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.
563 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
564 each band will be in separate stream:
566 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
570 Same as above, but with higher filter order:
572 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
576 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
578 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
584 Reduce audio bit resolution.
586 This filter is bit crusher with enhanced functionality. A bit crusher
587 is used to audibly reduce number of bits an audio signal is sampled
588 with. This doesn't change the bit depth at all, it just produces the
589 effect. Material reduced in bit depth sounds more harsh and "digital".
590 This filter is able to even round to continuous values instead of discrete
592 Additionally it has a D/C offset which results in different crushing of
593 the lower and the upper half of the signal.
594 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
596 Another feature of this filter is the logarithmic mode.
597 This setting switches from linear distances between bits to logarithmic ones.
598 The result is a much more "natural" sounding crusher which doesn't gate low
599 signals for example. The human ear has a logarithmic perception,
600 so this kind of crushing is much more pleasant.
601 Logarithmic crushing is also able to get anti-aliased.
603 The filter accepts the following options:
619 Can be linear: @code{lin} or logarithmic: @code{log}.
628 Set sample reduction.
631 Enable LFO. By default disabled.
642 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
646 Remove impulsive noise from input audio.
648 Samples detected as impulsive noise are replaced by interpolated samples using
649 autoregressive modelling.
653 Set window size, in milliseconds. Allowed range is from @code{10} to
654 @code{100}. Default value is @code{55} milliseconds.
655 This sets size of window which will be processed at once.
658 Set window overlap, in percentage of window size. Allowed range is from
659 @code{50} to @code{95}. Default value is @code{75} percent.
660 Setting this to a very high value increases impulsive noise removal but makes
661 whole process much slower.
664 Set autoregression order, in percentage of window size. Allowed range is from
665 @code{0} to @code{25}. Default value is @code{2} percent. This option also
666 controls quality of interpolated samples using neighbour good samples.
669 Set threshold value. Allowed range is from @code{1} to @code{100}.
670 Default value is @code{2}.
671 This controls the strength of impulsive noise which is going to be removed.
672 The lower value, the more samples will be detected as impulsive noise.
675 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
676 @code{10}. Default value is @code{2}.
677 If any two samples detected as noise are spaced less than this value then any
678 sample between those two samples will be also detected as noise.
683 It accepts the following values:
686 Select overlap-add method. Even not interpolated samples are slightly
687 changed with this method.
690 Select overlap-save method. Not interpolated samples remain unchanged.
693 Default value is @code{a}.
697 Remove clipped samples from input audio.
699 Samples detected as clipped are replaced by interpolated samples using
700 autoregressive modelling.
704 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
705 Default value is @code{55} milliseconds.
706 This sets size of window which will be processed at once.
709 Set window overlap, in percentage of window size. Allowed range is from @code{50}
710 to @code{95}. Default value is @code{75} percent.
713 Set autoregression order, in percentage of window size. Allowed range is from
714 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
715 quality of interpolated samples using neighbour good samples.
718 Set threshold value. Allowed range is from @code{1} to @code{100}.
719 Default value is @code{10}. Higher values make clip detection less aggressive.
722 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
723 Default value is @code{1000}. Higher values make clip detection less aggressive.
728 It accepts the following values:
731 Select overlap-add method. Even not interpolated samples are slightly changed
735 Select overlap-save method. Not interpolated samples remain unchanged.
738 Default value is @code{a}.
743 Delay one or more audio channels.
745 Samples in delayed channel are filled with silence.
747 The filter accepts the following option:
751 Set list of delays in milliseconds for each channel separated by '|'.
752 Unused delays will be silently ignored. If number of given delays is
753 smaller than number of channels all remaining channels will not be delayed.
754 If you want to delay exact number of samples, append 'S' to number.
755 If you want instead to delay in seconds, append 's' to number.
758 Use last set delay for all remaining channels. By default is disabled.
759 This option if enabled changes how option @code{delays} is interpreted.
766 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
767 the second channel (and any other channels that may be present) unchanged.
773 Delay second channel by 500 samples, the third channel by 700 samples and leave
774 the first channel (and any other channels that may be present) unchanged.
780 Delay all channels by same number of samples:
782 adelay=delays=64S:all=1
787 Remedy denormals in audio by adding extremely low-level noise.
789 This filter shall be placed before any filter that can produce denormals.
791 A description of the accepted parameters follows.
795 Set level of added noise in dB. Default is @code{-351}.
796 Allowed range is from -451 to -90.
799 Set type of added noise.
812 Default is @code{dc}.
817 This filter supports the all above options as @ref{commands}.
819 @section aderivative, aintegral
821 Compute derivative/integral of audio stream.
823 Applying both filters one after another produces original audio.
827 Apply echoing to the input audio.
829 Echoes are reflected sound and can occur naturally amongst mountains
830 (and sometimes large buildings) when talking or shouting; digital echo
831 effects emulate this behaviour and are often used to help fill out the
832 sound of a single instrument or vocal. The time difference between the
833 original signal and the reflection is the @code{delay}, and the
834 loudness of the reflected signal is the @code{decay}.
835 Multiple echoes can have different delays and decays.
837 A description of the accepted parameters follows.
841 Set input gain of reflected signal. Default is @code{0.6}.
844 Set output gain of reflected signal. Default is @code{0.3}.
847 Set list of time intervals in milliseconds between original signal and reflections
848 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
849 Default is @code{1000}.
852 Set list of loudness of reflected signals separated by '|'.
853 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
854 Default is @code{0.5}.
861 Make it sound as if there are twice as many instruments as are actually playing:
863 aecho=0.8:0.88:60:0.4
867 If delay is very short, then it sounds like a (metallic) robot playing music:
873 A longer delay will sound like an open air concert in the mountains:
875 aecho=0.8:0.9:1000:0.3
879 Same as above but with one more mountain:
881 aecho=0.8:0.9:1000|1800:0.3|0.25
886 Audio emphasis filter creates or restores material directly taken from LPs or
887 emphased CDs with different filter curves. E.g. to store music on vinyl the
888 signal has to be altered by a filter first to even out the disadvantages of
889 this recording medium.
890 Once the material is played back the inverse filter has to be applied to
891 restore the distortion of the frequency response.
893 The filter accepts the following options:
903 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
904 use @code{production} mode. Default is @code{reproduction} mode.
907 Set filter type. Selects medium. Can be one of the following:
919 select Compact Disc (CD).
925 select 50µs (FM-KF).
927 select 75µs (FM-KF).
933 This filter supports the all above options as @ref{commands}.
937 Modify an audio signal according to the specified expressions.
939 This filter accepts one or more expressions (one for each channel),
940 which are evaluated and used to modify a corresponding audio signal.
942 It accepts the following parameters:
946 Set the '|'-separated expressions list for each separate channel. If
947 the number of input channels is greater than the number of
948 expressions, the last specified expression is used for the remaining
951 @item channel_layout, c
952 Set output channel layout. If not specified, the channel layout is
953 specified by the number of expressions. If set to @samp{same}, it will
954 use by default the same input channel layout.
957 Each expression in @var{exprs} can contain the following constants and functions:
961 channel number of the current expression
964 number of the evaluated sample, starting from 0
970 time of the evaluated sample expressed in seconds
973 @item nb_out_channels
974 input and output number of channels
977 the value of input channel with number @var{CH}
980 Note: this filter is slow. For faster processing you should use a
989 aeval=val(ch)/2:c=same
993 Invert phase of the second channel:
1002 Apply fade-in/out effect to input audio.
1004 A description of the accepted parameters follows.
1008 Specify the effect type, can be either @code{in} for fade-in, or
1009 @code{out} for a fade-out effect. Default is @code{in}.
1011 @item start_sample, ss
1012 Specify the number of the start sample for starting to apply the fade
1013 effect. Default is 0.
1015 @item nb_samples, ns
1016 Specify the number of samples for which the fade effect has to last. At
1017 the end of the fade-in effect the output audio will have the same
1018 volume as the input audio, at the end of the fade-out transition
1019 the output audio will be silence. Default is 44100.
1021 @item start_time, st
1022 Specify the start time of the fade effect. Default is 0.
1023 The value must be specified as a time duration; see
1024 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1025 for the accepted syntax.
1026 If set this option is used instead of @var{start_sample}.
1029 Specify the duration of the fade effect. See
1030 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1031 for the accepted syntax.
1032 At the end of the fade-in effect the output audio will have the same
1033 volume as the input audio, at the end of the fade-out transition
1034 the output audio will be silence.
1035 By default the duration is determined by @var{nb_samples}.
1036 If set this option is used instead of @var{nb_samples}.
1039 Set curve for fade transition.
1041 It accepts the following values:
1044 select triangular, linear slope (default)
1046 select quarter of sine wave
1048 select half of sine wave
1050 select exponential sine wave
1054 select inverted parabola
1068 select inverted quarter of sine wave
1070 select inverted half of sine wave
1072 select double-exponential seat
1074 select double-exponential sigmoid
1076 select logistic sigmoid
1078 select sine cardinal function
1080 select inverted sine cardinal function
1086 @subsection Examples
1090 Fade in first 15 seconds of audio:
1092 afade=t=in:ss=0:d=15
1096 Fade out last 25 seconds of a 900 seconds audio:
1098 afade=t=out:st=875:d=25
1103 Denoise audio samples with FFT.
1105 A description of the accepted parameters follows.
1109 Set the noise reduction in dB, allowed range is 0.01 to 97.
1110 Default value is 12 dB.
1113 Set the noise floor in dB, allowed range is -80 to -20.
1114 Default value is -50 dB.
1119 It accepts the following values:
1128 Select shellac noise.
1131 Select custom noise, defined in @code{bn} option.
1133 Default value is white noise.
1137 Set custom band noise for every one of 15 bands.
1138 Bands are separated by ' ' or '|'.
1141 Set the residual floor in dB, allowed range is -80 to -20.
1142 Default value is -38 dB.
1145 Enable noise tracking. By default is disabled.
1146 With this enabled, noise floor is automatically adjusted.
1149 Enable residual tracking. By default is disabled.
1152 Set the output mode.
1154 It accepts the following values:
1157 Pass input unchanged.
1160 Pass noise filtered out.
1165 Default value is @var{o}.
1169 @subsection Commands
1171 This filter supports the following commands:
1173 @item sample_noise, sn
1174 Start or stop measuring noise profile.
1175 Syntax for the command is : "start" or "stop" string.
1176 After measuring noise profile is stopped it will be
1177 automatically applied in filtering.
1179 @item noise_reduction, nr
1180 Change noise reduction. Argument is single float number.
1181 Syntax for the command is : "@var{noise_reduction}"
1183 @item noise_floor, nf
1184 Change noise floor. Argument is single float number.
1185 Syntax for the command is : "@var{noise_floor}"
1187 @item output_mode, om
1188 Change output mode operation.
1189 Syntax for the command is : "i", "o" or "n" string.
1193 Apply arbitrary expressions to samples in frequency domain.
1197 Set frequency domain real expression for each separate channel separated
1198 by '|'. Default is "re".
1199 If the number of input channels is greater than the number of
1200 expressions, the last specified expression is used for the remaining
1204 Set frequency domain imaginary expression for each separate channel
1205 separated by '|'. Default is "im".
1207 Each expression in @var{real} and @var{imag} can contain the following
1208 constants and functions:
1215 current frequency bin number
1218 number of available bins
1221 channel number of the current expression
1230 current real part of frequency bin of current channel
1233 current imaginary part of frequency bin of current channel
1236 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1239 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1243 Set window size. Allowed range is from 16 to 131072.
1244 Default is @code{4096}
1247 Set window function. Default is @code{hann}.
1250 Set window overlap. If set to 1, the recommended overlap for selected
1251 window function will be picked. Default is @code{0.75}.
1254 @subsection Examples
1258 Leave almost only low frequencies in audio:
1260 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1264 Apply robotize effect:
1266 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1270 Apply whisper effect:
1272 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"
1279 Apply an arbitrary Finite Impulse Response filter.
1281 This filter is designed for applying long FIR filters,
1282 up to 60 seconds long.
1284 It can be used as component for digital crossover filters,
1285 room equalization, cross talk cancellation, wavefield synthesis,
1286 auralization, ambiophonics, ambisonics and spatialization.
1288 This filter uses the streams higher than first one as FIR coefficients.
1289 If the non-first stream holds a single channel, it will be used
1290 for all input channels in the first stream, otherwise
1291 the number of channels in the non-first stream must be same as
1292 the number of channels in the first stream.
1294 It accepts the following parameters:
1298 Set dry gain. This sets input gain.
1301 Set wet gain. This sets final output gain.
1304 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1307 Enable applying gain measured from power of IR.
1309 Set which approach to use for auto gain measurement.
1313 Do not apply any gain.
1316 select peak gain, very conservative approach. This is default value.
1319 select DC gain, limited application.
1322 select gain to noise approach, this is most popular one.
1326 Set gain to be applied to IR coefficients before filtering.
1327 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1330 Set format of IR stream. Can be @code{mono} or @code{input}.
1331 Default is @code{input}.
1334 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1335 Allowed range is 0.1 to 60 seconds.
1338 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1339 By default it is disabled.
1342 Set for which IR channel to display frequency response. By default is first channel
1343 displayed. This option is used only when @var{response} is enabled.
1346 Set video stream size. This option is used only when @var{response} is enabled.
1349 Set video stream frame rate. This option is used only when @var{response} is enabled.
1352 Set minimal partition size used for convolution. Default is @var{8192}.
1353 Allowed range is from @var{1} to @var{32768}.
1354 Lower values decreases latency at cost of higher CPU usage.
1357 Set maximal partition size used for convolution. Default is @var{8192}.
1358 Allowed range is from @var{8} to @var{32768}.
1359 Lower values may increase CPU usage.
1362 Set number of input impulse responses streams which will be switchable at runtime.
1363 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1366 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1367 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1368 This option can be changed at runtime via @ref{commands}.
1371 @subsection Examples
1375 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1377 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1384 Set output format constraints for the input audio. The framework will
1385 negotiate the most appropriate format to minimize conversions.
1387 It accepts the following parameters:
1390 @item sample_fmts, f
1391 A '|'-separated list of requested sample formats.
1393 @item sample_rates, r
1394 A '|'-separated list of requested sample rates.
1396 @item channel_layouts, cl
1397 A '|'-separated list of requested channel layouts.
1399 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1400 for the required syntax.
1403 If a parameter is omitted, all values are allowed.
1405 Force the output to either unsigned 8-bit or signed 16-bit stereo
1407 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1411 Apply frequency shift to input audio samples.
1413 The filter accepts the following options:
1417 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1418 Default value is 0.0.
1421 @subsection Commands
1423 This filter supports the above option as @ref{commands}.
1427 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1428 processing reduces disturbing noise between useful signals.
1430 Gating is done by detecting the volume below a chosen level @var{threshold}
1431 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1432 floor is set via @var{range}. Because an exact manipulation of the signal
1433 would cause distortion of the waveform the reduction can be levelled over
1434 time. This is done by setting @var{attack} and @var{release}.
1436 @var{attack} determines how long the signal has to fall below the threshold
1437 before any reduction will occur and @var{release} sets the time the signal
1438 has to rise above the threshold to reduce the reduction again.
1439 Shorter signals than the chosen attack time will be left untouched.
1443 Set input level before filtering.
1444 Default is 1. Allowed range is from 0.015625 to 64.
1447 Set the mode of operation. Can be @code{upward} or @code{downward}.
1448 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1449 will be amplified, expanding dynamic range in upward direction.
1450 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1453 Set the level of gain reduction when the signal is below the threshold.
1454 Default is 0.06125. Allowed range is from 0 to 1.
1455 Setting this to 0 disables reduction and then filter behaves like expander.
1458 If a signal rises above this level the gain reduction is released.
1459 Default is 0.125. Allowed range is from 0 to 1.
1462 Set a ratio by which the signal is reduced.
1463 Default is 2. Allowed range is from 1 to 9000.
1466 Amount of milliseconds the signal has to rise above the threshold before gain
1468 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1471 Amount of milliseconds the signal has to fall below the threshold before the
1472 reduction is increased again. Default is 250 milliseconds.
1473 Allowed range is from 0.01 to 9000.
1476 Set amount of amplification of signal after processing.
1477 Default is 1. Allowed range is from 1 to 64.
1480 Curve the sharp knee around the threshold to enter gain reduction more softly.
1481 Default is 2.828427125. Allowed range is from 1 to 8.
1484 Choose if exact signal should be taken for detection or an RMS like one.
1485 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1488 Choose if the average level between all channels or the louder channel affects
1490 Default is @code{average}. Can be @code{average} or @code{maximum}.
1493 @subsection Commands
1495 This filter supports the all above options as @ref{commands}.
1499 Apply an arbitrary Infinite Impulse Response filter.
1501 It accepts the following parameters:
1505 Set B/numerator/zeros/reflection coefficients.
1508 Set A/denominator/poles/ladder coefficients.
1520 Set coefficients format.
1524 lattice-ladder function
1526 analog transfer function
1528 digital transfer function
1530 Z-plane zeros/poles, cartesian (default)
1532 Z-plane zeros/poles, polar radians
1534 Z-plane zeros/poles, polar degrees
1540 Set type of processing.
1552 Set filtering precision.
1556 double-precision floating-point (default)
1558 single-precision floating-point
1566 Normalize filter coefficients, by default is enabled.
1567 Enabling it will normalize magnitude response at DC to 0dB.
1570 How much to use filtered signal in output. Default is 1.
1571 Range is between 0 and 1.
1574 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1575 By default it is disabled.
1578 Set for which IR channel to display frequency response. By default is first channel
1579 displayed. This option is used only when @var{response} is enabled.
1582 Set video stream size. This option is used only when @var{response} is enabled.
1585 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1588 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1589 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1592 Different coefficients and gains can be provided for every channel, in such case
1593 use '|' to separate coefficients or gains. Last provided coefficients will be
1594 used for all remaining channels.
1596 @subsection Examples
1600 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1602 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
1606 Same as above but in @code{zp} format:
1608 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
1612 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1614 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1620 The limiter prevents an input signal from rising over a desired threshold.
1621 This limiter uses lookahead technology to prevent your signal from distorting.
1622 It means that there is a small delay after the signal is processed. Keep in mind
1623 that the delay it produces is the attack time you set.
1625 The filter accepts the following options:
1629 Set input gain. Default is 1.
1632 Set output gain. Default is 1.
1635 Don't let signals above this level pass the limiter. Default is 1.
1638 The limiter will reach its attenuation level in this amount of time in
1639 milliseconds. Default is 5 milliseconds.
1642 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1643 Default is 50 milliseconds.
1646 When gain reduction is always needed ASC takes care of releasing to an
1647 average reduction level rather than reaching a reduction of 0 in the release
1651 Select how much the release time is affected by ASC, 0 means nearly no changes
1652 in release time while 1 produces higher release times.
1655 Auto level output signal. Default is enabled.
1656 This normalizes audio back to 0dB if enabled.
1659 Depending on picked setting it is recommended to upsample input 2x or 4x times
1660 with @ref{aresample} before applying this filter.
1664 Apply a two-pole all-pass filter with central frequency (in Hz)
1665 @var{frequency}, and filter-width @var{width}.
1666 An all-pass filter changes the audio's frequency to phase relationship
1667 without changing its frequency to amplitude relationship.
1669 The filter accepts the following options:
1673 Set frequency in Hz.
1676 Set method to specify band-width of filter.
1691 Specify the band-width of a filter in width_type units.
1694 How much to use filtered signal in output. Default is 1.
1695 Range is between 0 and 1.
1698 Specify which channels to filter, by default all available are filtered.
1701 Normalize biquad coefficients, by default is disabled.
1702 Enabling it will normalize magnitude response at DC to 0dB.
1705 Set the filter order, can be 1 or 2. Default is 2.
1708 Set transform type of IIR filter.
1717 @subsection Commands
1719 This filter supports the following commands:
1722 Change allpass frequency.
1723 Syntax for the command is : "@var{frequency}"
1726 Change allpass width_type.
1727 Syntax for the command is : "@var{width_type}"
1730 Change allpass width.
1731 Syntax for the command is : "@var{width}"
1735 Syntax for the command is : "@var{mix}"
1742 The filter accepts the following options:
1746 Set the number of loops. Setting this value to -1 will result in infinite loops.
1750 Set maximal number of samples. Default is 0.
1753 Set first sample of loop. Default is 0.
1759 Merge two or more audio streams into a single multi-channel stream.
1761 The filter accepts the following options:
1766 Set the number of inputs. Default is 2.
1770 If the channel layouts of the inputs are disjoint, and therefore compatible,
1771 the channel layout of the output will be set accordingly and the channels
1772 will be reordered as necessary. If the channel layouts of the inputs are not
1773 disjoint, the output will have all the channels of the first input then all
1774 the channels of the second input, in that order, and the channel layout of
1775 the output will be the default value corresponding to the total number of
1778 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1779 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1780 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1781 first input, b1 is the first channel of the second input).
1783 On the other hand, if both input are in stereo, the output channels will be
1784 in the default order: a1, a2, b1, b2, and the channel layout will be
1785 arbitrarily set to 4.0, which may or may not be the expected value.
1787 All inputs must have the same sample rate, and format.
1789 If inputs do not have the same duration, the output will stop with the
1792 @subsection Examples
1796 Merge two mono files into a stereo stream:
1798 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1802 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1804 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
1810 Mixes multiple audio inputs into a single output.
1812 Note that this filter only supports float samples (the @var{amerge}
1813 and @var{pan} audio filters support many formats). If the @var{amix}
1814 input has integer samples then @ref{aresample} will be automatically
1815 inserted to perform the conversion to float samples.
1819 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1821 will mix 3 input audio streams to a single output with the same duration as the
1822 first input and a dropout transition time of 3 seconds.
