:h=3/5*ih
17077 Increase the size, making the size a multiple of the chroma
17080 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17084 Increase the width to a maximum of 500 pixels,
17085 keeping the same aspect ratio as the input:
17087 scale=w='min(500\, iw*3/2):h=-1'
17091 Make pixels square by combining scale and setsar:
17093 scale='trunc(ih*dar):ih',setsar=1/1
17097 Make pixels square by combining scale and setsar,
17098 making sure the resulting resolution is even (required by some codecs):
17100 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17104 @subsection Commands
17106 This filter supports the following commands:
17110 Set the output video dimension expression.
17111 The command accepts the same syntax of the corresponding option.
17113 If the specified expression is not valid, it is kept at its current
17119 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
17120 format conversion on CUDA video frames. Setting the output width and height
17121 works in the same way as for the @var{scale} filter.
17123 The following additional options are accepted:
17126 The pixel format of the output CUDA frames. If set to the string "same" (the
17127 default), the input format will be kept. Note that automatic format negotiation
17128 and conversion is not yet supported for hardware frames
17131 The interpolation algorithm used for resizing. One of the following:
17138 @item cubic2p_bspline
17139 2-parameter cubic (B=1, C=0)
17141 @item cubic2p_catmullrom
17142 2-parameter cubic (B=0, C=1/2)
17144 @item cubic2p_b05c03
17145 2-parameter cubic (B=1/2, C=3/10)
17153 @item force_original_aspect_ratio
17154 Enable decreasing or increasing output video width or height if necessary to
17155 keep the original aspect ratio. Possible values:
17159 Scale the video as specified and disable this feature.
17162 The output video dimensions will automatically be decreased if needed.
17165 The output video dimensions will automatically be increased if needed.
17169 One useful instance of this option is that when you know a specific device's
17170 maximum allowed resolution, you can use this to limit the output video to
17171 that, while retaining the aspect ratio. For example, device A allows
17172 1280x720 playback, and your video is 1920x800. Using this option (set it to
17173 decrease) and specifying 1280x720 to the command line makes the output
17176 Please note that this is a different thing than specifying -1 for @option{w}
17177 or @option{h}, you still need to specify the output resolution for this option
17180 @item force_divisible_by
17181 Ensures that both the output dimensions, width and height, are divisible by the
17182 given integer when used together with @option{force_original_aspect_ratio}. This
17183 works similar to using @code{-n} in the @option{w} and @option{h} options.
17185 This option respects the value set for @option{force_original_aspect_ratio},
17186 increasing or decreasing the resolution accordingly. The video's aspect ratio
17187 may be slightly modified.
17189 This option can be handy if you need to have a video fit within or exceed
17190 a defined resolution using @option{force_original_aspect_ratio} but also have
17191 encoder restrictions on width or height divisibility.
17197 Scale (resize) the input video, based on a reference video.
17199 See the scale filter for available options, scale2ref supports the same but
17200 uses the reference video instead of the main input as basis. scale2ref also
17201 supports the following additional constants for the @option{w} and
17202 @option{h} options:
17207 The main input video's width and height
17210 The same as @var{main_w} / @var{main_h}
17213 The main input video's sample aspect ratio
17215 @item main_dar, mdar
17216 The main input video's display aspect ratio. Calculated from
17217 @code{(main_w / main_h) * main_sar}.
17221 The main input video's horizontal and vertical chroma subsample values.
17222 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
17226 The (sequential) number of the main input frame, starting from 0.
17227 Only available with @code{eval=frame}.
17230 The presentation timestamp of the main input frame, expressed as a number of
17231 seconds. Only available with @code{eval=frame}.
17234 The position (byte offset) of the frame in the main input stream, or NaN if
17235 this information is unavailable and/or meaningless (for example in case of synthetic video).
17236 Only available with @code{eval=frame}.
17239 @subsection Examples
17243 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
17245 'scale2ref[b][a];[a][b]overlay'
17249 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
17251 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17255 @subsection Commands
17257 This filter supports the following commands:
17261 Set the output video dimension expression.
17262 The command accepts the same syntax of the corresponding option.
17264 If the specified expression is not valid, it is kept at its current
17269 Scroll input video horizontally and/or vertically by constant speed.
17271 The filter accepts the following options:
17273 @item horizontal, h
17274 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
17275 Negative values changes scrolling direction.
17278 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
17279 Negative values changes scrolling direction.
17282 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
17285 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
17288 @subsection Commands
17290 This filter supports the following @ref{commands}:
17292 @item horizontal, h
17293 Set the horizontal scrolling speed.
17295 Set the vertical scrolling speed.
17301 Detect video scene change.
17303 This filter sets frame metadata with mafd between frame, the scene score, and
17304 forward the frame to the next filter, so they can use these metadata to detect
17305 scene change or others.
17307 In addition, this filter logs a message and sets frame metadata when it detects
17308 a scene change by @option{threshold}.
17310 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17312 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17313 to detect scene change.
17315 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17316 detect scene change with @option{threshold}.
17318 The filter accepts the following options:
17322 Set the scene change detection threshold as a percentage of maximum change. Good
17323 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17326 Default value is @code{10.}.
17329 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17330 You can enable it if you want to get snapshot of scene change frames only.
17333 @anchor{selectivecolor}
17334 @section selectivecolor
17336 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17337 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17338 by the "purity" of the color (that is, how saturated it already is).
17340 This filter is similar to the Adobe Photoshop Selective Color tool.
17342 The filter accepts the following options:
17345 @item correction_method
17346 Select color correction method.
17348 Available values are:
17351 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17354 Specified adjustments are relative to the original component value.
17356 Default is @code{absolute}.
17358 Adjustments for red pixels (pixels where the red component is the maximum)
17360 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17362 Adjustments for green pixels (pixels where the green component is the maximum)
17364 Adjustments for cyan pixels (pixels where the red component is the minimum)
17366 Adjustments for blue pixels (pixels where the blue component is the maximum)
17368 Adjustments for magenta pixels (pixels where the green component is the minimum)
17370 Adjustments for white pixels (pixels where all components are greater than 128)
17372 Adjustments for all pixels except pure black and pure white
17374 Adjustments for black pixels (pixels where all components are lesser than 128)
17376 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17379 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17380 4 space separated floating point adjustment values in the [-1,1] range,
17381 respectively to adjust the amount of cyan, magenta, yellow and black for the
17382 pixels of its range.
17384 @subsection Examples
17388 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17389 increase magenta by 27% in blue areas:
17391 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17395 Use a Photoshop selective color preset:
17397 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17401 @anchor{separatefields}
17402 @section separatefields
17404 The @code{separatefields} takes a frame-based video input and splits
17405 each frame into its components fields, producing a new half height clip
17406 with twice the frame rate and twice the frame count.
17408 This filter use field-dominance information in frame to decide which
17409 of each pair of fields to place first in the output.
17410 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17412 @section setdar, setsar
17414 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17417 This is done by changing the specified Sample (aka Pixel) Aspect
17418 Ratio, according to the following equation:
17420 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17423 Keep in mind that the @code{setdar} filter does not modify the pixel
17424 dimensions of the video frame. Also, the display aspect ratio set by
17425 this filter may be changed by later filters in the filterchain,
17426 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17429 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17430 the filter output video.
17432 Note that as a consequence of the application of this filter, the
17433 output display aspect ratio will change according to the equation
17436 Keep in mind that the sample aspect ratio set by the @code{setsar}
17437 filter may be changed by later filters in the filterchain, e.g. if
17438 another "setsar" or a "setdar" filter is applied.
17440 It accepts the following parameters:
17443 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17444 Set the aspect ratio used by the filter.
17446 The parameter can be a floating point number string, an expression, or
17447 a string of the form @var{num}:@var{den}, where @var{num} and
17448 @var{den} are the numerator and denominator of the aspect ratio. If
17449 the parameter is not specified, it is assumed the value "0".
17450 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17454 Set the maximum integer value to use for expressing numerator and
17455 denominator when reducing the expressed aspect ratio to a rational.
17456 Default value is @code{100}.
17460 The parameter @var{sar} is an expression containing
17461 the following constants:
17465 These are approximated values for the mathematical constants e
17466 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17469 The input width and height.
17472 These are the same as @var{w} / @var{h}.
17475 The input sample aspect ratio.
17478 The input display aspect ratio. It is the same as
17479 (@var{w} / @var{h}) * @var{sar}.
17482 Horizontal and vertical chroma subsample values. For example, for the
17483 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17486 @subsection Examples
17491 To change the display aspect ratio to 16:9, specify one of the following:
17498 To change the sample aspect ratio to 10:11, specify:
17504 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17505 1000 in the aspect ratio reduction, use the command:
17507 setdar=ratio=16/9:max=1000
17515 Force field for the output video frame.
17517 The @code{setfield} filter marks the interlace type field for the
17518 output frames. It does not change the input frame, but only sets the
17519 corresponding property, which affects how the frame is treated by
17520 following filters (e.g. @code{fieldorder} or @code{yadif}).
17522 The filter accepts the following options:
17527 Available values are:
17531 Keep the same field property.
17534 Mark the frame as bottom-field-first.
17537 Mark the frame as top-field-first.
17540 Mark the frame as progressive.
17547 Force frame parameter for the output video frame.
17549 The @code{setparams} filter marks interlace and color range for the
17550 output frames. It does not change the input frame, but only sets the
17551 corresponding property, which affects how the frame is treated by
17556 Available values are:
17560 Keep the same field property (default).
17563 Mark the frame as bottom-field-first.
17566 Mark the frame as top-field-first.
17569 Mark the frame as progressive.
17573 Available values are:
17577 Keep the same color range property (default).
17579 @item unspecified, unknown
17580 Mark the frame as unspecified color range.
17582 @item limited, tv, mpeg
17583 Mark the frame as limited range.
17585 @item full, pc, jpeg
17586 Mark the frame as full range.
17589 @item color_primaries
17590 Set the color primaries.
17591 Available values are:
17595 Keep the same color primaries property (default).
17612 Set the color transfer.
17613 Available values are:
17617 Keep the same color trc property (default).
17639 Set the colorspace.
17640 Available values are:
17644 Keep the same colorspace property (default).
17657 @item chroma-derived-nc
17658 @item chroma-derived-c
17665 Show a line containing various information for each input video frame.
17666 The input video is not modified.
17668 This filter supports the following options:
17672 Calculate checksums of each plane. By default enabled.
17675 The shown line contains a sequence of key/value pairs of the form
17676 @var{key}:@var{value}.
17678 The following values are shown in the output:
17682 The (sequential) number of the input frame, starting from 0.
17685 The Presentation TimeStamp of the input frame, expressed as a number of
17686 time base units. The time base unit depends on the filter input pad.
17689 The Presentation TimeStamp of the input frame, expressed as a number of
17693 The position of the frame in the input stream, or -1 if this information is
17694 unavailable and/or meaningless (for example in case of synthetic video).
17697 The pixel format name.
17700 The sample aspect ratio of the input frame, expressed in the form
17701 @var{num}/@var{den}.
17704 The size of the input frame. For the syntax of this option, check the
17705 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17708 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17709 for bottom field first).
17712 This is 1 if the frame is a key frame, 0 otherwise.
17715 The picture type of the input frame ("I" for an I-frame, "P" for a
17716 P-frame, "B" for a B-frame, or "?" for an unknown type).
17717 Also refer to the documentation of the @code{AVPictureType} enum and of
17718 the @code{av_get_picture_type_char} function defined in
17719 @file{libavutil/avutil.h}.
17722 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17724 @item plane_checksum
17725 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17726 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17729 The mean value of pixels in each plane of the input frame, expressed in the form
17730 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17733 The standard deviation of pixel values in each plane of the input frame, expressed
17734 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17738 @section showpalette
17740 Displays the 256 colors palette of each frame. This filter is only relevant for
17741 @var{pal8} pixel format frames.
17743 It accepts the following option:
17747 Set the size of the box used to represent one palette color entry. Default is
17748 @code{30} (for a @code{30x30} pixel box).
17751 @section shuffleframes
17753 Reorder and/or duplicate and/or drop video frames.
17755 It accepts the following parameters:
17759 Set the destination indexes of input frames.
17760 This is space or '|' separated list of indexes that maps input frames to output
17761 frames. Number of indexes also sets maximal value that each index may have.
17762 '-1' index have special meaning and that is to drop frame.
17765 The first frame has the index 0. The default is to keep the input unchanged.
17767 @subsection Examples
17771 Swap second and third frame of every three frames of the input:
17773 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17777 Swap 10th and 1st frame of every ten frames of the input:
17779 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17783 @section shuffleplanes
17785 Reorder and/or duplicate video planes.
17787 It accepts the following parameters:
17792 The index of the input plane to be used as the first output plane.
17795 The index of the input plane to be used as the second output plane.
17798 The index of the input plane to be used as the third output plane.
17801 The index of the input plane to be used as the fourth output plane.
17805 The first plane has the index 0. The default is to keep the input unchanged.
17807 @subsection Examples
17811 Swap the second and third planes of the input:
17813 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17817 @anchor{signalstats}
17818 @section signalstats
17819 Evaluate various visual metrics that assist in determining issues associated
17820 with the digitization of analog video media.
17822 By default the filter will log these metadata values:
17826 Display the minimal Y value contained within the input frame. Expressed in
17830 Display the Y value at the 10% percentile within the input frame. Expressed in
17834 Display the average Y value within the input frame. Expressed in range of
17838 Display the Y value at the 90% percentile within the input frame. Expressed in
17842 Display the maximum Y value contained within the input frame. Expressed in
17846 Display the minimal U value contained within the input frame. Expressed in
17850 Display the U value at the 10% percentile within the input frame. Expressed in
17854 Display the average U value within the input frame. Expressed in range of
17858 Display the U value at the 90% percentile within the input frame. Expressed in
17862 Display the maximum U value contained within the input frame. Expressed in
17866 Display the minimal V value contained within the input frame. Expressed in
17870 Display the V value at the 10% percentile within the input frame. Expressed in
17874 Display the average V value within the input frame. Expressed in range of
17878 Display the V value at the 90% percentile within the input frame. Expressed in
17882 Display the maximum V value contained within the input frame. Expressed in
17886 Display the minimal saturation value contained within the input frame.
17887 Expressed in range of [0-~181.02].
17890 Display the saturation value at the 10% percentile within the input frame.
17891 Expressed in range of [0-~181.02].
17894 Display the average saturation value within the input frame. Expressed in range
17898 Display the saturation value at the 90% percentile within the input frame.
17899 Expressed in range of [0-~181.02].
17902 Display the maximum saturation value contained within the input frame.
17903 Expressed in range of [0-~181.02].
17906 Display the median value for hue within the input frame. Expressed in range of
17910 Display the average value for hue within the input frame. Expressed in range of
17914 Display the average of sample value difference between all values of the Y
17915 plane in the current frame and corresponding values of the previous input frame.
17916 Expressed in range of [0-255].
17919 Display the average of sample value difference between all values of the U
17920 plane in the current frame and corresponding values of the previous input frame.
17921 Expressed in range of [0-255].
17924 Display the average of sample value difference between all values of the V
17925 plane in the current frame and corresponding values of the previous input frame.
17926 Expressed in range of [0-255].
17929 Display bit depth of Y plane in current frame.
17930 Expressed in range of [0-16].
17933 Display bit depth of U plane in current frame.
17934 Expressed in range of [0-16].
17937 Display bit depth of V plane in current frame.
17938 Expressed in range of [0-16].
17941 The filter accepts the following options:
17947 @option{stat} specify an additional form of image analysis.
17948 @option{out} output video with the specified type of pixel highlighted.
17950 Both options accept the following values:
17954 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17955 unlike the neighboring pixels of the same field. Examples of temporal outliers
17956 include the results of video dropouts, head clogs, or tape tracking issues.
17959 Identify @var{vertical line repetition}. Vertical line repetition includes
17960 similar rows of pixels within a frame. In born-digital video vertical line
17961 repetition is common, but this pattern is uncommon in video digitized from an
17962 analog source. When it occurs in video that results from the digitization of an
17963 analog source it can indicate concealment from a dropout compensator.
17966 Identify pixels that fall outside of legal broadcast range.
17970 Set the highlight color for the @option{out} option. The default color is
17974 @subsection Examples
17978 Output data of various video metrics:
17980 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17984 Output specific data about the minimum and maximum values of the Y plane per frame:
17986 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17990 Playback video while highlighting pixels that are outside of broadcast range in red.
17992 ffplay example.mov -vf signalstats="out=brng:color=red"
17996 Playback video with signalstats metadata drawn over the frame.
17998 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18001 The contents of signalstat_drawtext.txt used in the command are:
18004 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
18005 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
18006 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
18007 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
18015 Calculates the MPEG-7 Video Signature. The filter can handle more than one
18016 input. In this case the matching between the inputs can be calculated additionally.
18017 The filter always passes through the first input. The signature of each stream can
18018 be written into a file.
18020 It accepts the following options:
18024 Enable or disable the matching process.
18026 Available values are:
18030 Disable the calculation of a matching (default).
18032 Calculate the matching for the whole video and output whether the whole video
18033 matches or only parts.
18035 Calculate only until a matching is found or the video ends. Should be faster in
18040 Set the number of inputs. The option value must be a non negative integer.
18041 Default value is 1.
18044 Set the path to which the output is written. If there is more than one input,
18045 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
18046 integer), that will be replaced with the input number. If no filename is
18047 specified, no output will be written. This is the default.
18050 Choose the output format.
18052 Available values are:
18056 Use the specified binary representation (default).
18058 Use the specified xml representation.
18062 Set threshold to detect one word as similar. The option value must be an integer
18063 greater than zero. The default value is 9000.
18066 Set threshold to detect all words as similar. The option value must be an integer
18067 greater than zero. The default value is 60000.
18070 Set threshold to detect frames as similar. The option value must be an integer
18071 greater than zero. The default value is 116.
18074 Set the minimum length of a sequence in frames to recognize it as matching
18075 sequence. The option value must be a non negative integer value.
18076 The default value is 0.
18079 Set the minimum relation, that matching frames to all frames must have.
18080 The option value must be a double value between 0 and 1. The default value is 0.5.
18083 @subsection Examples
18087 To calculate the signature of an input video and store it in signature.bin:
18089 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18093 To detect whether two videos match and store the signatures in XML format in
18094 signature0.xml and signature1.xml:
18096 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 -
18104 Blur the input video without impacting the outlines.
18106 It accepts the following options:
18109 @item luma_radius, lr
18110 Set the luma radius. The option value must be a float number in
18111 the range [0.1,5.0] that specifies the variance of the gaussian filter
18112 used to blur the image (slower if larger). Default value is 1.0.
18114 @item luma_strength, ls
18115 Set the luma strength. The option value must be a float number
18116 in the range [-1.0,1.0] that configures the blurring. A value included
18117 in [0.0,1.0] will blur the image whereas a value included in
18118 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18120 @item luma_threshold, lt
18121 Set the luma threshold used as a coefficient to determine
18122 whether a pixel should be blurred or not. The option value must be an
18123 integer in the range [-30,30]. A value of 0 will filter all the image,
18124 a value included in [0,30] will filter flat areas and a value included
18125 in [-30,0] will filter edges. Default value is 0.
18127 @item chroma_radius, cr
18128 Set the chroma radius. The option value must be a float number in
18129 the range [0.1,5.0] that specifies the variance of the gaussian filter
18130 used to blur the image (slower if larger). Default value is @option{luma_radius}.
18132 @item chroma_strength, cs
18133 Set the chroma strength. The option value must be a float number
18134 in the range [-1.0,1.0] that configures the blurring. A value included
18135 in [0.0,1.0] will blur the image whereas a value included in
18136 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
18138 @item chroma_threshold, ct
18139 Set the chroma threshold used as a coefficient to determine
18140 whether a pixel should be blurred or not. The option value must be an
18141 integer in the range [-30,30]. A value of 0 will filter all the image,
18142 a value included in [0,30] will filter flat areas and a value included
18143 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
18146 If a chroma option is not explicitly set, the corresponding luma value
18150 Apply sobel operator to input video stream.
18152 The filter accepts the following option:
18156 Set which planes will be processed, unprocessed planes will be copied.
18157 By default value 0xf, all planes will be processed.
18160 Set value which will be multiplied with filtered result.
18163 Set value which will be added to filtered result.
18166 @subsection Commands
18168 This filter supports the all above options as @ref{commands}.
18173 Apply a simple postprocessing filter that compresses and decompresses the image
18174 at several (or - in the case of @option{quality} level @code{6} - all) shifts
18175 and average the results.