1824 It accepts the following parameters:
1828 The number of inputs. If unspecified, it defaults to 2.
1831 How to determine the end-of-stream.
1835 The duration of the longest input. (default)
1838 The duration of the shortest input.
1841 The duration of the first input.
1845 @item dropout_transition
1846 The transition time, in seconds, for volume renormalization when an input
1847 stream ends. The default value is 2 seconds.
1850 Specify weight of each input audio stream as sequence.
1851 Each weight is separated by space. By default all inputs have same weight.
1854 @subsection Commands
1856 This filter supports the following commands:
1859 Syntax is same as option with same name.
1864 Multiply first audio stream with second audio stream and store result
1865 in output audio stream. Multiplication is done by multiplying each
1866 sample from first stream with sample at same position from second stream.
1868 With this element-wise multiplication one can create amplitude fades and
1869 amplitude modulations.
1871 @section anequalizer
1873 High-order parametric multiband equalizer for each channel.
1875 It accepts the following parameters:
1879 This option string is in format:
1880 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1881 Each equalizer band is separated by '|'.
1885 Set channel number to which equalization will be applied.
1886 If input doesn't have that channel the entry is ignored.
1889 Set central frequency for band.
1890 If input doesn't have that frequency the entry is ignored.
1893 Set band width in Hertz.
1896 Set band gain in dB.
1899 Set filter type for band, optional, can be:
1903 Butterworth, this is default.
1914 With this option activated frequency response of anequalizer is displayed
1918 Set video stream size. Only useful if curves option is activated.
1921 Set max gain that will be displayed. Only useful if curves option is activated.
1922 Setting this to a reasonable value makes it possible to display gain which is derived from
1923 neighbour bands which are too close to each other and thus produce higher gain
1924 when both are activated.
1927 Set frequency scale used to draw frequency response in video output.
1928 Can be linear or logarithmic. Default is logarithmic.
1931 Set color for each channel curve which is going to be displayed in video stream.
1932 This is list of color names separated by space or by '|'.
1933 Unrecognised or missing colors will be replaced by white color.
1936 @subsection Examples
1940 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1941 for first 2 channels using Chebyshev type 1 filter:
1943 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1947 @subsection Commands
1949 This filter supports the following commands:
1952 Alter existing filter parameters.
1953 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1955 @var{fN} is existing filter number, starting from 0, if no such filter is available
1957 @var{freq} set new frequency parameter.
1958 @var{width} set new width parameter in Hertz.
1959 @var{gain} set new gain parameter in dB.
1961 Full filter invocation with asendcmd may look like this:
1962 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1967 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1969 Each sample is adjusted by looking for other samples with similar contexts. This
1970 context similarity is defined by comparing their surrounding patches of size
1971 @option{p}. Patches are searched in an area of @option{r} around the sample.
1973 The filter accepts the following options:
1977 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1980 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1981 Default value is 2 milliseconds.
1984 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1985 Default value is 6 milliseconds.
1988 Set the output mode.
1990 It accepts the following values:
1993 Pass input unchanged.
1996 Pass noise filtered out.
2001 Default value is @var{o}.
2005 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
2008 @subsection Commands
2010 This filter supports the all above options as @ref{commands}.
2013 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
2015 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
2016 relate to producing the least mean square of the error signal (difference between the desired,
2017 2nd input audio stream and the actual signal, the 1st input audio stream).
2019 A description of the accepted options follows.
2032 Set the filter leakage.
2035 It accepts the following values:
2044 Pass filtered samples.
2047 Pass difference between desired and filtered samples.
2049 Default value is @var{o}.
2053 @subsection Examples
2057 One of many usages of this filter is noise reduction, input audio is filtered
2058 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2060 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2064 @subsection Commands
2066 This filter supports the same commands as options, excluding option @code{order}.
2070 Pass the audio source unchanged to the output.
2074 Pad the end of an audio stream with silence.
2076 This can be used together with @command{ffmpeg} @option{-shortest} to
2077 extend audio streams to the same length as the video stream.
2079 A description of the accepted options follows.
2083 Set silence packet size. Default value is 4096.
2086 Set the number of samples of silence to add to the end. After the
2087 value is reached, the stream is terminated. This option is mutually
2088 exclusive with @option{whole_len}.
2091 Set the minimum total number of samples in the output audio stream. If
2092 the value is longer than the input audio length, silence is added to
2093 the end, until the value is reached. This option is mutually exclusive
2094 with @option{pad_len}.
2097 Specify the duration of samples of silence to add. See
2098 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2099 for the accepted syntax. Used only if set to non-zero value.
2102 Specify the minimum total duration in the output audio stream. See
2103 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2104 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2105 the input audio length, silence is added to the end, until the value is reached.
2106 This option is mutually exclusive with @option{pad_dur}
2109 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2110 nor @option{whole_dur} option is set, the filter will add silence to the end of
2111 the input stream indefinitely.
2113 @subsection Examples
2117 Add 1024 samples of silence to the end of the input:
2123 Make sure the audio output will contain at least 10000 samples, pad
2124 the input with silence if required:
2126 apad=whole_len=10000
2130 Use @command{ffmpeg} to pad the audio input with silence, so that the
2131 video stream will always result the shortest and will be converted
2132 until the end in the output file when using the @option{shortest}
2135 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2140 Add a phasing effect to the input audio.
2142 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2143 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2145 A description of the accepted parameters follows.
2149 Set input gain. Default is 0.4.
2152 Set output gain. Default is 0.74
2155 Set delay in milliseconds. Default is 3.0.
2158 Set decay. Default is 0.4.
2161 Set modulation speed in Hz. Default is 0.5.
2164 Set modulation type. Default is triangular.
2166 It accepts the following values:
2173 @section aphaseshift
2174 Apply phase shift to input audio samples.
2176 The filter accepts the following options:
2180 Specify phase shift. Allowed range is from -1.0 to 1.0.
2181 Default value is 0.0.
2184 @subsection Commands
2186 This filter supports the above option as @ref{commands}.
2190 Audio pulsator is something between an autopanner and a tremolo.
2191 But it can produce funny stereo effects as well. Pulsator changes the volume
2192 of the left and right channel based on a LFO (low frequency oscillator) with
2193 different waveforms and shifted phases.
2194 This filter have the ability to define an offset between left and right
2195 channel. An offset of 0 means that both LFO shapes match each other.
2196 The left and right channel are altered equally - a conventional tremolo.
2197 An offset of 50% means that the shape of the right channel is exactly shifted
2198 in phase (or moved backwards about half of the frequency) - pulsator acts as
2199 an autopanner. At 1 both curves match again. Every setting in between moves the
2200 phase shift gapless between all stages and produces some "bypassing" sounds with
2201 sine and triangle waveforms. The more you set the offset near 1 (starting from
2202 the 0.5) the faster the signal passes from the left to the right speaker.
2204 The filter accepts the following options:
2208 Set input gain. By default it is 1. Range is [0.015625 - 64].
2211 Set output gain. By default it is 1. Range is [0.015625 - 64].
2214 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2215 sawup or sawdown. Default is sine.
2218 Set modulation. Define how much of original signal is affected by the LFO.
2221 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2224 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2227 Set pulse width. Default is 1. Allowed range is [0 - 2].
2230 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2233 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2237 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2241 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2242 if timing is set to hz.
2248 Resample the input audio to the specified parameters, using the
2249 libswresample library. If none are specified then the filter will
2250 automatically convert between its input and output.
2252 This filter is also able to stretch/squeeze the audio data to make it match
2253 the timestamps or to inject silence / cut out audio to make it match the
2254 timestamps, do a combination of both or do neither.
2256 The filter accepts the syntax
2257 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2258 expresses a sample rate and @var{resampler_options} is a list of
2259 @var{key}=@var{value} pairs, separated by ":". See the
2260 @ref{Resampler Options,,"Resampler Options" section in the
2261 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2262 for the complete list of supported options.
2264 @subsection Examples
2268 Resample the input audio to 44100Hz:
2274 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2275 samples per second compensation:
2277 aresample=async=1000
2283 Reverse an audio clip.
2285 Warning: This filter requires memory to buffer the entire clip, so trimming
2288 @subsection Examples
2292 Take the first 5 seconds of a clip, and reverse it.
2294 atrim=end=5,areverse
2300 Reduce noise from speech using Recurrent Neural Networks.
2302 This filter accepts the following options:
2306 Set train model file to load. This option is always required.
2309 @section asetnsamples
2311 Set the number of samples per each output audio frame.
2313 The last output packet may contain a different number of samples, as
2314 the filter will flush all the remaining samples when the input audio
2317 The filter accepts the following options:
2321 @item nb_out_samples, n
2322 Set the number of frames per each output audio frame. The number is
2323 intended as the number of samples @emph{per each channel}.
2324 Default value is 1024.
2327 If set to 1, the filter will pad the last audio frame with zeroes, so
2328 that the last frame will contain the same number of samples as the
2329 previous ones. Default value is 1.
2332 For example, to set the number of per-frame samples to 1234 and
2333 disable padding for the last frame, use:
2335 asetnsamples=n=1234:p=0
2340 Set the sample rate without altering the PCM data.
2341 This will result in a change of speed and pitch.
2343 The filter accepts the following options:
2346 @item sample_rate, r
2347 Set the output sample rate. Default is 44100 Hz.
2352 Show a line containing various information for each input audio frame.
2353 The input audio is not modified.
2355 The shown line contains a sequence of key/value pairs of the form
2356 @var{key}:@var{value}.
2358 The following values are shown in the output:
2362 The (sequential) number of the input frame, starting from 0.
2365 The presentation timestamp of the input frame, in time base units; the time base
2366 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2369 The presentation timestamp of the input frame in seconds.
2372 position of the frame in the input stream, -1 if this information in
2373 unavailable and/or meaningless (for example in case of synthetic audio)
2382 The sample rate for the audio frame.
2385 The number of samples (per channel) in the frame.
2388 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2389 audio, the data is treated as if all the planes were concatenated.
2391 @item plane_checksums
2392 A list of Adler-32 checksums for each data plane.
2396 Apply audio soft clipping.
2398 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2399 along a smooth curve, rather than the abrupt shape of hard-clipping.
2401 This filter accepts the following options:
2405 Set type of soft-clipping.
2407 It accepts the following values:
2421 Set additional parameter which controls sigmoid function.
2424 Set oversampling factor.
2427 @subsection Commands
2429 This filter supports the all above options as @ref{commands}.
2432 Automatic Speech Recognition
2434 This filter uses PocketSphinx for speech recognition. To enable
2435 compilation of this filter, you need to configure FFmpeg with
2436 @code{--enable-pocketsphinx}.
2438 It accepts the following options:
2442 Set sampling rate of input audio. Defaults is @code{16000}.
2443 This need to match speech models, otherwise one will get poor results.
2446 Set dictionary containing acoustic model files.
2449 Set pronunciation dictionary.
2452 Set language model file.
2455 Set language model set.
2458 Set which language model to use.
2461 Set output for log messages.
2464 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2469 Display time domain statistical information about the audio channels.
2470 Statistics are calculated and displayed for each audio channel and,
2471 where applicable, an overall figure is also given.
2473 It accepts the following option:
2476 Short window length in seconds, used for peak and trough RMS measurement.
2477 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2481 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2482 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2485 Available keys for each channel are:
2531 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2532 this @code{lavfi.astats.Overall.Peak_count}.
2534 For description what each key means read below.
2537 Set number of frame after which stats are going to be recalculated.
2538 Default is disabled.
2540 @item measure_perchannel
2541 Select the entries which need to be measured per channel. The metadata keys can
2542 be used as flags, default is @option{all} which measures everything.
2543 @option{none} disables all per channel measurement.
2545 @item measure_overall
2546 Select the entries which need to be measured overall. The metadata keys can
2547 be used as flags, default is @option{all} which measures everything.
2548 @option{none} disables all overall measurement.
2552 A description of each shown parameter follows:
2556 Mean amplitude displacement from zero.
2559 Minimal sample level.
2562 Maximal sample level.
2564 @item Min difference
2565 Minimal difference between two consecutive samples.
2567 @item Max difference
2568 Maximal difference between two consecutive samples.
2570 @item Mean difference
2571 Mean difference between two consecutive samples.
2572 The average of each difference between two consecutive samples.
2574 @item RMS difference
2575 Root Mean Square difference between two consecutive samples.
2579 Standard peak and RMS level measured in dBFS.
2583 Peak and trough values for RMS level measured over a short window.
2586 Standard ratio of peak to RMS level (note: not in dB).
2589 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2590 (i.e. either @var{Min level} or @var{Max level}).
2593 Number of occasions (not the number of samples) that the signal attained either
2594 @var{Min level} or @var{Max level}.
2596 @item Noise floor dB
2597 Minimum local peak measured in dBFS over a short window.
2599 @item Noise floor count
2600 Number of occasions (not the number of samples) that the signal attained
2604 Overall bit depth of audio. Number of bits used for each sample.
2607 Measured dynamic range of audio in dB.
2609 @item Zero crossings
2610 Number of points where the waveform crosses the zero level axis.
2612 @item Zero crossings rate
2613 Rate of Zero crossings and number of audio samples.
2617 Boost subwoofer frequencies.
2619 The filter accepts the following options:
2623 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2624 Default value is 0.7.
2627 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2628 Default value is 0.7.
2631 Set delay line decay gain value. Allowed range is from 0 to 1.
2632 Default value is 0.7.
2635 Set delay line feedback gain value. Allowed range is from 0 to 1.
2636 Default value is 0.9.
2639 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2640 Default value is 100.
2643 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2644 Default value is 0.5.
2647 Set delay. Allowed range is from 1 to 100.
2648 Default value is 20.
2651 @subsection Commands
2653 This filter supports the all above options as @ref{commands}.
2656 Cut super frequencies.
2658 The filter accepts the following options:
2662 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2663 Default value is 20000.
2666 Set filter order. Available values are from 3 to 20.
2667 Default value is 10.
2670 @subsection Commands
2672 This filter supports the all above options as @ref{commands}.
2678 The filter accepts exactly one parameter, the audio tempo. If not
2679 specified then the filter will assume nominal 1.0 tempo. Tempo must
2680 be in the [0.5, 100.0] range.
2682 Note that tempo greater than 2 will skip some samples rather than
2683 blend them in. If for any reason this is a concern it is always
2684 possible to daisy-chain several instances of atempo to achieve the
2685 desired product tempo.
2687 @subsection Examples
2691 Slow down audio to 80% tempo:
2697 To speed up audio to 300% tempo:
2703 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2705 atempo=sqrt(3),atempo=sqrt(3)
2709 @subsection Commands
2711 This filter supports the following commands:
2714 Change filter tempo scale factor.
2715 Syntax for the command is : "@var{tempo}"
2720 Trim the input so that the output contains one continuous subpart of the input.
2722 It accepts the following parameters:
2725 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2726 sample with the timestamp @var{start} will be the first sample in the output.
2729 Specify time of the first audio sample that will be dropped, i.e. the
2730 audio sample immediately preceding the one with the timestamp @var{end} will be
2731 the last sample in the output.
2734 Same as @var{start}, except this option sets the start timestamp in samples
2738 Same as @var{end}, except this option sets the end timestamp in samples instead
2742 The maximum duration of the output in seconds.
2745 The number of the first sample that should be output.
2748 The number of the first sample that should be dropped.
2751 @option{start}, @option{end}, and @option{duration} are expressed as time
2752 duration specifications; see
2753 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2755 Note that the first two sets of the start/end options and the @option{duration}
2756 option look at the frame timestamp, while the _sample options simply count the
2757 samples that pass through the filter. So start/end_pts and start/end_sample will
2758 give different results when the timestamps are wrong, inexact or do not start at
2759 zero. Also note that this filter does not modify the timestamps. If you wish
2760 to have the output timestamps start at zero, insert the asetpts filter after the
2763 If multiple start or end options are set, this filter tries to be greedy and
2764 keep all samples that match at least one of the specified constraints. To keep
2765 only the part that matches all the constraints at once, chain multiple atrim
2768 The defaults are such that all the input is kept. So it is possible to set e.g.
2769 just the end values to keep everything before the specified time.
2774 Drop everything except the second minute of input:
2776 ffmpeg -i INPUT -af atrim=60:120
2780 Keep only the first 1000 samples:
2782 ffmpeg -i INPUT -af atrim=end_sample=1000
2787 @section axcorrelate
2788 Calculate normalized cross-correlation between two input audio streams.
2790 Resulted samples are always between -1 and 1 inclusive.
2791 If result is 1 it means two input samples are highly correlated in that selected segment.
2792 Result 0 means they are not correlated at all.
2793 If result is -1 it means two input samples are out of phase, which means they cancel each
2796 The filter accepts the following options:
2800 Set size of segment over which cross-correlation is calculated.
2801 Default is 256. Allowed range is from 2 to 131072.
2804 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2805 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2806 are always zero and thus need much less calculations to make.
2807 This is generally not true, but is valid for typical audio streams.
2810 @subsection Examples
2814 Calculate correlation between channels in stereo audio stream:
2816 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2822 Apply a two-pole Butterworth band-pass filter with central
2823 frequency @var{frequency}, and (3dB-point) band-width width.
2824 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2825 instead of the default: constant 0dB peak gain.
2826 The filter roll off at 6dB per octave (20dB per decade).
2828 The filter accepts the following options:
2832 Set the filter's central frequency. Default is @code{3000}.
2835 Constant skirt gain if set to 1. Defaults to 0.
2838 Set method to specify band-width of filter.
2853 Specify the band-width of a filter in width_type units.
2856 How much to use filtered signal in output. Default is 1.
2857 Range is between 0 and 1.
2860 Specify which channels to filter, by default all available are filtered.
2863 Normalize biquad coefficients, by default is disabled.
2864 Enabling it will normalize magnitude response at DC to 0dB.
2867 Set transform type of IIR filter.
2876 @subsection Commands
2878 This filter supports the following commands:
2881 Change bandpass frequency.
2882 Syntax for the command is : "@var{frequency}"
2885 Change bandpass width_type.
2886 Syntax for the command is : "@var{width_type}"
2889 Change bandpass width.
2890 Syntax for the command is : "@var{width}"
2893 Change bandpass mix.
2894 Syntax for the command is : "@var{mix}"
2899 Apply a two-pole Butterworth band-reject filter with central
2900 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2901 The filter roll off at 6dB per octave (20dB per decade).
2903 The filter accepts the following options:
2907 Set the filter's central frequency. Default is @code{3000}.
2910 Set method to specify band-width of filter.
2925 Specify the band-width of a filter in width_type units.
2928 How much to use filtered signal in output. Default is 1.
2929 Range is between 0 and 1.
2932 Specify which channels to filter, by default all available are filtered.
2935 Normalize biquad coefficients, by default is disabled.
2936 Enabling it will normalize magnitude response at DC to 0dB.
2939 Set transform type of IIR filter.
2948 @subsection Commands
2950 This filter supports the following commands:
2953 Change bandreject frequency.
2954 Syntax for the command is : "@var{frequency}"
2957 Change bandreject width_type.
2958 Syntax for the command is : "@var{width_type}"
2961 Change bandreject width.
2962 Syntax for the command is : "@var{width}"
2965 Change bandreject mix.
2966 Syntax for the command is : "@var{mix}"
2969 @section bass, lowshelf
2971 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2972 shelving filter with a response similar to that of a standard
2973 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2975 The filter accepts the following options:
2979 Give the gain at 0 Hz. Its useful range is about -20
2980 (for a large cut) to +20 (for a large boost).
2981 Beware of clipping when using a positive gain.
2984 Set the filter's central frequency and so can be used
2985 to extend or reduce the frequency range to be boosted or cut.
2986 The default value is @code{100} Hz.
2989 Set method to specify band-width of filter.
3004 Determine how steep is the filter's shelf transition.
3007 How much to use filtered signal in output. Default is 1.
3008 Range is between 0 and 1.
3011 Specify which channels to filter, by default all available are filtered.
3014 Normalize biquad coefficients, by default is disabled.
3015 Enabling it will normalize magnitude response at DC to 0dB.
3018 Set transform type of IIR filter.
3027 @subsection Commands
3029 This filter supports the following commands:
3032 Change bass frequency.
3033 Syntax for the command is : "@var{frequency}"
3036 Change bass width_type.
3037 Syntax for the command is : "@var{width_type}"
3041 Syntax for the command is : "@var{width}"
3045 Syntax for the command is : "@var{gain}"
3049 Syntax for the command is : "@var{mix}"
3054 Apply a biquad IIR filter with the given coefficients.
3055 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3056 are the numerator and denominator coefficients respectively.
3057 and @var{channels}, @var{c} specify which channels to filter, by default all
3058 available are filtered.
3060 @subsection Commands
3062 This filter supports the following commands:
3070 Change biquad parameter.
3071 Syntax for the command is : "@var{value}"
3074 How much to use filtered signal in output. Default is 1.
3075 Range is between 0 and 1.
3078 Specify which channels to filter, by default all available are filtered.
3081 Normalize biquad coefficients, by default is disabled.
3082 Enabling it will normalize magnitude response at DC to 0dB.
3085 Set transform type of IIR filter.
3095 Bauer stereo to binaural transformation, which improves headphone listening of
3096 stereo audio records.
3098 To enable compilation of this filter you need to configure FFmpeg with
3099 @code{--enable-libbs2b}.
3101 It accepts the following parameters:
3105 Pre-defined crossfeed level.
3109 Default level (fcut=700, feed=50).
3112 Chu Moy circuit (fcut=700, feed=60).
3115 Jan Meier circuit (fcut=650, feed=95).
3120 Cut frequency (in Hz).
3129 Remap input channels to new locations.
3131 It accepts the following parameters:
3134 Map channels from input to output. The argument is a '|'-separated list of
3135 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3136 @var{in_channel} form. @var{in_channel} can be either the name of the input
3137 channel (e.g. FL for front left) or its index in the input channel layout.
3138 @var{out_channel} is the name of the output channel or its index in the output
3139 channel layout. If @var{out_channel} is not given then it is implicitly an
3140 index, starting with zero and increasing by one for each mapping.