18177 The filter accepts the following options:
18181 Set quality. This option defines the number of levels for averaging. It accepts
18182 an integer in the range 0-6. If set to @code{0}, the filter will have no
18183 effect. A value of @code{6} means the higher quality. For each increment of
18184 that value the speed drops by a factor of approximately 2. Default value is
18188 Force a constant quantization parameter. If not set, the filter will use the QP
18189 from the video stream (if available).
18192 Set thresholding mode. Available modes are:
18196 Set hard thresholding (default).
18198 Set soft thresholding (better de-ringing effect, but likely blurrier).
18201 @item use_bframe_qp
18202 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
18203 option may cause flicker since the B-Frames have often larger QP. Default is
18204 @code{0} (not enabled).
18207 @subsection Commands
18209 This filter supports the following commands:
18211 @item quality, level
18212 Set quality level. The value @code{max} can be used to set the maximum level,
18213 currently @code{6}.
18219 Scale the input by applying one of the super-resolution methods based on
18220 convolutional neural networks. Supported models:
18224 Super-Resolution Convolutional Neural Network model (SRCNN).
18225 See @url{https://arxiv.org/abs/1501.00092}.
18228 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18229 See @url{https://arxiv.org/abs/1609.05158}.
18232 Training scripts as well as scripts for model file (.pb) saving can be found at
18233 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
18234 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
18236 Native model files (.model) can be generated from TensorFlow model
18237 files (.pb) by using tools/python/convert.py
18239 The filter accepts the following options:
18243 Specify which DNN backend to use for model loading and execution. This option accepts
18244 the following values:
18248 Native implementation of DNN loading and execution.
18251 TensorFlow backend. To enable this backend you
18252 need to install the TensorFlow for C library (see
18253 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
18254 @code{--enable-libtensorflow}
18257 Default value is @samp{native}.
18260 Set path to model file specifying network architecture and its parameters.
18261 Note that different backends use different file formats. TensorFlow backend
18262 can load files for both formats, while native backend can load files for only
18266 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
18267 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
18268 input upscaled using bicubic upscaling with proper scale factor.
18271 This feature can also be finished with @ref{dnn_processing} filter.
18275 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
18277 This filter takes in input two input videos, the first input is
18278 considered the "main" source and is passed unchanged to the
18279 output. The second input is used as a "reference" video for computing
18282 Both video inputs must have the same resolution and pixel format for
18283 this filter to work correctly. Also it assumes that both inputs
18284 have the same number of frames, which are compared one by one.
18286 The filter stores the calculated SSIM of each frame.
18288 The description of the accepted parameters follows.
18291 @item stats_file, f
18292 If specified the filter will use the named file to save the SSIM of
18293 each individual frame. When filename equals "-" the data is sent to
18297 The file printed if @var{stats_file} is selected, contains a sequence of
18298 key/value pairs of the form @var{key}:@var{value} for each compared
18301 A description of each shown parameter follows:
18305 sequential number of the input frame, starting from 1
18307 @item Y, U, V, R, G, B
18308 SSIM of the compared frames for the component specified by the suffix.
18311 SSIM of the compared frames for the whole frame.
18314 Same as above but in dB representation.
18317 This filter also supports the @ref{framesync} options.
18319 @subsection Examples
18324 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18325 [main][ref] ssim="stats_file=stats.log" [out]
18328 On this example the input file being processed is compared with the
18329 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18330 is stored in @file{stats.log}.
18333 Another example with both psnr and ssim at same time:
18335 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18339 Another example with different containers:
18341 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 -
18347 Convert between different stereoscopic image formats.
18349 The filters accept the following options:
18353 Set stereoscopic image format of input.
18355 Available values for input image formats are:
18358 side by side parallel (left eye left, right eye right)
18361 side by side crosseye (right eye left, left eye right)
18364 side by side parallel with half width resolution
18365 (left eye left, right eye right)
18368 side by side crosseye with half width resolution
18369 (right eye left, left eye right)
18373 above-below (left eye above, right eye below)
18377 above-below (right eye above, left eye below)
18381 above-below with half height resolution
18382 (left eye above, right eye below)
18386 above-below with half height resolution
18387 (right eye above, left eye below)
18390 alternating frames (left eye first, right eye second)
18393 alternating frames (right eye first, left eye second)
18396 interleaved rows (left eye has top row, right eye starts on next row)
18399 interleaved rows (right eye has top row, left eye starts on next row)
18402 interleaved columns, left eye first
18405 interleaved columns, right eye first
18407 Default value is @samp{sbsl}.
18411 Set stereoscopic image format of output.
18415 side by side parallel (left eye left, right eye right)
18418 side by side crosseye (right eye left, left eye right)
18421 side by side parallel with half width resolution
18422 (left eye left, right eye right)
18425 side by side crosseye with half width resolution
18426 (right eye left, left eye right)
18430 above-below (left eye above, right eye below)
18434 above-below (right eye above, left eye below)
18438 above-below with half height resolution
18439 (left eye above, right eye below)
18443 above-below with half height resolution
18444 (right eye above, left eye below)
18447 alternating frames (left eye first, right eye second)
18450 alternating frames (right eye first, left eye second)
18453 interleaved rows (left eye has top row, right eye starts on next row)
18456 interleaved rows (right eye has top row, left eye starts on next row)
18459 anaglyph red/blue gray
18460 (red filter on left eye, blue filter on right eye)
18463 anaglyph red/green gray
18464 (red filter on left eye, green filter on right eye)
18467 anaglyph red/cyan gray
18468 (red filter on left eye, cyan filter on right eye)
18471 anaglyph red/cyan half colored
18472 (red filter on left eye, cyan filter on right eye)
18475 anaglyph red/cyan color
18476 (red filter on left eye, cyan filter on right eye)
18479 anaglyph red/cyan color optimized with the least squares projection of dubois
18480 (red filter on left eye, cyan filter on right eye)
18483 anaglyph green/magenta gray
18484 (green filter on left eye, magenta filter on right eye)
18487 anaglyph green/magenta half colored
18488 (green filter on left eye, magenta filter on right eye)
18491 anaglyph green/magenta colored
18492 (green filter on left eye, magenta filter on right eye)
18495 anaglyph green/magenta color optimized with the least squares projection of dubois
18496 (green filter on left eye, magenta filter on right eye)
18499 anaglyph yellow/blue gray
18500 (yellow filter on left eye, blue filter on right eye)
18503 anaglyph yellow/blue half colored
18504 (yellow filter on left eye, blue filter on right eye)
18507 anaglyph yellow/blue colored
18508 (yellow filter on left eye, blue filter on right eye)
18511 anaglyph yellow/blue color optimized with the least squares projection of dubois
18512 (yellow filter on left eye, blue filter on right eye)
18515 mono output (left eye only)
18518 mono output (right eye only)
18521 checkerboard, left eye first
18524 checkerboard, right eye first
18527 interleaved columns, left eye first
18530 interleaved columns, right eye first
18536 Default value is @samp{arcd}.
18539 @subsection Examples
18543 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18549 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18555 @section streamselect, astreamselect
18556 Select video or audio streams.
18558 The filter accepts the following options:
18562 Set number of inputs. Default is 2.
18565 Set input indexes to remap to outputs.
18568 @subsection Commands
18570 The @code{streamselect} and @code{astreamselect} filter supports the following
18575 Set input indexes to remap to outputs.
18578 @subsection Examples
18582 Select first 5 seconds 1st stream and rest of time 2nd stream:
18584 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18588 Same as above, but for audio:
18590 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18597 Draw subtitles on top of input video using the libass library.
18599 To enable compilation of this filter you need to configure FFmpeg with
18600 @code{--enable-libass}. This filter also requires a build with libavcodec and
18601 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18602 Alpha) subtitles format.
18604 The filter accepts the following options:
18608 Set the filename of the subtitle file to read. It must be specified.
18610 @item original_size
18611 Specify the size of the original video, the video for which the ASS file
18612 was composed. For the syntax of this option, check the
18613 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18614 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18615 correctly scale the fonts if the aspect ratio has been changed.
18618 Set a directory path containing fonts that can be used by the filter.
18619 These fonts will be used in addition to whatever the font provider uses.
18622 Process alpha channel, by default alpha channel is untouched.
18625 Set subtitles input character encoding. @code{subtitles} filter only. Only
18626 useful if not UTF-8.
18628 @item stream_index, si
18629 Set subtitles stream index. @code{subtitles} filter only.
18632 Override default style or script info parameters of the subtitles. It accepts a
18633 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18636 If the first key is not specified, it is assumed that the first value
18637 specifies the @option{filename}.
18639 For example, to render the file @file{sub.srt} on top of the input
18640 video, use the command:
18645 which is equivalent to:
18647 subtitles=filename=sub.srt
18650 To render the default subtitles stream from file @file{video.mkv}, use:
18652 subtitles=video.mkv
18655 To render the second subtitles stream from that file, use:
18657 subtitles=video.mkv:si=1
18660 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18661 @code{DejaVu Serif}, use:
18663 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18666 @section super2xsai
18668 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18669 Interpolate) pixel art scaling algorithm.
18671 Useful for enlarging pixel art images without reducing sharpness.
18675 Swap two rectangular objects in video.
18677 This filter accepts the following options:
18687 Set 1st rect x coordinate.
18690 Set 1st rect y coordinate.
18693 Set 2nd rect x coordinate.
18696 Set 2nd rect y coordinate.
18698 All expressions are evaluated once for each frame.
18701 The all options are expressions containing the following constants:
18706 The input width and height.
18709 same as @var{w} / @var{h}
18712 input sample aspect ratio
18715 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18718 The number of the input frame, starting from 0.
18721 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18724 the position in the file of the input frame, NAN if unknown
18731 Blend successive video frames.
18737 Apply telecine process to the video.
18739 This filter accepts the following options:
18748 The default value is @code{top}.
18752 A string of numbers representing the pulldown pattern you wish to apply.
18753 The default value is @code{23}.
18757 Some typical patterns:
18762 24p: 2332 (preferred)
18769 24p: 222222222223 ("Euro pulldown")
18774 @section thistogram
18776 Compute and draw a color distribution histogram for the input video across time.
18778 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18779 at certain time, this filter shows also past histograms of number of frames defined
18780 by @code{width} option.
18782 The computed histogram is a representation of the color component
18783 distribution in an image.
18785 The filter accepts the following options:
18789 Set width of single color component output. Default value is @code{0}.
18790 Value of @code{0} means width will be picked from input video.
18791 This also set number of passed histograms to keep.
18792 Allowed range is [0, 8192].
18794 @item display_mode, d
18796 It accepts the following values:
18799 Per color component graphs are placed below each other.
18802 Per color component graphs are placed side by side.
18805 Presents information identical to that in the @code{parade}, except
18806 that the graphs representing color components are superimposed directly
18809 Default is @code{stack}.
18811 @item levels_mode, m
18812 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18813 Default is @code{linear}.
18815 @item components, c
18816 Set what color components to display.
18817 Default is @code{7}.
18820 Set background opacity. Default is @code{0.9}.
18823 Show envelope. Default is disabled.
18826 Set envelope color. Default is @code{gold}.
18831 Available values for slide is:
18834 Draw new frame when right border is reached.
18837 Replace old columns with new ones.
18840 Scroll from right to left.
18843 Scroll from left to right.
18846 Draw single picture.
18849 Default is @code{replace}.
18854 Apply threshold effect to video stream.
18856 This filter needs four video streams to perform thresholding.
18857 First stream is stream we are filtering.
18858 Second stream is holding threshold values, third stream is holding min values,
18859 and last, fourth stream is holding max values.
18861 The filter accepts the following option:
18865 Set which planes will be processed, unprocessed planes will be copied.
18866 By default value 0xf, all planes will be processed.
18869 For example if first stream pixel's component value is less then threshold value
18870 of pixel component from 2nd threshold stream, third stream value will picked,
18871 otherwise fourth stream pixel component value will be picked.
18873 Using color source filter one can perform various types of thresholding:
18875 @subsection Examples
18879 Binary threshold, using gray color as threshold:
18881 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18885 Inverted binary threshold, using gray color as threshold:
18887 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18891 Truncate binary threshold, using gray color as threshold:
18893 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18897 Threshold to zero, using gray color as threshold:
18899 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18903 Inverted threshold to zero, using gray color as threshold:
18905 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18910 Select the most representative frame in a given sequence of consecutive frames.
18912 The filter accepts the following options:
18916 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18917 will pick one of them, and then handle the next batch of @var{n} frames until
18918 the end. Default is @code{100}.
18921 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18922 value will result in a higher memory usage, so a high value is not recommended.
18924 @subsection Examples
18928 Extract one picture each 50 frames:
18934 Complete example of a thumbnail creation with @command{ffmpeg}:
18936 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18943 Tile several successive frames together.
18945 The @ref{untile} filter can do the reverse.
18947 The filter accepts the following options:
18952 Set the grid size (i.e. the number of lines and columns). For the syntax of
18953 this option, check the
18954 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18957 Set the maximum number of frames to render in the given area. It must be less
18958 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18959 the area will be used.
18962 Set the outer border margin in pixels.
18965 Set the inner border thickness (i.e. the number of pixels between frames). For
18966 more advanced padding options (such as having different values for the edges),
18967 refer to the pad video filter.
18970 Specify the color of the unused area. For the syntax of this option, check the
18971 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18972 The default value of @var{color} is "black".
18975 Set the number of frames to overlap when tiling several successive frames together.
18976 The value must be between @code{0} and @var{nb_frames - 1}.
18979 Set the number of frames to initially be empty before displaying first output frame.
18980 This controls how soon will one get first output frame.
18981 The value must be between @code{0} and @var{nb_frames - 1}.
18984 @subsection Examples
18988 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18990 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18992 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18993 duplicating each output frame to accommodate the originally detected frame
18997 Display @code{5} pictures in an area of @code{3x2} frames,
18998 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18999 mixed flat and named options:
19001 tile=3x2:nb_frames=5:padding=7:margin=2
19005 @section tinterlace
19007 Perform various types of temporal field interlacing.
19009 Frames are counted starting from 1, so the first input frame is
19012 The filter accepts the following options:
19017 Specify the mode of the interlacing. This option can also be specified
19018 as a value alone. See below for a list of values for this option.
19020 Available values are:
19024 Move odd frames into the upper field, even into the lower field,
19025 generating a double height frame at half frame rate.
19029 Frame 1 Frame 2 Frame 3 Frame 4
19031 11111 22222 33333 44444
19032 11111 22222 33333 44444
19033 11111 22222 33333 44444
19034 11111 22222 33333 44444
19048 Only output odd frames, even frames are dropped, generating a frame with
19049 unchanged height at half frame rate.
19054 Frame 1 Frame 2 Frame 3 Frame 4
19056 11111 22222 33333 44444
19057 11111 22222 33333 44444
19058 11111 22222 33333 44444
19059 11111 22222 33333 44444
19069 Only output even frames, odd frames are dropped, generating a frame with
19070 unchanged height at half frame rate.
19075 Frame 1 Frame 2 Frame 3 Frame 4
19077 11111 22222 33333 44444
19078 11111 22222 33333 44444
19079 11111 22222 33333 44444
19080 11111 22222 33333 44444
19090 Expand each frame to full height, but pad alternate lines with black,
19091 generating a frame with double height at the same input frame rate.
19096 Frame 1 Frame 2 Frame 3 Frame 4
19098 11111 22222 33333 44444
19099 11111 22222 33333 44444
19100 11111 22222 33333 44444
19101 11111 22222 33333 44444
19104 11111 ..... 33333 .....
19105 ..... 22222 ..... 44444
19106 11111 ..... 33333 .....
19107 ..... 22222 ..... 44444
19108 11111 ..... 33333 .....
19109 ..... 22222 ..... 44444
19110 11111 ..... 33333 .....
19111 ..... 22222 ..... 44444
19115 @item interleave_top, 4
19116 Interleave the upper field from odd frames with the lower field from
19117 even frames, generating a frame with unchanged height at half frame rate.
19122 Frame 1 Frame 2 Frame 3 Frame 4
19124 11111<- 22222 33333<- 44444
19125 11111 22222<- 33333 44444<-
19126 11111<- 22222 33333<- 44444
19127 11111 22222<- 33333 44444<-
19137 @item interleave_bottom, 5
19138 Interleave the lower field from odd frames with the upper field from
19139 even frames, generating a frame with unchanged height at half frame rate.
19144 Frame 1 Frame 2 Frame 3 Frame 4
19146 11111 22222<- 33333 44444<-
19147 11111<- 22222 33333<- 44444
19148 11111 22222<- 33333 44444<-
19149 11111<- 22222 33333<- 44444
19159 @item interlacex2, 6
19160 Double frame rate with unchanged height. Frames are inserted each
19161 containing the second temporal field from the previous input frame and
19162 the first temporal field from the next input frame. This mode relies on
19163 the top_field_first flag. Useful for interlaced video displays with no
19164 field synchronisation.
19169 Frame 1 Frame 2 Frame 3 Frame 4
19171 11111 22222 33333 44444
19172 11111 22222 33333 44444
19173 11111 22222 33333 44444
19174 11111 22222 33333 44444
19177 11111 22222 22222 33333 33333 44444 44444
19178 11111 11111 22222 22222 33333 33333 44444
19179 11111 22222 22222 33333 33333 44444 44444
19180 11111 11111 22222 22222 33333 33333 44444
19185 Move odd frames into the upper field, even into the lower field,
19186 generating a double height frame at same frame rate.
19191 Frame 1 Frame 2 Frame 3 Frame 4
19193 11111 22222 33333 44444
19194 11111 22222 33333 44444
19195 11111 22222 33333 44444
19196 11111 22222 33333 44444
19199 11111 33333 33333 55555
19200 22222 22222 44444 44444
19201 11111 33333 33333 55555
19202 22222 22222 44444 44444
19203 11111 33333 33333 55555
19204 22222 22222 44444 44444
19205 11111 33333 33333 55555
19206 22222 22222 44444 44444
19211 Numeric values are deprecated but are accepted for backward
19212 compatibility reasons.
19214 Default mode is @code{merge}.
19217 Specify flags influencing the filter process.
19219 Available value for @var{flags} is:
19222 @item low_pass_filter, vlpf
19223 Enable linear vertical low-pass filtering in the filter.
19224 Vertical low-pass filtering is required when creating an interlaced
19225 destination from a progressive source which contains high-frequency
19226 vertical detail. Filtering will reduce interlace 'twitter' and Moire
19229 @item complex_filter, cvlpf
19230 Enable complex vertical low-pass filtering.
19231 This will slightly less reduce interlace 'twitter' and Moire
19232 patterning but better retain detail and subjective sharpness impression.
19235 Bypass already interlaced frames, only adjust the frame rate.
19238 Vertical low-pass filtering and bypassing already interlaced frames can only be
19239 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
19244 Pick median pixels from several successive input video frames.
19246 The filter accepts the following options:
19250 Set radius of median filter.
19251 Default is 1. Allowed range is from 1 to 127.
19254 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19257 Set median percentile. Default value is @code{0.5}.
19258 Default value of @code{0.5} will pick always median values, while @code{0} will pick
19259 minimum values, and @code{1} maximum values.
19264 Mix successive video frames.
19266 A description of the accepted options follows.
19270 The number of successive frames to mix. If unspecified, it defaults to 3.
19273 Specify weight of each input video frame.
19274 Each weight is separated by space. If number of weights is smaller than
19275 number of @var{frames} last specified weight will be used for all remaining
19279 Specify scale, if it is set it will be multiplied with sum
19280 of each weight multiplied with pixel values to give final destination
19281 pixel value. By default @var{scale} is auto scaled to sum of weights.
19284 @subsection Examples
19288 Average 7 successive frames:
19290 tmix=frames=7:weights="1 1 1 1 1 1 1"
19294 Apply simple temporal convolution:
19296 tmix=frames=3:weights="-1 3 -1"
19300 Similar as above but only showing temporal differences:
19302 tmix=frames=3:weights="-1 2 -1":scale=1
19308 Tone map colors from different dynamic ranges.
19310 This filter expects data in single precision floating point, as it needs to
19311 operate on (and can output) out-of-range values. Another filter, such as
19312 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19314 The tonemapping algorithms implemented only work on linear light, so input
19315 data should be linearized beforehand (and possibly correctly tagged).
19318 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19321 @subsection Options
19322 The filter accepts the following options.
19326 Set the tone map algorithm to use.
19328 Possible values are:
19331 Do not apply any tone map, only desaturate overbright pixels.
19334 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19335 in-range values, while distorting out-of-range values.
19338 Stretch the entire reference gamut to a linear multiple of the display.
19341 Fit a logarithmic transfer between the tone curves.
19344 Preserve overall image brightness with a simple curve, using nonlinear
19345 contrast, which results in flattening details and degrading color accuracy.
19348 Preserve both dark and bright details better than @var{reinhard}, at the cost
19349 of slightly darkening everything. Use it when detail preservation is more
19350 important than color and brightness accuracy.
19353 Smoothly map out-of-range values, while retaining contrast and colors for
19354 in-range material as much as possible. Use it when color accuracy is more
19355 important than detail preservation.
19361 Tune the tone mapping algorithm.
19363 This affects the following algorithms:
19369 Specifies the scale factor to use while stretching.
19373 Specifies the exponent of the function.
19377 Specify an extra linear coefficient to multiply into the signal before clipping.
19381 Specify the local contrast coefficient at the display peak.
19382 Default to 0.5, which means that in-gamut values will be about half as bright
19389 Specify the transition point from linear to mobius transform. Every value
19390 below this point is guaranteed to be mapped 1:1. The higher the value, the
19391 more accurate the result will be, at the cost of losing bright details.
19392 Default to 0.3, which due to the steep initial slope still preserves in-range
19393 colors fairly accurately.
19397 Apply desaturation for highlights that exceed this level of brightness. The
19398 higher the parameter, the more color information will be preserved. This
19399 setting helps prevent unnaturally blown-out colors for super-highlights, by
19400 (smoothly) turning into white instead. This makes images feel more natural,
19401 at the cost of reducing information about out-of-range colors.
19403 The default of 2.0 is somewhat conservative and will mostly just apply to
19404 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19406 This option works only if the input frame has a supported color tag.
19409 Override signal/nominal/reference peak with this value. Useful when the
19410 embedded peak information in display metadata is not reliable or when tone
19411 mapping from a lower range to a higher range.
19416 Temporarily pad video frames.
19418 The filter accepts the following options:
19422 Specify number of delay frames before input video stream. Default is 0.
19425 Specify number of padding frames after input video stream.
19426 Set to -1 to pad indefinitely. Default is 0.
19429 Set kind of frames added to beginning of stream.
19430 Can be either @var{add} or @var{clone}.
19431 With @var{add} frames of solid-color are added.
19432 With @var{clone} frames are clones of first frame.
19433 Default is @var{add}.
19436 Set kind of frames added to end of stream.
19437 Can be either @var{add} or @var{clone}.
19438 With @var{add} frames of solid-color are added.
19439 With @var{clone} frames are clones of last frame.
19440 Default is @var{add}.
19442 @item start_duration, stop_duration
19443 Specify the duration of the start/stop delay. See
19444 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19445 for the accepted syntax.
19446 These options override @var{start} and @var{stop}. Default is 0.
19449 Specify the color of the padded area. For the syntax of this option,
19450 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19451 manual,ffmpeg-utils}.
19453 The default value of @var{color} is "black".
19459 Transpose rows with columns in the input video and optionally flip it.
19461 It accepts the following parameters:
19466 Specify the transposition direction.
19468 Can assume the following values:
19470 @item 0, 4, cclock_flip
19471 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19479 Rotate by 90 degrees clockwise, that is:
19487 Rotate by 90 degrees counterclockwise, that is:
19494 @item 3, 7, clock_flip
19495 Rotate by 90 degrees clockwise and vertically flip, that is:
19503 For values between 4-7, the transposition is only done if the input
19504 video geometry is portrait and not landscape. These values are
19505 deprecated, the @code{passthrough} option should be used instead.
19507 Numerical values are deprecated, and should be dropped in favor of
19508 symbolic constants.
19511 Do not apply the transposition if the input geometry matches the one
19512 specified by the specified value. It accepts the following values:
19515 Always apply transposition.
19517 Preserve portrait geometry (when @var{height} >= @var{width}).
19519 Preserve landscape geometry (when @var{width} >= @var{height}).
19522 Default value is @code{none}.
19525 For example to rotate by 90 degrees clockwise and preserve portrait
19528 transpose=dir=1:passthrough=portrait
19531 The command above can also be specified as:
19533 transpose=1:portrait
19536 @section transpose_npp
19538 Transpose rows with columns in the input video and optionally flip it.
19539 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19541 It accepts the following parameters:
19546 Specify the transposition direction.
19548 Can assume the following values:
19551 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19554 Rotate by 90 degrees clockwise.
19557 Rotate by 90 degrees counterclockwise.
19560 Rotate by 90 degrees clockwise and vertically flip.
19564 Do not apply the transposition if the input geometry matches the one
19565 specified by the specified value. It accepts the following values:
19568 Always apply transposition. (default)
19570 Preserve portrait geometry (when @var{height} >= @var{width}).
19572 Preserve landscape geometry (when @var{width} >= @var{height}).
19578 Trim the input so that the output contains one continuous subpart of the input.
19580 It accepts the following parameters:
19583 Specify the time of the start of the kept section, i.e. the frame with the
19584 timestamp @var{start} will be the first frame in the output.
19587 Specify the time of the first frame that will be dropped, i.e. the frame
19588 immediately preceding the one with the timestamp @var{end} will be the last
19589 frame in the output.
19592 This is the same as @var{start}, except this option sets the start timestamp
19593 in timebase units instead of seconds.
19596 This is the same as @var{end}, except this option sets the end timestamp
19597 in timebase units instead of seconds.
19600 The maximum duration of the output in seconds.
19603 The number of the first frame that should be passed to the output.
19606 The number of the first frame that should be dropped.
19609 @option{start}, @option{end}, and @option{duration} are expressed as time
19610 duration specifications; see
19611 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19612 for the accepted syntax.
19614 Note that the first two sets of the start/end options and the @option{duration}
19615 option look at the frame timestamp, while the _frame variants simply count the
19616 frames that pass through the filter. Also note that this filter does not modify
19617 the timestamps. If you wish for the output timestamps to start at zero, insert a
19618 setpts filter after the trim filter.
19620 If multiple start or end options are set, this filter tries to be greedy and
19621 keep all the frames that match at least one of the specified constraints. To keep
19622 only the part that matches all the constraints at once, chain multiple trim
19625 The defaults are such that all the input is kept. So it is possible to set e.g.
19626 just the end values to keep everything before the specified time.
19631 Drop everything except the second minute of input:
19633 ffmpeg -i INPUT -vf trim=60:120
19637 Keep only the first second:
19639 ffmpeg -i INPUT -vf trim=duration=1
19644 @section unpremultiply
19645 Apply alpha unpremultiply effect to input video stream using first plane
19646 of second stream as alpha.
19648 Both streams must have same dimensions and same pixel format.
19650 The filter accepts the following option:
19654 Set which planes will be processed, unprocessed planes will be copied.
19655 By default value 0xf, all planes will be processed.
19657 If the format has 1 or 2 components, then luma is bit 0.
19658 If the format has 3 or 4 components:
19659 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19660 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19661 If present, the alpha channel is always the last bit.
19664 Do not require 2nd input for processing, instead use alpha plane from input stream.
19670 Sharpen or blur the input video.
19672 It accepts the following parameters:
19675 @item luma_msize_x, lx
19676 Set the luma matrix horizontal size. It must be an odd integer between
19677 3 and 23. The default value is 5.
19679 @item luma_msize_y, ly
19680 Set the luma matrix vertical size. It must be an odd integer between 3
19681 and 23. The default value is 5.
19683 @item luma_amount, la
19684 Set the luma effect strength. It must be a floating point number, reasonable
19685 values lay between -1.5 and 1.5.
19687 Negative values will blur the input video, while positive values will
19688 sharpen it, a value of zero will disable the effect.
19690 Default value is 1.0.
19692 @item chroma_msize_x, cx
19693 Set the chroma matrix horizontal size. It must be an odd integer
19694 between 3 and 23. The default value is 5.
19696 @item chroma_msize_y, cy
19697 Set the chroma matrix vertical size. It must be an odd integer
19698 between 3 and 23. The default value is 5.
19700 @item chroma_amount, ca
19701 Set the chroma effect strength. It must be a floating point number, reasonable
19702 values lay between -1.5 and 1.5.
19704 Negative values will blur the input video, while positive values will
19705 sharpen it, a value of zero will disable the effect.
19707 Default value is 0.0.
19711 All parameters are optional and default to the equivalent of the
19712 string '5:5:1.0:5:5:0.0'.
19714 @subsection Examples
19718 Apply strong luma sharpen effect:
19720 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19724 Apply a strong blur of both luma and chroma parameters:
19726 unsharp=7:7:-2:7:7:-2
19733 Decompose a video made of tiled images into the individual images.
19735 The frame rate of the output video is the frame rate of the input video
19736 multiplied by the number of tiles.
19738 This filter does the reverse of @ref{tile}.
19740 The filter accepts the following options:
19745 Set the grid size (i.e. the number of lines and columns). For the syntax of
19746 this option, check the
19747 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19750 @subsection Examples
19754 Produce a 1-second video from a still image file made of 25 frames stacked
19755 vertically, like an analogic film reel:
19757 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19763 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19764 the image at several (or - in the case of @option{quality} level @code{8} - all)
19765 shifts and average the results.
19767 The way this differs from the behavior of spp is that uspp actually encodes &
19768 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19769 DCT similar to MJPEG.
19771 The filter accepts the following options:
19775 Set quality. This option defines the number of levels for averaging. It accepts
19776 an integer in the range 0-8. If set to @code{0}, the filter will have no
19777 effect. A value of @code{8} means the higher quality. For each increment of
19778 that value the speed drops by a factor of approximately 2. Default value is
19782 Force a constant quantization parameter. If not set, the filter will use the QP
19783 from the video stream (if available).
19788 Convert 360 videos between various formats.
19790 The filter accepts the following options:
19796 Set format of the input/output video.
19804 Equirectangular projection.
19809 Cubemap with 3x2/6x1/1x6 layout.
19811 Format specific options:
19816 Set padding proportion for the input/output cubemap. Values in decimals.
19823 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)
19826 Default value is @b{@samp{0}}.
19827 Maximum value is @b{@samp{0.1}}.
19831 Set fixed padding for the input/output cubemap. Values in pixels.
19833 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19837 Set order of faces for the input/output cubemap. Choose one direction for each position.
19839 Designation of directions:
19855 Default value is @b{@samp{rludfb}}.
19859 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19861 Designation of angles:
19864 0 degrees clockwise
19866 90 degrees clockwise
19868 180 degrees clockwise
19870 270 degrees clockwise
19873 Default value is @b{@samp{000000}}.
19877 Equi-Angular Cubemap.
19884 Format specific options:
19889 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19891 If diagonal field of view is set it overrides horizontal and vertical field of view.
19896 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19898 If diagonal field of view is set it overrides horizontal and vertical field of view.
19904 Format specific options:
19909 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19911 If diagonal field of view is set it overrides horizontal and vertical field of view.
19916 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19918 If diagonal field of view is set it overrides horizontal and vertical field of view.
19924 Facebook's 360 formats.
19927 Stereographic format.
19929 Format specific options:
19934 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19936 If diagonal field of view is set it overrides horizontal and vertical field of view.
19941 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19943 If diagonal field of view is set it overrides horizontal and vertical field of view.
19950 Ball format, gives significant distortion toward the back.
19953 Hammer-Aitoff map projection format.
19956 Sinusoidal map projection format.
19959 Fisheye projection.
19961 Format specific options:
19966 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19968 If diagonal field of view is set it overrides horizontal and vertical field of view.
19973 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19975 If diagonal field of view is set it overrides horizontal and vertical field of view.
19979 Pannini projection.
19981 Format specific options:
19984 Set output pannini parameter.
19987 Set input pannini parameter.
19991 Cylindrical projection.
19993 Format specific options:
19998 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20000 If diagonal field of view is set it overrides horizontal and vertical field of view.
20005 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20007 If diagonal field of view is set it overrides horizontal and vertical field of view.
20011 Perspective projection. @i{(output only)}
20013 Format specific options:
20016 Set perspective parameter.
20020 Tetrahedron projection.
20023 Truncated square pyramid projection.
20027 Half equirectangular projection.
20032 Format specific options:
20037 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20039 If diagonal field of view is set it overrides horizontal and vertical field of view.
20044 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20046 If diagonal field of view is set it overrides horizontal and vertical field of view.
20050 Orthographic format.
20052 Format specific options:
20057 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20059 If diagonal field of view is set it overrides horizontal and vertical field of view.
20064 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20066 If diagonal field of view is set it overrides horizontal and vertical field of view.
20070 Octahedron projection.
20074 Set interpolation method.@*
20075 @i{Note: more complex interpolation methods require much more memory to run.}
20085 Bilinear interpolation.
20087 Lagrange9 interpolation.
20090 Bicubic interpolation.
20093 Lanczos interpolation.
20096 Spline16 interpolation.
20099 Gaussian interpolation.
20101 Mitchell interpolation.
20104 Default value is @b{@samp{line}}.
20108 Set the output video resolution.
20110 Default resolution depends on formats.
20114 Set the input/output stereo format.
20125 Default value is @b{@samp{2d}} for input and output format.
20130 Set rotation for the output video. Values in degrees.
20133 Set rotation order for the output video. Choose one item for each position.
20144 Default value is @b{@samp{ypr}}.
20149 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
20153 Set if input video is flipped horizontally/vertically. Boolean values.
20156 Set if input video is transposed. Boolean value, by default disabled.
20159 Set if output video needs to be transposed. Boolean value, by default disabled.
20162 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
20165 @subsection Examples
20169 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
20171 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20174 Extract back view of Equi-Angular Cubemap:
20176 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20179 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
20181 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20185 @subsection Commands
20187 This filter supports subset of above options as @ref{commands}.
20189 @section vaguedenoiser
20191 Apply a wavelet based denoiser.
20193 It transforms each frame from the video input into the wavelet domain,
20194 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20195 the obtained coefficients. It does an inverse wavelet transform after.
20196 Due to wavelet properties, it should give a nice smoothed result, and
20197 reduced noise, without blurring picture features.
20199 This filter accepts the following options:
20203 The filtering strength. The higher, the more filtered the video will be.
20204 Hard thresholding can use a higher threshold than soft thresholding
20205 before the video looks overfiltered. Default value is 2.
20208 The filtering method the filter will use.
20210 It accepts the following values:
20213 All values under the threshold will be zeroed.
20216 All values under the threshold will be zeroed. All values above will be
20217 reduced by the threshold.
20220 Scales or nullifies coefficients - intermediary between (more) soft and
20221 (less) hard thresholding.
20224 Default is garrote.
20227 Number of times, the wavelet will decompose the picture. Picture can't
20228 be decomposed beyond a particular point (typically, 8 for a 640x480
20229 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
20232 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
20235 A list of the planes to process. By default all planes are processed.
20238 The threshold type the filter will use.
20240 It accepts the following values:
20243 Threshold used is same for all decompositions.
20246 Threshold used depends also on each decomposition coefficients.
20249 Default is universal.
20252 @section vectorscope
20254 Display 2 color component values in the two dimensional graph (which is called
20257 This filter accepts the following options:
20261 Set vectorscope mode.
20263 It accepts the following values:
20267 Gray values are displayed on graph, higher brightness means more pixels have
20268 same component color value on location in graph. This is the default mode.
20271 Gray values are displayed on graph. Surrounding pixels values which are not
20272 present in video frame are drawn in gradient of 2 color components which are
20273 set by option @code{x} and @code{y}. The 3rd color component is static.
20276 Actual color components values present in video frame are displayed on graph.
20279 Similar as color2 but higher frequency of same values @code{x} and @code{y}
20280 on graph increases value of another color component, which is luminance by
20281 default values of @code{x} and @code{y}.
20284 Actual colors present in video frame are displayed on graph. If two different
20285 colors map to same position on graph then color with higher value of component
20286 not present in graph is picked.
20289 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
20290 component picked from radial gradient.
20294 Set which color component will be represented on X-axis. Default is @code{1}.
20297 Set which color component will be represented on Y-axis. Default is @code{2}.
20300 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20301 of color component which represents frequency of (X, Y) location in graph.
20306 No envelope, this is default.
20309 Instant envelope, even darkest single pixel will be clearly highlighted.
20312 Hold maximum and minimum values presented in graph over time. This way you
20313 can still spot out of range values without constantly looking at vectorscope.
20316 Peak and instant envelope combined together.
20320 Set what kind of graticule to draw.
20329 Set graticule opacity.
20332 Set graticule flags.
20336 Draw graticule for white point.
20339 Draw graticule for black point.
20342 Draw color points short names.
20346 Set background opacity.
20348 @item lthreshold, l
20349 Set low threshold for color component not represented on X or Y axis.
20350 Values lower than this value will be ignored. Default is 0.
20351 Note this value is multiplied with actual max possible value one pixel component
20352 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20355 @item hthreshold, h
20356 Set high threshold for color component not represented on X or Y axis.
20357 Values higher than this value will be ignored. Default is 1.
20358 Note this value is multiplied with actual max possible value one pixel component
20359 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20360 is 0.9 * 255 = 230.
20362 @item colorspace, c
20363 Set what kind of colorspace to use when drawing graticule.
20373 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20374 This means no tint, and output will remain gray.
20377 @anchor{vidstabdetect}
20378 @section vidstabdetect
20380 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20381 @ref{vidstabtransform} for pass 2.
20383 This filter generates a file with relative translation and rotation
20384 transform information about subsequent frames, which is then used by
20385 the @ref{vidstabtransform} filter.
20387 To enable compilation of this filter you need to configure FFmpeg with
20388 @code{--enable-libvidstab}.
20390 This filter accepts the following options:
20394 Set the path to the file used to write the transforms information.
20395 Default value is @file{transforms.trf}.
20398 Set how shaky the video is and how quick the camera is. It accepts an
20399 integer in the range 1-10, a value of 1 means little shakiness, a
20400 value of 10 means strong shakiness. Default value is 5.
20403 Set the accuracy of the detection process. It must be a value in the
20404 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20405 accuracy. Default value is 15.
20408 Set stepsize of the search process. The region around minimum is
20409 scanned with 1 pixel resolution. Default value is 6.
20412 Set minimum contrast. Below this value a local measurement field is
20413 discarded. Must be a floating point value in the range 0-1. Default
20417 Set reference frame number for tripod mode.
20419 If enabled, the motion of the frames is compared to a reference frame
20420 in the filtered stream, identified by the specified number. The idea
20421 is to compensate all movements in a more-or-less static scene and keep
20422 the camera view absolutely still.
20424 If set to 0, it is disabled. The frames are counted starting from 1.
20427 Show fields and transforms in the resulting frames. It accepts an
20428 integer in the range 0-2. Default value is 0, which disables any
20432 @subsection Examples
20436 Use default values:
20442 Analyze strongly shaky movie and put the results in file
20443 @file{mytransforms.trf}:
20445 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20449 Visualize the result of internal transformations in the resulting
20452 vidstabdetect=show=1
20456 Analyze a video with medium shakiness using @command{ffmpeg}:
20458 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20462 @anchor{vidstabtransform}
20463 @section vidstabtransform
20465 Video stabilization/deshaking: pass 2 of 2,
20466 see @ref{vidstabdetect} for pass 1.
20468 Read a file with transform information for each frame and
20469 apply/compensate them. Together with the @ref{vidstabdetect}
20470 filter this can be used to deshake videos. See also
20471 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20472 the @ref{unsharp} filter, see below.
20474 To enable compilation of this filter you need to configure FFmpeg with
20475 @code{--enable-libvidstab}.
20477 @subsection Options
20481 Set path to the file used to read the transforms. Default value is
20482 @file{transforms.trf}.
20485 Set the number of frames (value*2 + 1) used for lowpass filtering the
20486 camera movements. Default value is 10.
20488 For example a number of 10 means that 21 frames are used (10 in the
20489 past and 10 in the future) to smoothen the motion in the video. A
20490 larger value leads to a smoother video, but limits the acceleration of
20491 the camera (pan/tilt movements). 0 is a special case where a static
20492 camera is simulated.
20495 Set the camera path optimization algorithm.
20497 Accepted values are:
20500 gaussian kernel low-pass filter on camera motion (default)
20502 averaging on transformations
20506 Set maximal number of pixels to translate frames. Default value is -1,
20510 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20511 value is -1, meaning no limit.
20514 Specify how to deal with borders that may be visible due to movement
20517 Available values are:
20520 keep image information from previous frame (default)
20522 fill the border black
20526 Invert transforms if set to 1. Default value is 0.
20529 Consider transforms as relative to previous frame if set to 1,
20530 absolute if set to 0. Default value is 0.