3142 @item channel_layout
3143 The channel layout of the output stream.
3146 If no mapping is present, the filter will implicitly map input channels to
3147 output channels, preserving indices.
3149 @subsection Examples
3153 For example, assuming a 5.1+downmix input MOV file,
3155 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3157 will create an output WAV file tagged as stereo from the downmix channels of
3161 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3163 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3167 @section channelsplit
3169 Split each channel from an input audio stream into a separate output stream.
3171 It accepts the following parameters:
3173 @item channel_layout
3174 The channel layout of the input stream. The default is "stereo".
3176 A channel layout describing the channels to be extracted as separate output streams
3177 or "all" to extract each input channel as a separate stream. The default is "all".
3179 Choosing channels not present in channel layout in the input will result in an error.
3182 @subsection Examples
3186 For example, assuming a stereo input MP3 file,
3188 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3190 will create an output Matroska file with two audio streams, one containing only
3191 the left channel and the other the right channel.
3194 Split a 5.1 WAV file into per-channel files:
3196 ffmpeg -i in.wav -filter_complex
3197 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3198 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3199 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3204 Extract only LFE from a 5.1 WAV file:
3206 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3207 -map '[LFE]' lfe.wav
3212 Add a chorus effect to the audio.
3214 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3216 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3217 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3218 The modulation depth defines the range the modulated delay is played before or after
3219 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3220 sound tuned around the original one, like in a chorus where some vocals are slightly
3223 It accepts the following parameters:
3226 Set input gain. Default is 0.4.
3229 Set output gain. Default is 0.4.
3232 Set delays. A typical delay is around 40ms to 60ms.
3244 @subsection Examples
3250 chorus=0.7:0.9:55:0.4:0.25:2
3256 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3260 Fuller sounding chorus with three delays:
3262 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
3267 Compress or expand the audio's dynamic range.
3269 It accepts the following parameters:
3275 A list of times in seconds for each channel over which the instantaneous level
3276 of the input signal is averaged to determine its volume. @var{attacks} refers to
3277 increase of volume and @var{decays} refers to decrease of volume. For most
3278 situations, the attack time (response to the audio getting louder) should be
3279 shorter than the decay time, because the human ear is more sensitive to sudden
3280 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3281 a typical value for decay is 0.8 seconds.
3282 If specified number of attacks & decays is lower than number of channels, the last
3283 set attack/decay will be used for all remaining channels.
3286 A list of points for the transfer function, specified in dB relative to the
3287 maximum possible signal amplitude. Each key points list must be defined using
3288 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3289 @code{x0/y0 x1/y1 x2/y2 ....}
3291 The input values must be in strictly increasing order but the transfer function
3292 does not have to be monotonically rising. The point @code{0/0} is assumed but
3293 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3294 function are @code{-70/-70|-60/-20|1/0}.
3297 Set the curve radius in dB for all joints. It defaults to 0.01.
3300 Set the additional gain in dB to be applied at all points on the transfer
3301 function. This allows for easy adjustment of the overall gain.
3305 Set an initial volume, in dB, to be assumed for each channel when filtering
3306 starts. This permits the user to supply a nominal level initially, so that, for
3307 example, a very large gain is not applied to initial signal levels before the
3308 companding has begun to operate. A typical value for audio which is initially
3309 quiet is -90 dB. It defaults to 0.
3312 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3313 delayed before being fed to the volume adjuster. Specifying a delay
3314 approximately equal to the attack/decay times allows the filter to effectively
3315 operate in predictive rather than reactive mode. It defaults to 0.
3319 @subsection Examples
3323 Make music with both quiet and loud passages suitable for listening to in a
3326 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3329 Another example for audio with whisper and explosion parts:
3331 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3335 A noise gate for when the noise is at a lower level than the signal:
3337 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3341 Here is another noise gate, this time for when the noise is at a higher level
3342 than the signal (making it, in some ways, similar to squelch):
3344 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3348 2:1 compression starting at -6dB:
3350 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3354 2:1 compression starting at -9dB:
3356 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3360 2:1 compression starting at -12dB:
3362 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3366 2:1 compression starting at -18dB:
3368 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3372 3:1 compression starting at -15dB:
3374 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3380 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3386 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
3390 Hard limiter at -6dB:
3392 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3396 Hard limiter at -12dB:
3398 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3402 Hard noise gate at -35 dB:
3404 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3410 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3414 @section compensationdelay
3416 Compensation Delay Line is a metric based delay to compensate differing
3417 positions of microphones or speakers.
3419 For example, you have recorded guitar with two microphones placed in
3420 different locations. Because the front of sound wave has fixed speed in
3421 normal conditions, the phasing of microphones can vary and depends on
3422 their location and interposition. The best sound mix can be achieved when
3423 these microphones are in phase (synchronized). Note that a distance of
3424 ~30 cm between microphones makes one microphone capture the signal in
3425 antiphase to the other microphone. That makes the final mix sound moody.
3426 This filter helps to solve phasing problems by adding different delays
3427 to each microphone track and make them synchronized.
3429 The best result can be reached when you take one track as base and
3430 synchronize other tracks one by one with it.
3431 Remember that synchronization/delay tolerance depends on sample rate, too.
3432 Higher sample rates will give more tolerance.
3434 The filter accepts the following parameters:
3438 Set millimeters distance. This is compensation distance for fine tuning.
3442 Set cm distance. This is compensation distance for tightening distance setup.
3446 Set meters distance. This is compensation distance for hard distance setup.
3450 Set dry amount. Amount of unprocessed (dry) signal.
3454 Set wet amount. Amount of processed (wet) signal.
3458 Set temperature in degrees Celsius. This is the temperature of the environment.
3463 Apply headphone crossfeed filter.
3465 Crossfeed is the process of blending the left and right channels of stereo
3467 It is mainly used to reduce extreme stereo separation of low frequencies.
3469 The intent is to produce more speaker like sound to the listener.
3471 The filter accepts the following options:
3475 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3476 This sets gain of low shelf filter for side part of stereo image.
3477 Default is -6dB. Max allowed is -30db when strength is set to 1.
3480 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3481 This sets cut off frequency of low shelf filter. Default is cut off near
3482 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3485 Set curve slope of low shelf filter. Default is 0.5.
3486 Allowed range is from 0.01 to 1.
3489 Set input gain. Default is 0.9.
3492 Set output gain. Default is 1.
3495 @subsection Commands
3497 This filter supports the all above options as @ref{commands}.
3499 @section crystalizer
3500 Simple algorithm to expand audio dynamic range.
3502 The filter accepts the following options:
3506 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3507 (unchanged sound) to 10.0 (maximum effect).
3510 Enable clipping. By default is enabled.
3513 @subsection Commands
3515 This filter supports the all above options as @ref{commands}.
3518 Apply a DC shift to the audio.
3520 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3521 in the recording chain) from the audio. The effect of a DC offset is reduced
3522 headroom and hence volume. The @ref{astats} filter can be used to determine if
3523 a signal has a DC offset.
3527 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3531 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3532 used to prevent clipping.
3537 Apply de-essing to the audio samples.
3541 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3545 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3549 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3553 Set the output mode.
3555 It accepts the following values:
3558 Pass input unchanged.
3561 Pass ess filtered out.
3566 Default value is @var{o}.
3572 Measure audio dynamic range.
3574 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3575 is found in transition material. And anything less that 8 have very poor dynamics
3576 and is very compressed.
3578 The filter accepts the following options:
3582 Set window length in seconds used to split audio into segments of equal length.
3583 Default is 3 seconds.
3587 Dynamic Audio Normalizer.
3589 This filter applies a certain amount of gain to the input audio in order
3590 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3591 contrast to more "simple" normalization algorithms, the Dynamic Audio
3592 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3593 This allows for applying extra gain to the "quiet" sections of the audio
3594 while avoiding distortions or clipping the "loud" sections. In other words:
3595 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3596 sections, in the sense that the volume of each section is brought to the
3597 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3598 this goal *without* applying "dynamic range compressing". It will retain 100%
3599 of the dynamic range *within* each section of the audio file.
3603 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3604 Default is 500 milliseconds.
3605 The Dynamic Audio Normalizer processes the input audio in small chunks,
3606 referred to as frames. This is required, because a peak magnitude has no
3607 meaning for just a single sample value. Instead, we need to determine the
3608 peak magnitude for a contiguous sequence of sample values. While a "standard"
3609 normalizer would simply use the peak magnitude of the complete file, the
3610 Dynamic Audio Normalizer determines the peak magnitude individually for each
3611 frame. The length of a frame is specified in milliseconds. By default, the
3612 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3613 been found to give good results with most files.
3614 Note that the exact frame length, in number of samples, will be determined
3615 automatically, based on the sampling rate of the individual input audio file.
3618 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3619 number. Default is 31.
3620 Probably the most important parameter of the Dynamic Audio Normalizer is the
3621 @code{window size} of the Gaussian smoothing filter. The filter's window size
3622 is specified in frames, centered around the current frame. For the sake of
3623 simplicity, this must be an odd number. Consequently, the default value of 31
3624 takes into account the current frame, as well as the 15 preceding frames and
3625 the 15 subsequent frames. Using a larger window results in a stronger
3626 smoothing effect and thus in less gain variation, i.e. slower gain
3627 adaptation. Conversely, using a smaller window results in a weaker smoothing
3628 effect and thus in more gain variation, i.e. faster gain adaptation.
3629 In other words, the more you increase this value, the more the Dynamic Audio
3630 Normalizer will behave like a "traditional" normalization filter. On the
3631 contrary, the more you decrease this value, the more the Dynamic Audio
3632 Normalizer will behave like a dynamic range compressor.
3635 Set the target peak value. This specifies the highest permissible magnitude
3636 level for the normalized audio input. This filter will try to approach the
3637 target peak magnitude as closely as possible, but at the same time it also
3638 makes sure that the normalized signal will never exceed the peak magnitude.
3639 A frame's maximum local gain factor is imposed directly by the target peak
3640 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3641 It is not recommended to go above this value.
3644 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3645 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3646 factor for each input frame, i.e. the maximum gain factor that does not
3647 result in clipping or distortion. The maximum gain factor is determined by
3648 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3649 additionally bounds the frame's maximum gain factor by a predetermined
3650 (global) maximum gain factor. This is done in order to avoid excessive gain
3651 factors in "silent" or almost silent frames. By default, the maximum gain
3652 factor is 10.0, For most inputs the default value should be sufficient and
3653 it usually is not recommended to increase this value. Though, for input
3654 with an extremely low overall volume level, it may be necessary to allow even
3655 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3656 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3657 Instead, a "sigmoid" threshold function will be applied. This way, the
3658 gain factors will smoothly approach the threshold value, but never exceed that
3662 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3663 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3664 This means that the maximum local gain factor for each frame is defined
3665 (only) by the frame's highest magnitude sample. This way, the samples can
3666 be amplified as much as possible without exceeding the maximum signal
3667 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3668 Normalizer can also take into account the frame's root mean square,
3669 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3670 determine the power of a time-varying signal. It is therefore considered
3671 that the RMS is a better approximation of the "perceived loudness" than
3672 just looking at the signal's peak magnitude. Consequently, by adjusting all
3673 frames to a constant RMS value, a uniform "perceived loudness" can be
3674 established. If a target RMS value has been specified, a frame's local gain
3675 factor is defined as the factor that would result in exactly that RMS value.
3676 Note, however, that the maximum local gain factor is still restricted by the
3677 frame's highest magnitude sample, in order to prevent clipping.
3680 Enable channels coupling. By default is enabled.
3681 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3682 amount. This means the same gain factor will be applied to all channels, i.e.
3683 the maximum possible gain factor is determined by the "loudest" channel.
3684 However, in some recordings, it may happen that the volume of the different
3685 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3686 In this case, this option can be used to disable the channel coupling. This way,
3687 the gain factor will be determined independently for each channel, depending
3688 only on the individual channel's highest magnitude sample. This allows for
3689 harmonizing the volume of the different channels.
3692 Enable DC bias correction. By default is disabled.
3693 An audio signal (in the time domain) is a sequence of sample values.
3694 In the Dynamic Audio Normalizer these sample values are represented in the
3695 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3696 audio signal, or "waveform", should be centered around the zero point.
3697 That means if we calculate the mean value of all samples in a file, or in a
3698 single frame, then the result should be 0.0 or at least very close to that
3699 value. If, however, there is a significant deviation of the mean value from
3700 0.0, in either positive or negative direction, this is referred to as a
3701 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3702 Audio Normalizer provides optional DC bias correction.
3703 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3704 the mean value, or "DC correction" offset, of each input frame and subtract
3705 that value from all of the frame's sample values which ensures those samples
3706 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3707 boundaries, the DC correction offset values will be interpolated smoothly
3708 between neighbouring frames.
3710 @item altboundary, b
3711 Enable alternative boundary mode. By default is disabled.
3712 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3713 around each frame. This includes the preceding frames as well as the
3714 subsequent frames. However, for the "boundary" frames, located at the very
3715 beginning and at the very end of the audio file, not all neighbouring
3716 frames are available. In particular, for the first few frames in the audio
3717 file, the preceding frames are not known. And, similarly, for the last few
3718 frames in the audio file, the subsequent frames are not known. Thus, the
3719 question arises which gain factors should be assumed for the missing frames
3720 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3721 to deal with this situation. The default boundary mode assumes a gain factor
3722 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3723 "fade out" at the beginning and at the end of the input, respectively.
3726 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3727 By default, the Dynamic Audio Normalizer does not apply "traditional"
3728 compression. This means that signal peaks will not be pruned and thus the
3729 full dynamic range will be retained within each local neighbourhood. However,
3730 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3731 normalization algorithm with a more "traditional" compression.
3732 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3733 (thresholding) function. If (and only if) the compression feature is enabled,
3734 all input frames will be processed by a soft knee thresholding function prior
3735 to the actual normalization process. Put simply, the thresholding function is
3736 going to prune all samples whose magnitude exceeds a certain threshold value.
3737 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3738 value. Instead, the threshold value will be adjusted for each individual
3740 In general, smaller parameters result in stronger compression, and vice versa.
3741 Values below 3.0 are not recommended, because audible distortion may appear.
3744 Set the target threshold value. This specifies the lowest permissible
3745 magnitude level for the audio input which will be normalized.
3746 If input frame volume is above this value frame will be normalized.
3747 Otherwise frame may not be normalized at all. The default value is set
3748 to 0, which means all input frames will be normalized.
3749 This option is mostly useful if digital noise is not wanted to be amplified.
3752 @subsection Commands
3754 This filter supports the all above options as @ref{commands}.
3758 Make audio easier to listen to on headphones.
3760 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3761 so that when listened to on headphones the stereo image is moved from
3762 inside your head (standard for headphones) to outside and in front of
3763 the listener (standard for speakers).
3769 Apply a two-pole peaking equalisation (EQ) filter. With this
3770 filter, the signal-level at and around a selected frequency can
3771 be increased or decreased, whilst (unlike bandpass and bandreject
3772 filters) that at all other frequencies is unchanged.
3774 In order to produce complex equalisation curves, this filter can
3775 be given several times, each with a different central frequency.
3777 The filter accepts the following options:
3781 Set the filter's central frequency in Hz.
3784 Set method to specify band-width of filter.
3799 Specify the band-width of a filter in width_type units.
3802 Set the required gain or attenuation in dB.
3803 Beware of clipping when using a positive gain.
3806 How much to use filtered signal in output. Default is 1.
3807 Range is between 0 and 1.
3810 Specify which channels to filter, by default all available are filtered.
3813 Normalize biquad coefficients, by default is disabled.
3814 Enabling it will normalize magnitude response at DC to 0dB.
3817 Set transform type of IIR filter.
3826 @subsection Examples
3829 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3831 equalizer=f=1000:t=h:width=200:g=-10
3835 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3837 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3841 @subsection Commands
3843 This filter supports the following commands:
3846 Change equalizer frequency.
3847 Syntax for the command is : "@var{frequency}"
3850 Change equalizer width_type.
3851 Syntax for the command is : "@var{width_type}"
3854 Change equalizer width.
3855 Syntax for the command is : "@var{width}"
3858 Change equalizer gain.
3859 Syntax for the command is : "@var{gain}"
3862 Change equalizer mix.
3863 Syntax for the command is : "@var{mix}"
3866 @section extrastereo
3868 Linearly increases the difference between left and right channels which
3869 adds some sort of "live" effect to playback.
3871 The filter accepts the following options:
3875 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3876 (average of both channels), with 1.0 sound will be unchanged, with
3877 -1.0 left and right channels will be swapped.
3880 Enable clipping. By default is enabled.
3883 @subsection Commands
3885 This filter supports the all above options as @ref{commands}.
3887 @section firequalizer
3888 Apply FIR Equalization using arbitrary frequency response.
3890 The filter accepts the following option:
3894 Set gain curve equation (in dB). The expression can contain variables:
3897 the evaluated frequency
3901 channel number, set to 0 when multichannels evaluation is disabled
3903 channel id, see libavutil/channel_layout.h, set to the first channel id when
3904 multichannels evaluation is disabled
3908 channel_layout, see libavutil/channel_layout.h
3913 @item gain_interpolate(f)
3914 interpolate gain on frequency f based on gain_entry
3915 @item cubic_interpolate(f)
3916 same as gain_interpolate, but smoother
3918 This option is also available as command. Default is @code{gain_interpolate(f)}.
3921 Set gain entry for gain_interpolate function. The expression can
3925 store gain entry at frequency f with value g
3927 This option is also available as command.
3930 Set filter delay in seconds. Higher value means more accurate.
3931 Default is @code{0.01}.
3934 Set filter accuracy in Hz. Lower value means more accurate.
3935 Default is @code{5}.
3938 Set window function. Acceptable values are:
3941 rectangular window, useful when gain curve is already smooth
3943 hann window (default)
3949 3-terms continuous 1st derivative nuttall window
3951 minimum 3-terms discontinuous nuttall window
3953 4-terms continuous 1st derivative nuttall window
3955 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3957 blackman-harris window
3963 If enabled, use fixed number of audio samples. This improves speed when
3964 filtering with large delay. Default is disabled.
3967 Enable multichannels evaluation on gain. Default is disabled.
3970 Enable zero phase mode by subtracting timestamp to compensate delay.
3971 Default is disabled.
3974 Set scale used by gain. Acceptable values are:
3977 linear frequency, linear gain
3979 linear frequency, logarithmic (in dB) gain (default)
3981 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3983 logarithmic frequency, logarithmic gain
3987 Set file for dumping, suitable for gnuplot.
3990 Set scale for dumpfile. Acceptable values are same with scale option.
3994 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3995 Default is disabled.
3998 Enable minimum phase impulse response. Default is disabled.
4001 @subsection Examples
4006 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4009 lowpass at 1000 Hz with gain_entry:
4011 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4014 custom equalization:
4016 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4019 higher delay with zero phase to compensate delay:
4021 firequalizer=delay=0.1:fixed=on:zero_phase=on
4024 lowpass on left channel, highpass on right channel:
4026 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4027 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4032 Apply a flanging effect to the audio.
4034 The filter accepts the following options:
4038 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4041 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4044 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4048 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4049 Default value is 71.
4052 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4055 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4056 Default value is @var{sinusoidal}.
4059 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4060 Default value is 25.
4063 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4064 Default is @var{linear}.
4068 Apply Haas effect to audio.
4070 Note that this makes most sense to apply on mono signals.
4071 With this filter applied to mono signals it give some directionality and
4072 stretches its stereo image.
4074 The filter accepts the following options:
4078 Set input level. By default is @var{1}, or 0dB
4081 Set output level. By default is @var{1}, or 0dB.
4084 Set gain applied to side part of signal. By default is @var{1}.
4087 Set kind of middle source. Can be one of the following:
4097 Pick middle part signal of stereo image.
4100 Pick side part signal of stereo image.
4104 Change middle phase. By default is disabled.
4107 Set left channel delay. By default is @var{2.05} milliseconds.
4110 Set left channel balance. By default is @var{-1}.
4113 Set left channel gain. By default is @var{1}.
4116 Change left phase. By default is disabled.
4119 Set right channel delay. By defaults is @var{2.12} milliseconds.
4122 Set right channel balance. By default is @var{1}.
4125 Set right channel gain. By default is @var{1}.
4128 Change right phase. By default is enabled.
4133 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4134 embedded HDCD codes is expanded into a 20-bit PCM stream.
4136 The filter supports the Peak Extend and Low-level Gain Adjustment features
4137 of HDCD, and detects the Transient Filter flag.
4140 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4143 When using the filter with wav, note the default encoding for wav is 16-bit,
4144 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4145 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4147 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4148 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4151 The filter accepts the following options:
4154 @item disable_autoconvert
4155 Disable any automatic format conversion or resampling in the filter graph.
4157 @item process_stereo
4158 Process the stereo channels together. If target_gain does not match between
4159 channels, consider it invalid and use the last valid target_gain.
4162 Set the code detect timer period in ms.
4165 Always extend peaks above -3dBFS even if PE isn't signaled.
4168 Replace audio with a solid tone and adjust the amplitude to signal some
4169 specific aspect of the decoding process. The output file can be loaded in
4170 an audio editor alongside the original to aid analysis.
4172 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4179 Gain adjustment level at each sample
4181 Samples where peak extend occurs
4183 Samples where the code detect timer is active
4185 Samples where the target gain does not match between channels
4191 Apply head-related transfer functions (HRTFs) to create virtual
4192 loudspeakers around the user for binaural listening via headphones.
4193 The HRIRs are provided via additional streams, for each channel
4194 one stereo input stream is needed.
4196 The filter accepts the following options:
4200 Set mapping of input streams for convolution.
4201 The argument is a '|'-separated list of channel names in order as they
4202 are given as additional stream inputs for filter.
4203 This also specify number of input streams. Number of input streams
4204 must be not less than number of channels in first stream plus one.
4207 Set gain applied to audio. Value is in dB. Default is 0.
4210 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4211 processing audio in time domain which is slow.