20533 Set percentage to zoom. A positive value will result in a zoom-in
20534 effect, a negative value in a zoom-out effect. Default value is 0 (no
20538 Set optimal zooming to avoid borders.
20540 Accepted values are:
20545 optimal static zoom value is determined (only very strong movements
20546 will lead to visible borders) (default)
20548 optimal adaptive zoom value is determined (no borders will be
20549 visible), see @option{zoomspeed}
20552 Note that the value given at zoom is added to the one calculated here.
20555 Set percent to zoom maximally each frame (enabled when
20556 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20560 Specify type of interpolation.
20562 Available values are:
20567 linear only horizontal
20569 linear in both directions (default)
20571 cubic in both directions (slow)
20575 Enable virtual tripod mode if set to 1, which is equivalent to
20576 @code{relative=0:smoothing=0}. Default value is 0.
20578 Use also @code{tripod} option of @ref{vidstabdetect}.
20581 Increase log verbosity if set to 1. Also the detected global motions
20582 are written to the temporary file @file{global_motions.trf}. Default
20586 @subsection Examples
20590 Use @command{ffmpeg} for a typical stabilization with default values:
20592 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20595 Note the use of the @ref{unsharp} filter which is always recommended.
20598 Zoom in a bit more and load transform data from a given file:
20600 vidstabtransform=zoom=5:input="mytransforms.trf"
20604 Smoothen the video even more:
20606 vidstabtransform=smoothing=30
20612 Flip the input video vertically.
20614 For example, to vertically flip a video with @command{ffmpeg}:
20616 ffmpeg -i in.avi -vf "vflip" out.avi
20621 Detect variable frame rate video.
20623 This filter tries to detect if the input is variable or constant frame rate.
20625 At end it will output number of frames detected as having variable delta pts,
20626 and ones with constant delta pts.
20627 If there was frames with variable delta, than it will also show min, max and
20628 average delta encountered.
20632 Boost or alter saturation.
20634 The filter accepts the following options:
20637 Set strength of boost if positive value or strength of alter if negative value.
20638 Default is 0. Allowed range is from -2 to 2.
20641 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20644 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20647 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20650 Set the red luma coefficient.
20653 Set the green luma coefficient.
20656 Set the blue luma coefficient.
20659 If @code{intensity} is negative and this is set to 1, colors will change,
20660 otherwise colors will be less saturated, more towards gray.
20663 @subsection Commands
20665 This filter supports the all above options as @ref{commands}.
20670 Make or reverse a natural vignetting effect.
20672 The filter accepts the following options:
20676 Set lens angle expression as a number of radians.
20678 The value is clipped in the @code{[0,PI/2]} range.
20680 Default value: @code{"PI/5"}
20684 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20688 Set forward/backward mode.
20690 Available modes are:
20693 The larger the distance from the central point, the darker the image becomes.
20696 The larger the distance from the central point, the brighter the image becomes.
20697 This can be used to reverse a vignette effect, though there is no automatic
20698 detection to extract the lens @option{angle} and other settings (yet). It can
20699 also be used to create a burning effect.
20702 Default value is @samp{forward}.
20705 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20707 It accepts the following values:
20710 Evaluate expressions only once during the filter initialization.
20713 Evaluate expressions for each incoming frame. This is way slower than the
20714 @samp{init} mode since it requires all the scalers to be re-computed, but it
20715 allows advanced dynamic expressions.
20718 Default value is @samp{init}.
20721 Set dithering to reduce the circular banding effects. Default is @code{1}
20725 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20726 Setting this value to the SAR of the input will make a rectangular vignetting
20727 following the dimensions of the video.
20729 Default is @code{1/1}.
20732 @subsection Expressions
20734 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20735 following parameters.
20740 input width and height
20743 the number of input frame, starting from 0
20746 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20747 @var{TB} units, NAN if undefined
20750 frame rate of the input video, NAN if the input frame rate is unknown
20753 the PTS (Presentation TimeStamp) of the filtered video frame,
20754 expressed in seconds, NAN if undefined
20757 time base of the input video
20761 @subsection Examples
20765 Apply simple strong vignetting effect:
20771 Make a flickering vignetting:
20773 vignette='PI/4+random(1)*PI/50':eval=frame
20778 @section vmafmotion
20780 Obtain the average VMAF motion score of a video.
20781 It is one of the component metrics of VMAF.
20783 The obtained average motion score is printed through the logging system.
20785 The filter accepts the following options:
20789 If specified, the filter will use the named file to save the motion score of
20790 each frame with respect to the previous frame.
20791 When filename equals "-" the data is sent to standard output.
20796 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20800 Stack input videos vertically.
20802 All streams must be of same pixel format and of same width.
20804 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20805 to create same output.
20807 The filter accepts the following options:
20811 Set number of input streams. Default is 2.
20814 If set to 1, force the output to terminate when the shortest input
20815 terminates. Default value is 0.
20820 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20821 Deinterlacing Filter").
20823 Based on the process described by Martin Weston for BBC R&D, and
20824 implemented based on the de-interlace algorithm written by Jim
20825 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20826 uses filter coefficients calculated by BBC R&D.
20828 This filter uses field-dominance information in frame to decide which
20829 of each pair of fields to place first in the output.
20830 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20832 There are two sets of filter coefficients, so called "simple"
20833 and "complex". Which set of filter coefficients is used can
20834 be set by passing an optional parameter:
20838 Set the interlacing filter coefficients. Accepts one of the following values:
20842 Simple filter coefficient set.
20844 More-complex filter coefficient set.
20846 Default value is @samp{complex}.
20849 Specify which frames to deinterlace. Accepts one of the following values:
20853 Deinterlace all frames,
20855 Only deinterlace frames marked as interlaced.
20858 Default value is @samp{all}.
20862 Video waveform monitor.
20864 The waveform monitor plots color component intensity. By default luminance
20865 only. Each column of the waveform corresponds to a column of pixels in the
20868 It accepts the following options:
20872 Can be either @code{row}, or @code{column}. Default is @code{column}.
20873 In row mode, the graph on the left side represents color component value 0 and
20874 the right side represents value = 255. In column mode, the top side represents
20875 color component value = 0 and bottom side represents value = 255.
20878 Set intensity. Smaller values are useful to find out how many values of the same
20879 luminance are distributed across input rows/columns.
20880 Default value is @code{0.04}. Allowed range is [0, 1].
20883 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20884 In mirrored mode, higher values will be represented on the left
20885 side for @code{row} mode and at the top for @code{column} mode. Default is
20886 @code{1} (mirrored).
20890 It accepts the following values:
20893 Presents information identical to that in the @code{parade}, except
20894 that the graphs representing color components are superimposed directly
20897 This display mode makes it easier to spot relative differences or similarities
20898 in overlapping areas of the color components that are supposed to be identical,
20899 such as neutral whites, grays, or blacks.
20902 Display separate graph for the color components side by side in
20903 @code{row} mode or one below the other in @code{column} mode.
20906 Display separate graph for the color components side by side in
20907 @code{column} mode or one below the other in @code{row} mode.
20909 Using this display mode makes it easy to spot color casts in the highlights
20910 and shadows of an image, by comparing the contours of the top and the bottom
20911 graphs of each waveform. Since whites, grays, and blacks are characterized
20912 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20913 should display three waveforms of roughly equal width/height. If not, the
20914 correction is easy to perform by making level adjustments the three waveforms.
20916 Default is @code{stack}.
20918 @item components, c
20919 Set which color components to display. Default is 1, which means only luminance
20920 or red color component if input is in RGB colorspace. If is set for example to
20921 7 it will display all 3 (if) available color components.
20926 No envelope, this is default.
20929 Instant envelope, minimum and maximum values presented in graph will be easily
20930 visible even with small @code{step} value.
20933 Hold minimum and maximum values presented in graph across time. This way you
20934 can still spot out of range values without constantly looking at waveforms.
20937 Peak and instant envelope combined together.
20943 No filtering, this is default.
20946 Luma and chroma combined together.
20949 Similar as above, but shows difference between blue and red chroma.
20952 Similar as above, but use different colors.
20955 Similar as above, but again with different colors.
20958 Displays only chroma.
20961 Displays actual color value on waveform.
20964 Similar as above, but with luma showing frequency of chroma values.
20968 Set which graticule to display.
20972 Do not display graticule.
20975 Display green graticule showing legal broadcast ranges.
20978 Display orange graticule showing legal broadcast ranges.
20981 Display invert graticule showing legal broadcast ranges.
20985 Set graticule opacity.
20988 Set graticule flags.
20992 Draw numbers above lines. By default enabled.
20995 Draw dots instead of lines.
20999 Set scale used for displaying graticule.
21006 Default is digital.
21009 Set background opacity.
21013 Set tint for output.
21014 Only used with lowpass filter and when display is not overlay and input
21015 pixel formats are not RGB.
21018 @section weave, doubleweave
21020 The @code{weave} takes a field-based video input and join
21021 each two sequential fields into single frame, producing a new double
21022 height clip with half the frame rate and half the frame count.
21024 The @code{doubleweave} works same as @code{weave} but without
21025 halving frame rate and frame count.
21027 It accepts the following option:
21031 Set first field. Available values are:
21035 Set the frame as top-field-first.
21038 Set the frame as bottom-field-first.
21042 @subsection Examples
21046 Interlace video using @ref{select} and @ref{separatefields} filter:
21048 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21053 Apply the xBR high-quality magnification filter which is designed for pixel
21054 art. It follows a set of edge-detection rules, see
21055 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
21057 It accepts the following option:
21061 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
21062 @code{3xBR} and @code{4} for @code{4xBR}.
21063 Default is @code{3}.
21068 Apply cross fade from one input video stream to another input video stream.
21069 The cross fade is applied for specified duration.
21071 The filter accepts the following options:
21075 Set one of available transition effects:
21123 Default transition effect is fade.
21126 Set cross fade duration in seconds.
21127 Default duration is 1 second.
21130 Set cross fade start relative to first input stream in seconds.
21131 Default offset is 0.
21134 Set expression for custom transition effect.
21136 The expressions can use the following variables and functions:
21141 The coordinates of the current sample.
21145 The width and height of the image.
21148 Progress of transition effect.
21151 Currently processed plane.
21154 Return value of first input at current location and plane.
21157 Return value of second input at current location and plane.
21163 Return the value of the pixel at location (@var{x},@var{y}) of the
21164 first/second/third/fourth component of first input.
21170 Return the value of the pixel at location (@var{x},@var{y}) of the
21171 first/second/third/fourth component of second input.
21175 @subsection Examples
21179 Cross fade from one input video to another input video, with fade transition and duration of transition
21180 of 2 seconds starting at offset of 5 seconds:
21182 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21187 Pick median pixels from several input videos.
21189 The filter accepts the following options:
21193 Set number of inputs.
21194 Default is 3. Allowed range is from 3 to 255.
21195 If number of inputs is even number, than result will be mean value between two median values.
21198 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
21201 Set median percentile. Default value is @code{0.5}.
21202 Default value of @code{0.5} will pick always median values, while @code{0} will pick
21203 minimum values, and @code{1} maximum values.
21207 Stack video inputs into custom layout.
21209 All streams must be of same pixel format.
21211 The filter accepts the following options:
21215 Set number of input streams. Default is 2.
21218 Specify layout of inputs.
21219 This option requires the desired layout configuration to be explicitly set by the user.
21220 This sets position of each video input in output. Each input
21221 is separated by '|'.
21222 The first number represents the column, and the second number represents the row.
21223 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
21224 where X is video input from which to take width or height.
21225 Multiple values can be used when separated by '+'. In such
21226 case values are summed together.
21228 Note that if inputs are of different sizes gaps may appear, as not all of
21229 the output video frame will be filled. Similarly, videos can overlap each
21230 other if their position doesn't leave enough space for the full frame of
21233 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
21234 a layout must be set by the user.
21237 If set to 1, force the output to terminate when the shortest input
21238 terminates. Default value is 0.
21241 If set to valid color, all unused pixels will be filled with that color.
21242 By default fill is set to none, so it is disabled.
21245 @subsection Examples
21249 Display 4 inputs into 2x2 grid.
21253 input1(0, 0) | input3(w0, 0)
21254 input2(0, h0) | input4(w0, h0)
21258 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21261 Note that if inputs are of different sizes, gaps or overlaps may occur.
21264 Display 4 inputs into 1x4 grid.
21271 input4(0, h0+h1+h2)
21275 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21278 Note that if inputs are of different widths, unused space will appear.
21281 Display 9 inputs into 3x3 grid.
21285 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21286 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21287 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21291 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
21294 Note that if inputs are of different sizes, gaps or overlaps may occur.
21297 Display 16 inputs into 4x4 grid.
21301 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21302 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21303 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21304 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21308 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|
21309 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
21312 Note that if inputs are of different sizes, gaps or overlaps may occur.
21319 Deinterlace the input video ("yadif" means "yet another deinterlacing
21322 It accepts the following parameters:
21328 The interlacing mode to adopt. It accepts one of the following values:
21331 @item 0, send_frame
21332 Output one frame for each frame.
21333 @item 1, send_field
21334 Output one frame for each field.
21335 @item 2, send_frame_nospatial
21336 Like @code{send_frame}, but it skips the spatial interlacing check.
21337 @item 3, send_field_nospatial
21338 Like @code{send_field}, but it skips the spatial interlacing check.
21341 The default value is @code{send_frame}.
21344 The picture field parity assumed for the input interlaced video. It accepts one
21345 of the following values:
21349 Assume the top field is first.
21351 Assume the bottom field is first.
21353 Enable automatic detection of field parity.
21356 The default value is @code{auto}.
21357 If the interlacing is unknown or the decoder does not export this information,
21358 top field first will be assumed.
21361 Specify which frames to deinterlace. Accepts one of the following
21366 Deinterlace all frames.
21367 @item 1, interlaced
21368 Only deinterlace frames marked as interlaced.
21371 The default value is @code{all}.
21374 @section yadif_cuda
21376 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21377 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21380 It accepts the following parameters:
21386 The interlacing mode to adopt. It accepts one of the following values:
21389 @item 0, send_frame
21390 Output one frame for each frame.
21391 @item 1, send_field
21392 Output one frame for each field.
21393 @item 2, send_frame_nospatial
21394 Like @code{send_frame}, but it skips the spatial interlacing check.
21395 @item 3, send_field_nospatial
21396 Like @code{send_field}, but it skips the spatial interlacing check.
21399 The default value is @code{send_frame}.
21402 The picture field parity assumed for the input interlaced video. It accepts one
21403 of the following values:
21407 Assume the top field is first.
21409 Assume the bottom field is first.
21411 Enable automatic detection of field parity.
21414 The default value is @code{auto}.
21415 If the interlacing is unknown or the decoder does not export this information,
21416 top field first will be assumed.
21419 Specify which frames to deinterlace. Accepts one of the following
21424 Deinterlace all frames.
21425 @item 1, interlaced
21426 Only deinterlace frames marked as interlaced.
21429 The default value is @code{all}.
21434 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21435 The algorithm is described in
21436 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21438 It accepts the following parameters:
21442 Set the window radius. Default value is 3.
21445 Set which planes to filter. Default is only the first plane.
21448 Set blur strength. Default value is 128.
21451 @subsection Commands
21452 This filter supports same @ref{commands} as options.
21456 Apply Zoom & Pan effect.
21458 This filter accepts the following options:
21462 Set the zoom expression. Range is 1-10. Default is 1.
21466 Set the x and y expression. Default is 0.
21469 Set the duration expression in number of frames.
21470 This sets for how many number of frames effect will last for
21471 single input image.
21474 Set the output image size, default is 'hd720'.
21477 Set the output frame rate, default is '25'.
21480 Each expression can contain the following constants:
21499 Output frame count.
21502 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21504 @item out_time, time, ot
21505 The output timestamp expressed in seconds.
21509 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21510 for current input frame.
21514 'x' and 'y' of last output frame of previous input frame or 0 when there was
21515 not yet such frame (first input frame).
21518 Last calculated zoom from 'z' expression for current input frame.
21521 Last calculated zoom of last output frame of previous input frame.
21524 Number of output frames for current input frame. Calculated from 'd' expression
21525 for each input frame.
21528 number of output frames created for previous input frame
21531 Rational number: input width / input height
21534 sample aspect ratio
21537 display aspect ratio
21541 @subsection Examples
21545 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21547 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
21551 Zoom in up to 1.5x and pan always at center of picture:
21553 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21557 Same as above but without pausing:
21559 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21563 Zoom in 2x into center of picture only for the first second of the input video:
21565 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21572 Scale (resize) the input video, using the z.lib library:
21573 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21574 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21576 The zscale filter forces the output display aspect ratio to be the same
21577 as the input, by changing the output sample aspect ratio.
21579 If the input image format is different from the format requested by
21580 the next filter, the zscale filter will convert the input to the
21583 @subsection Options
21584 The filter accepts the following options.
21589 Set the output video dimension expression. Default value is the input
21592 If the @var{width} or @var{w} value is 0, the input width is used for
21593 the output. If the @var{height} or @var{h} value is 0, the input height
21594 is used for the output.
21596 If one and only one of the values is -n with n >= 1, the zscale filter
21597 will use a value that maintains the aspect ratio of the input image,
21598 calculated from the other specified dimension. After that it will,
21599 however, make sure that the calculated dimension is divisible by n and
21600 adjust the value if necessary.
21602 If both values are -n with n >= 1, the behavior will be identical to
21603 both values being set to 0 as previously detailed.
21605 See below for the list of accepted constants for use in the dimension
21609 Set the video size. For the syntax of this option, check the
21610 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21613 Set the dither type.
21615 Possible values are:
21620 @item error_diffusion
21626 Set the resize filter type.
21628 Possible values are:
21638 Default is bilinear.
21641 Set the color range.
21643 Possible values are:
21650 Default is same as input.
21653 Set the color primaries.
21655 Possible values are:
21665 Default is same as input.
21668 Set the transfer characteristics.
21670 Possible values are:
21684 Default is same as input.
21687 Set the colorspace matrix.
21689 Possible value are:
21700 Default is same as input.
21703 Set the input color range.
21705 Possible values are:
21712 Default is same as input.
21714 @item primariesin, pin
21715 Set the input color primaries.
21717 Possible values are:
21727 Default is same as input.
21729 @item transferin, tin
21730 Set the input transfer characteristics.
21732 Possible values are:
21743 Default is same as input.
21745 @item matrixin, min
21746 Set the input colorspace matrix.
21748 Possible value are:
21760 Set the output chroma location.
21762 Possible values are:
21773 @item chromalin, cin
21774 Set the input chroma location.
21776 Possible values are:
21788 Set the nominal peak luminance.
21791 The values of the @option{w} and @option{h} options are expressions
21792 containing the following constants:
21797 The input width and height
21801 These are the same as @var{in_w} and @var{in_h}.
21805 The output (scaled) width and height
21809 These are the same as @var{out_w} and @var{out_h}
21812 The same as @var{iw} / @var{ih}
21815 input sample aspect ratio
21818 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21822 horizontal and vertical input chroma subsample values. For example for the
21823 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21827 horizontal and vertical output chroma subsample values. For example for the
21828 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21831 @subsection Commands
21833 This filter supports the following commands:
21837 Set the output video dimension expression.
21838 The command accepts the same syntax of the corresponding option.
21840 If the specified expression is not valid, it is kept at its current
21844 @c man end VIDEO FILTERS
21846 @chapter OpenCL Video Filters
21847 @c man begin OPENCL VIDEO FILTERS
21849 Below is a description of the currently available OpenCL video filters.
21851 To enable compilation of these filters you need to configure FFmpeg with
21852 @code{--enable-opencl}.
21854 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21857 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21858 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21859 given device parameters.
21861 @item -filter_hw_device @var{name}
21862 Pass the hardware device called @var{name} to all filters in any filter graph.
21866 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21870 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21872 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21876 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.
21878 @section avgblur_opencl
21880 Apply average blur filter.
21882 The filter accepts the following options:
21886 Set horizontal radius size.
21887 Range is @code{[1, 1024]} and default value is @code{1}.
21890 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21893 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21896 @subsection Example
21900 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.
21902 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21906 @section boxblur_opencl
21908 Apply a boxblur algorithm to the input video.
21910 It accepts the following parameters:
21914 @item luma_radius, lr
21915 @item luma_power, lp
21916 @item chroma_radius, cr
21917 @item chroma_power, cp
21918 @item alpha_radius, ar
21919 @item alpha_power, ap
21923 A description of the accepted options follows.
21926 @item luma_radius, lr
21927 @item chroma_radius, cr
21928 @item alpha_radius, ar
21929 Set an expression for the box radius in pixels used for blurring the
21930 corresponding input plane.
21932 The radius value must be a non-negative number, and must not be
21933 greater than the value of the expression @code{min(w,h)/2} for the
21934 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21937 Default value for @option{luma_radius} is "2". If not specified,
21938 @option{chroma_radius} and @option{alpha_radius} default to the
21939 corresponding value set for @option{luma_radius}.
21941 The expressions can contain the following constants:
21945 The input width and height in pixels.
21949 The input chroma image width and height in pixels.
21953 The horizontal and vertical chroma subsample values. For example, for the
21954 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21957 @item luma_power, lp
21958 @item chroma_power, cp
21959 @item alpha_power, ap
21960 Specify how many times the boxblur filter is applied to the
21961 corresponding plane.
21963 Default value for @option{luma_power} is 2. If not specified,
21964 @option{chroma_power} and @option{alpha_power} default to the
21965 corresponding value set for @option{luma_power}.
21967 A value of 0 will disable the effect.
21970 @subsection Examples
21972 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.
21976 Apply a boxblur filter with the luma, chroma, and alpha radius
21977 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.
21979 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21980 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21984 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.
21986 For the luma plane, a 2x2 box radius will be run once.
21988 For the chroma plane, a 4x4 box radius will be run 5 times.
21990 For the alpha plane, a 3x3 box radius will be run 7 times.
21992 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21996 @section colorkey_opencl
21997 RGB colorspace color keying.
21999 The filter accepts the following options:
22003 The color which will be replaced with transparency.
22006 Similarity percentage with the key color.
22008 0.01 matches only the exact key color, while 1.0 matches everything.
22013 0.0 makes pixels either fully transparent, or not transparent at all.
22015 Higher values result in semi-transparent pixels, with a higher transparency
22016 the more similar the pixels color is to the key color.
22019 @subsection Examples
22023 Make every semi-green pixel in the input transparent with some slight blending:
22025 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22029 @section convolution_opencl
22031 Apply convolution of 3x3, 5x5, 7x7 matrix.
22033 The filter accepts the following options:
22040 Set matrix for each plane.
22041 Matrix is sequence of 9, 25 or 49 signed numbers.
22042 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
22048 Set multiplier for calculated value for each plane.
22049 If unset or 0, it will be sum of all matrix elements.
22050 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
22056 Set bias for each plane. This value is added to the result of the multiplication.
22057 Useful for making the overall image brighter or darker.
22058 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
22062 @subsection Examples
22068 -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
22074 -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
22078 Apply edge enhance:
22080 -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
22086 -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
22090 Apply laplacian edge detector which includes diagonals:
22092 -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
22098 -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
22102 @section erosion_opencl
22104 Apply erosion effect to the video.
22106 This filter replaces the pixel by the local(3x3) minimum.
22108 It accepts the following options:
22115 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22116 If @code{0}, plane will remain unchanged.
22119 Flag which specifies the pixel to refer to.
22120 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22122 Flags to local 3x3 coordinates region centered on @code{x}:
22131 @subsection Example
22135 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.
22137 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22141 @section deshake_opencl
22142 Feature-point based video stabilization filter.
22144 The filter accepts the following options:
22148 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
22151 Whether or not additional debug info should be displayed, both in the processed output and in the console.
22153 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
22155 Viewing point matches in the output video is only supported for RGB input.
22157 Defaults to @code{0}.
22159 @item adaptive_crop
22160 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
22162 Defaults to @code{1}.
22164 @item refine_features
22165 Whether or not feature points should be refined at a sub-pixel level.
22167 This can be turned off for a slight performance gain at the cost of precision.
22169 Defaults to @code{1}.
22171 @item smooth_strength
22172 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
22174 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
22176 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
22178 Defaults to @code{0.0}.
22180 @item smooth_window_multiplier
22181 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
22183 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
22185 Acceptable values range from @code{0.1} to @code{10.0}.
22187 Larger values increase the amount of motion data available for determining how to smooth the camera path,
22188 potentially improving smoothness, but also increase latency and memory usage.
22190 Defaults to @code{2.0}.
22194 @subsection Examples
22198 Stabilize a video with a fixed, medium smoothing strength:
22200 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22204 Stabilize a video with debugging (both in console and in rendered video):
22206 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22210 @section dilation_opencl
22212 Apply dilation effect to the video.
22214 This filter replaces the pixel by the local(3x3) maximum.
22216 It accepts the following options:
22223 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22224 If @code{0}, plane will remain unchanged.
22227 Flag which specifies the pixel to refer to.
22228 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22230 Flags to local 3x3 coordinates region centered on @code{x}:
22239 @subsection Example
22243 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.
22245 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22249 @section nlmeans_opencl
22251 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
22253 @section overlay_opencl
22255 Overlay one video on top of another.
22257 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
22258 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
22260 The filter accepts the following options:
22265 Set the x coordinate of the overlaid video on the main video.
22266 Default value is @code{0}.
22269 Set the y coordinate of the overlaid video on the main video.
22270 Default value is @code{0}.
22274 @subsection Examples
22278 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
22280 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22283 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
22285 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22290 @section pad_opencl
22292 Add paddings to the input image, and place the original input at the
22293 provided @var{x}, @var{y} coordinates.
22295 It accepts the following options:
22300 Specify an expression for the size of the output image with the
22301 paddings added. If the value for @var{width} or @var{height} is 0, the
22302 corresponding input size is used for the output.
22304 The @var{width} expression can reference the value set by the
22305 @var{height} expression, and vice versa.
22307 The default value of @var{width} and @var{height} is 0.
22311 Specify the offsets to place the input image at within the padded area,
22312 with respect to the top/left border of the output image.
22314 The @var{x} expression can reference the value set by the @var{y}
22315 expression, and vice versa.
22317 The default value of @var{x} and @var{y} is 0.
22319 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22320 so the input image is centered on the padded area.
22323 Specify the color of the padded area. For the syntax of this option,
22324 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22325 manual,ffmpeg-utils}.
22328 Pad to an aspect instead to a resolution.
22331 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22332 options are expressions containing the following constants:
22337 The input video width and height.
22341 These are the same as @var{in_w} and @var{in_h}.
22345 The output width and height (the size of the padded area), as
22346 specified by the @var{width} and @var{height} expressions.
22350 These are the same as @var{out_w} and @var{out_h}.
22354 The x and y offsets as specified by the @var{x} and @var{y}
22355 expressions, or NAN if not yet specified.
22358 same as @var{iw} / @var{ih}
22361 input sample aspect ratio
22364 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22367 @section prewitt_opencl
22369 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22371 The filter accepts the following option:
22375 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22378 Set value which will be multiplied with filtered result.
22379 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22382 Set value which will be added to filtered result.
22383 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22386 @subsection Example
22390 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22392 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22396 @anchor{program_opencl}
22397 @section program_opencl
22399 Filter video using an OpenCL program.
22404 OpenCL program source file.
22407 Kernel name in program.
22410 Number of inputs to the filter. Defaults to 1.
22413 Size of output frames. Defaults to the same as the first input.
22417 The @code{program_opencl} filter also supports the @ref{framesync} options.
22419 The program source file must contain a kernel function with the given name,
22420 which will be run once for each plane of the output. Each run on a plane
22421 gets enqueued as a separate 2D global NDRange with one work-item for each
22422 pixel to be generated. The global ID offset for each work-item is therefore
22423 the coordinates of a pixel in the destination image.
22425 The kernel function needs to take the following arguments:
22428 Destination image, @var{__write_only image2d_t}.
22430 This image will become the output; the kernel should write all of it.
22432 Frame index, @var{unsigned int}.
22434 This is a counter starting from zero and increasing by one for each frame.
22436 Source images, @var{__read_only image2d_t}.
22438 These are the most recent images on each input. The kernel may read from
22439 them to generate the output, but they can't be written to.
22446 Copy the input to the output (output must be the same size as the input).
22448 __kernel void copy(__write_only image2d_t destination,
22449 unsigned int index,
22450 __read_only image2d_t source)
22452 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22454 int2 location = (int2)(get_global_id(0), get_global_id(1));
22456 float4 value = read_imagef(source, sampler, location);
22458 write_imagef(destination, location, value);
22463 Apply a simple transformation, rotating the input by an amount increasing
22464 with the index counter. Pixel values are linearly interpolated by the
22465 sampler, and the output need not have the same dimensions as the input.
22467 __kernel void rotate_image(__write_only image2d_t dst,
22468 unsigned int index,
22469 __read_only image2d_t src)
22471 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22472 CLK_FILTER_LINEAR);
22474 float angle = (float)index / 100.0f;
22476 float2 dst_dim = convert_float2(get_image_dim(dst));
22477 float2 src_dim = convert_float2(get_image_dim(src));
22479 float2 dst_cen = dst_dim / 2.0f;
22480 float2 src_cen = src_dim / 2.0f;
22482 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22484 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22486 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22487 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22489 src_pos = src_pos * src_dim / dst_dim;
22491 float2 src_loc = src_pos + src_cen;
22493 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22494 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22495 write_imagef(dst, dst_loc, 0.5f);
22497 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22502 Blend two inputs together, with the amount of each input used varying
22503 with the index counter.
22505 __kernel void blend_images(__write_only image2d_t dst,
22506 unsigned int index,
22507 __read_only image2d_t src1,
22508 __read_only image2d_t src2)
22510 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22511 CLK_FILTER_LINEAR);
22513 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22515 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22516 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22517 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22519 float4 val1 = read_imagef(src1, sampler, src1_loc);
22520 float4 val2 = read_imagef(src2, sampler, src2_loc);
22522 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22528 @section roberts_opencl
22529 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22531 The filter accepts the following option:
22535 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22538 Set value which will be multiplied with filtered result.
22539 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22542 Set value which will be added to filtered result.
22543 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22546 @subsection Example
22550 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22552 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22556 @section sobel_opencl
22558 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22560 The filter accepts the following option:
22564 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22567 Set value which will be multiplied with filtered result.
22568 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22571 Set value which will be added to filtered result.
22572 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22575 @subsection Example
22579 Apply sobel operator with scale set to 2 and delta set to 10
22581 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22585 @section tonemap_opencl
22587 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22589 It accepts the following parameters:
22593 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22596 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22599 Apply desaturation for highlights that exceed this level of brightness. The
22600 higher the parameter, the more color information will be preserved. This
22601 setting helps prevent unnaturally blown-out colors for super-highlights, by
22602 (smoothly) turning into white instead. This makes images feel more natural,
22603 at the cost of reducing information about out-of-range colors.
22605 The default value is 0.5, and the algorithm here is a little different from
22606 the cpu version tonemap currently. A setting of 0.0 disables this option.
22609 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22610 is used to detect whether the scene has changed or not. If the distance between
22611 the current frame average brightness and the current running average exceeds
22612 a threshold value, we would re-calculate scene average and peak brightness.
22613 The default value is 0.2.
22616 Specify the output pixel format.
22618 Currently supported formats are:
22625 Set the output color range.
22627 Possible values are:
22633 Default is same as input.
22636 Set the output color primaries.
22638 Possible values are:
22644 Default is same as input.
22647 Set the output transfer characteristics.
22649 Possible values are:
22658 Set the output colorspace matrix.
22660 Possible value are:
22666 Default is same as input.
22670 @subsection Example
22674 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22676 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22680 @section unsharp_opencl
22682 Sharpen or blur the input video.
22684 It accepts the following parameters:
22687 @item luma_msize_x, lx
22688 Set the luma matrix horizontal size.
22689 Range is @code{[1, 23]} and default value is @code{5}.
22691 @item luma_msize_y, ly
22692 Set the luma matrix vertical size.
22693 Range is @code{[1, 23]} and default value is @code{5}.
22695 @item luma_amount, la
22696 Set the luma effect strength.
22697 Range is @code{[-10, 10]} and default value is @code{1.0}.
22699 Negative values will blur the input video, while positive values will
22700 sharpen it, a value of zero will disable the effect.
22702 @item chroma_msize_x, cx
22703 Set the chroma matrix horizontal size.
22704 Range is @code{[1, 23]} and default value is @code{5}.
22706 @item chroma_msize_y, cy
22707 Set the chroma matrix vertical size.
22708 Range is @code{[1, 23]} and default value is @code{5}.
22710 @item chroma_amount, ca
22711 Set the chroma effect strength.
22712 Range is @code{[-10, 10]} and default value is @code{0.0}.
22714 Negative values will blur the input video, while positive values will
22715 sharpen it, a value of zero will disable the effect.
22719 All parameters are optional and default to the equivalent of the
22720 string '5:5:1.0:5:5:0.0'.
22722 @subsection Examples
22726 Apply strong luma sharpen effect:
22728 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22732 Apply a strong blur of both luma and chroma parameters:
22734 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22738 @section xfade_opencl
22740 Cross fade two videos with custom transition effect by using OpenCL.
22742 It accepts the following options:
22746 Set one of possible transition effects.
22750 Select custom transition effect, the actual transition description
22751 will be picked from source and kernel options.
22763 Default transition is fade.
22767 OpenCL program source file for custom transition.
22770 Set name of kernel to use for custom transition from program source file.
22773 Set duration of video transition.
22776 Set time of start of transition relative to first video.
22779 The program source file must contain a kernel function with the given name,
22780 which will be run once for each plane of the output. Each run on a plane
22781 gets enqueued as a separate 2D global NDRange with one work-item for each
22782 pixel to be generated. The global ID offset for each work-item is therefore
22783 the coordinates of a pixel in the destination image.
22785 The kernel function needs to take the following arguments:
22788 Destination image, @var{__write_only image2d_t}.
22790 This image will become the output; the kernel should write all of it.
22793 First Source image, @var{__read_only image2d_t}.
22794 Second Source image, @var{__read_only image2d_t}.
22796 These are the most recent images on each input. The kernel may read from
22797 them to generate the output, but they can't be written to.
22800 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22807 Apply dots curtain transition effect:
22809 __kernel void blend_images(__write_only image2d_t dst,
22810 __read_only image2d_t src1,
22811 __read_only image2d_t src2,
22814 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22815 CLK_FILTER_LINEAR);
22816 int2 p = (int2)(get_global_id(0), get_global_id(1));
22817 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22818 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22821 float2 dots = (float2)(20.0, 20.0);
22822 float2 center = (float2)(0,0);
22825 float4 val1 = read_imagef(src1, sampler, p);
22826 float4 val2 = read_imagef(src2, sampler, p);
22827 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22829 write_imagef(dst, p, next ? val1 : val2);
22835 @c man end OPENCL VIDEO FILTERS
22837 @chapter VAAPI Video Filters
22838 @c man begin VAAPI VIDEO FILTERS
22840 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22842 To enable compilation of these filters you need to configure FFmpeg with
22843 @code{--enable-vaapi}.
22845 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}
22847 @section tonemap_vaapi
22849 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22850 It maps the dynamic range of HDR10 content to the SDR content.
22851 It currently only accepts HDR10 as input.
22853 It accepts the following parameters:
22857 Specify the output pixel format.
22859 Currently supported formats are:
22868 Set the output color primaries.
22870 Default is same as input.
22873 Set the output transfer characteristics.
22878 Set the output colorspace matrix.
22880 Default is same as input.
22884 @subsection Example
22888 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22890 tonemap_vaapi=format=p010:t=bt2020-10
22894 @c man end VAAPI VIDEO FILTERS
22896 @chapter Video Sources
22897 @c man begin VIDEO SOURCES
22899 Below is a description of the currently available video sources.
22903 Buffer video frames, and make them available to the filter chain.
22905 This source is mainly intended for a programmatic use, in particular
22906 through the interface defined in @file{libavfilter/buffersrc.h}.
22908 It accepts the following parameters:
22913 Specify the size (width and height) of the buffered video frames. For the
22914 syntax of this option, check the
22915 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22918 The input video width.
22921 The input video height.
22924 A string representing the pixel format of the buffered video frames.
22925 It may be a number corresponding to a pixel format, or a pixel format
22929 Specify the timebase assumed by the timestamps of the buffered frames.
22932 Specify the frame rate expected for the video stream.
22934 @item pixel_aspect, sar
22935 The sample (pixel) aspect ratio of the input video.
22938 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22939 to the filtergraph description to specify swscale flags for automatically
22940 inserted scalers. See @ref{Filtergraph syntax}.
22942 @item hw_frames_ctx
22943 When using a hardware pixel format, this should be a reference to an
22944 AVHWFramesContext describing input frames.
22949 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22952 will instruct the source to accept video frames with size 320x240 and
22953 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22954 square pixels (1:1 sample aspect ratio).
22955 Since the pixel format with name "yuv410p" corresponds to the number 6
22956 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22957 this example corresponds to:
22959 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22962 Alternatively, the options can be specified as a flat string, but this
22963 syntax is deprecated:
22965 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22969 Create a pattern generated by an elementary cellular automaton.
22971 The initial state of the cellular automaton can be defined through the
22972 @option{filename} and @option{pattern} options. If such options are
22973 not specified an initial state is created randomly.
22975 At each new frame a new row in the video is filled with the result of
22976 the cellular automaton next generation. The behavior when the whole
22977 frame is filled is defined by the @option{scroll} option.
22979 This source accepts the following options:
22983 Read the initial cellular automaton state, i.e. the starting row, from
22984 the specified file.
22985 In the file, each non-whitespace character is considered an alive
22986 cell, a newline will terminate the row, and further characters in the
22987 file will be ignored.
22990 Read the initial cellular automaton state, i.e. the starting row, from
22991 the specified string.
22993 Each non-whitespace character in the string is considered an alive
22994 cell, a newline will terminate the row, and further characters in the
22995 string will be ignored.
22998 Set the video rate, that is the number of frames generated per second.
23001 @item random_fill_ratio, ratio
23002 Set the random fill ratio for the initial cellular automaton row. It
23003 is a floating point number value ranging from 0 to 1, defaults to
23006 This option is ignored when a file or a pattern is specified.
23008 @item random_seed, seed
23009 Set the seed for filling randomly the initial row, must be an integer
23010 included between 0 and UINT32_MAX. If not specified, or if explicitly
23011 set to -1, the filter will try to use a good random seed on a best
23015 Set the cellular automaton rule, it is a number ranging from 0 to 255.
23016 Default value is 110.
23019 Set the size of the output video. For the syntax of this option, check the
23020 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23022 If @option{filename} or @option{pattern} is specified, the size is set
23023 by default to the width of the specified initial state row, and the
23024 height is set to @var{width} * PHI.
23026 If @option{size} is set, it must contain the width of the specified
23027 pattern string, and the specified pattern will be centered in the
23030 If a filename or a pattern string is not specified, the size value
23031 defaults to "320x518" (used for a randomly generated initial state).
23034 If set to 1, scroll the output upward when all the rows in the output
23035 have been already filled. If set to 0, the new generated row will be
23036 written over the top row just after the bottom row is filled.
23039 @item start_full, full
23040 If set to 1, completely fill the output with generated rows before
23041 outputting the first frame.
23042 This is the default behavior, for disabling set the value to 0.
23045 If set to 1, stitch the left and right row edges together.
23046 This is the default behavior, for disabling set the value to 0.
23049 @subsection Examples
23053 Read the initial state from @file{pattern}, and specify an output of
23056 cellauto=f=pattern:s=200x400
23060 Generate a random initial row with a width of 200 cells, with a fill
23063 cellauto=ratio=2/3:s=200x200
23067 Create a pattern generated by rule 18 starting by a single alive cell
23068 centered on an initial row with width 100:
23070 cellauto=p=@@:s=100x400:full=0:rule=18
23074 Specify a more elaborated initial pattern:
23076 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
23081 @anchor{coreimagesrc}
23082 @section coreimagesrc
23083 Video source generated on GPU using Apple's CoreImage API on OSX.
23085 This video source is a specialized version of the @ref{coreimage} video filter.
23086 Use a core image generator at the beginning of the applied filterchain to
23087 generate the content.
23089 The coreimagesrc video source accepts the following options:
23091 @item list_generators
23092 List all available generators along with all their respective options as well as
23093 possible minimum and maximum values along with the default values.
23095 list_generators=true
23099 Specify the size of the sourced video. For the syntax of this option, check the
23100 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23101 The default value is @code{320x240}.
23104 Specify the frame rate of the sourced video, as the number of frames
23105 generated per second. It has to be a string in the format
23106 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23107 number or a valid video frame rate abbreviation. The default value is
23111 Set the sample aspect ratio of the sourced video.
23114 Set the duration of the sourced video. See
23115 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23116 for the accepted syntax.
23118 If not specified, or the expressed duration is negative, the video is
23119 supposed to be generated forever.
23122 Additionally, all options of the @ref{coreimage} video filter are accepted.
23123 A complete filterchain can be used for further processing of the
23124 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
23125 and examples for details.
23127 @subsection Examples
23132 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23133 given as complete and escaped command-line for Apple's standard bash shell:
23135 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
23137 This example is equivalent to the QRCode example of @ref{coreimage} without the
23138 need for a nullsrc video source.