4212 @var{freq} is processing audio in frequency domain which is fast.
4213 Default is @var{freq}.
4216 Set custom gain for LFE channels. Value is in dB. Default is 0.
4219 Set size of frame in number of samples which will be processed at once.
4220 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4223 Set format of hrir stream.
4224 Default value is @var{stereo}. Alternative value is @var{multich}.
4225 If value is set to @var{stereo}, number of additional streams should
4226 be greater or equal to number of input channels in first input stream.
4227 Also each additional stream should have stereo number of channels.
4228 If value is set to @var{multich}, number of additional streams should
4229 be exactly one. Also number of input channels of additional stream
4230 should be equal or greater than twice number of channels of first input
4234 @subsection Examples
4238 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4239 each amovie filter use stereo file with IR coefficients as input.
4240 The files give coefficients for each position of virtual loudspeaker:
4243 -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"
4248 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4249 but now in @var{multich} @var{hrir} format.
4251 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"
4258 Apply a high-pass filter with 3dB point frequency.
4259 The filter can be either single-pole, or double-pole (the default).
4260 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4262 The filter accepts the following options:
4266 Set frequency in Hz. Default is 3000.
4269 Set number of poles. Default is 2.
4272 Set method to specify band-width of filter.
4287 Specify the band-width of a filter in width_type units.
4288 Applies only to double-pole filter.
4289 The default is 0.707q and gives a Butterworth response.
4292 How much to use filtered signal in output. Default is 1.
4293 Range is between 0 and 1.
4296 Specify which channels to filter, by default all available are filtered.
4299 Normalize biquad coefficients, by default is disabled.
4300 Enabling it will normalize magnitude response at DC to 0dB.
4303 Set transform type of IIR filter.
4312 @subsection Commands
4314 This filter supports the following commands:
4317 Change highpass frequency.
4318 Syntax for the command is : "@var{frequency}"
4321 Change highpass width_type.
4322 Syntax for the command is : "@var{width_type}"
4325 Change highpass width.
4326 Syntax for the command is : "@var{width}"
4329 Change highpass mix.
4330 Syntax for the command is : "@var{mix}"
4335 Join multiple input streams into one multi-channel stream.
4337 It accepts the following parameters:
4341 The number of input streams. It defaults to 2.
4343 @item channel_layout
4344 The desired output channel layout. It defaults to stereo.
4347 Map channels from inputs to output. The argument is a '|'-separated list of
4348 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4349 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4350 can be either the name of the input channel (e.g. FL for front left) or its
4351 index in the specified input stream. @var{out_channel} is the name of the output
4355 The filter will attempt to guess the mappings when they are not specified
4356 explicitly. It does so by first trying to find an unused matching input channel
4357 and if that fails it picks the first unused input channel.
4359 Join 3 inputs (with properly set channel layouts):
4361 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4364 Build a 5.1 output from 6 single-channel streams:
4366 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4367 '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'
4373 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4375 To enable compilation of this filter you need to configure FFmpeg with
4376 @code{--enable-ladspa}.
4380 Specifies the name of LADSPA plugin library to load. If the environment
4381 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4382 each one of the directories specified by the colon separated list in
4383 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4384 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4385 @file{/usr/lib/ladspa/}.
4388 Specifies the plugin within the library. Some libraries contain only
4389 one plugin, but others contain many of them. If this is not set filter
4390 will list all available plugins within the specified library.
4393 Set the '|' separated list of controls which are zero or more floating point
4394 values that determine the behavior of the loaded plugin (for example delay,
4396 Controls need to be defined using the following syntax:
4397 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4398 @var{valuei} is the value set on the @var{i}-th control.
4399 Alternatively they can be also defined using the following syntax:
4400 @var{value0}|@var{value1}|@var{value2}|..., where
4401 @var{valuei} is the value set on the @var{i}-th control.
4402 If @option{controls} is set to @code{help}, all available controls and
4403 their valid ranges are printed.
4405 @item sample_rate, s
4406 Specify the sample rate, default to 44100. Only used if plugin have
4410 Set the number of samples per channel per each output frame, default
4411 is 1024. Only used if plugin have zero inputs.
4414 Set the minimum duration of the sourced audio. See
4415 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4416 for the accepted syntax.
4417 Note that the resulting duration may be greater than the specified duration,
4418 as the generated audio is always cut at the end of a complete frame.
4419 If not specified, or the expressed duration is negative, the audio is
4420 supposed to be generated forever.
4421 Only used if plugin have zero inputs.
4424 Enable latency compensation, by default is disabled.
4425 Only used if plugin have inputs.
4428 @subsection Examples
4432 List all available plugins within amp (LADSPA example plugin) library:
4438 List all available controls and their valid ranges for @code{vcf_notch}
4439 plugin from @code{VCF} library:
4441 ladspa=f=vcf:p=vcf_notch:c=help
4445 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4448 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4452 Add reverberation to the audio using TAP-plugins
4453 (Tom's Audio Processing plugins):
4455 ladspa=file=tap_reverb:tap_reverb
4459 Generate white noise, with 0.2 amplitude:
4461 ladspa=file=cmt:noise_source_white:c=c0=.2
4465 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4466 @code{C* Audio Plugin Suite} (CAPS) library:
4468 ladspa=file=caps:Click:c=c1=20'
4472 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4474 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4478 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4479 @code{SWH Plugins} collection:
4481 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4485 Attenuate low frequencies using Multiband EQ from Steve Harris
4486 @code{SWH Plugins} collection:
4488 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4492 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4495 ladspa=caps:Narrower
4499 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4501 ladspa=caps:White:.2
4505 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4507 ladspa=caps:Fractal:c=c1=1
4511 Dynamic volume normalization using @code{VLevel} plugin:
4513 ladspa=vlevel-ladspa:vlevel_mono
4517 @subsection Commands
4519 This filter supports the following commands:
4522 Modify the @var{N}-th control value.
4524 If the specified value is not valid, it is ignored and prior one is kept.
4529 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4530 Support for both single pass (livestreams, files) and double pass (files) modes.
4531 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4532 detect true peaks, the audio stream will be upsampled to 192 kHz.
4533 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4535 The filter accepts the following options:
4539 Set integrated loudness target.
4540 Range is -70.0 - -5.0. Default value is -24.0.
4543 Set loudness range target.
4544 Range is 1.0 - 20.0. Default value is 7.0.
4547 Set maximum true peak.
4548 Range is -9.0 - +0.0. Default value is -2.0.
4550 @item measured_I, measured_i
4551 Measured IL of input file.
4552 Range is -99.0 - +0.0.
4554 @item measured_LRA, measured_lra
4555 Measured LRA of input file.
4556 Range is 0.0 - 99.0.
4558 @item measured_TP, measured_tp
4559 Measured true peak of input file.
4560 Range is -99.0 - +99.0.
4562 @item measured_thresh
4563 Measured threshold of input file.
4564 Range is -99.0 - +0.0.
4567 Set offset gain. Gain is applied before the true-peak limiter.
4568 Range is -99.0 - +99.0. Default is +0.0.
4571 Normalize by linearly scaling the source audio.
4572 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4573 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4574 be lower than source LRA and the change in integrated loudness shouldn't
4575 result in a true peak which exceeds the target TP. If any of these
4576 conditions aren't met, normalization mode will revert to @var{dynamic}.
4577 Options are @code{true} or @code{false}. Default is @code{true}.
4580 Treat mono input files as "dual-mono". If a mono file is intended for playback
4581 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4582 If set to @code{true}, this option will compensate for this effect.
4583 Multi-channel input files are not affected by this option.
4584 Options are true or false. Default is false.
4587 Set print format for stats. Options are summary, json, or none.
4588 Default value is none.
4593 Apply a low-pass filter with 3dB point frequency.
4594 The filter can be either single-pole or double-pole (the default).
4595 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4597 The filter accepts the following options:
4601 Set frequency in Hz. Default is 500.
4604 Set number of poles. Default is 2.
4607 Set method to specify band-width of filter.
4622 Specify the band-width of a filter in width_type units.
4623 Applies only to double-pole filter.
4624 The default is 0.707q and gives a Butterworth response.
4627 How much to use filtered signal in output. Default is 1.
4628 Range is between 0 and 1.
4631 Specify which channels to filter, by default all available are filtered.
4634 Normalize biquad coefficients, by default is disabled.
4635 Enabling it will normalize magnitude response at DC to 0dB.
4638 Set transform type of IIR filter.
4647 @subsection Examples
4650 Lowpass only LFE channel, it LFE is not present it does nothing:
4656 @subsection Commands
4658 This filter supports the following commands:
4661 Change lowpass frequency.
4662 Syntax for the command is : "@var{frequency}"
4665 Change lowpass width_type.
4666 Syntax for the command is : "@var{width_type}"
4669 Change lowpass width.
4670 Syntax for the command is : "@var{width}"
4674 Syntax for the command is : "@var{mix}"
4679 Load a LV2 (LADSPA Version 2) plugin.
4681 To enable compilation of this filter you need to configure FFmpeg with
4682 @code{--enable-lv2}.
4686 Specifies the plugin URI. You may need to escape ':'.
4689 Set the '|' separated list of controls which are zero or more floating point
4690 values that determine the behavior of the loaded plugin (for example delay,
4692 If @option{controls} is set to @code{help}, all available controls and
4693 their valid ranges are printed.
4695 @item sample_rate, s
4696 Specify the sample rate, default to 44100. Only used if plugin have
4700 Set the number of samples per channel per each output frame, default
4701 is 1024. Only used if plugin have zero inputs.
4704 Set the minimum duration of the sourced audio. See
4705 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4706 for the accepted syntax.
4707 Note that the resulting duration may be greater than the specified duration,
4708 as the generated audio is always cut at the end of a complete frame.
4709 If not specified, or the expressed duration is negative, the audio is
4710 supposed to be generated forever.
4711 Only used if plugin have zero inputs.
4714 @subsection Examples
4718 Apply bass enhancer plugin from Calf:
4720 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4724 Apply vinyl plugin from Calf:
4726 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4730 Apply bit crusher plugin from ArtyFX:
4732 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4737 Multiband Compress or expand the audio's dynamic range.
4739 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4740 This is akin to the crossover of a loudspeaker, and results in flat frequency
4741 response when absent compander action.
4743 It accepts the following parameters:
4747 This option syntax is:
4748 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4749 For explanation of each item refer to compand filter documentation.
4755 Mix channels with specific gain levels. The filter accepts the output
4756 channel layout followed by a set of channels definitions.
4758 This filter is also designed to efficiently remap the channels of an audio
4761 The filter accepts parameters of the form:
4762 "@var{l}|@var{outdef}|@var{outdef}|..."
4766 output channel layout or number of channels
4769 output channel specification, of the form:
4770 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4773 output channel to define, either a channel name (FL, FR, etc.) or a channel
4774 number (c0, c1, etc.)
4777 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4780 input channel to use, see out_name for details; it is not possible to mix
4781 named and numbered input channels
4784 If the `=' in a channel specification is replaced by `<', then the gains for
4785 that specification will be renormalized so that the total is 1, thus
4786 avoiding clipping noise.
4788 @subsection Mixing examples
4790 For example, if you want to down-mix from stereo to mono, but with a bigger
4791 factor for the left channel:
4793 pan=1c|c0=0.9*c0+0.1*c1
4796 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4797 7-channels surround:
4799 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4802 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4803 that should be preferred (see "-ac" option) unless you have very specific
4806 @subsection Remapping examples
4808 The channel remapping will be effective if, and only if:
4811 @item gain coefficients are zeroes or ones,
4812 @item only one input per channel output,
4815 If all these conditions are satisfied, the filter will notify the user ("Pure
4816 channel mapping detected"), and use an optimized and lossless method to do the
4819 For example, if you have a 5.1 source and want a stereo audio stream by
4820 dropping the extra channels:
4822 pan="stereo| c0=FL | c1=FR"
4825 Given the same source, you can also switch front left and front right channels
4826 and keep the input channel layout:
4828 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4831 If the input is a stereo audio stream, you can mute the front left channel (and
4832 still keep the stereo channel layout) with:
4837 Still with a stereo audio stream input, you can copy the right channel in both
4838 front left and right:
4840 pan="stereo| c0=FR | c1=FR"
4845 ReplayGain scanner filter. This filter takes an audio stream as an input and
4846 outputs it unchanged.
4847 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4851 Convert the audio sample format, sample rate and channel layout. It is
4852 not meant to be used directly.
4855 Apply time-stretching and pitch-shifting with librubberband.
4857 To enable compilation of this filter, you need to configure FFmpeg with
4858 @code{--enable-librubberband}.
4860 The filter accepts the following options:
4864 Set tempo scale factor.
4867 Set pitch scale factor.
4870 Set transients detector.
4871 Possible values are:
4880 Possible values are:
4889 Possible values are:
4896 Set processing window size.
4897 Possible values are:
4906 Possible values are:
4913 Enable formant preservation when shift pitching.
4914 Possible values are:
4922 Possible values are:
4931 Possible values are:
4938 @subsection Commands
4940 This filter supports the following commands:
4943 Change filter tempo scale factor.
4944 Syntax for the command is : "@var{tempo}"
4947 Change filter pitch scale factor.
4948 Syntax for the command is : "@var{pitch}"
4951 @section sidechaincompress
4953 This filter acts like normal compressor but has the ability to compress
4954 detected signal using second input signal.
4955 It needs two input streams and returns one output stream.
4956 First input stream will be processed depending on second stream signal.
4957 The filtered signal then can be filtered with other filters in later stages of
4958 processing. See @ref{pan} and @ref{amerge} filter.
4960 The filter accepts the following options:
4964 Set input gain. Default is 1. Range is between 0.015625 and 64.
4967 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4968 Default is @code{downward}.
4971 If a signal of second stream raises above this level it will affect the gain
4972 reduction of first stream.
4973 By default is 0.125. Range is between 0.00097563 and 1.
4976 Set a ratio about which the signal is reduced. 1:2 means that if the level
4977 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4978 Default is 2. Range is between 1 and 20.
4981 Amount of milliseconds the signal has to rise above the threshold before gain
4982 reduction starts. Default is 20. Range is between 0.01 and 2000.
4985 Amount of milliseconds the signal has to fall below the threshold before
4986 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4989 Set the amount by how much signal will be amplified after processing.
4990 Default is 1. Range is from 1 to 64.
4993 Curve the sharp knee around the threshold to enter gain reduction more softly.
4994 Default is 2.82843. Range is between 1 and 8.
4997 Choose if the @code{average} level between all channels of side-chain stream
4998 or the louder(@code{maximum}) channel of side-chain stream affects the
4999 reduction. Default is @code{average}.
5002 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5003 of @code{rms}. Default is @code{rms} which is mainly smoother.
5006 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5009 How much to use compressed signal in output. Default is 1.
5010 Range is between 0 and 1.
5013 @subsection Commands
5015 This filter supports the all above options as @ref{commands}.
5017 @subsection Examples
5021 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5022 depending on the signal of 2nd input and later compressed signal to be
5023 merged with 2nd input:
5025 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5029 @section sidechaingate
5031 A sidechain gate acts like a normal (wideband) gate but has the ability to
5032 filter the detected signal before sending it to the gain reduction stage.
5033 Normally a gate uses the full range signal to detect a level above the
5035 For example: If you cut all lower frequencies from your sidechain signal
5036 the gate will decrease the volume of your track only if not enough highs
5037 appear. With this technique you are able to reduce the resonation of a
5038 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5040 It needs two input streams and returns one output stream.
5041 First input stream will be processed depending on second stream signal.
5043 The filter accepts the following options:
5047 Set input level before filtering.
5048 Default is 1. Allowed range is from 0.015625 to 64.
5051 Set the mode of operation. Can be @code{upward} or @code{downward}.
5052 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5053 will be amplified, expanding dynamic range in upward direction.
5054 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5057 Set the level of gain reduction when the signal is below the threshold.
5058 Default is 0.06125. Allowed range is from 0 to 1.
5059 Setting this to 0 disables reduction and then filter behaves like expander.
5062 If a signal rises above this level the gain reduction is released.
5063 Default is 0.125. Allowed range is from 0 to 1.
5066 Set a ratio about which the signal is reduced.
5067 Default is 2. Allowed range is from 1 to 9000.
5070 Amount of milliseconds the signal has to rise above the threshold before gain
5072 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5075 Amount of milliseconds the signal has to fall below the threshold before the
5076 reduction is increased again. Default is 250 milliseconds.
5077 Allowed range is from 0.01 to 9000.
5080 Set amount of amplification of signal after processing.
5081 Default is 1. Allowed range is from 1 to 64.
5084 Curve the sharp knee around the threshold to enter gain reduction more softly.
5085 Default is 2.828427125. Allowed range is from 1 to 8.
5088 Choose if exact signal should be taken for detection or an RMS like one.
5089 Default is rms. Can be peak or rms.
5092 Choose if the average level between all channels or the louder channel affects
5094 Default is average. Can be average or maximum.
5097 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5100 @subsection Commands
5102 This filter supports the all above options as @ref{commands}.
5104 @section silencedetect
5106 Detect silence in an audio stream.
5108 This filter logs a message when it detects that the input audio volume is less
5109 or equal to a noise tolerance value for a duration greater or equal to the
5110 minimum detected noise duration.
5112 The printed times and duration are expressed in seconds. The
5113 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5114 is set on the first frame whose timestamp equals or exceeds the detection
5115 duration and it contains the timestamp of the first frame of the silence.
5117 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5118 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5119 keys are set on the first frame after the silence. If @option{mono} is
5120 enabled, and each channel is evaluated separately, the @code{.X}
5121 suffixed keys are used, and @code{X} corresponds to the channel number.
5123 The filter accepts the following options:
5127 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5128 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5131 Set silence duration until notification (default is 2 seconds). See
5132 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5133 for the accepted syntax.
5136 Process each channel separately, instead of combined. By default is disabled.
5139 @subsection Examples
5143 Detect 5 seconds of silence with -50dB noise tolerance:
5145 silencedetect=n=-50dB:d=5
5149 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5150 tolerance in @file{silence.mp3}:
5152 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5156 @section silenceremove
5158 Remove silence from the beginning, middle or end of the audio.
5160 The filter accepts the following options:
5164 This value is used to indicate if audio should be trimmed at beginning of
5165 the audio. A value of zero indicates no silence should be trimmed from the
5166 beginning. When specifying a non-zero value, it trims audio up until it
5167 finds non-silence. Normally, when trimming silence from beginning of audio
5168 the @var{start_periods} will be @code{1} but it can be increased to higher
5169 values to trim all audio up to specific count of non-silence periods.
5170 Default value is @code{0}.
5172 @item start_duration
5173 Specify the amount of time that non-silence must be detected before it stops
5174 trimming audio. By increasing the duration, bursts of noises can be treated
5175 as silence and trimmed off. Default value is @code{0}.
5177 @item start_threshold
5178 This indicates what sample value should be treated as silence. For digital
5179 audio, a value of @code{0} may be fine but for audio recorded from analog,
5180 you may wish to increase the value to account for background noise.
5181 Can be specified in dB (in case "dB" is appended to the specified value)
5182 or amplitude ratio. Default value is @code{0}.
5185 Specify max duration of silence at beginning that will be kept after
5186 trimming. Default is 0, which is equal to trimming all samples detected
5190 Specify mode of detection of silence end in start of multi-channel audio.
5191 Can be @var{any} or @var{all}. Default is @var{any}.
5192 With @var{any}, any sample that is detected as non-silence will cause
5193 stopped trimming of silence.
5194 With @var{all}, only if all channels are detected as non-silence will cause
5195 stopped trimming of silence.
5198 Set the count for trimming silence from the end of audio.
5199 To remove silence from the middle of a file, specify a @var{stop_periods}
5200 that is negative. This value is then treated as a positive value and is
5201 used to indicate the effect should restart processing as specified by
5202 @var{start_periods}, making it suitable for removing periods of silence
5203 in the middle of the audio.
5204 Default value is @code{0}.
5207 Specify a duration of silence that must exist before audio is not copied any
5208 more. By specifying a higher duration, silence that is wanted can be left in
5210 Default value is @code{0}.
5212 @item stop_threshold
5213 This is the same as @option{start_threshold} but for trimming silence from
5215 Can be specified in dB (in case "dB" is appended to the specified value)
5216 or amplitude ratio. Default value is @code{0}.
5219 Specify max duration of silence at end that will be kept after
5220 trimming. Default is 0, which is equal to trimming all samples detected
5224 Specify mode of detection of silence start in end of multi-channel audio.
5225 Can be @var{any} or @var{all}. Default is @var{any}.
5226 With @var{any}, any sample that is detected as non-silence will cause
5227 stopped trimming of silence.
5228 With @var{all}, only if all channels are detected as non-silence will cause
5229 stopped trimming of silence.
5232 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5233 and works better with digital silence which is exactly 0.
5234 Default value is @code{rms}.
5237 Set duration in number of seconds used to calculate size of window in number
5238 of samples for detecting silence.
5239 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5242 @subsection Examples
5246 The following example shows how this filter can be used to start a recording
5247 that does not contain the delay at the start which usually occurs between
5248 pressing the record button and the start of the performance:
5250 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5254 Trim all silence encountered from beginning to end where there is more than 1
5255 second of silence in audio:
5257 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5261 Trim all digital silence samples, using peak detection, from beginning to end
5262 where there is more than 0 samples of digital silence in audio and digital
5263 silence is detected in all channels at same positions in stream:
5265 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5271 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5272 loudspeakers around the user for binaural listening via headphones (audio
5273 formats up to 9 channels supported).
5274 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5275 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5276 Austrian Academy of Sciences.
5278 To enable compilation of this filter you need to configure FFmpeg with
5279 @code{--enable-libmysofa}.