23143 Generate several gradients.
23147 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23148 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23151 Set frame rate, expressed as number of frames per second. Default
23154 @item c0, c1, c2, c3, c4, c5, c6, c7
23155 Set 8 colors. Default values for colors is to pick random one.
23157 @item x0, y0, y0, y1
23158 Set gradient line source and destination points. If negative or out of range, random ones
23162 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
23165 Set seed for picking gradient line points.
23168 Set the duration of the sourced video. See
23169 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23170 for the accepted syntax.
23172 If not specified, or the expressed duration is negative, the video is
23173 supposed to be generated forever.
23176 Set speed of gradients rotation.
23180 @section mandelbrot
23182 Generate a Mandelbrot set fractal, and progressively zoom towards the
23183 point specified with @var{start_x} and @var{start_y}.
23185 This source accepts the following options:
23190 Set the terminal pts value. Default value is 400.
23193 Set the terminal scale value.
23194 Must be a floating point value. Default value is 0.3.
23197 Set the inner coloring mode, that is the algorithm used to draw the
23198 Mandelbrot fractal internal region.
23200 It shall assume one of the following values:
23205 Show time until convergence.
23207 Set color based on point closest to the origin of the iterations.
23212 Default value is @var{mincol}.
23215 Set the bailout value. Default value is 10.0.
23218 Set the maximum of iterations performed by the rendering
23219 algorithm. Default value is 7189.
23222 Set outer coloring mode.
23223 It shall assume one of following values:
23225 @item iteration_count
23226 Set iteration count mode.
23227 @item normalized_iteration_count
23228 set normalized iteration count mode.
23230 Default value is @var{normalized_iteration_count}.
23233 Set frame rate, expressed as number of frames per second. Default
23237 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23238 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23241 Set the initial scale value. Default value is 3.0.
23244 Set the initial x position. Must be a floating point value between
23245 -100 and 100. Default value is -0.743643887037158704752191506114774.
23248 Set the initial y position. Must be a floating point value between
23249 -100 and 100. Default value is -0.131825904205311970493132056385139.
23254 Generate various test patterns, as generated by the MPlayer test filter.
23256 The size of the generated video is fixed, and is 256x256.
23257 This source is useful in particular for testing encoding features.
23259 This source accepts the following options:
23264 Specify the frame rate of the sourced video, as the number of frames
23265 generated per second. It has to be a string in the format
23266 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23267 number or a valid video frame rate abbreviation. The default value is
23271 Set the duration of the sourced video. See
23272 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23273 for the accepted syntax.
23275 If not specified, or the expressed duration is negative, the video is
23276 supposed to be generated forever.
23280 Set the number or the name of the test to perform. Supported tests are:
23294 @item max_frames, m
23295 Set the maximum number of frames generated for each test, default value is 30.
23299 Default value is "all", which will cycle through the list of all tests.
23304 mptestsrc=t=dc_luma
23307 will generate a "dc_luma" test pattern.
23309 @section frei0r_src
23311 Provide a frei0r source.
23313 To enable compilation of this filter you need to install the frei0r
23314 header and configure FFmpeg with @code{--enable-frei0r}.
23316 This source accepts the following parameters:
23321 The size of the video to generate. For the syntax of this option, check the
23322 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23325 The framerate of the generated video. It may be a string of the form
23326 @var{num}/@var{den} or a frame rate abbreviation.
23329 The name to the frei0r source to load. For more information regarding frei0r and
23330 how to set the parameters, read the @ref{frei0r} section in the video filters
23333 @item filter_params
23334 A '|'-separated list of parameters to pass to the frei0r source.
23338 For example, to generate a frei0r partik0l source with size 200x200
23339 and frame rate 10 which is overlaid on the overlay filter main input:
23341 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23346 Generate a life pattern.
23348 This source is based on a generalization of John Conway's life game.
23350 The sourced input represents a life grid, each pixel represents a cell
23351 which can be in one of two possible states, alive or dead. Every cell
23352 interacts with its eight neighbours, which are the cells that are
23353 horizontally, vertically, or diagonally adjacent.
23355 At each interaction the grid evolves according to the adopted rule,
23356 which specifies the number of neighbor alive cells which will make a
23357 cell stay alive or born. The @option{rule} option allows one to specify
23360 This source accepts the following options:
23364 Set the file from which to read the initial grid state. In the file,
23365 each non-whitespace character is considered an alive cell, and newline
23366 is used to delimit the end of each row.
23368 If this option is not specified, the initial grid is generated
23372 Set the video rate, that is the number of frames generated per second.
23375 @item random_fill_ratio, ratio
23376 Set the random fill ratio for the initial random grid. It is a
23377 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23378 It is ignored when a file is specified.
23380 @item random_seed, seed
23381 Set the seed for filling the initial random grid, must be an integer
23382 included between 0 and UINT32_MAX. If not specified, or if explicitly
23383 set to -1, the filter will try to use a good random seed on a best
23389 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23390 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23391 @var{NS} specifies the number of alive neighbor cells which make a
23392 live cell stay alive, and @var{NB} the number of alive neighbor cells
23393 which make a dead cell to become alive (i.e. to "born").
23394 "s" and "b" can be used in place of "S" and "B", respectively.
23396 Alternatively a rule can be specified by an 18-bits integer. The 9
23397 high order bits are used to encode the next cell state if it is alive
23398 for each number of neighbor alive cells, the low order bits specify
23399 the rule for "borning" new cells. Higher order bits encode for an
23400 higher number of neighbor cells.
23401 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23402 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23404 Default value is "S23/B3", which is the original Conway's game of life
23405 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23406 cells, and will born a new cell if there are three alive cells around
23410 Set the size of the output video. For the syntax of this option, check the
23411 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23413 If @option{filename} is specified, the size is set by default to the
23414 same size of the input file. If @option{size} is set, it must contain
23415 the size specified in the input file, and the initial grid defined in
23416 that file is centered in the larger resulting area.
23418 If a filename is not specified, the size value defaults to "320x240"
23419 (used for a randomly generated initial grid).
23422 If set to 1, stitch the left and right grid edges together, and the
23423 top and bottom edges also. Defaults to 1.
23426 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23427 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23428 value from 0 to 255.
23431 Set the color of living (or new born) cells.
23434 Set the color of dead cells. If @option{mold} is set, this is the first color
23435 used to represent a dead cell.
23438 Set mold color, for definitely dead and moldy cells.
23440 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23441 ffmpeg-utils manual,ffmpeg-utils}.
23444 @subsection Examples
23448 Read a grid from @file{pattern}, and center it on a grid of size
23451 life=f=pattern:s=300x300
23455 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23457 life=ratio=2/3:s=200x200
23461 Specify a custom rule for evolving a randomly generated grid:
23467 Full example with slow death effect (mold) using @command{ffplay}:
23469 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23476 @anchor{haldclutsrc}
23479 @anchor{pal100bars}
23480 @anchor{rgbtestsrc}
23482 @anchor{smptehdbars}
23485 @anchor{yuvtestsrc}
23486 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23488 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23490 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23492 The @code{color} source provides an uniformly colored input.
23494 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23495 @ref{haldclut} filter.
23497 The @code{nullsrc} source returns unprocessed video frames. It is
23498 mainly useful to be employed in analysis / debugging tools, or as the
23499 source for filters which ignore the input data.
23501 The @code{pal75bars} source generates a color bars pattern, based on
23502 EBU PAL recommendations with 75% color levels.
23504 The @code{pal100bars} source generates a color bars pattern, based on
23505 EBU PAL recommendations with 100% color levels.
23507 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23508 detecting RGB vs BGR issues. You should see a red, green and blue
23509 stripe from top to bottom.
23511 The @code{smptebars} source generates a color bars pattern, based on
23512 the SMPTE Engineering Guideline EG 1-1990.
23514 The @code{smptehdbars} source generates a color bars pattern, based on
23515 the SMPTE RP 219-2002.
23517 The @code{testsrc} source generates a test video pattern, showing a
23518 color pattern, a scrolling gradient and a timestamp. This is mainly
23519 intended for testing purposes.
23521 The @code{testsrc2} source is similar to testsrc, but supports more
23522 pixel formats instead of just @code{rgb24}. This allows using it as an
23523 input for other tests without requiring a format conversion.
23525 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23526 see a y, cb and cr stripe from top to bottom.
23528 The sources accept the following parameters:
23533 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23534 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23535 pixels to be used as identity matrix for 3D lookup tables. Each component is
23536 coded on a @code{1/(N*N)} scale.
23539 Specify the color of the source, only available in the @code{color}
23540 source. For the syntax of this option, check the
23541 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23544 Specify the size of the sourced video. For the syntax of this option, check the
23545 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23546 The default value is @code{320x240}.
23548 This option is not available with the @code{allrgb}, @code{allyuv}, and
23549 @code{haldclutsrc} filters.
23552 Specify the frame rate of the sourced video, as the number of frames
23553 generated per second. It has to be a string in the format
23554 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23555 number or a valid video frame rate abbreviation. The default value is
23559 Set the duration of the sourced video. See
23560 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23561 for the accepted syntax.
23563 If not specified, or the expressed duration is negative, the video is
23564 supposed to be generated forever.
23566 Since the frame rate is used as time base, all frames including the last one
23567 will have their full duration. If the specified duration is not a multiple
23568 of the frame duration, it will be rounded up.
23571 Set the sample aspect ratio of the sourced video.
23574 Specify the alpha (opacity) of the background, only available in the
23575 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23576 255 (fully opaque, the default).
23579 Set the number of decimals to show in the timestamp, only available in the
23580 @code{testsrc} source.
23582 The displayed timestamp value will correspond to the original
23583 timestamp value multiplied by the power of 10 of the specified
23584 value. Default value is 0.
23587 @subsection Examples
23591 Generate a video with a duration of 5.3 seconds, with size
23592 176x144 and a frame rate of 10 frames per second:
23594 testsrc=duration=5.3:size=qcif:rate=10
23598 The following graph description will generate a red source
23599 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23602 color=c=red@@0.2:s=qcif:r=10
23606 If the input content is to be ignored, @code{nullsrc} can be used. The
23607 following command generates noise in the luminance plane by employing
23608 the @code{geq} filter:
23610 nullsrc=s=256x256, geq=random(1)*255:128:128
23614 @subsection Commands
23616 The @code{color} source supports the following commands:
23620 Set the color of the created image. Accepts the same syntax of the
23621 corresponding @option{color} option.
23626 Generate video using an OpenCL program.
23631 OpenCL program source file.
23634 Kernel name in program.
23637 Size of frames to generate. This must be set.
23640 Pixel format to use for the generated frames. This must be set.
23643 Number of frames generated every second. Default value is '25'.
23647 For details of how the program loading works, see the @ref{program_opencl}
23654 Generate a colour ramp by setting pixel values from the position of the pixel
23655 in the output image. (Note that this will work with all pixel formats, but
23656 the generated output will not be the same.)
23658 __kernel void ramp(__write_only image2d_t dst,
23659 unsigned int index)
23661 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23664 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23666 write_imagef(dst, loc, val);
23671 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23673 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23674 unsigned int index)
23676 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23678 float4 value = 0.0f;
23679 int x = loc.x + index;
23680 int y = loc.y + index;
23681 while (x > 0 || y > 0) {
23682 if (x % 3 == 1 && y % 3 == 1) {
23690 write_imagef(dst, loc, value);
23696 @section sierpinski
23698 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23700 This source accepts the following options:
23704 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23705 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23708 Set frame rate, expressed as number of frames per second. Default
23712 Set seed which is used for random panning.
23715 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23718 Set fractal type, can be default @code{carpet} or @code{triangle}.
23721 @c man end VIDEO SOURCES
23723 @chapter Video Sinks
23724 @c man begin VIDEO SINKS
23726 Below is a description of the currently available video sinks.
23728 @section buffersink
23730 Buffer video frames, and make them available to the end of the filter
23733 This sink is mainly intended for programmatic use, in particular
23734 through the interface defined in @file{libavfilter/buffersink.h}
23735 or the options system.
23737 It accepts a pointer to an AVBufferSinkContext structure, which
23738 defines the incoming buffers' formats, to be passed as the opaque
23739 parameter to @code{avfilter_init_filter} for initialization.
23743 Null video sink: do absolutely nothing with the input video. It is
23744 mainly useful as a template and for use in analysis / debugging
23747 @c man end VIDEO SINKS
23749 @chapter Multimedia Filters
23750 @c man begin MULTIMEDIA FILTERS
23752 Below is a description of the currently available multimedia filters.
23756 Convert input audio to a video output, displaying the audio bit scope.
23758 The filter accepts the following options:
23762 Set frame rate, expressed as number of frames per second. Default
23766 Specify the video size for the output. For the syntax of this option, check the
23767 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23768 Default value is @code{1024x256}.
23771 Specify list of colors separated by space or by '|' which will be used to
23772 draw channels. Unrecognized or missing colors will be replaced
23776 @section adrawgraph
23777 Draw a graph using input audio metadata.
23779 See @ref{drawgraph}
23781 @section agraphmonitor
23783 See @ref{graphmonitor}.
23785 @section ahistogram
23787 Convert input audio to a video output, displaying the volume histogram.
23789 The filter accepts the following options:
23793 Specify how histogram is calculated.
23795 It accepts the following values:
23798 Use single histogram for all channels.
23800 Use separate histogram for each channel.
23802 Default is @code{single}.
23805 Set frame rate, expressed as number of frames per second. Default
23809 Specify the video size for the output. For the syntax of this option, check the
23810 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23811 Default value is @code{hd720}.
23816 It accepts the following values:
23827 reverse logarithmic
23829 Default is @code{log}.
23832 Set amplitude scale.
23834 It accepts the following values:
23841 Default is @code{log}.
23844 Set how much frames to accumulate in histogram.
23845 Default is 1. Setting this to -1 accumulates all frames.
23848 Set histogram ratio of window height.
23851 Set sonogram sliding.
23853 It accepts the following values:
23856 replace old rows with new ones.
23858 scroll from top to bottom.
23860 Default is @code{replace}.
23863 @section aphasemeter
23865 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23866 representing mean phase of current audio frame. A video output can also be produced and is
23867 enabled by default. The audio is passed through as first output.
23869 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23870 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23871 and @code{1} means channels are in phase.
23873 The filter accepts the following options, all related to its video output:
23877 Set the output frame rate. Default value is @code{25}.
23880 Set the video size for the output. For the syntax of this option, check the
23881 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23882 Default value is @code{800x400}.
23887 Specify the red, green, blue contrast. Default values are @code{2},
23888 @code{7} and @code{1}.
23889 Allowed range is @code{[0, 255]}.
23892 Set color which will be used for drawing median phase. If color is
23893 @code{none} which is default, no median phase value will be drawn.
23896 Enable video output. Default is enabled.
23899 @subsection phasing detection
23901 The filter also detects out of phase and mono sequences in stereo streams.
23902 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23904 The filter accepts the following options for this detection:
23908 Enable mono and out of phase detection. Default is disabled.
23911 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23912 Allowed range is @code{[0, 1]}.
23915 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23916 Allowed range is @code{[90, 180]}.
23919 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23922 @subsection Examples
23926 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23928 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23932 @section avectorscope
23934 Convert input audio to a video output, representing the audio vector
23937 The filter is used to measure the difference between channels of stereo
23938 audio stream. A monaural signal, consisting of identical left and right
23939 signal, results in straight vertical line. Any stereo separation is visible
23940 as a deviation from this line, creating a Lissajous figure.
23941 If the straight (or deviation from it) but horizontal line appears this
23942 indicates that the left and right channels are out of phase.
23944 The filter accepts the following options:
23948 Set the vectorscope mode.
23950 Available values are:
23953 Lissajous rotated by 45 degrees.
23956 Same as above but not rotated.
23959 Shape resembling half of circle.
23962 Default value is @samp{lissajous}.
23965 Set the video size for the output. For the syntax of this option, check the
23966 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23967 Default value is @code{400x400}.
23970 Set the output frame rate. Default value is @code{25}.
23976 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23977 @code{160}, @code{80} and @code{255}.
23978 Allowed range is @code{[0, 255]}.
23984 Specify the red, green, blue and alpha fade. Default values are @code{15},
23985 @code{10}, @code{5} and @code{5}.
23986 Allowed range is @code{[0, 255]}.
23989 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23990 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23993 Set the vectorscope drawing mode.
23995 Available values are:
23998 Draw dot for each sample.
24001 Draw line between previous and current sample.
24004 Default value is @samp{dot}.
24007 Specify amplitude scale of audio samples.
24009 Available values are:
24025 Swap left channel axis with right channel axis.
24035 Mirror only x axis.
24038 Mirror only y axis.
24046 @subsection Examples
24050 Complete example using @command{ffplay}:
24052 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24053 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24057 @section bench, abench
24059 Benchmark part of a filtergraph.
24061 The filter accepts the following options:
24065 Start or stop a timer.
24067 Available values are:
24070 Get the current time, set it as frame metadata (using the key
24071 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
24074 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
24075 the input frame metadata to get the time difference. Time difference, average,
24076 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
24077 @code{min}) are then printed. The timestamps are expressed in seconds.
24081 @subsection Examples
24085 Benchmark @ref{selectivecolor} filter:
24087 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24093 Concatenate audio and video streams, joining them together one after the
24096 The filter works on segments of synchronized video and audio streams. All
24097 segments must have the same number of streams of each type, and that will
24098 also be the number of streams at output.
24100 The filter accepts the following options:
24105 Set the number of segments. Default is 2.
24108 Set the number of output video streams, that is also the number of video
24109 streams in each segment. Default is 1.
24112 Set the number of output audio streams, that is also the number of audio
24113 streams in each segment. Default is 0.
24116 Activate unsafe mode: do not fail if segments have a different format.
24120 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
24121 @var{a} audio outputs.
24123 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
24124 segment, in the same order as the outputs, then the inputs for the second
24127 Related streams do not always have exactly the same duration, for various
24128 reasons including codec frame size or sloppy authoring. For that reason,
24129 related synchronized streams (e.g. a video and its audio track) should be
24130 concatenated at once. The concat filter will use the duration of the longest
24131 stream in each segment (except the last one), and if necessary pad shorter
24132 audio streams with silence.
24134 For this filter to work correctly, all segments must start at timestamp 0.
24136 All corresponding streams must have the same parameters in all segments; the
24137 filtering system will automatically select a common pixel format for video
24138 streams, and a common sample format, sample rate and channel layout for
24139 audio streams, but other settings, such as resolution, must be converted
24140 explicitly by the user.
24142 Different frame rates are acceptable but will result in variable frame rate
24143 at output; be sure to configure the output file to handle it.
24145 @subsection Examples
24149 Concatenate an opening, an episode and an ending, all in bilingual version
24150 (video in stream 0, audio in streams 1 and 2):
24152 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24153 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24154 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24155 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24159 Concatenate two parts, handling audio and video separately, using the
24160 (a)movie sources, and adjusting the resolution:
24162 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24163 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24164 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24166 Note that a desync will happen at the stitch if the audio and video streams
24167 do not have exactly the same duration in the first file.
24171 @subsection Commands
24173 This filter supports the following commands:
24176 Close the current segment and step to the next one
24182 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
24183 level. By default, it logs a message at a frequency of 10Hz with the
24184 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
24185 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
24187 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
24188 sample format is double-precision floating point. The input stream will be converted to
24189 this specification, if needed. Users may need to insert aformat and/or aresample filters
24190 after this filter to obtain the original parameters.
24192 The filter also has a video output (see the @var{video} option) with a real
24193 time graph to observe the loudness evolution. The graphic contains the logged
24194 message mentioned above, so it is not printed anymore when this option is set,
24195 unless the verbose logging is set. The main graphing area contains the
24196 short-term loudness (3 seconds of analysis), and the gauge on the right is for
24197 the momentary loudness (400 milliseconds), but can optionally be configured
24198 to instead display short-term loudness (see @var{gauge}).
24200 The green area marks a +/- 1LU target range around the target loudness
24201 (-23LUFS by default, unless modified through @var{target}).
24203 More information about the Loudness Recommendation EBU R128 on
24204 @url{http://tech.ebu.ch/loudness}.
24206 The filter accepts the following options:
24211 Activate the video output. The audio stream is passed unchanged whether this
24212 option is set or no. The video stream will be the first output stream if
24213 activated. Default is @code{0}.
24216 Set the video size. This option is for video only. For the syntax of this
24218 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24219 Default and minimum resolution is @code{640x480}.
24222 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
24223 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
24224 other integer value between this range is allowed.