5281 The filter accepts the following options:
5285 Set the SOFA file used for rendering.
5288 Set gain applied to audio. Value is in dB. Default is 0.
5291 Set rotation of virtual loudspeakers in deg. Default is 0.
5294 Set elevation of virtual speakers in deg. Default is 0.
5297 Set distance in meters between loudspeakers and the listener with near-field
5298 HRTFs. Default is 1.
5301 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5302 processing audio in time domain which is slow.
5303 @var{freq} is processing audio in frequency domain which is fast.
5304 Default is @var{freq}.
5307 Set custom positions of virtual loudspeakers. Syntax for this option is:
5308 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5309 Each virtual loudspeaker is described with short channel name following with
5310 azimuth and elevation in degrees.
5311 Each virtual loudspeaker description is separated by '|'.
5312 For example to override front left and front right channel positions use:
5313 'speakers=FL 45 15|FR 345 15'.
5314 Descriptions with unrecognised channel names are ignored.
5317 Set custom gain for LFE channels. Value is in dB. Default is 0.
5320 Set custom frame size in number of samples. Default is 1024.
5321 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5322 is set to @var{freq}.
5325 Should all IRs be normalized upon importing SOFA file.
5326 By default is enabled.
5329 Should nearest IRs be interpolated with neighbor IRs if exact position
5330 does not match. By default is disabled.
5333 Minphase all IRs upon loading of SOFA file. By default is disabled.
5336 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5339 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5342 @subsection Examples
5346 Using ClubFritz6 sofa file:
5348 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5352 Using ClubFritz12 sofa file and bigger radius with small rotation:
5354 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5358 Similar as above but with custom speaker positions for front left, front right, back left and back right
5359 and also with custom gain:
5361 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5368 This filter expands or compresses each half-cycle of audio samples
5369 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5370 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5372 The filter accepts the following options:
5376 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5377 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5380 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5381 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5382 would be such that local peak value reaches target peak value but never to surpass it and that
5383 ratio between new and previous peak value does not surpass this option value.
5385 @item compression, c
5386 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5387 This option controls maximum local half-cycle of samples compression. This option is used
5388 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5389 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5390 that peak's half-cycle will be compressed by current compression factor.
5393 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5394 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5395 Any half-cycle samples with their local peak value below or same as this option value will be
5396 compressed by current compression factor, otherwise, if greater than threshold value they will be
5397 expanded with expansion factor so that it could reach peak target value but never surpass it.
5400 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5401 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5402 each new half-cycle until it reaches @option{expansion} value.
5403 Setting this options too high may lead to distortions.
5406 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5407 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5408 each new half-cycle until it reaches @option{compression} value.
5411 Specify which channels to filter, by default all available channels are filtered.
5414 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5415 option. When enabled any half-cycle of samples with their local peak value below or same as
5416 @option{threshold} option will be expanded otherwise it will be compressed.
5419 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5420 When disabled each filtered channel gain calculation is independent, otherwise when this option
5421 is enabled the minimum of all possible gains for each filtered channel is used.
5424 @subsection Commands
5426 This filter supports the all above options as @ref{commands}.
5428 @section stereotools
5430 This filter has some handy utilities to manage stereo signals, for converting
5431 M/S stereo recordings to L/R signal while having control over the parameters
5432 or spreading the stereo image of master track.
5434 The filter accepts the following options:
5438 Set input level before filtering for both channels. Defaults is 1.
5439 Allowed range is from 0.015625 to 64.
5442 Set output level after filtering for both channels. Defaults is 1.
5443 Allowed range is from 0.015625 to 64.
5446 Set input balance between both channels. Default is 0.
5447 Allowed range is from -1 to 1.
5450 Set output balance between both channels. Default is 0.
5451 Allowed range is from -1 to 1.
5454 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5455 clipping. Disabled by default.
5458 Mute the left channel. Disabled by default.
5461 Mute the right channel. Disabled by default.
5464 Change the phase of the left channel. Disabled by default.
5467 Change the phase of the right channel. Disabled by default.
5470 Set stereo mode. Available values are:
5474 Left/Right to Left/Right, this is default.
5477 Left/Right to Mid/Side.
5480 Mid/Side to Left/Right.
5483 Left/Right to Left/Left.
5486 Left/Right to Right/Right.
5489 Left/Right to Left + Right.
5492 Left/Right to Right/Left.
5495 Mid/Side to Left/Left.
5498 Mid/Side to Right/Right.
5502 Set level of side signal. Default is 1.
5503 Allowed range is from 0.015625 to 64.
5506 Set balance of side signal. Default is 0.
5507 Allowed range is from -1 to 1.
5510 Set level of the middle signal. Default is 1.
5511 Allowed range is from 0.015625 to 64.
5514 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5517 Set stereo base between mono and inversed channels. Default is 0.
5518 Allowed range is from -1 to 1.
5521 Set delay in milliseconds how much to delay left from right channel and
5522 vice versa. Default is 0. Allowed range is from -20 to 20.
5525 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5528 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5530 @item bmode_in, bmode_out
5531 Set balance mode for balance_in/balance_out option.
5533 Can be one of the following:
5537 Classic balance mode. Attenuate one channel at time.
5538 Gain is raised up to 1.
5541 Similar as classic mode above but gain is raised up to 2.
5544 Equal power distribution, from -6dB to +6dB range.
5548 @subsection Examples
5552 Apply karaoke like effect:
5554 stereotools=mlev=0.015625
5558 Convert M/S signal to L/R:
5560 "stereotools=mode=ms>lr"
5564 @section stereowiden
5566 This filter enhance the stereo effect by suppressing signal common to both
5567 channels and by delaying the signal of left into right and vice versa,
5568 thereby widening the stereo effect.
5570 The filter accepts the following options:
5574 Time in milliseconds of the delay of left signal into right and vice versa.
5575 Default is 20 milliseconds.
5578 Amount of gain in delayed signal into right and vice versa. Gives a delay
5579 effect of left signal in right output and vice versa which gives widening
5580 effect. Default is 0.3.
5583 Cross feed of left into right with inverted phase. This helps in suppressing
5584 the mono. If the value is 1 it will cancel all the signal common to both
5585 channels. Default is 0.3.
5588 Set level of input signal of original channel. Default is 0.8.
5591 @subsection Commands
5593 This filter supports the all above options except @code{delay} as @ref{commands}.
5595 @section superequalizer
5596 Apply 18 band equalizer.
5598 The filter accepts the following options:
5605 Set 131Hz band gain.
5607 Set 185Hz band gain.
5609 Set 262Hz band gain.
5611 Set 370Hz band gain.
5613 Set 523Hz band gain.
5615 Set 740Hz band gain.
5617 Set 1047Hz band gain.
5619 Set 1480Hz band gain.
5621 Set 2093Hz band gain.
5623 Set 2960Hz band gain.
5625 Set 4186Hz band gain.
5627 Set 5920Hz band gain.
5629 Set 8372Hz band gain.
5631 Set 11840Hz band gain.
5633 Set 16744Hz band gain.
5635 Set 20000Hz band gain.
5639 Apply audio surround upmix filter.
5641 This filter allows to produce multichannel output from audio stream.
5643 The filter accepts the following options:
5647 Set output channel layout. By default, this is @var{5.1}.
5649 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5650 for the required syntax.
5653 Set input channel layout. By default, this is @var{stereo}.
5655 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5656 for the required syntax.
5659 Set input volume level. By default, this is @var{1}.
5662 Set output volume level. By default, this is @var{1}.
5665 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5668 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5671 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5674 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5675 In @var{add} mode, LFE channel is created from input audio and added to output.
5676 In @var{sub} mode, LFE channel is created from input audio and added to output but
5677 also all non-LFE output channels are subtracted with output LFE channel.
5680 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5681 Default is @var{90}.
5684 Set front center input volume. By default, this is @var{1}.
5687 Set front center output volume. By default, this is @var{1}.
5690 Set front left input volume. By default, this is @var{1}.
5693 Set front left output volume. By default, this is @var{1}.
5696 Set front right input volume. By default, this is @var{1}.
5699 Set front right output volume. By default, this is @var{1}.
5702 Set side left input volume. By default, this is @var{1}.
5705 Set side left output volume. By default, this is @var{1}.
5708 Set side right input volume. By default, this is @var{1}.
5711 Set side right output volume. By default, this is @var{1}.
5714 Set back left input volume. By default, this is @var{1}.
5717 Set back left output volume. By default, this is @var{1}.
5720 Set back right input volume. By default, this is @var{1}.
5723 Set back right output volume. By default, this is @var{1}.
5726 Set back center input volume. By default, this is @var{1}.
5729 Set back center output volume. By default, this is @var{1}.
5732 Set LFE input volume. By default, this is @var{1}.
5735 Set LFE output volume. By default, this is @var{1}.
5738 Set spread usage of stereo image across X axis for all channels.
5741 Set spread usage of stereo image across Y axis for all channels.
5743 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5744 Set spread usage of stereo image across X axis for each channel.
5746 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5747 Set spread usage of stereo image across Y axis for each channel.
5750 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5753 Set window function.
5755 It accepts the following values:
5778 Default is @code{hann}.
5781 Set window overlap. If set to 1, the recommended overlap for selected
5782 window function will be picked. Default is @code{0.5}.
5785 @section treble, highshelf
5787 Boost or cut treble (upper) frequencies of the audio using a two-pole
5788 shelving filter with a response similar to that of a standard
5789 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5791 The filter accepts the following options:
5795 Give the gain at whichever is the lower of ~22 kHz and the
5796 Nyquist frequency. Its useful range is about -20 (for a large cut)
5797 to +20 (for a large boost). Beware of clipping when using a positive gain.
5800 Set the filter's central frequency and so can be used
5801 to extend or reduce the frequency range to be boosted or cut.
5802 The default value is @code{3000} Hz.
5805 Set method to specify band-width of filter.
5820 Determine how steep is the filter's shelf transition.
5823 How much to use filtered signal in output. Default is 1.
5824 Range is between 0 and 1.
5827 Specify which channels to filter, by default all available are filtered.
5830 Normalize biquad coefficients, by default is disabled.
5831 Enabling it will normalize magnitude response at DC to 0dB.
5834 Set transform type of IIR filter.
5843 @subsection Commands
5845 This filter supports the following commands:
5848 Change treble frequency.
5849 Syntax for the command is : "@var{frequency}"
5852 Change treble width_type.
5853 Syntax for the command is : "@var{width_type}"
5856 Change treble width.
5857 Syntax for the command is : "@var{width}"
5861 Syntax for the command is : "@var{gain}"
5865 Syntax for the command is : "@var{mix}"
5870 Sinusoidal amplitude modulation.
5872 The filter accepts the following options:
5876 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5877 (20 Hz or lower) will result in a tremolo effect.
5878 This filter may also be used as a ring modulator by specifying
5879 a modulation frequency higher than 20 Hz.
5880 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5883 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5884 Default value is 0.5.
5889 Sinusoidal phase modulation.
5891 The filter accepts the following options:
5895 Modulation frequency in Hertz.
5896 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5899 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5900 Default value is 0.5.
5905 Adjust the input audio volume.
5907 It accepts the following parameters:
5911 Set audio volume expression.
5913 Output values are clipped to the maximum value.
5915 The output audio volume is given by the relation:
5917 @var{output_volume} = @var{volume} * @var{input_volume}
5920 The default value for @var{volume} is "1.0".
5923 This parameter represents the mathematical precision.
5925 It determines which input sample formats will be allowed, which affects the
5926 precision of the volume scaling.
5930 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5932 32-bit floating-point; this limits input sample format to FLT. (default)
5934 64-bit floating-point; this limits input sample format to DBL.
5938 Choose the behaviour on encountering ReplayGain side data in input frames.
5942 Remove ReplayGain side data, ignoring its contents (the default).
5945 Ignore ReplayGain side data, but leave it in the frame.
5948 Prefer the track gain, if present.
5951 Prefer the album gain, if present.
5954 @item replaygain_preamp
5955 Pre-amplification gain in dB to apply to the selected replaygain gain.
5957 Default value for @var{replaygain_preamp} is 0.0.
5959 @item replaygain_noclip
5960 Prevent clipping by limiting the gain applied.
5962 Default value for @var{replaygain_noclip} is 1.
5965 Set when the volume expression is evaluated.
5967 It accepts the following values:
5970 only evaluate expression once during the filter initialization, or
5971 when the @samp{volume} command is sent
5974 evaluate expression for each incoming frame
5977 Default value is @samp{once}.
5980 The volume expression can contain the following parameters.
5984 frame number (starting at zero)
5987 @item nb_consumed_samples
5988 number of samples consumed by the filter
5990 number of samples in the current frame
5992 original frame position in the file
5998 PTS at start of stream
6000 time at start of stream
6006 last set volume value
6009 Note that when @option{eval} is set to @samp{once} only the
6010 @var{sample_rate} and @var{tb} variables are available, all other
6011 variables will evaluate to NAN.
6013 @subsection Commands
6015 This filter supports the following commands:
6018 Modify the volume expression.
6019 The command accepts the same syntax of the corresponding option.
6021 If the specified expression is not valid, it is kept at its current
6025 @subsection Examples
6029 Halve the input audio volume:
6033 volume=volume=-6.0206dB
6036 In all the above example the named key for @option{volume} can be
6037 omitted, for example like in:
6043 Increase input audio power by 6 decibels using fixed-point precision:
6045 volume=volume=6dB:precision=fixed
6049 Fade volume after time 10 with an annihilation period of 5 seconds:
6051 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6055 @section volumedetect
6057 Detect the volume of the input video.
6059 The filter has no parameters. The input is not modified. Statistics about
6060 the volume will be printed in the log when the input stream end is reached.
6062 In particular it will show the mean volume (root mean square), maximum
6063 volume (on a per-sample basis), and the beginning of a histogram of the
6064 registered volume values (from the maximum value to a cumulated 1/1000 of
6067 All volumes are in decibels relative to the maximum PCM value.
6069 @subsection Examples
6071 Here is an excerpt of the output:
6073 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6074 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6075 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6076 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6077 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6078 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6079 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6080 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6081 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6087 The mean square energy is approximately -27 dB, or 10^-2.7.
6089 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6091 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6094 In other words, raising the volume by +4 dB does not cause any clipping,
6095 raising it by +5 dB causes clipping for 6 samples, etc.
6097 @c man end AUDIO FILTERS
6099 @chapter Audio Sources
6100 @c man begin AUDIO SOURCES
6102 Below is a description of the currently available audio sources.
6106 Buffer audio frames, and make them available to the filter chain.
6108 This source is mainly intended for a programmatic use, in particular
6109 through the interface defined in @file{libavfilter/buffersrc.h}.
6111 It accepts the following parameters:
6115 The timebase which will be used for timestamps of submitted frames. It must be
6116 either a floating-point number or in @var{numerator}/@var{denominator} form.
6119 The sample rate of the incoming audio buffers.
6122 The sample format of the incoming audio buffers.
6123 Either a sample format name or its corresponding integer representation from
6124 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6126 @item channel_layout
6127 The channel layout of the incoming audio buffers.
6128 Either a channel layout name from channel_layout_map in
6129 @file{libavutil/channel_layout.c} or its corresponding integer representation
6130 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6133 The number of channels of the incoming audio buffers.
6134 If both @var{channels} and @var{channel_layout} are specified, then they
6139 @subsection Examples
6142 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6145 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6146 Since the sample format with name "s16p" corresponds to the number
6147 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6150 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6155 Generate an audio signal specified by an expression.
6157 This source accepts in input one or more expressions (one for each
6158 channel), which are evaluated and used to generate a corresponding
6161 This source accepts the following options:
6165 Set the '|'-separated expressions list for each separate channel. In case the
6166 @option{channel_layout} option is not specified, the selected channel layout
6167 depends on the number of provided expressions. Otherwise the last
6168 specified expression is applied to the remaining output channels.
6170 @item channel_layout, c
6171 Set the channel layout. The number of channels in the specified layout
6172 must be equal to the number of specified expressions.
6175 Set the minimum duration of the sourced audio. See
6176 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6177 for the accepted syntax.
6178 Note that the resulting duration may be greater than the specified
6179 duration, as the generated audio is always cut at the end of a
6182 If not specified, or the expressed duration is negative, the audio is
6183 supposed to be generated forever.
6186 Set the number of samples per channel per each output frame,
6189 @item sample_rate, s
6190 Specify the sample rate, default to 44100.
6193 Each expression in @var{exprs} can contain the following constants:
6197 number of the evaluated sample, starting from 0
6200 time of the evaluated sample expressed in seconds, starting from 0
6207 @subsection Examples
6217 Generate a sin signal with frequency of 440 Hz, set sample rate to
6220 aevalsrc="sin(440*2*PI*t):s=8000"
6224 Generate a two channels signal, specify the channel layout (Front
6225 Center + Back Center) explicitly:
6227 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6231 Generate white noise:
6233 aevalsrc="-2+random(0)"
6237 Generate an amplitude modulated signal:
6239 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6243 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6245 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6252 Generate a FIR coefficients using frequency sampling method.
6254 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6256 The filter accepts the following options:
6260 Set number of filter coefficents in output audio stream.
6261 Default value is 1025.
6264 Set frequency points from where magnitude and phase are set.
6265 This must be in non decreasing order, and first element must be 0, while last element
6266 must be 1. Elements are separated by white spaces.
6269 Set magnitude value for every frequency point set by @option{frequency}.
6270 Number of values must be same as number of frequency points.
6271 Values are separated by white spaces.
6274 Set phase value for every frequency point set by @option{frequency}.
6275 Number of values must be same as number of frequency points.
6276 Values are separated by white spaces.
6278 @item sample_rate, r
6279 Set sample rate, default is 44100.
6282 Set number of samples per each frame. Default is 1024.
6285 Set window function. Default is blackman.
6290 The null audio source, return unprocessed audio frames. It is mainly useful
6291 as a template and to be employed in analysis / debugging tools, or as
6292 the source for filters which ignore the input data (for example the sox
6295 This source accepts the following options:
6299 @item channel_layout, cl
6301 Specifies the channel layout, and can be either an integer or a string
6302 representing a channel layout. The default value of @var{channel_layout}
6305 Check the channel_layout_map definition in
6306 @file{libavutil/channel_layout.c} for the mapping between strings and
6307 channel layout values.
6309 @item sample_rate, r
6310 Specifies the sample rate, and defaults to 44100.
6313 Set the number of samples per requested frames.
6316 Set the duration of the sourced audio. See
6317 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6318 for the accepted syntax.
6320 If not specified, or the expressed duration is negative, the audio is
6321 supposed to be generated forever.
6324 @subsection Examples
6328 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6330 anullsrc=r=48000:cl=4
6334 Do the same operation with a more obvious syntax:
6336 anullsrc=r=48000:cl=mono
6340 All the parameters need to be explicitly defined.
6344 Synthesize a voice utterance using the libflite library.
6346 To enable compilation of this filter you need to configure FFmpeg with
6347 @code{--enable-libflite}.
6349 Note that versions of the flite library prior to 2.0 are not thread-safe.
6351 The filter accepts the following options:
6356 If set to 1, list the names of the available voices and exit
6357 immediately. Default value is 0.
6360 Set the maximum number of samples per frame. Default value is 512.
6363 Set the filename containing the text to speak.
6366 Set the text to speak.
6369 Set the voice to use for the speech synthesis. Default value is
6370 @code{kal}. See also the @var{list_voices} option.
6373 @subsection Examples
6377 Read from file @file{speech.txt}, and synthesize the text using the
6378 standard flite voice:
6380 flite=textfile=speech.txt
6384 Read the specified text selecting the @code{slt} voice:
6386 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6390 Input text to ffmpeg:
6392 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6396 Make @file{ffplay} speak the specified text, using @code{flite} and
6397 the @code{lavfi} device:
6399 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6403 For more information about libflite, check:
6404 @url{http://www.festvox.org/flite/}
6408 Generate a noise audio signal.
6410 The filter accepts the following options:
6413 @item sample_rate, r
6414 Specify the sample rate. Default value is 48000 Hz.
6417 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6421 Specify the duration of the generated audio stream. Not specifying this option
6422 results in noise with an infinite length.
6424 @item color, colour, c
6425 Specify the color of noise. Available noise colors are white, pink, brown,
6426 blue, violet and velvet. Default color is white.
6429 Specify a value used to seed the PRNG.
6432 Set the number of samples per each output frame, default is 1024.
6435 @subsection Examples
6440 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6442 anoisesrc=d=60:c=pink:r=44100:a=0.5
6448 Generate odd-tap Hilbert transform FIR coefficients.
6450 The resulting stream can be used with @ref{afir} filter for phase-shifting
6451 the signal by 90 degrees.
6453 This is used in many matrix coding schemes and for analytic signal generation.
6454 The process is often written as a multiplication by i (or j), the imaginary unit.
6456 The filter accepts the following options:
6460 @item sample_rate, s
6461 Set sample rate, default is 44100.
6464 Set length of FIR filter, default is 22051.
6467 Set number of samples per each frame.
6470 Set window function to be used when generating FIR coefficients.
6475 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6477 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6479 The filter accepts the following options:
6482 @item sample_rate, r
6483 Set sample rate, default is 44100.
6486 Set number of samples per each frame. Default is 1024.
6489 Set high-pass frequency. Default is 0.
6492 Set low-pass frequency. Default is 0.
6493 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6494 is higher than 0 then filter will create band-pass filter coefficients,
6495 otherwise band-reject filter coefficients.
6498 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6501 Set Kaiser window beta.
6504 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6507 Enable rounding, by default is disabled.
6510 Set number of taps for high-pass filter.
6513 Set number of taps for low-pass filter.
6518 Generate an audio signal made of a sine wave with amplitude 1/8.
6520 The audio signal is bit-exact.
6522 The filter accepts the following options:
6527 Set the carrier frequency. Default is 440 Hz.
6529 @item beep_factor, b
6530 Enable a periodic beep every second with frequency @var{beep_factor} times
6531 the carrier frequency. Default is 0, meaning the beep is disabled.