24227 Set metadata injection. If set to @code{1}, the audio input will be segmented
24228 into 100ms output frames, each of them containing various loudness information
24229 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
24231 Default is @code{0}.
24234 Force the frame logging level.
24236 Available values are:
24239 information logging level
24241 verbose logging level
24244 By default, the logging level is set to @var{info}. If the @option{video} or
24245 the @option{metadata} options are set, it switches to @var{verbose}.
24250 Available modes can be cumulated (the option is a @code{flag} type). Possible
24254 Disable any peak mode (default).
24256 Enable sample-peak mode.
24258 Simple peak mode looking for the higher sample value. It logs a message
24259 for sample-peak (identified by @code{SPK}).
24261 Enable true-peak mode.
24263 If enabled, the peak lookup is done on an over-sampled version of the input
24264 stream for better peak accuracy. It logs a message for true-peak.
24265 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
24266 This mode requires a build with @code{libswresample}.
24270 Treat mono input files as "dual mono". If a mono file is intended for playback
24271 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
24272 If set to @code{true}, this option will compensate for this effect.
24273 Multi-channel input files are not affected by this option.
24276 Set a specific pan law to be used for the measurement of dual mono files.
24277 This parameter is optional, and has a default value of -3.01dB.
24280 Set a specific target level (in LUFS) used as relative zero in the visualization.
24281 This parameter is optional and has a default value of -23LUFS as specified
24282 by EBU R128. However, material published online may prefer a level of -16LUFS
24283 (e.g. for use with podcasts or video platforms).
24286 Set the value displayed by the gauge. Valid values are @code{momentary} and s
24287 @code{shortterm}. By default the momentary value will be used, but in certain
24288 scenarios it may be more useful to observe the short term value instead (e.g.
24292 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24293 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24294 video output, not the summary or continuous log output.
24297 @subsection Examples
24301 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24303 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24307 Run an analysis with @command{ffmpeg}:
24309 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24313 @section interleave, ainterleave
24315 Temporally interleave frames from several inputs.
24317 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24319 These filters read frames from several inputs and send the oldest
24320 queued frame to the output.
24322 Input streams must have well defined, monotonically increasing frame
24325 In order to submit one frame to output, these filters need to enqueue
24326 at least one frame for each input, so they cannot work in case one
24327 input is not yet terminated and will not receive incoming frames.
24329 For example consider the case when one input is a @code{select} filter
24330 which always drops input frames. The @code{interleave} filter will keep
24331 reading from that input, but it will never be able to send new frames
24332 to output until the input sends an end-of-stream signal.
24334 Also, depending on inputs synchronization, the filters will drop
24335 frames in case one input receives more frames than the other ones, and
24336 the queue is already filled.
24338 These filters accept the following options:
24342 Set the number of different inputs, it is 2 by default.
24345 How to determine the end-of-stream.
24349 The duration of the longest input. (default)
24352 The duration of the shortest input.
24355 The duration of the first input.
24360 @subsection Examples
24364 Interleave frames belonging to different streams using @command{ffmpeg}:
24366 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24370 Add flickering blur effect:
24372 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24376 @section metadata, ametadata
24378 Manipulate frame metadata.
24380 This filter accepts the following options:
24384 Set mode of operation of the filter.
24386 Can be one of the following:
24390 If both @code{value} and @code{key} is set, select frames
24391 which have such metadata. If only @code{key} is set, select
24392 every frame that has such key in metadata.
24395 Add new metadata @code{key} and @code{value}. If key is already available
24399 Modify value of already present key.
24402 If @code{value} is set, delete only keys that have such value.
24403 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24407 Print key and its value if metadata was found. If @code{key} is not set print all
24408 metadata values available in frame.
24412 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24415 Set metadata value which will be used. This option is mandatory for
24416 @code{modify} and @code{add} mode.
24419 Which function to use when comparing metadata value and @code{value}.
24421 Can be one of following:
24425 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24428 Values are interpreted as strings, returns true if metadata value starts with
24429 the @code{value} option string.
24432 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24435 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24438 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24441 Values are interpreted as floats, returns true if expression from option @code{expr}
24445 Values are interpreted as strings, returns true if metadata value ends with
24446 the @code{value} option string.
24450 Set expression which is used when @code{function} is set to @code{expr}.
24451 The expression is evaluated through the eval API and can contain the following
24456 Float representation of @code{value} from metadata key.
24459 Float representation of @code{value} as supplied by user in @code{value} option.
24463 If specified in @code{print} mode, output is written to the named file. Instead of
24464 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24465 for standard output. If @code{file} option is not set, output is written to the log
24466 with AV_LOG_INFO loglevel.
24469 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24473 @subsection Examples
24477 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24480 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24483 Print silencedetect output to file @file{metadata.txt}.
24485 silencedetect,ametadata=mode=print:file=metadata.txt
24488 Direct all metadata to a pipe with file descriptor 4.
24490 metadata=mode=print:file='pipe\:4'
24494 @section perms, aperms
24496 Set read/write permissions for the output frames.
24498 These filters are mainly aimed at developers to test direct path in the
24499 following filter in the filtergraph.
24501 The filters accept the following options:
24505 Select the permissions mode.
24507 It accepts the following values:
24510 Do nothing. This is the default.
24512 Set all the output frames read-only.
24514 Set all the output frames directly writable.
24516 Make the frame read-only if writable, and writable if read-only.
24518 Set each output frame read-only or writable randomly.
24522 Set the seed for the @var{random} mode, must be an integer included between
24523 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24524 @code{-1}, the filter will try to use a good random seed on a best effort
24528 Note: in case of auto-inserted filter between the permission filter and the
24529 following one, the permission might not be received as expected in that
24530 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24531 perms/aperms filter can avoid this problem.
24533 @section realtime, arealtime
24535 Slow down filtering to match real time approximately.
24537 These filters will pause the filtering for a variable amount of time to
24538 match the output rate with the input timestamps.
24539 They are similar to the @option{re} option to @code{ffmpeg}.
24541 They accept the following options:
24545 Time limit for the pauses. Any pause longer than that will be considered
24546 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24548 Speed factor for processing. The value must be a float larger than zero.
24549 Values larger than 1.0 will result in faster than realtime processing,
24550 smaller will slow processing down. The @var{limit} is automatically adapted
24551 accordingly. Default is 1.0.
24553 A processing speed faster than what is possible without these filters cannot
24558 @section select, aselect
24560 Select frames to pass in output.
24562 This filter accepts the following options:
24567 Set expression, which is evaluated for each input frame.
24569 If the expression is evaluated to zero, the frame is discarded.
24571 If the evaluation result is negative or NaN, the frame is sent to the
24572 first output; otherwise it is sent to the output with index
24573 @code{ceil(val)-1}, assuming that the input index starts from 0.
24575 For example a value of @code{1.2} corresponds to the output with index
24576 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24579 Set the number of outputs. The output to which to send the selected
24580 frame is based on the result of the evaluation. Default value is 1.
24583 The expression can contain the following constants:
24587 The (sequential) number of the filtered frame, starting from 0.
24590 The (sequential) number of the selected frame, starting from 0.
24592 @item prev_selected_n
24593 The sequential number of the last selected frame. It's NAN if undefined.
24596 The timebase of the input timestamps.
24599 The PTS (Presentation TimeStamp) of the filtered video frame,
24600 expressed in @var{TB} units. It's NAN if undefined.
24603 The PTS of the filtered video frame,
24604 expressed in seconds. It's NAN if undefined.
24607 The PTS of the previously filtered video frame. It's NAN if undefined.
24609 @item prev_selected_pts
24610 The PTS of the last previously filtered video frame. It's NAN if undefined.
24612 @item prev_selected_t
24613 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24616 The PTS of the first video frame in the video. It's NAN if undefined.
24619 The time of the first video frame in the video. It's NAN if undefined.
24621 @item pict_type @emph{(video only)}
24622 The type of the filtered frame. It can assume one of the following
24634 @item interlace_type @emph{(video only)}
24635 The frame interlace type. It can assume one of the following values:
24638 The frame is progressive (not interlaced).
24640 The frame is top-field-first.
24642 The frame is bottom-field-first.
24645 @item consumed_sample_n @emph{(audio only)}
24646 the number of selected samples before the current frame
24648 @item samples_n @emph{(audio only)}
24649 the number of samples in the current frame
24651 @item sample_rate @emph{(audio only)}
24652 the input sample rate
24655 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24658 the position in the file of the filtered frame, -1 if the information
24659 is not available (e.g. for synthetic video)
24661 @item scene @emph{(video only)}
24662 value between 0 and 1 to indicate a new scene; a low value reflects a low
24663 probability for the current frame to introduce a new scene, while a higher
24664 value means the current frame is more likely to be one (see the example below)
24666 @item concatdec_select
24667 The concat demuxer can select only part of a concat input file by setting an
24668 inpoint and an outpoint, but the output packets may not be entirely contained
24669 in the selected interval. By using this variable, it is possible to skip frames
24670 generated by the concat demuxer which are not exactly contained in the selected
24673 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24674 and the @var{lavf.concat.duration} packet metadata values which are also
24675 present in the decoded frames.
24677 The @var{concatdec_select} variable is -1 if the frame pts is at least
24678 start_time and either the duration metadata is missing or the frame pts is less
24679 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24682 That basically means that an input frame is selected if its pts is within the
24683 interval set by the concat demuxer.
24687 The default value of the select expression is "1".
24689 @subsection Examples
24693 Select all frames in input:
24698 The example above is the same as:
24710 Select only I-frames:
24712 select='eq(pict_type\,I)'
24716 Select one frame every 100:
24718 select='not(mod(n\,100))'
24722 Select only frames contained in the 10-20 time interval:
24724 select=between(t\,10\,20)
24728 Select only I-frames contained in the 10-20 time interval:
24730 select=between(t\,10\,20)*eq(pict_type\,I)
24734 Select frames with a minimum distance of 10 seconds:
24736 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24740 Use aselect to select only audio frames with samples number > 100:
24742 aselect='gt(samples_n\,100)'
24746 Create a mosaic of the first scenes:
24748 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24751 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24755 Send even and odd frames to separate outputs, and compose them:
24757 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24761 Select useful frames from an ffconcat file which is using inpoints and
24762 outpoints but where the source files are not intra frame only.
24764 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24768 @section sendcmd, asendcmd
24770 Send commands to filters in the filtergraph.
24772 These filters read commands to be sent to other filters in the
24775 @code{sendcmd} must be inserted between two video filters,
24776 @code{asendcmd} must be inserted between two audio filters, but apart
24777 from that they act the same way.
24779 The specification of commands can be provided in the filter arguments
24780 with the @var{commands} option, or in a file specified by the
24781 @var{filename} option.
24783 These filters accept the following options:
24786 Set the commands to be read and sent to the other filters.
24788 Set the filename of the commands to be read and sent to the other
24792 @subsection Commands syntax
24794 A commands description consists of a sequence of interval
24795 specifications, comprising a list of commands to be executed when a
24796 particular event related to that interval occurs. The occurring event
24797 is typically the current frame time entering or leaving a given time
24800 An interval is specified by the following syntax:
24802 @var{START}[-@var{END}] @var{COMMANDS};
24805 The time interval is specified by the @var{START} and @var{END} times.
24806 @var{END} is optional and defaults to the maximum time.
24808 The current frame time is considered within the specified interval if
24809 it is included in the interval [@var{START}, @var{END}), that is when
24810 the time is greater or equal to @var{START} and is lesser than
24813 @var{COMMANDS} consists of a sequence of one or more command
24814 specifications, separated by ",", relating to that interval. The
24815 syntax of a command specification is given by:
24817 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24820 @var{FLAGS} is optional and specifies the type of events relating to
24821 the time interval which enable sending the specified command, and must
24822 be a non-null sequence of identifier flags separated by "+" or "|" and
24823 enclosed between "[" and "]".
24825 The following flags are recognized:
24828 The command is sent when the current frame timestamp enters the
24829 specified interval. In other words, the command is sent when the
24830 previous frame timestamp was not in the given interval, and the
24834 The command is sent when the current frame timestamp leaves the
24835 specified interval. In other words, the command is sent when the
24836 previous frame timestamp was in the given interval, and the
24840 The command @var{ARG} is interpreted as expression and result of
24841 expression is passed as @var{ARG}.
24843 The expression is evaluated through the eval API and can contain the following
24848 Original position in the file of the frame, or undefined if undefined
24849 for the current frame.
24852 The presentation timestamp in input.
24855 The count of the input frame for video or audio, starting from 0.
24858 The time in seconds of the current frame.
24861 The start time in seconds of the current command interval.
24864 The end time in seconds of the current command interval.
24867 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24872 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24875 @var{TARGET} specifies the target of the command, usually the name of
24876 the filter class or a specific filter instance name.
24878 @var{COMMAND} specifies the name of the command for the target filter.
24880 @var{ARG} is optional and specifies the optional list of argument for
24881 the given @var{COMMAND}.
24883 Between one interval specification and another, whitespaces, or
24884 sequences of characters starting with @code{#} until the end of line,
24885 are ignored and can be used to annotate comments.
24887 A simplified BNF description of the commands specification syntax
24890 @var{COMMAND_FLAG} ::= "enter" | "leave"
24891 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24892 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24893 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24894 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24895 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24898 @subsection Examples
24902 Specify audio tempo change at second 4:
24904 asendcmd=c='4.0 atempo tempo 1.5',atempo
24908 Target a specific filter instance:
24910 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24914 Specify a list of drawtext and hue commands in a file.
24916 # show text in the interval 5-10
24917 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24918 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24920 # desaturate the image in the interval 15-20
24921 15.0-20.0 [enter] hue s 0,
24922 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24924 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24926 # apply an exponential saturation fade-out effect, starting from time 25
24927 25 [enter] hue s exp(25-t)
24930 A filtergraph allowing to read and process the above command list
24931 stored in a file @file{test.cmd}, can be specified with:
24933 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24938 @section setpts, asetpts
24940 Change the PTS (presentation timestamp) of the input frames.
24942 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24944 This filter accepts the following options:
24949 The expression which is evaluated for each frame to construct its timestamp.
24953 The expression is evaluated through the eval API and can contain the following
24957 @item FRAME_RATE, FR
24958 frame rate, only defined for constant frame-rate video
24961 The presentation timestamp in input
24964 The count of the input frame for video or the number of consumed samples,
24965 not including the current frame for audio, starting from 0.
24967 @item NB_CONSUMED_SAMPLES
24968 The number of consumed samples, not including the current frame (only
24971 @item NB_SAMPLES, S
24972 The number of samples in the current frame (only audio)
24974 @item SAMPLE_RATE, SR
24975 The audio sample rate.
24978 The PTS of the first frame.
24981 the time in seconds of the first frame
24984 State whether the current frame is interlaced.
24987 the time in seconds of the current frame
24990 original position in the file of the frame, or undefined if undefined
24991 for the current frame
24994 The previous input PTS.
24997 previous input time in seconds
25000 The previous output PTS.
25003 previous output time in seconds
25006 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
25010 The wallclock (RTC) time at the start of the movie in microseconds.
25013 The timebase of the input timestamps.
25017 @subsection Examples
25021 Start counting PTS from zero
25023 setpts=PTS-STARTPTS
25027 Apply fast motion effect:
25033 Apply slow motion effect:
25039 Set fixed rate of 25 frames per second:
25045 Set fixed rate 25 fps with some jitter:
25047 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25051 Apply an offset of 10 seconds to the input PTS:
25057 Generate timestamps from a "live source" and rebase onto the current timebase:
25059 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25063 Generate timestamps by counting samples:
25072 Force color range for the output video frame.
25074 The @code{setrange} filter marks the color range property for the
25075 output frames. It does not change the input frame, but only sets the
25076 corresponding property, which affects how the frame is treated by
25079 The filter accepts the following options:
25084 Available values are:
25088 Keep the same color range property.
25090 @item unspecified, unknown
25091 Set the color range as unspecified.
25093 @item limited, tv, mpeg
25094 Set the color range as limited.
25096 @item full, pc, jpeg
25097 Set the color range as full.
25101 @section settb, asettb
25103 Set the timebase to use for the output frames timestamps.
25104 It is mainly useful for testing timebase configuration.
25106 It accepts the following parameters:
25111 The expression which is evaluated into the output timebase.
25115 The value for @option{tb} is an arithmetic expression representing a
25116 rational. The expression can contain the constants "AVTB" (the default
25117 timebase), "intb" (the input timebase) and "sr" (the sample rate,
25118 audio only). Default value is "intb".
25120 @subsection Examples
25124 Set the timebase to 1/25:
25130 Set the timebase to 1/10:
25136 Set the timebase to 1001/1000:
25142 Set the timebase to 2*intb:
25148 Set the default timebase value:
25155 Convert input audio to a video output representing frequency spectrum
25156 logarithmically using Brown-Puckette constant Q transform algorithm with
25157 direct frequency domain coefficient calculation (but the transform itself
25158 is not really constant Q, instead the Q factor is actually variable/clamped),
25159 with musical tone scale, from E0 to D#10.
25161 The filter accepts the following options:
25165 Specify the video size for the output. It must be even. For the syntax of this option,
25166 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25167 Default value is @code{1920x1080}.
25170 Set the output frame rate. Default value is @code{25}.
25173 Set the bargraph height. It must be even. Default value is @code{-1} which
25174 computes the bargraph height automatically.
25177 Set the axis height. It must be even. Default value is @code{-1} which computes
25178 the axis height automatically.
25181 Set the sonogram height. It must be even. Default value is @code{-1} which
25182 computes the sonogram height automatically.
25185 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
25186 instead. Default value is @code{1}.
25188 @item sono_v, volume
25189 Specify the sonogram volume expression. It can contain variables:
25192 the @var{bar_v} evaluated expression
25193 @item frequency, freq, f
25194 the frequency where it is evaluated
25195 @item timeclamp, tc
25196 the value of @var{timeclamp} option
25200 @item a_weighting(f)
25201 A-weighting of equal loudness
25202 @item b_weighting(f)
25203 B-weighting of equal loudness
25204 @item c_weighting(f)
25205 C-weighting of equal loudness.
25207 Default value is @code{16}.
25209 @item bar_v, volume2
25210 Specify the bargraph volume expression. It can contain variables:
25213 the @var{sono_v} evaluated expression
25214 @item frequency, freq, f
25215 the frequency where it is evaluated
25216 @item timeclamp, tc
25217 the value of @var{timeclamp} option
25221 @item a_weighting(f)
25222 A-weighting of equal loudness
25223 @item b_weighting(f)
25224 B-weighting of equal loudness
25225 @item c_weighting(f)
25226 C-weighting of equal loudness.
25228 Default value is @code{sono_v}.
25230 @item sono_g, gamma
25231 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
25232 higher gamma makes the spectrum having more range. Default value is @code{3}.
25233 Acceptable range is @code{[1, 7]}.
25235 @item bar_g, gamma2
25236 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
25240 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
25241 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
25243 @item timeclamp, tc
25244 Specify the transform timeclamp. At low frequency, there is trade-off between
25245 accuracy in time domain and frequency domain. If timeclamp is lower,
25246 event in time domain is represented more accurately (such as fast bass drum),
25247 otherwise event in frequency domain is represented more accurately
25248 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
25251 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
25252 limits future samples by applying asymmetric windowing in time domain, useful
25253 when low latency is required. Accepted range is @code{[0, 1]}.
25256 Specify the transform base frequency. Default value is @code{20.01523126408007475},
25257 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
25260 Specify the transform end frequency. Default value is @code{20495.59681441799654},
25261 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
25264 This option is deprecated and ignored.
25267 Specify the transform length in time domain. Use this option to control accuracy
25268 trade-off between time domain and frequency domain at every frequency sample.
25269 It can contain variables:
25271 @item frequency, freq, f
25272 the frequency where it is evaluated
25273 @item timeclamp, tc
25274 the value of @var{timeclamp} option.
25276 Default value is @code{384*tc/(384+tc*f)}.
25279 Specify the transform count for every video frame. Default value is @code{6}.
25280 Acceptable range is @code{[1, 30]}.
25283 Specify the transform count for every single pixel. Default value is @code{0},
25284 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
25287 Specify font file for use with freetype to draw the axis. If not specified,
25288 use embedded font. Note that drawing with font file or embedded font is not
25289 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25293 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25294 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25298 Specify font color expression. This is arithmetic expression that should return
25299 integer value 0xRRGGBB. It can contain variables:
25301 @item frequency, freq, f
25302 the frequency where it is evaluated
25303 @item timeclamp, tc
25304 the value of @var{timeclamp} option
25309 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25310 @item r(x), g(x), b(x)
25311 red, green, and blue value of intensity x.
25313 Default value is @code{st(0, (midi(f)-59.5)/12);
25314 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25315 r(1-ld(1)) + b(ld(1))}.
25318 Specify image file to draw the axis. This option override @var{fontfile} and
25319 @var{fontcolor} option.
25322 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25323 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25324 Default value is @code{1}.
25327 Set colorspace. The accepted values are:
25330 Unspecified (default)
25339 BT.470BG or BT.601-6 625
25342 SMPTE-170M or BT.601-6 525
25348 BT.2020 with non-constant luminance
25353 Set spectrogram color scheme. This is list of floating point values with format
25354 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25355 The default is @code{1|0.5|0|0|0.5|1}.
25359 @subsection Examples
25363 Playing audio while showing the spectrum:
25365 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25369 Same as above, but with frame rate 30 fps:
25371 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25375 Playing at 1280x720:
25377 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25381 Disable sonogram display:
25387 A1 and its harmonics: A1, A2, (near)E3, A3:
25389 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),
25390 asplit[a][out1]; [a] showcqt [out0]'
25394 Same as above, but with more accuracy in frequency domain:
25396 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),
25397 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25403 bar_v=10:sono_v=bar_v*a_weighting(f)
25407 Custom gamma, now spectrum is linear to the amplitude.
25413 Custom tlength equation:
25415 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)))'
25419 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25421 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25425 Custom font using fontconfig:
25427 font='Courier New,Monospace,mono|bold'
25431 Custom frequency range with custom axis using image file:
25433 axisfile=myaxis.png:basefreq=40:endfreq=10000
25439 Convert input audio to video output representing the audio power spectrum.
25440 Audio amplitude is on Y-axis while frequency is on X-axis.
25442 The filter accepts the following options:
25446 Specify size of video. For the syntax of this option, check the
25447 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25448 Default is @code{1024x512}.
25452 This set how each frequency bin will be represented.
25454 It accepts the following values:
25460 Default is @code{bar}.
25463 Set amplitude scale.
25465 It accepts the following values:
25479 Default is @code{log}.
25482 Set frequency scale.
25484 It accepts the following values:
25493 Reverse logarithmic scale.
25495 Default is @code{lin}.
25498 Set window size. Allowed range is from 16 to 65536.
25500 Default is @code{2048}
25503 Set windowing function.
25505 It accepts the following values:
25528 Default is @code{hanning}.
25531 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25532 which means optimal overlap for selected window function will be picked.
25535 Set time averaging. Setting this to 0 will display current maximal peaks.
25536 Default is @code{1}, which means time averaging is disabled.
25539 Specify list of colors separated by space or by '|' which will be used to
25540 draw channel frequencies. Unrecognized or missing colors will be replaced
25544 Set channel display mode.
25546 It accepts the following values:
25551 Default is @code{combined}.
25554 Set minimum amplitude used in @code{log} amplitude scaler.
25557 Set data display mode.
25559 It accepts the following values:
25565 Default is @code{magnitude}.
25568 @section showspatial
25570 Convert stereo input audio to a video output, representing the spatial relationship
25571 between two channels.
25573 The filter accepts the following options:
25577 Specify the video size for the output. For the syntax of this option, check the
25578 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25579 Default value is @code{512x512}.
25582 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25585 Set window function.
25587 It accepts the following values:
25612 Default value is @code{hann}.
25615 Set ratio of overlap window. Default value is @code{0.5}.
25616 When value is @code{1} overlap is set to recommended size for specific
25617 window function currently used.
25620 @anchor{showspectrum}
25621 @section showspectrum
25623 Convert input audio to a video output, representing the audio frequency
25626 The filter accepts the following options:
25630 Specify the video size for the output. For the syntax of this option, check the
25631 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25632 Default value is @code{640x512}.
25635 Specify how the spectrum should slide along the window.
25637 It accepts the following values:
25640 the samples start again on the left when they reach the right
25642 the samples scroll from right to left
25644 frames are only produced when the samples reach the right
25646 the samples scroll from left to right
25649 Default value is @code{replace}.
25652 Specify display mode.
25654 It accepts the following values:
25657 all channels are displayed in the same row
25659 all channels are displayed in separate rows
25662 Default value is @samp{combined}.
25665 Specify display color mode.
25667 It accepts the following values:
25670 each channel is displayed in a separate color
25672 each channel is displayed using the same color scheme
25674 each channel is displayed using the rainbow color scheme
25676 each channel is displayed using the moreland color scheme
25678 each channel is displayed using the nebulae color scheme
25680 each channel is displayed using the fire color scheme
25682 each channel is displayed using the fiery color scheme
25684 each channel is displayed using the fruit color scheme
25686 each channel is displayed using the cool color scheme
25688 each channel is displayed using the magma color scheme
25690 each channel is displayed using the green color scheme
25692 each channel is displayed using the viridis color scheme
25694 each channel is displayed using the plasma color scheme
25696 each channel is displayed using the cividis color scheme
25698 each channel is displayed using the terrain color scheme
25701 Default value is @samp{channel}.
25704 Specify scale used for calculating intensity color values.
25706 It accepts the following values:
25711 square root, default
25722 Default value is @samp{sqrt}.
25725 Specify frequency scale.
25727 It accepts the following values:
25735 Default value is @samp{lin}.
25738 Set saturation modifier for displayed colors. Negative values provide
25739 alternative color scheme. @code{0} is no saturation at all.
25740 Saturation must be in [-10.0, 10.0] range.
25741 Default value is @code{1}.
25744 Set window function.
25746 It accepts the following values:
25771 Default value is @code{hann}.
25774 Set orientation of time vs frequency axis. Can be @code{vertical} or
25775 @code{horizontal}. Default is @code{vertical}.
25778 Set ratio of overlap window. Default value is @code{0}.
25779 When value is @code{1} overlap is set to recommended size for specific
25780 window function currently used.
25783 Set scale gain for calculating intensity color values.
25784 Default value is @code{1}.
25787 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25790 Set color rotation, must be in [-1.0, 1.0] range.
25791 Default value is @code{0}.
25794 Set start frequency from which to display spectrogram. Default is @code{0}.
25797 Set stop frequency to which to display spectrogram. Default is @code{0}.
25800 Set upper frame rate limit. Default is @code{auto}, unlimited.
25803 Draw time and frequency axes and legends. Default is disabled.
25806 The usage is very similar to the showwaves filter; see the examples in that
25809 @subsection Examples
25813 Large window with logarithmic color scaling:
25815 showspectrum=s=1280x480:scale=log
25819 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25821 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25822 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25826 @section showspectrumpic
25828 Convert input audio to a single video frame, representing the audio frequency
25831 The filter accepts the following options:
25835 Specify the video size for the output. For the syntax of this option, check the
25836 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25837 Default value is @code{4096x2048}.
25840 Specify display mode.
25842 It accepts the following values:
25845 all channels are displayed in the same row
25847 all channels are displayed in separate rows
25849 Default value is @samp{combined}.
25852 Specify display color mode.
25854 It accepts the following values:
25857 each channel is displayed in a separate color
25859 each channel is displayed using the same color scheme
25861 each channel is displayed using the rainbow color scheme
25863 each channel is displayed using the moreland color scheme
25865 each channel is displayed using the nebulae color scheme
25867 each channel is displayed using the fire color scheme
25869 each channel is displayed using the fiery color scheme
25871 each channel is displayed using the fruit color scheme
25873 each channel is displayed using the cool color scheme
25875 each channel is displayed using the magma color scheme
25877 each channel is displayed using the green color scheme
25879 each channel is displayed using the viridis color scheme
25881 each channel is displayed using the plasma color scheme
25883 each channel is displayed using the cividis color scheme
25885 each channel is displayed using the terrain color scheme
25887 Default value is @samp{intensity}.
25890 Specify scale used for calculating intensity color values.
25892 It accepts the following values:
25897 square root, default
25907 Default value is @samp{log}.
25910 Specify frequency scale.
25912 It accepts the following values:
25920 Default value is @samp{lin}.
25923 Set saturation modifier for displayed colors. Negative values provide
25924 alternative color scheme. @code{0} is no saturation at all.
25925 Saturation must be in [-10.0, 10.0] range.
25926 Default value is @code{1}.
25929 Set window function.
25931 It accepts the following values:
25955 Default value is @code{hann}.
25958 Set orientation of time vs frequency axis. Can be @code{vertical} or
25959 @code{horizontal}. Default is @code{vertical}.
25962 Set scale gain for calculating intensity color values.
25963 Default value is @code{1}.
25966 Draw time and frequency axes and legends. Default is enabled.
25969 Set color rotation, must be in [-1.0, 1.0] range.
25970 Default value is @code{0}.
25973 Set start frequency from which to display spectrogram. Default is @code{0}.
25976 Set stop frequency to which to display spectrogram. Default is @code{0}.
25979 @subsection Examples
25983 Extract an audio spectrogram of a whole audio track
25984 in a 1024x1024 picture using @command{ffmpeg}:
25986 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25990 @section showvolume
25992 Convert input audio volume to a video output.
25994 The filter accepts the following options:
26001 Set border width, allowed range is [0, 5]. Default is 1.
26004 Set channel width, allowed range is [80, 8192]. Default is 400.
26007 Set channel height, allowed range is [1, 900]. Default is 20.
26010 Set fade, allowed range is [0, 1]. Default is 0.95.
26013 Set volume color expression.
26015 The expression can use the following variables:
26019 Current max volume of channel in dB.
26025 Current channel number, starting from 0.
26029 If set, displays channel names. Default is enabled.
26032 If set, displays volume values. Default is enabled.
26035 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
26036 default is @code{h}.
26039 Set step size, allowed range is [0, 5]. Default is 0, which means
26043 Set background opacity, allowed range is [0, 1]. Default is 0.
26046 Set metering mode, can be peak: @code{p} or rms: @code{r},
26047 default is @code{p}.
26050 Set display scale, can be linear: @code{lin} or log: @code{log},
26051 default is @code{lin}.
26055 If set to > 0., display a line for the max level
26056 in the previous seconds.
26057 default is disabled: @code{0.}
26060 The color of the max line. Use when @code{dm} option is set to > 0.
26061 default is: @code{orange}
26066 Convert input audio to a video output, representing the samples waves.
26068 The filter accepts the following options:
26072 Specify the video size for the output. For the syntax of this option, check the
26073 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26074 Default value is @code{600x240}.
26079 Available values are:
26082 Draw a point for each sample.
26085 Draw a vertical line for each sample.
26088 Draw a point for each sample and a line between them.
26091 Draw a centered vertical line for each sample.
26094 Default value is @code{point}.
26097 Set the number of samples which are printed on the same column. A
26098 larger value will decrease the frame rate. Must be a positive
26099 integer. This option can be set only if the value for @var{rate}
26100 is not explicitly specified.
26103 Set the (approximate) output frame rate. This is done by setting the
26104 option @var{n}. Default value is "25".
26106 @item split_channels
26107 Set if channels should be drawn separately or overlap. Default value is 0.
26110 Set colors separated by '|' which are going to be used for drawing of each channel.
26113 Set amplitude scale.
26115 Available values are:
26133 Set the draw mode. This is mostly useful to set for high @var{n}.
26135 Available values are:
26138 Scale pixel values for each drawn sample.
26141 Draw every sample directly.
26144 Default value is @code{scale}.
26147 @subsection Examples
26151 Output the input file audio and the corresponding video representation
26154 amovie=a.mp3,asplit[out0],showwaves[out1]
26158 Create a synthetic signal and show it with showwaves, forcing a
26159 frame rate of 30 frames per second:
26161 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26165 @section showwavespic
26167 Convert input audio to a single video frame, representing the samples waves.
26169 The filter accepts the following options:
26173 Specify the video size for the output. For the syntax of this option, check the
26174 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26175 Default value is @code{600x240}.
26177 @item split_channels
26178 Set if channels should be drawn separately or overlap. Default value is 0.
26181 Set colors separated by '|' which are going to be used for drawing of each channel.
26184 Set amplitude scale.
26186 Available values are:
26206 Available values are:
26209 Scale pixel values for each drawn sample.
26212 Draw every sample directly.
26215 Default value is @code{scale}.
26218 Set the filter mode.
26220 Available values are:
26223 Use average samples values for each drawn sample.
26226 Use peak samples values for each drawn sample.
26229 Default value is @code{average}.
26232 @subsection Examples
26236 Extract a channel split representation of the wave form of a whole audio track
26237 in a 1024x800 picture using @command{ffmpeg}:
26239 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26243 @section sidedata, asidedata
26245 Delete frame side data, or select frames based on it.
26247 This filter accepts the following options:
26251 Set mode of operation of the filter.
26253 Can be one of the following:
26257 Select every frame with side data of @code{type}.
26260 Delete side data of @code{type}. If @code{type} is not set, delete all side
26266 Set side data type used with all modes. Must be set for @code{select} mode. For
26267 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
26268 in @file{libavutil/frame.h}. For example, to choose
26269 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
26273 @section spectrumsynth
26275 Synthesize audio from 2 input video spectrums, first input stream represents
26276 magnitude across time and second represents phase across time.
26277 The filter will transform from frequency domain as displayed in videos back
26278 to time domain as presented in audio output.
26280 This filter is primarily created for reversing processed @ref{showspectrum}
26281 filter outputs, but can synthesize sound from other spectrograms too.
26282 But in such case results are going to be poor if the phase data is not
26283 available, because in such cases phase data need to be recreated, usually
26284 it's just recreated from random noise.
26285 For best results use gray only output (@code{channel} color mode in
26286 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
26287 @code{lin} scale for phase video. To produce phase, for 2nd video, use
26288 @code{data} option. Inputs videos should generally use @code{fullframe}
26289 slide mode as that saves resources needed for decoding video.
26291 The filter accepts the following options:
26295 Specify sample rate of output audio, the sample rate of audio from which
26296 spectrum was generated may differ.
26299 Set number of channels represented in input video spectrums.
26302 Set scale which was used when generating magnitude input spectrum.
26303 Can be @code{lin} or @code{log}. Default is @code{log}.
26306 Set slide which was used when generating inputs spectrums.
26307 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26308 Default is @code{fullframe}.
26311 Set window function used for resynthesis.
26314 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26315 which means optimal overlap for selected window function will be picked.
26318 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26319 Default is @code{vertical}.
26322 @subsection Examples
26326 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26327 then resynthesize videos back to audio with spectrumsynth:
26329 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
26330 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
26331 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26335 @section split, asplit
26337 Split input into several identical outputs.
26339 @code{asplit} works with audio input, @code{split} with video.
26341 The filter accepts a single parameter which specifies the number of outputs. If
26342 unspecified, it defaults to 2.
26344 @subsection Examples
26348 Create two separate outputs from the same input:
26350 [in] split [out0][out1]
26354 To create 3 or more outputs, you need to specify the number of
26357 [in] asplit=3 [out0][out1][out2]
26361 Create two separate outputs from the same input, one cropped and
26364 [in] split [splitout1][splitout2];
26365 [splitout1] crop=100:100:0:0 [cropout];
26366 [splitout2] pad=200:200:100:100 [padout];
26370 Create 5 copies of the input audio with @command{ffmpeg}:
26372 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26378 Receive commands sent through a libzmq client, and forward them to
26379 filters in the filtergraph.
26381 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26382 must be inserted between two video filters, @code{azmq} between two
26383 audio filters. Both are capable to send messages to any filter type.
26385 To enable these filters you need to install the libzmq library and
26386 headers and configure FFmpeg with @code{--enable-libzmq}.
26388 For more information about libzmq see:
26389 @url{http://www.zeromq.org/}
26391 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26392 receives messages sent through a network interface defined by the
26393 @option{bind_address} (or the abbreviation "@option{b}") option.
26394 Default value of this option is @file{tcp://localhost:5555}. You may
26395 want to alter this value to your needs, but do not forget to escape any
26396 ':' signs (see @ref{filtergraph escaping}).
26398 The received message must be in the form:
26400 @var{TARGET} @var{COMMAND} [@var{ARG}]
26403 @var{TARGET} specifies the target of the command, usually the name of
26404 the filter class or a specific filter instance name. The default
26405 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26406 but you can override this by using the @samp{filter_name@@id} syntax
26407 (see @ref{Filtergraph syntax}).
26409 @var{COMMAND} specifies the name of the command for the target filter.
26411 @var{ARG} is optional and specifies the optional argument list for the
26412 given @var{COMMAND}.
26414 Upon reception, the message is processed and the corresponding command
26415 is injected into the filtergraph. Depending on the result, the filter
26416 will send a reply to the client, adopting the format:
26418 @var{ERROR_CODE} @var{ERROR_REASON}
26422 @var{MESSAGE} is optional.
26424 @subsection Examples
26426 Look at @file{tools/zmqsend} for an example of a zmq client which can
26427 be used to send commands processed by these filters.
26429 Consider the following filtergraph generated by @command{ffplay}.
26430 In this example the last overlay filter has an instance name. All other
26431 filters will have default instance names.
26434 ffplay -dumpgraph 1 -f lavfi "
26435 color=s=100x100:c=red [l];
26436 color=s=100x100:c=blue [r];
26437 nullsrc=s=200x100, zmq [bg];
26438 [bg][l] overlay [bg+l];
26439 [bg+l][r] overlay@@my=x=100 "
26442 To change the color of the left side of the video, the following
26443 command can be used:
26445 echo Parsed_color_0 c yellow | tools/zmqsend
26448 To change the right side:
26450 echo Parsed_color_1 c pink | tools/zmqsend
26453 To change the position of the right side:
26455 echo overlay@@my x 150 | tools/zmqsend
26459 @c man end MULTIMEDIA FILTERS
26461 @chapter Multimedia Sources
26462 @c man begin MULTIMEDIA SOURCES
26464 Below is a description of the currently available multimedia sources.
26468 This is the same as @ref{movie} source, except it selects an audio
26474 Read audio and/or video stream(s) from a movie container.
26476 It accepts the following parameters:
26480 The name of the resource to read (not necessarily a file; it can also be a
26481 device or a stream accessed through some protocol).
26483 @item format_name, f
26484 Specifies the format assumed for the movie to read, and can be either
26485 the name of a container or an input device. If not specified, the
26486 format is guessed from @var{movie_name} or by probing.
26488 @item seek_point, sp
26489 Specifies the seek point in seconds. The frames will be output
26490 starting from this seek point. The parameter is evaluated with
26491 @code{av_strtod}, so the numerical value may be suffixed by an IS
26492 postfix. The default value is "0".
26495 Specifies the streams to read. Several streams can be specified,
26496 separated by "+". The source will then have as many outputs, in the
26497 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26498 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26499 respectively the default (best suited) video and audio stream. Default
26500 is "dv", or "da" if the filter is called as "amovie".
26502 @item stream_index, si
26503 Specifies the index of the video stream to read. If the value is -1,
26504 the most suitable video stream will be automatically selected. The default
26505 value is "-1". Deprecated. If the filter is called "amovie", it will select
26506 audio instead of video.
26509 Specifies how many times to read the stream in sequence.
26510 If the value is 0, the stream will be looped infinitely.
26511 Default value is "1".
26513 Note that when the movie is looped the source timestamps are not
26514 changed, so it will generate non monotonically increasing timestamps.
26516 @item discontinuity
26517 Specifies the time difference between frames above which the point is
26518 considered a timestamp discontinuity which is removed by adjusting the later
26522 It allows overlaying a second video on top of the main input of
26523 a filtergraph, as shown in this graph:
26525 input -----------> deltapts0 --> overlay --> output
26528 movie --> scale--> deltapts1 -------+
26530 @subsection Examples
26534 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26535 on top of the input labelled "in":
26537 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26538 [in] setpts=PTS-STARTPTS [main];
26539 [main][over] overlay=16:16 [out]
26543 Read from a video4linux2 device, and overlay it on top of the input
26546 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26547 [in] setpts=PTS-STARTPTS [main];
26548 [main][over] overlay=16:16 [out]
26552 Read the first video stream and the audio stream with id 0x81 from
26553 dvd.vob; the video is connected to the pad named "video" and the audio is
26554 connected to the pad named "audio":
26556 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26560 @subsection Commands
26562 Both movie and amovie support the following commands:
26565 Perform seek using "av_seek_frame".
26566 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26569 @var{stream_index}: If stream_index is -1, a default
26570 stream is selected, and @var{timestamp} is automatically converted
26571 from AV_TIME_BASE units to the stream specific time_base.
26573 @var{timestamp}: Timestamp in AVStream.time_base units
26574 or, if no stream is specified, in AV_TIME_BASE units.
26576 @var{flags}: Flags which select direction and seeking mode.
26580 Get movie duration in AV_TIME_BASE units.
26584 @c man end MULTIMEDIA SOURCES