6533 @item sample_rate, r
6534 Specify the sample rate, default is 44100.
6537 Specify the duration of the generated audio stream.
6539 @item samples_per_frame
6540 Set the number of samples per output frame.
6542 The expression can contain the following constants:
6546 The (sequential) number of the output audio frame, starting from 0.
6549 The PTS (Presentation TimeStamp) of the output audio frame,
6550 expressed in @var{TB} units.
6553 The PTS of the output audio frame, expressed in seconds.
6556 The timebase of the output audio frames.
6559 Default is @code{1024}.
6562 @subsection Examples
6567 Generate a simple 440 Hz sine wave:
6573 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6577 sine=frequency=220:beep_factor=4:duration=5
6581 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6584 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6588 @c man end AUDIO SOURCES
6590 @chapter Audio Sinks
6591 @c man begin AUDIO SINKS
6593 Below is a description of the currently available audio sinks.
6595 @section abuffersink
6597 Buffer audio frames, and make them available to the end of filter chain.
6599 This sink is mainly intended for programmatic use, in particular
6600 through the interface defined in @file{libavfilter/buffersink.h}
6601 or the options system.
6603 It accepts a pointer to an AVABufferSinkContext structure, which
6604 defines the incoming buffers' formats, to be passed as the opaque
6605 parameter to @code{avfilter_init_filter} for initialization.
6608 Null audio sink; do absolutely nothing with the input audio. It is
6609 mainly useful as a template and for use in analysis / debugging
6612 @c man end AUDIO SINKS
6614 @chapter Video Filters
6615 @c man begin VIDEO FILTERS
6617 When you configure your FFmpeg build, you can disable any of the
6618 existing filters using @code{--disable-filters}.
6619 The configure output will show the video filters included in your
6622 Below is a description of the currently available video filters.
6626 Mark a region of interest in a video frame.
6628 The frame data is passed through unchanged, but metadata is attached
6629 to the frame indicating regions of interest which can affect the
6630 behaviour of later encoding. Multiple regions can be marked by
6631 applying the filter multiple times.
6635 Region distance in pixels from the left edge of the frame.
6637 Region distance in pixels from the top edge of the frame.
6639 Region width in pixels.
6641 Region height in pixels.
6643 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6644 and may contain the following variables:
6647 Width of the input frame.
6649 Height of the input frame.
6653 Quantisation offset to apply within the region.
6655 This must be a real value in the range -1 to +1. A value of zero
6656 indicates no quality change. A negative value asks for better quality
6657 (less quantisation), while a positive value asks for worse quality
6658 (greater quantisation).
6660 The range is calibrated so that the extreme values indicate the
6661 largest possible offset - if the rest of the frame is encoded with the
6662 worst possible quality, an offset of -1 indicates that this region
6663 should be encoded with the best possible quality anyway. Intermediate
6664 values are then interpolated in some codec-dependent way.
6666 For example, in 10-bit H.264 the quantisation parameter varies between
6667 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6668 this region should be encoded with a QP around one-tenth of the full
6669 range better than the rest of the frame. So, if most of the frame
6670 were to be encoded with a QP of around 30, this region would get a QP
6671 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6672 An extreme value of -1 would indicate that this region should be
6673 encoded with the best possible quality regardless of the treatment of
6674 the rest of the frame - that is, should be encoded at a QP of -12.
6676 If set to true, remove any existing regions of interest marked on the
6677 frame before adding the new one.
6680 @subsection Examples
6684 Mark the centre quarter of the frame as interesting.
6686 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6689 Mark the 100-pixel-wide region on the left edge of the frame as very
6690 uninteresting (to be encoded at much lower quality than the rest of
6693 addroi=0:0:100:ih:+1/5
6697 @section alphaextract
6699 Extract the alpha component from the input as a grayscale video. This
6700 is especially useful with the @var{alphamerge} filter.
6704 Add or replace the alpha component of the primary input with the
6705 grayscale value of a second input. This is intended for use with
6706 @var{alphaextract} to allow the transmission or storage of frame
6707 sequences that have alpha in a format that doesn't support an alpha
6710 For example, to reconstruct full frames from a normal YUV-encoded video
6711 and a separate video created with @var{alphaextract}, you might use:
6713 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6718 Amplify differences between current pixel and pixels of adjacent frames in
6719 same pixel location.
6721 This filter accepts the following options:
6725 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6726 For example radius of 3 will instruct filter to calculate average of 7 frames.
6729 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6732 Set threshold for difference amplification. Any difference greater or equal to
6733 this value will not alter source pixel. Default is 10.
6734 Allowed range is from 0 to 65535.
6737 Set tolerance for difference amplification. Any difference lower to
6738 this value will not alter source pixel. Default is 0.
6739 Allowed range is from 0 to 65535.
6742 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6743 This option controls maximum possible value that will decrease source pixel value.
6746 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6747 This option controls maximum possible value that will increase source pixel value.
6750 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6753 @subsection Commands
6755 This filter supports the following @ref{commands} that corresponds to option of same name:
6767 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6768 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6769 Substation Alpha) subtitles files.
6771 This filter accepts the following option in addition to the common options from
6772 the @ref{subtitles} filter:
6776 Set the shaping engine
6778 Available values are:
6781 The default libass shaping engine, which is the best available.
6783 Fast, font-agnostic shaper that can do only substitutions
6785 Slower shaper using OpenType for substitutions and positioning
6788 The default is @code{auto}.
6792 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6794 The filter accepts the following options:
6798 Set threshold A for 1st plane. Default is 0.02.
6799 Valid range is 0 to 0.3.
6802 Set threshold B for 1st plane. Default is 0.04.
6803 Valid range is 0 to 5.
6806 Set threshold A for 2nd plane. Default is 0.02.
6807 Valid range is 0 to 0.3.
6810 Set threshold B for 2nd plane. Default is 0.04.
6811 Valid range is 0 to 5.
6814 Set threshold A for 3rd plane. Default is 0.02.
6815 Valid range is 0 to 0.3.
6818 Set threshold B for 3rd plane. Default is 0.04.
6819 Valid range is 0 to 5.
6821 Threshold A is designed to react on abrupt changes in the input signal and
6822 threshold B is designed to react on continuous changes in the input signal.
6825 Set number of frames filter will use for averaging. Default is 9. Must be odd
6826 number in range [5, 129].
6829 Set what planes of frame filter will use for averaging. Default is all.
6832 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6833 Alternatively can be set to @code{s} serial.
6835 Parallel can be faster then serial, while other way around is never true.
6836 Parallel will abort early on first change being greater then thresholds, while serial
6837 will continue processing other side of frames if they are equal or below thresholds.
6840 @subsection Commands
6841 This filter supports same @ref{commands} as options except option @code{s}.
6842 The command accepts the same syntax of the corresponding option.
6846 Apply average blur filter.
6848 The filter accepts the following options:
6852 Set horizontal radius size.
6855 Set which planes to filter. By default all planes are filtered.
6858 Set vertical radius size, if zero it will be same as @code{sizeX}.
6859 Default is @code{0}.
6862 @subsection Commands
6863 This filter supports same commands as options.
6864 The command accepts the same syntax of the corresponding option.
6866 If the specified expression is not valid, it is kept at its current
6871 Compute the bounding box for the non-black pixels in the input frame
6874 This filter computes the bounding box containing all the pixels with a
6875 luminance value greater than the minimum allowed value.
6876 The parameters describing the bounding box are printed on the filter
6879 The filter accepts the following option:
6883 Set the minimal luminance value. Default is @code{16}.
6887 Apply bilateral filter, spatial smoothing while preserving edges.
6889 The filter accepts the following options:
6892 Set sigma of gaussian function to calculate spatial weight.
6893 Allowed range is 0 to 512. Default is 0.1.
6896 Set sigma of gaussian function to calculate range weight.
6897 Allowed range is 0 to 1. Default is 0.1.
6900 Set planes to filter. Default is first only.
6903 @section bitplanenoise
6905 Show and measure bit plane noise.
6907 The filter accepts the following options:
6911 Set which plane to analyze. Default is @code{1}.
6914 Filter out noisy pixels from @code{bitplane} set above.
6915 Default is disabled.
6918 @section blackdetect
6920 Detect video intervals that are (almost) completely black. Can be
6921 useful to detect chapter transitions, commercials, or invalid
6924 The filter outputs its detection analysis to both the log as well as
6925 frame metadata. If a black segment of at least the specified minimum
6926 duration is found, a line with the start and end timestamps as well
6927 as duration is printed to the log with level @code{info}. In addition,
6928 a log line with level @code{debug} is printed per frame showing the
6929 black amount detected for that frame.
6931 The filter also attaches metadata to the first frame of a black
6932 segment with key @code{lavfi.black_start} and to the first frame
6933 after the black segment ends with key @code{lavfi.black_end}. The
6934 value is the frame's timestamp. This metadata is added regardless
6935 of the minimum duration specified.
6937 The filter accepts the following options:
6940 @item black_min_duration, d
6941 Set the minimum detected black duration expressed in seconds. It must
6942 be a non-negative floating point number.
6944 Default value is 2.0.
6946 @item picture_black_ratio_th, pic_th
6947 Set the threshold for considering a picture "black".
6948 Express the minimum value for the ratio:
6950 @var{nb_black_pixels} / @var{nb_pixels}
6953 for which a picture is considered black.
6954 Default value is 0.98.
6956 @item pixel_black_th, pix_th
6957 Set the threshold for considering a pixel "black".
6959 The threshold expresses the maximum pixel luminance value for which a
6960 pixel is considered "black". The provided value is scaled according to
6961 the following equation:
6963 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6966 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6967 the input video format, the range is [0-255] for YUV full-range
6968 formats and [16-235] for YUV non full-range formats.
6970 Default value is 0.10.
6973 The following example sets the maximum pixel threshold to the minimum
6974 value, and detects only black intervals of 2 or more seconds:
6976 blackdetect=d=2:pix_th=0.00
6981 Detect frames that are (almost) completely black. Can be useful to
6982 detect chapter transitions or commercials. Output lines consist of
6983 the frame number of the detected frame, the percentage of blackness,
6984 the position in the file if known or -1 and the timestamp in seconds.
6986 In order to display the output lines, you need to set the loglevel at
6987 least to the AV_LOG_INFO value.
6989 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6990 The value represents the percentage of pixels in the picture that
6991 are below the threshold value.
6993 It accepts the following parameters:
6998 The percentage of the pixels that have to be below the threshold; it defaults to
7001 @item threshold, thresh
7002 The threshold below which a pixel value is considered black; it defaults to
7010 Blend two video frames into each other.
7012 The @code{blend} filter takes two input streams and outputs one
7013 stream, the first input is the "top" layer and second input is
7014 "bottom" layer. By default, the output terminates when the longest input terminates.
7016 The @code{tblend} (time blend) filter takes two consecutive frames
7017 from one single stream, and outputs the result obtained by blending
7018 the new frame on top of the old frame.
7020 A description of the accepted options follows.
7028 Set blend mode for specific pixel component or all pixel components in case
7029 of @var{all_mode}. Default value is @code{normal}.
7031 Available values for component modes are:
7073 Set blend opacity for specific pixel component or all pixel components in case
7074 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7081 Set blend expression for specific pixel component or all pixel components in case
7082 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7084 The expressions can use the following variables:
7088 The sequential number of the filtered frame, starting from @code{0}.
7092 the coordinates of the current sample
7096 the width and height of currently filtered plane
7100 Width and height scale for the plane being filtered. It is the
7101 ratio between the dimensions of the current plane to the luma plane,
7102 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7103 the luma plane and @code{0.5,0.5} for the chroma planes.
7106 Time of the current frame, expressed in seconds.
7109 Value of pixel component at current location for first video frame (top layer).
7112 Value of pixel component at current location for second video frame (bottom layer).
7116 The @code{blend} filter also supports the @ref{framesync} options.
7118 @subsection Examples
7122 Apply transition from bottom layer to top layer in first 10 seconds:
7124 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7128 Apply linear horizontal transition from top layer to bottom layer:
7130 blend=all_expr='A*(X/W)+B*(1-X/W)'
7134 Apply 1x1 checkerboard effect:
7136 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7140 Apply uncover left effect:
7142 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7146 Apply uncover down effect:
7148 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7152 Apply uncover up-left effect:
7154 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7158 Split diagonally video and shows top and bottom layer on each side:
7160 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7164 Display differences between the current and the previous frame:
7166 tblend=all_mode=grainextract
7172 Denoise frames using Block-Matching 3D algorithm.
7174 The filter accepts the following options.
7178 Set denoising strength. Default value is 1.
7179 Allowed range is from 0 to 999.9.
7180 The denoising algorithm is very sensitive to sigma, so adjust it
7181 according to the source.
7184 Set local patch size. This sets dimensions in 2D.
7187 Set sliding step for processing blocks. Default value is 4.
7188 Allowed range is from 1 to 64.
7189 Smaller values allows processing more reference blocks and is slower.
7192 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7193 When set to 1, no block matching is done. Larger values allows more blocks
7195 Allowed range is from 1 to 256.
7198 Set radius for search block matching. Default is 9.
7199 Allowed range is from 1 to INT32_MAX.
7202 Set step between two search locations for block matching. Default is 1.
7203 Allowed range is from 1 to 64. Smaller is slower.
7206 Set threshold of mean square error for block matching. Valid range is 0 to
7210 Set thresholding parameter for hard thresholding in 3D transformed domain.
7211 Larger values results in stronger hard-thresholding filtering in frequency
7215 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7216 Default is @code{basic}.
7219 If enabled, filter will use 2nd stream for block matching.
7220 Default is disabled for @code{basic} value of @var{estim} option,
7221 and always enabled if value of @var{estim} is @code{final}.
7224 Set planes to filter. Default is all available except alpha.
7227 @subsection Examples
7231 Basic filtering with bm3d:
7233 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7237 Same as above, but filtering only luma:
7239 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7243 Same as above, but with both estimation modes:
7245 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
7249 Same as above, but prefilter with @ref{nlmeans} filter instead:
7251 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
7257 Apply a boxblur algorithm to the input video.
7259 It accepts the following parameters:
7263 @item luma_radius, lr
7264 @item luma_power, lp
7265 @item chroma_radius, cr
7266 @item chroma_power, cp
7267 @item alpha_radius, ar
7268 @item alpha_power, ap
7272 A description of the accepted options follows.
7275 @item luma_radius, lr
7276 @item chroma_radius, cr
7277 @item alpha_radius, ar
7278 Set an expression for the box radius in pixels used for blurring the
7279 corresponding input plane.
7281 The radius value must be a non-negative number, and must not be
7282 greater than the value of the expression @code{min(w,h)/2} for the
7283 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7286 Default value for @option{luma_radius} is "2". If not specified,
7287 @option{chroma_radius} and @option{alpha_radius} default to the
7288 corresponding value set for @option{luma_radius}.
7290 The expressions can contain the following constants:
7294 The input width and height in pixels.
7298 The input chroma image width and height in pixels.
7302 The horizontal and vertical chroma subsample values. For example, for the
7303 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7306 @item luma_power, lp
7307 @item chroma_power, cp
7308 @item alpha_power, ap
7309 Specify how many times the boxblur filter is applied to the
7310 corresponding plane.
7312 Default value for @option{luma_power} is 2. If not specified,
7313 @option{chroma_power} and @option{alpha_power} default to the
7314 corresponding value set for @option{luma_power}.
7316 A value of 0 will disable the effect.
7319 @subsection Examples
7323 Apply a boxblur filter with the luma, chroma, and alpha radii
7326 boxblur=luma_radius=2:luma_power=1
7331 Set the luma radius to 2, and alpha and chroma radius to 0:
7333 boxblur=2:1:cr=0:ar=0
7337 Set the luma and chroma radii to a fraction of the video dimension:
7339 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7345 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7346 Deinterlacing Filter").
7348 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7349 interpolation algorithms.
7350 It accepts the following parameters:
7354 The interlacing mode to adopt. It accepts one of the following values:
7358 Output one frame for each frame.
7360 Output one frame for each field.
7363 The default value is @code{send_field}.
7366 The picture field parity assumed for the input interlaced video. It accepts one
7367 of the following values:
7371 Assume the top field is first.
7373 Assume the bottom field is first.
7375 Enable automatic detection of field parity.
7378 The default value is @code{auto}.
7379 If the interlacing is unknown or the decoder does not export this information,
7380 top field first will be assumed.
7383 Specify which frames to deinterlace. Accepts one of the following
7388 Deinterlace all frames.
7390 Only deinterlace frames marked as interlaced.
7393 The default value is @code{all}.
7398 Apply Contrast Adaptive Sharpen filter to video stream.
7400 The filter accepts the following options:
7404 Set the sharpening strength. Default value is 0.
7407 Set planes to filter. Default value is to filter all
7408 planes except alpha plane.
7412 Remove all color information for all colors except for certain one.
7414 The filter accepts the following options:
7418 The color which will not be replaced with neutral chroma.
7421 Similarity percentage with the above color.
7422 0.01 matches only the exact key color, while 1.0 matches everything.
7426 0.0 makes pixels either fully gray, or not gray at all.
7427 Higher values result in more preserved color.
7430 Signals that the color passed is already in YUV instead of RGB.
7432 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7433 This can be used to pass exact YUV values as hexadecimal numbers.
7436 @subsection Commands
7437 This filter supports same @ref{commands} as options.
7438 The command accepts the same syntax of the corresponding option.
7440 If the specified expression is not valid, it is kept at its current
7444 YUV colorspace color/chroma keying.
7446 The filter accepts the following options:
7450 The color which will be replaced with transparency.
7453 Similarity percentage with the key color.
7455 0.01 matches only the exact key color, while 1.0 matches everything.
7460 0.0 makes pixels either fully transparent, or not transparent at all.
7462 Higher values result in semi-transparent pixels, with a higher transparency
7463 the more similar the pixels color is to the key color.
7466 Signals that the color passed is already in YUV instead of RGB.
7468 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7469 This can be used to pass exact YUV values as hexadecimal numbers.
7472 @subsection Commands
7473 This filter supports same @ref{commands} as options.
7474 The command accepts the same syntax of the corresponding option.
7476 If the specified expression is not valid, it is kept at its current
7479 @subsection Examples
7483 Make every green pixel in the input image transparent:
7485 ffmpeg -i input.png -vf chromakey=green out.png
7489 Overlay a greenscreen-video on top of a static black background.
7491 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
7496 Reduce chrominance noise.
7498 The filter accepts the following options:
7502 Set threshold for averaging chrominance values.
7503 Sum of absolute difference of U and V pixel components or current
7504 pixel and neighbour pixels lower than this threshold will be used in
7505 averaging. Luma component is left unchanged and is copied to output.
7506 Default value is 30. Allowed range is from 1 to 200.
7509 Set horizontal radius of rectangle used for averaging.
7510 Allowed range is from 1 to 100. Default value is 5.
7513 Set vertical radius of rectangle used for averaging.
7514 Allowed range is from 1 to 100. Default value is 5.
7517 Set horizontal step when averaging. Default value is 1.
7518 Allowed range is from 1 to 50.
7519 Mostly useful to speed-up filtering.
7522 Set vertical step when averaging. Default value is 1.
7523 Allowed range is from 1 to 50.
7524 Mostly useful to speed-up filtering.
7527 @subsection Commands
7528 This filter supports same @ref{commands} as options.
7529 The command accepts the same syntax of the corresponding option.
7531 @section chromashift
7532 Shift chroma pixels horizontally and/or vertically.
7534 The filter accepts the following options:
7537 Set amount to shift chroma-blue horizontally.
7539 Set amount to shift chroma-blue vertically.
7541 Set amount to shift chroma-red horizontally.
7543 Set amount to shift chroma-red vertically.
7545 Set edge mode, can be @var{smear}, default, or @var{warp}.
7548 @subsection Commands
7550 This filter supports the all above options as @ref{commands}.
7554 Display CIE color diagram with pixels overlaid onto it.
7556 The filter accepts the following options:
7571 @item uhdtv, rec2020
7585 Set what gamuts to draw.
7587 See @code{system} option for available values.
7590 Set ciescope size, by default set to 512.
7593 Set intensity used to map input pixel values to CIE diagram.
7596 Set contrast used to draw tongue colors that are out of active color system gamut.
7599 Correct gamma displayed on scope, by default enabled.
7602 Show white point on CIE diagram, by default disabled.
7605 Set input gamma. Used only with XYZ input color space.
7610 Visualize information exported by some codecs.
7612 Some codecs can export information through frames using side-data or other
7613 means. For example, some MPEG based codecs export motion vectors through the
7614 @var{export_mvs} flag in the codec @option{flags2} option.
7616 The filter accepts the following option:
7620 Set motion vectors to visualize.
7622 Available flags for @var{mv} are:
7626 forward predicted MVs of P-frames
7628 forward predicted MVs of B-frames
7630 backward predicted MVs of B-frames
7634 Display quantization parameters using the chroma planes.
7637 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7639 Available flags for @var{mv_type} are:
7643 forward predicted MVs
7645 backward predicted MVs
7648 @item frame_type, ft
7649 Set frame type to visualize motion vectors of.
7651 Available flags for @var{frame_type} are:
7655 intra-coded frames (I-frames)
7657 predicted frames (P-frames)
7659 bi-directionally predicted frames (B-frames)
7663 @subsection Examples
7667 Visualize forward predicted MVs of all frames using @command{ffplay}:
7669 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7673 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7675 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7679 @section colorbalance
7680 Modify intensity of primary colors (red, green and blue) of input frames.
7682 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7683 regions for the red-cyan, green-magenta or blue-yellow balance.
7685 A positive adjustment value shifts the balance towards the primary color, a negative
7686 value towards the complementary color.
7688 The filter accepts the following options:
7694 Adjust red, green and blue shadows (darkest pixels).
7699 Adjust red, green and blue midtones (medium pixels).
7704 Adjust red, green and blue highlights (brightest pixels).
7706 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7709 Preserve lightness when changing color balance. Default is disabled.
7712 @subsection Examples
7716 Add red color cast to shadows:
7722 @subsection Commands
7724 This filter supports the all above options as @ref{commands}.
7726 @section colorchannelmixer
7728 Adjust video input frames by re-mixing color channels.
7730 This filter modifies a color channel by adding the values associated to
7731 the other channels of the same pixels. For example if the value to
7732 modify is red, the output value will be:
7734 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7737 The filter accepts the following options:
7744 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7745 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7751 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7752 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7758 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7759 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7765 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7766 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7768 Allowed ranges for options are @code{[-2.0, 2.0]}.
7771 @subsection Examples
7775 Convert source to grayscale:
7777 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7780 Simulate sepia tones:
7782 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7786 @subsection Commands
7788 This filter supports the all above options as @ref{commands}.
7791 RGB colorspace color keying.
7793 The filter accepts the following options:
7797 The color which will be replaced with transparency.
7800 Similarity percentage with the key color.
7802 0.01 matches only the exact key color, while 1.0 matches everything.
7807 0.0 makes pixels either fully transparent, or not transparent at all.
7809 Higher values result in semi-transparent pixels, with a higher transparency
7810 the more similar the pixels color is to the key color.
7813 @subsection Examples
7817 Make every green pixel in the input image transparent:
7819 ffmpeg -i input.png -vf colorkey=green out.png
7823 Overlay a greenscreen-video on top of a static background image.
7825 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
7829 @subsection Commands
7830 This filter supports same @ref{commands} as options.
7831 The command accepts the same syntax of the corresponding option.
7833 If the specified expression is not valid, it is kept at its current
7837 Remove all color information for all RGB colors except for certain one.
7839 The filter accepts the following options:
7843 The color which will not be replaced with neutral gray.
7846 Similarity percentage with the above color.
7847 0.01 matches only the exact key color, while 1.0 matches everything.
7850 Blend percentage. 0.0 makes pixels fully gray.
7851 Higher values result in more preserved color.
7854 @subsection Commands
7855 This filter supports same @ref{commands} as options.
7856 The command accepts the same syntax of the corresponding option.
7858 If the specified expression is not valid, it is kept at its current
7861 @section colorlevels
7863 Adjust video input frames using levels.
7865 The filter accepts the following options:
7872 Adjust red, green, blue and alpha input black point.
7873 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7879 Adjust red, green, blue and alpha input white point.
7880 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7882 Input levels are used to lighten highlights (bright tones), darken shadows
7883 (dark tones), change the balance of bright and dark tones.
7889 Adjust red, green, blue and alpha output black point.
7890 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7896 Adjust red, green, blue and alpha output white point.
7897 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7899 Output levels allows manual selection of a constrained output level range.
7902 @subsection Examples
7906 Make video output darker:
7908 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7914 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7918 Make video output lighter:
7920 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7924 Increase brightness:
7926 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7930 @subsection Commands
7932 This filter supports the all above options as @ref{commands}.
7934 @section colormatrix
7936 Convert color matrix.
7938 The filter accepts the following options:
7943 Specify the source and destination color matrix. Both values must be
7946 The accepted values are:
7974 For example to convert from BT.601 to SMPTE-240M, use the command:
7976 colormatrix=bt601:smpte240m
7981 Convert colorspace, transfer characteristics or color primaries.
7982 Input video needs to have an even size.
7984 The filter accepts the following options:
7989 Specify all color properties at once.
7991 The accepted values are:
8021 Specify output colorspace.
8023 The accepted values are:
8032 BT.470BG or BT.601-6 625
8035 SMPTE-170M or BT.601-6 525
8044 BT.2020 with non-constant luminance
8050 Specify output transfer characteristics.
8052 The accepted values are:
8064 Constant gamma of 2.2
8067 Constant gamma of 2.8
8070 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8088 BT.2020 for 10-bits content
8091 BT.2020 for 12-bits content
8097 Specify output color primaries.
8099 The accepted values are:
8108 BT.470BG or BT.601-6 625
8111 SMPTE-170M or BT.601-6 525
8135 Specify output color range.
8137 The accepted values are:
8140 TV (restricted) range
8143 MPEG (restricted) range
8154 Specify output color format.
8156 The accepted values are:
8159 YUV 4:2:0 planar 8-bits
8162 YUV 4:2:0 planar 10-bits
8165 YUV 4:2:0 planar 12-bits
8168 YUV 4:2:2 planar 8-bits
8171 YUV 4:2:2 planar 10-bits
8174 YUV 4:2:2 planar 12-bits
8177 YUV 4:4:4 planar 8-bits
8180 YUV 4:4:4 planar 10-bits
8183 YUV 4:4:4 planar 12-bits
8188 Do a fast conversion, which skips gamma/primary correction. This will take
8189 significantly less CPU, but will be mathematically incorrect. To get output
8190 compatible with that produced by the colormatrix filter, use fast=1.
8193 Specify dithering mode.
8195 The accepted values are:
8201 Floyd-Steinberg dithering
8205 Whitepoint adaptation mode.
8207 The accepted values are:
8210 Bradford whitepoint adaptation
8213 von Kries whitepoint adaptation
8216 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8220 Override all input properties at once. Same accepted values as @ref{all}.
8223 Override input colorspace. Same accepted values as @ref{space}.
8226 Override input color primaries. Same accepted values as @ref{primaries}.
8229 Override input transfer characteristics. Same accepted values as @ref{trc}.
8232 Override input color range. Same accepted values as @ref{range}.
8236 The filter converts the transfer characteristics, color space and color
8237 primaries to the specified user values. The output value, if not specified,
8238 is set to a default value based on the "all" property. If that property is
8239 also not specified, the filter will log an error. The output color range and
8240 format default to the same value as the input color range and format. The
8241 input transfer characteristics, color space, color primaries and color range
8242 should be set on the input data. If any of these are missing, the filter will
8243 log an error and no conversion will take place.
8245 For example to convert the input to SMPTE-240M, use the command:
8247 colorspace=smpte240m
8250 @section convolution
8252 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8254 The filter accepts the following options:
8261 Set matrix for each plane.
8262 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8263 and from 1 to 49 odd number of signed integers in @var{row} mode.
8269 Set multiplier for calculated value for each plane.
8270 If unset or 0, it will be sum of all matrix elements.
8276 Set bias for each plane. This value is added to the result of the multiplication.
8277 Useful for making the overall image brighter or darker. Default is 0.0.
8283 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8284 Default is @var{square}.
8287 @subsection Examples
8293 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"
8299 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"
8305 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"
8311 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"
8315 Apply laplacian edge detector which includes diagonals:
8317 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"
8323 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"
8329 Apply 2D convolution of video stream in frequency domain using second stream
8332 The filter accepts the following options:
8336 Set which planes to process.
8339 Set which impulse video frames will be processed, can be @var{first}
8340 or @var{all}. Default is @var{all}.
8343 The @code{convolve} filter also supports the @ref{framesync} options.
8347 Copy the input video source unchanged to the output. This is mainly useful for
8352 Video filtering on GPU using Apple's CoreImage API on OSX.
8354 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8355 processed by video hardware. However, software-based OpenGL implementations
8356 exist which means there is no guarantee for hardware processing. It depends on
8359 There are many filters and image generators provided by Apple that come with a
8360 large variety of options. The filter has to be referenced by its name along
8363 The coreimage filter accepts the following options:
8366 List all available filters and generators along with all their respective
8367 options as well as possible minimum and maximum values along with the default
8374 Specify all filters by their respective name and options.
8375 Use @var{list_filters} to determine all valid filter names and options.
8376 Numerical options are specified by a float value and are automatically clamped
8377 to their respective value range. Vector and color options have to be specified
8378 by a list of space separated float values. Character escaping has to be done.
8379 A special option name @code{default} is available to use default options for a
8382 It is required to specify either @code{default} or at least one of the filter options.
8383 All omitted options are used with their default values.
8384 The syntax of the filter string is as follows:
8386 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8390 Specify a rectangle where the output of the filter chain is copied into the
8391 input image. It is given by a list of space separated float values:
8393 output_rect=x\ y\ width\ height
8395 If not given, the output rectangle equals the dimensions of the input image.
8396 The output rectangle is automatically cropped at the borders of the input
8397 image. Negative values are valid for each component.
8399 output_rect=25\ 25\ 100\ 100
8403 Several filters can be chained for successive processing without GPU-HOST
8404 transfers allowing for fast processing of complex filter chains.
8405 Currently, only filters with zero (generators) or exactly one (filters) input
8406 image and one output image are supported. Also, transition filters are not yet
8409 Some filters generate output images with additional padding depending on the
8410 respective filter kernel. The padding is automatically removed to ensure the
8411 filter output has the same size as the input image.
8413 For image generators, the size of the output image is determined by the
8414 previous output image of the filter chain or the input image of the whole
8415 filterchain, respectively. The generators do not use the pixel information of
8416 this image to generate their output. However, the generated output is
8417 blended onto this image, resulting in partial or complete coverage of the
8420 The @ref{coreimagesrc} video source can be used for generating input images
8421 which are directly fed into the filter chain. By using it, providing input
8422 images by another video source or an input video is not required.
8424 @subsection Examples
8429 List all filters available:
8431 coreimage=list_filters=true
8435 Use the CIBoxBlur filter with default options to blur an image:
8437 coreimage=filter=CIBoxBlur@@default
8441 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8442 its center at 100x100 and a radius of 50 pixels:
8444 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8448 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8449 given as complete and escaped command-line for Apple's standard bash shell:
8451 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8457 Cover a rectangular object
8459 It accepts the following options:
8463 Filepath of the optional cover image, needs to be in yuv420.
8468 It accepts the following values:
8471 cover it by the supplied image
8473 cover it by interpolating the surrounding pixels
8476 Default value is @var{blur}.
8479 @subsection Examples
8483 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8485 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8491 Crop the input video to given dimensions.
8493 It accepts the following parameters:
8497 The width of the output video. It defaults to @code{iw}.
8498 This expression is evaluated only once during the filter
8499 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8502 The height of the output video. It defaults to @code{ih}.
8503 This expression is evaluated only once during the filter
8504 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8507 The horizontal position, in the input video, of the left edge of the output
8508 video. It defaults to @code{(in_w-out_w)/2}.
8509 This expression is evaluated per-frame.
8512 The vertical position, in the input video, of the top edge of the output video.
8513 It defaults to @code{(in_h-out_h)/2}.
8514 This expression is evaluated per-frame.
8517 If set to 1 will force the output display aspect ratio
8518 to be the same of the input, by changing the output sample aspect
8519 ratio. It defaults to 0.
8522 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8523 width/height/x/y as specified and will not be rounded to nearest smaller value.
8527 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8528 expressions containing the following constants:
8533 The computed values for @var{x} and @var{y}. They are evaluated for
8538 The input width and height.
8542 These are the same as @var{in_w} and @var{in_h}.
8546 The output (cropped) width and height.
8550 These are the same as @var{out_w} and @var{out_h}.
8553 same as @var{iw} / @var{ih}
8556 input sample aspect ratio
8559 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8563 horizontal and vertical chroma subsample values. For example for the
8564 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8567 The number of the input frame, starting from 0.
8570 the position in the file of the input frame, NAN if unknown
8573 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8577 The expression for @var{out_w} may depend on the value of @var{out_h},
8578 and the expression for @var{out_h} may depend on @var{out_w}, but they
8579 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8580 evaluated after @var{out_w} and @var{out_h}.
8582 The @var{x} and @var{y} parameters specify the expressions for the
8583 position of the top-left corner of the output (non-cropped) area. They
8584 are evaluated for each frame. If the evaluated value is not valid, it
8585 is approximated to the nearest valid value.
8587 The expression for @var{x} may depend on @var{y}, and the expression
8588 for @var{y} may depend on @var{x}.
8590 @subsection Examples
8594 Crop area with size 100x100 at position (12,34).
8599 Using named options, the example above becomes:
8601 crop=w=100:h=100:x=12:y=34
8605 Crop the central input area with size 100x100:
8611 Crop the central input area with size 2/3 of the input video:
8613 crop=2/3*in_w:2/3*in_h
8617 Crop the input video central square:
8624 Delimit the rectangle with the top-left corner placed at position
8625 100:100 and the right-bottom corner corresponding to the right-bottom
8626 corner of the input image.
8628 crop=in_w-100:in_h-100:100:100
8632 Crop 10 pixels from the left and right borders, and 20 pixels from
8633 the top and bottom borders
8635 crop=in_w-2*10:in_h-2*20
8639 Keep only the bottom right quarter of the input image:
8641 crop=in_w/2:in_h/2:in_w/2:in_h/2
8645 Crop height for getting Greek harmony:
8647 crop=in_w:1/PHI*in_w
8651 Apply trembling effect:
8653 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)
8657 Apply erratic camera effect depending on timestamp:
8659 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)"
8663 Set x depending on the value of y:
8665 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8669 @subsection Commands
8671 This filter supports the following commands:
8677 Set width/height of the output video and the horizontal/vertical position
8679 The command accepts the same syntax of the corresponding option.
8681 If the specified expression is not valid, it is kept at its current
8687 Auto-detect the crop size.
8689 It calculates the necessary cropping parameters and prints the
8690 recommended parameters via the logging system. The detected dimensions
8691 correspond to the non-black area of the input video.
8693 It accepts the following parameters:
8698 Set higher black value threshold, which can be optionally specified
8699 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8700 value greater to the set value is considered non-black. It defaults to 24.
8701 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8702 on the bitdepth of the pixel format.
8705 The value which the width/height should be divisible by. It defaults to
8706 16. The offset is automatically adjusted to center the video. Use 2 to
8707 get only even dimensions (needed for 4:2:2 video). 16 is best when
8708 encoding to most video codecs.
8710 @item reset_count, reset
8711 Set the counter that determines after how many frames cropdetect will
8712 reset the previously detected largest video area and start over to
8713 detect the current optimal crop area. Default value is 0.
8715 This can be useful when channel logos distort the video area. 0
8716 indicates 'never reset', and returns the largest area encountered during
8723 Delay video filtering until a given wallclock timestamp. The filter first
8724 passes on @option{preroll} amount of frames, then it buffers at most
8725 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8726 it forwards the buffered frames and also any subsequent frames coming in its
8729 The filter can be used synchronize the output of multiple ffmpeg processes for
8730 realtime output devices like decklink. By putting the delay in the filtering
8731 chain and pre-buffering frames the process can pass on data to output almost
8732 immediately after the target wallclock timestamp is reached.
8734 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8740 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8743 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8746 The maximum duration of content to buffer before waiting for the cue expressed
8747 in seconds. Default is 0.
8754 Apply color adjustments using curves.
8756 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8757 component (red, green and blue) has its values defined by @var{N} key points
8758 tied from each other using a smooth curve. The x-axis represents the pixel
8759 values from the input frame, and the y-axis the new pixel values to be set for
8762 By default, a component curve is defined by the two points @var{(0;0)} and
8763 @var{(1;1)}. This creates a straight line where each original pixel value is
8764 "adjusted" to its own value, which means no change to the image.
8766 The filter allows you to redefine these two points and add some more. A new
8767 curve (using a natural cubic spline interpolation) will be define to pass
8768 smoothly through all these new coordinates. The new defined points needs to be
8769 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8770 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8771 the vector spaces, the values will be clipped accordingly.
8773 The filter accepts the following options:
8777 Select one of the available color presets. This option can be used in addition
8778 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8779 options takes priority on the preset values.
8780 Available presets are:
8783 @item color_negative
8786 @item increase_contrast
8788 @item linear_contrast
8789 @item medium_contrast
8791 @item strong_contrast
8794 Default is @code{none}.
8796 Set the master key points. These points will define a second pass mapping. It
8797 is sometimes called a "luminance" or "value" mapping. It can be used with
8798 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8799 post-processing LUT.
8801 Set the key points for the red component.
8803 Set the key points for the green component.
8805 Set the key points for the blue component.
8807 Set the key points for all components (not including master).
8808 Can be used in addition to the other key points component
8809 options. In this case, the unset component(s) will fallback on this
8810 @option{all} setting.
8812 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8814 Save Gnuplot script of the curves in specified file.
8817 To avoid some filtergraph syntax conflicts, each key points list need to be
8818 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8820 @subsection Examples
8824 Increase slightly the middle level of blue:
8826 curves=blue='0/0 0.5/0.58 1/1'
8832 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'
8834 Here we obtain the following coordinates for each components:
8837 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8839 @code{(0;0) (0.50;0.48) (1;1)}
8841 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8845 The previous example can also be achieved with the associated built-in preset:
8847 curves=preset=vintage
8857 Use a Photoshop preset and redefine the points of the green component:
8859 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8863 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8864 and @command{gnuplot}:
8866 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8867 gnuplot -p /tmp/curves.plt
8873 Video data analysis filter.
8875 This filter shows hexadecimal pixel values of part of video.
8877 The filter accepts the following options:
8881 Set output video size.
8884 Set x offset from where to pick pixels.
8887 Set y offset from where to pick pixels.
8890 Set scope mode, can be one of the following:
8893 Draw hexadecimal pixel values with white color on black background.
8896 Draw hexadecimal pixel values with input video pixel color on black
8900 Draw hexadecimal pixel values on color background picked from input video,
8901 the text color is picked in such way so its always visible.
8905 Draw rows and columns numbers on left and top of video.
8908 Set background opacity.
8911 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
8915 Apply Directional blur filter.
8917 The filter accepts the following options:
8921 Set angle of directional blur. Default is @code{45}.
8924 Set radius of directional blur. Default is @code{5}.
8927 Set which planes to filter. By default all planes are filtered.
8930 @subsection Commands
8931 This filter supports same @ref{commands} as options.
8932 The command accepts the same syntax of the corresponding option.
8934 If the specified expression is not valid, it is kept at its current
8939 Denoise frames using 2D DCT (frequency domain filtering).
8941 This filter is not designed for real time.
8943 The filter accepts the following options:
8947 Set the noise sigma constant.
8949 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8950 coefficient (absolute value) below this threshold with be dropped.
8952 If you need a more advanced filtering, see @option{expr}.
8954 Default is @code{0}.
8957 Set number overlapping pixels for each block. Since the filter can be slow, you
8958 may want to reduce this value, at the cost of a less effective filter and the
8959 risk of various artefacts.
8961 If the overlapping value doesn't permit processing the whole input width or
8962 height, a warning will be displayed and according borders won't be denoised.
8964 Default value is @var{blocksize}-1, which is the best possible setting.
8967 Set the coefficient factor expression.
8969 For each coefficient of a DCT block, this expression will be evaluated as a
8970 multiplier value for the coefficient.
8972 If this is option is set, the @option{sigma} option will be ignored.
8974 The absolute value of the coefficient can be accessed through the @var{c}
8978 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8979 @var{blocksize}, which is the width and height of the processed blocks.
8981 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8982 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8983 on the speed processing. Also, a larger block size does not necessarily means a
8987 @subsection Examples
8989 Apply a denoise with a @option{sigma} of @code{4.5}:
8994 The same operation can be achieved using the expression system:
8996 dctdnoiz=e='gte(c, 4.5*3)'
8999 Violent denoise using a block size of @code{16x16}:
9006 Remove banding artifacts from input video.
9007 It works by replacing banded pixels with average value of referenced pixels.
9009 The filter accepts the following options:
9016 Set banding detection threshold for each plane. Default is 0.02.
9017 Valid range is 0.00003 to 0.5.
9018 If difference between current pixel and reference pixel is less than threshold,
9019 it will be considered as banded.
9022 Banding detection range in pixels. Default is 16. If positive, random number
9023 in range 0 to set value will be used. If negative, exact absolute value
9025 The range defines square of four pixels around current pixel.
9028 Set direction in radians from which four pixel will be compared. If positive,
9029 random direction from 0 to set direction will be picked. If negative, exact of
9030 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9031 will pick only pixels on same row and -PI/2 will pick only pixels on same
9035 If enabled, current pixel is compared with average value of all four
9036 surrounding pixels. The default is enabled. If disabled current pixel is
9037 compared with all four surrounding pixels. The pixel is considered banded
9038 if only all four differences with surrounding pixels are less than threshold.
9041 If enabled, current pixel is changed if and only if all pixel components are banded,
9042 e.g. banding detection threshold is triggered for all color components.
9043 The default is disabled.
9048 Remove blocking artifacts from input video.
9050 The filter accepts the following options:
9054 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9055 This controls what kind of deblocking is applied.
9058 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9064 Set blocking detection thresholds. Allowed range is 0 to 1.
9065 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9066 Using higher threshold gives more deblocking strength.
9067 Setting @var{alpha} controls threshold detection at exact edge of block.
9068 Remaining options controls threshold detection near the edge. Each one for
9069 below/above or left/right. Setting any of those to @var{0} disables
9073 Set planes to filter. Default is to filter all available planes.
9076 @subsection Examples
9080 Deblock using weak filter and block size of 4 pixels.
9082 deblock=filter=weak:block=4
9086 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9087 deblocking more edges.
9089 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9093 Similar as above, but filter only first plane.
9095 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9099 Similar as above, but filter only second and third plane.
9101 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9108 Drop duplicated frames at regular intervals.
9110 The filter accepts the following options:
9114 Set the number of frames from which one will be dropped. Setting this to
9115 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9116 Default is @code{5}.
9119 Set the threshold for duplicate detection. If the difference metric for a frame
9120 is less than or equal to this value, then it is declared as duplicate. Default
9124 Set scene change threshold. Default is @code{15}.
9128 Set the size of the x and y-axis blocks used during metric calculations.
9129 Larger blocks give better noise suppression, but also give worse detection of
9130 small movements. Must be a power of two. Default is @code{32}.
9133 Mark main input as a pre-processed input and activate clean source input
9134 stream. This allows the input to be pre-processed with various filters to help
9135 the metrics calculation while keeping the frame selection lossless. When set to
9136 @code{1}, the first stream is for the pre-processed input, and the second
9137 stream is the clean source from where the kept frames are chosen. Default is
9141 Set whether or not chroma is considered in the metric calculations. Default is
9147 Apply 2D deconvolution of video stream in frequency domain using second stream
9150 The filter accepts the following options:
9154 Set which planes to process.
9157 Set which impulse video frames will be processed, can be @var{first}
9158 or @var{all}. Default is @var{all}.
9161 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9162 and height are not same and not power of 2 or if stream prior to convolving
9166 The @code{deconvolve} filter also supports the @ref{framesync} options.
9170 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9172 It accepts the following options:
9176 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9177 @var{rainbows} for cross-color reduction.
9180 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9183 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9186 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9189 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9194 Apply deflate effect to the video.
9196 This filter replaces the pixel by the local(3x3) average by taking into account
9197 only values lower than the pixel.
9199 It accepts the following options:
9206 Limit the maximum change for each plane, default is 65535.
9207 If 0, plane will remain unchanged.
9210 @subsection Commands
9212 This filter supports the all above options as @ref{commands}.
9216 Remove temporal frame luminance variations.
9218 It accepts the following options:
9222 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9225 Set averaging mode to smooth temporal luminance variations.
9227 Available values are:
9252 Do not actually modify frame. Useful when one only wants metadata.
9257 Remove judder produced by partially interlaced telecined content.
9259 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9260 source was partially telecined content then the output of @code{pullup,dejudder}
9261 will have a variable frame rate. May change the recorded frame rate of the
9262 container. Aside from that change, this filter will not affect constant frame
9265 The option available in this filter is:
9269 Specify the length of the window over which the judder repeats.
9271 Accepts any integer greater than 1. Useful values are:
9275 If the original was telecined from 24 to 30 fps (Film to NTSC).
9278 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9281 If a mixture of the two.
9284 The default is @samp{4}.
9289 Suppress a TV station logo by a simple interpolation of the surrounding
9290 pixels. Just set a rectangle covering the logo and watch it disappear
9291 (and sometimes something even uglier appear - your mileage may vary).
9293 It accepts the following parameters:
9298 Specify the top left corner coordinates of the logo. They must be
9303 Specify the width and height of the logo to clear. They must be
9307 Specify the thickness of the fuzzy edge of the rectangle (added to
9308 @var{w} and @var{h}). The default value is 1. This option is
9309 deprecated, setting higher values should no longer be necessary and
9313 When set to 1, a green rectangle is drawn on the screen to simplify
9314 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9315 The default value is 0.
9317 The rectangle is drawn on the outermost pixels which will be (partly)
9318 replaced with interpolated values. The values of the next pixels
9319 immediately outside this rectangle in each direction will be used to
9320 compute the interpolated pixel values inside the rectangle.
9324 @subsection Examples
9328 Set a rectangle covering the area with top left corner coordinates 0,0
9329 and size 100x77, and a band of size 10:
9331 delogo=x=0:y=0:w=100:h=77:band=10
9339 Remove the rain in the input image/video by applying the derain methods based on
9340 convolutional neural networks. Supported models:
9344 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9345 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9348 Training as well as model generation scripts are provided in
9349 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9351 Native model files (.model) can be generated from TensorFlow model
9352 files (.pb) by using tools/python/convert.py
9354 The filter accepts the following options:
9358 Specify which filter to use. This option accepts the following values:
9362 Derain filter. To conduct derain filter, you need to use a derain model.
9365 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9367 Default value is @samp{derain}.
9370 Specify which DNN backend to use for model loading and execution. This option accepts
9371 the following values:
9375 Native implementation of DNN loading and execution.
9378 TensorFlow backend. To enable this backend you
9379 need to install the TensorFlow for C library (see
9380 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9381 @code{--enable-libtensorflow}
9383 Default value is @samp{native}.
9386 Set path to model file specifying network architecture and its parameters.
9387 Note that different backends use different file formats. TensorFlow and native
9388 backend can load files for only its format.
9391 It can also be finished with @ref{dnn_processing} filter.
9395 Attempt to fix small changes in horizontal and/or vertical shift. This
9396 filter helps remove camera shake from hand-holding a camera, bumping a
9397 tripod, moving on a vehicle, etc.
9399 The filter accepts the following options:
9407 Specify a rectangular area where to limit the search for motion
9409 If desired the search for motion vectors can be limited to a
9410 rectangular area of the frame defined by its top left corner, width
9411 and height. These parameters have the same meaning as the drawbox
9412 filter which can be used to visualise the position of the bounding
9415 This is useful when simultaneous movement of subjects within the frame
9416 might be confused for camera motion by the motion vector search.
9418 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9419 then the full frame is used. This allows later options to be set
9420 without specifying the bounding box for the motion vector search.
9422 Default - search the whole frame.
9426 Specify the maximum extent of movement in x and y directions in the
9427 range 0-64 pixels. Default 16.
9430 Specify how to generate pixels to fill blanks at the edge of the
9431 frame. Available values are:
9434 Fill zeroes at blank locations
9436 Original image at blank locations
9438 Extruded edge value at blank locations
9440 Mirrored edge at blank locations
9442 Default value is @samp{mirror}.
9445 Specify the blocksize to use for motion search. Range 4-128 pixels,
9449 Specify the contrast threshold for blocks. Only blocks with more than
9450 the specified contrast (difference between darkest and lightest
9451 pixels) will be considered. Range 1-255, default 125.
9454 Specify the search strategy. Available values are:
9457 Set exhaustive search
9459 Set less exhaustive search.
9461 Default value is @samp{exhaustive}.
9464 If set then a detailed log of the motion search is written to the
9471 Remove unwanted contamination of foreground colors, caused by reflected color of
9472 greenscreen or bluescreen.
9474 This filter accepts the following options:
9478 Set what type of despill to use.
9481 Set how spillmap will be generated.
9484 Set how much to get rid of still remaining spill.
9487 Controls amount of red in spill area.
9490 Controls amount of green in spill area.
9491 Should be -1 for greenscreen.
9494 Controls amount of blue in spill area.
9495 Should be -1 for bluescreen.
9498 Controls brightness of spill area, preserving colors.
9501 Modify alpha from generated spillmap.
9504 @subsection Commands
9506 This filter supports the all above options as @ref{commands}.
9510 Apply an exact inverse of the telecine operation. It requires a predefined
9511 pattern specified using the pattern option which must be the same as that passed
9512 to the telecine filter.
9514 This filter accepts the following options:
9523 The default value is @code{top}.
9527 A string of numbers representing the pulldown pattern you wish to apply.
9528 The default value is @code{23}.
9531 A number representing position of the first frame with respect to the telecine
9532 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9537 Apply dilation effect to the video.
9539 This filter replaces the pixel by the local(3x3) maximum.
9541 It accepts the following options:
9548 Limit the maximum change for each plane, default is 65535.
9549 If 0, plane will remain unchanged.
9552 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9555 Flags to local 3x3 coordinates maps like this:
9562 @subsection Commands
9564 This filter supports the all above options as @ref{commands}.
9568 Displace pixels as indicated by second and third input stream.
9570 It takes three input streams and outputs one stream, the first input is the
9571 source, and second and third input are displacement maps.
9573 The second input specifies how much to displace pixels along the
9574 x-axis, while the third input specifies how much to displace pixels
9576 If one of displacement map streams terminates, last frame from that
9577 displacement map will be used.
9579 Note that once generated, displacements maps can be reused over and over again.
9581 A description of the accepted options follows.
9585 Set displace behavior for pixels that are out of range.
9587 Available values are:
9590 Missing pixels are replaced by black pixels.
9593 Adjacent pixels will spread out to replace missing pixels.
9596 Out of range pixels are wrapped so they point to pixels of other side.
9599 Out of range pixels will be replaced with mirrored pixels.
9601 Default is @samp{smear}.
9605 @subsection Examples
9609 Add ripple effect to rgb input of video size hd720:
9611 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
9615 Add wave effect to rgb input of video size hd720:
9617 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
9621 @anchor{dnn_processing}
9622 @section dnn_processing
9624 Do image processing with deep neural networks. It works together with another filter
9625 which converts the pixel format of the Frame to what the dnn network requires.
9627 The filter accepts the following options:
9631 Specify which DNN backend to use for model loading and execution. This option accepts
9632 the following values:
9636 Native implementation of DNN loading and execution.
9639 TensorFlow backend. To enable this backend you
9640 need to install the TensorFlow for C library (see
9641 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9642 @code{--enable-libtensorflow}
9645 OpenVINO backend. To enable this backend you
9646 need to build and install the OpenVINO for C library (see
9647 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9648 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9649 be needed if the header files and libraries are not installed into system path)
9653 Default value is @samp{native}.
9656 Set path to model file specifying network architecture and its parameters.
9657 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9658 backend can load files for only its format.
9660 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9663 Set the input name of the dnn network.
9666 Set the output name of the dnn network.
9670 @subsection Examples
9674 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9676 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9680 Halve the pixel value of the frame with format gray32f:
9682 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
9686 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9688 ./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
9692 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9694 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9701 Draw a colored box on the input image.
9703 It accepts the following parameters:
9708 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9712 The expressions which specify the width and height of the box; if 0 they are interpreted as
9713 the input width and height. It defaults to 0.
9716 Specify the color of the box to write. For the general syntax of this option,
9717 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9718 value @code{invert} is used, the box edge color is the same as the
9719 video with inverted luma.
9722 The expression which sets the thickness of the box edge.
9723 A value of @code{fill} will create a filled box. Default value is @code{3}.
9725 See below for the list of accepted constants.
9728 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9729 will overwrite the video's color and alpha pixels.
9730 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9733 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9734 following constants:
9738 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9742 horizontal and vertical chroma subsample values. For example for the
9743 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9747 The input width and height.
9750 The input sample aspect ratio.
9754 The x and y offset coordinates where the box is drawn.
9758 The width and height of the drawn box.
9761 The thickness of the drawn box.
9763 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9764 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9768 @subsection Examples
9772 Draw a black box around the edge of the input image:
9778 Draw a box with color red and an opacity of 50%:
9780 drawbox=10:20:200:60:red@@0.5
9783 The previous example can be specified as:
9785 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9789 Fill the box with pink color:
9791 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9795 Draw a 2-pixel red 2.40:1 mask:
9797 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
9801 @subsection Commands
9802 This filter supports same commands as options.
9803 The command accepts the same syntax of the corresponding option.
9805 If the specified expression is not valid, it is kept at its current
9810 Draw a graph using input video metadata.
9812 It accepts the following parameters:
9816 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9819 Set 1st foreground color expression.
9822 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9825 Set 2nd foreground color expression.
9828 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9831 Set 3rd foreground color expression.
9834 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9837 Set 4th foreground color expression.
9840 Set minimal value of metadata value.
9843 Set maximal value of metadata value.
9846 Set graph background color. Default is white.
9851 Available values for mode is:
9858 Default is @code{line}.
9863 Available values for slide is:
9866 Draw new frame when right border is reached.
9869 Replace old columns with new ones.
9872 Scroll from right to left.
9875 Scroll from left to right.
9878 Draw single picture.
9881 Default is @code{frame}.
9884 Set size of graph video. For the syntax of this option, check the
9885 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9886 The default value is @code{900x256}.
9889 Set the output frame rate. Default value is @code{25}.
9891 The foreground color expressions can use the following variables:
9894 Minimal value of metadata value.
9897 Maximal value of metadata value.
9900 Current metadata key value.
9903 The color is defined as 0xAABBGGRR.
9906 Example using metadata from @ref{signalstats} filter:
9908 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9911 Example using metadata from @ref{ebur128} filter:
9913 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9918 Draw a grid on the input image.
9920 It accepts the following parameters:
9925 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9929 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9930 input width and height, respectively, minus @code{thickness}, so image gets
9931 framed. Default to 0.
9934 Specify the color of the grid. For the general syntax of this option,
9935 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9936 value @code{invert} is used, the grid color is the same as the
9937 video with inverted luma.
9940 The expression which sets the thickness of the grid line. Default value is @code{1}.
9942 See below for the list of accepted constants.
9945 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9946 will overwrite the video's color and alpha pixels.
9947 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9950 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9951 following constants:
9955 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9959 horizontal and vertical chroma subsample values. For example for the
9960 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9964 The input grid cell width and height.
9967 The input sample aspect ratio.
9971 The x and y coordinates of some point of grid intersection (meant to configure offset).
9975 The width and height of the drawn cell.
9978 The thickness of the drawn cell.
9980 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9981 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9985 @subsection Examples
9989 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9991 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9995 Draw a white 3x3 grid with an opacity of 50%:
9997 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10001 @subsection Commands
10002 This filter supports same commands as options.
10003 The command accepts the same syntax of the corresponding option.
10005 If the specified expression is not valid, it is kept at its current
10011 Draw a text string or text from a specified file on top of a video, using the
10012 libfreetype library.
10014 To enable compilation of this filter, you need to configure FFmpeg with
10015 @code{--enable-libfreetype}.
10016 To enable default font fallback and the @var{font} option you need to
10017 configure FFmpeg with @code{--enable-libfontconfig}.
10018 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10019 @code{--enable-libfribidi}.
10023 It accepts the following parameters:
10028 Used to draw a box around text using the background color.
10029 The value must be either 1 (enable) or 0 (disable).
10030 The default value of @var{box} is 0.
10033 Set the width of the border to be drawn around the box using @var{boxcolor}.
10034 The default value of @var{boxborderw} is 0.
10037 The color to be used for drawing box around text. For the syntax of this
10038 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10040 The default value of @var{boxcolor} is "white".
10043 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10044 The default value of @var{line_spacing} is 0.
10047 Set the width of the border to be drawn around the text using @var{bordercolor}.
10048 The default value of @var{borderw} is 0.
10051 Set the color to be used for drawing border around text. For the syntax of this
10052 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10054 The default value of @var{bordercolor} is "black".
10057 Select how the @var{text} is expanded. Can be either @code{none},
10058 @code{strftime} (deprecated) or
10059 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10063 Set a start time for the count. Value is in microseconds. Only applied
10064 in the deprecated strftime expansion mode. To emulate in normal expansion
10065 mode use the @code{pts} function, supplying the start time (in seconds)
10066 as the second argument.
10069 If true, check and fix text coords to avoid clipping.
10072 The color to be used for drawing fonts. For the syntax of this option, check
10073 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10075 The default value of @var{fontcolor} is "black".
10077 @item fontcolor_expr
10078 String which is expanded the same way as @var{text} to obtain dynamic
10079 @var{fontcolor} value. By default this option has empty value and is not
10080 processed. When this option is set, it overrides @var{fontcolor} option.
10083 The font family to be used for drawing text. By default Sans.
10086 The font file to be used for drawing text. The path must be included.
10087 This parameter is mandatory if the fontconfig support is disabled.
10090 Draw the text applying alpha blending. The value can
10091 be a number between 0.0 and 1.0.
10092 The expression accepts the same variables @var{x, y} as well.
10093 The default value is 1.
10094 Please see @var{fontcolor_expr}.
10097 The font size to be used for drawing text.
10098 The default value of @var{fontsize} is 16.
10101 If set to 1, attempt to shape the text (for example, reverse the order of
10102 right-to-left text and join Arabic characters) before drawing it.
10103 Otherwise, just draw the text exactly as given.
10104 By default 1 (if supported).
10106 @item ft_load_flags
10107 The flags to be used for loading the fonts.
10109 The flags map the corresponding flags supported by libfreetype, and are
10110 a combination of the following values:
10117 @item vertical_layout
10118 @item force_autohint
10121 @item ignore_global_advance_width
10123 @item ignore_transform
10125 @item linear_design
10129 Default value is "default".
10131 For more information consult the documentation for the FT_LOAD_*
10135 The color to be used for drawing a shadow behind the drawn text. For the
10136 syntax of this option, check the @ref{color syntax,,"Color" section in the
10137 ffmpeg-utils manual,ffmpeg-utils}.
10139 The default value of @var{shadowcolor} is "black".
10143 The x and y offsets for the text shadow position with respect to the
10144 position of the text. They can be either positive or negative
10145 values. The default value for both is "0".
10148 The starting frame number for the n/frame_num variable. The default value
10152 The size in number of spaces to use for rendering the tab.
10153 Default value is 4.
10156 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10157 format. It can be used with or without text parameter. @var{timecode_rate}
10158 option must be specified.
10160 @item timecode_rate, rate, r
10161 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10162 integer. Minimum value is "1".
10163 Drop-frame timecode is supported for frame rates 30 & 60.
10166 If set to 1, the output of the timecode option will wrap around at 24 hours.
10167 Default is 0 (disabled).
10170 The text string to be drawn. The text must be a sequence of UTF-8
10171 encoded characters.
10172 This parameter is mandatory if no file is specified with the parameter
10176 A text file containing text to be drawn. The text must be a sequence
10177 of UTF-8 encoded characters.
10179 This parameter is mandatory if no text string is specified with the
10180 parameter @var{text}.
10182 If both @var{text} and @var{textfile} are specified, an error is thrown.
10185 If set to 1, the @var{textfile} will be reloaded before each frame.
10186 Be sure to update it atomically, or it may be read partially, or even fail.
10190 The expressions which specify the offsets where text will be drawn
10191 within the video frame. They are relative to the top/left border of the
10194 The default value of @var{x} and @var{y} is "0".
10196 See below for the list of accepted constants and functions.
10199 The parameters for @var{x} and @var{y} are expressions containing the
10200 following constants and functions:
10204 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10208 horizontal and vertical chroma subsample values. For example for the
10209 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10212 the height of each text line
10220 @item max_glyph_a, ascent
10221 the maximum distance from the baseline to the highest/upper grid
10222 coordinate used to place a glyph outline point, for all the rendered
10224 It is a positive value, due to the grid's orientation with the Y axis
10227 @item max_glyph_d, descent
10228 the maximum distance from the baseline to the lowest grid coordinate
10229 used to place a glyph outline point, for all the rendered glyphs.
10230 This is a negative value, due to the grid's orientation, with the Y axis
10234 maximum glyph height, that is the maximum height for all the glyphs
10235 contained in the rendered text, it is equivalent to @var{ascent} -
10239 maximum glyph width, that is the maximum width for all the glyphs
10240 contained in the rendered text
10243 the number of input frame, starting from 0
10245 @item rand(min, max)
10246 return a random number included between @var{min} and @var{max}
10249 The input sample aspect ratio.
10252 timestamp expressed in seconds, NAN if the input timestamp is unknown
10255 the height of the rendered text
10258 the width of the rendered text
10262 the x and y offset coordinates where the text is drawn.
10264 These parameters allow the @var{x} and @var{y} expressions to refer
10265 to each other, so you can for example specify @code{y=x/dar}.
10268 A one character description of the current frame's picture type.
10271 The current packet's position in the input file or stream
10272 (in bytes, from the start of the input). A value of -1 indicates
10273 this info is not available.
10276 The current packet's duration, in seconds.
10279 The current packet's size (in bytes).
10282 @anchor{drawtext_expansion}
10283 @subsection Text expansion
10285 If @option{expansion} is set to @code{strftime},
10286 the filter recognizes strftime() sequences in the provided text and
10287 expands them accordingly. Check the documentation of strftime(). This
10288 feature is deprecated.
10290 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10292 If @option{expansion} is set to @code{normal} (which is the default),
10293 the following expansion mechanism is used.
10295 The backslash character @samp{\}, followed by any character, always expands to
10296 the second character.
10298 Sequences of the form @code{%@{...@}} are expanded. The text between the
10299 braces is a function name, possibly followed by arguments separated by ':'.
10300 If the arguments contain special characters or delimiters (':' or '@}'),
10301 they should be escaped.
10303 Note that they probably must also be escaped as the value for the
10304 @option{text} option in the filter argument string and as the filter
10305 argument in the filtergraph description, and possibly also for the shell,
10306 that makes up to four levels of escaping; using a text file avoids these
10309 The following functions are available:
10314 The expression evaluation result.
10316 It must take one argument specifying the expression to be evaluated,
10317 which accepts the same constants and functions as the @var{x} and
10318 @var{y} values. Note that not all constants should be used, for
10319 example the text size is not known when evaluating the expression, so
10320 the constants @var{text_w} and @var{text_h} will have an undefined
10323 @item expr_int_format, eif
10324 Evaluate the expression's value and output as formatted integer.
10326 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10327 The second argument specifies the output format. Allowed values are @samp{x},
10328 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10329 @code{printf} function.
10330 The third parameter is optional and sets the number of positions taken by the output.
10331 It can be used to add padding with zeros from the left.
10334 The time at which the filter is running, expressed in UTC.
10335 It can accept an argument: a strftime() format string.
10338 The time at which the filter is running, expressed in the local time zone.
10339 It can accept an argument: a strftime() format string.
10342 Frame metadata. Takes one or two arguments.
10344 The first argument is mandatory and specifies the metadata key.
10346 The second argument is optional and specifies a default value, used when the
10347 metadata key is not found or empty.
10349 Available metadata can be identified by inspecting entries
10350 starting with TAG included within each frame section
10351 printed by running @code{ffprobe -show_frames}.
10353 String metadata generated in filters leading to
10354 the drawtext filter are also available.
10357 The frame number, starting from 0.
10360 A one character description of the current picture type.
10363 The timestamp of the current frame.
10364 It can take up to three arguments.
10366 The first argument is the format of the timestamp; it defaults to @code{flt}
10367 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10368 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10369 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10370 @code{localtime} stands for the timestamp of the frame formatted as
10371 local time zone time.
10373 The second argument is an offset added to the timestamp.
10375 If the format is set to @code{hms}, a third argument @code{24HH} may be
10376 supplied to present the hour part of the formatted timestamp in 24h format
10379 If the format is set to @code{localtime} or @code{gmtime},
10380 a third argument may be supplied: a strftime() format string.
10381 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10384 @subsection Commands
10386 This filter supports altering parameters via commands:
10389 Alter existing filter parameters.
10391 Syntax for the argument is the same as for filter invocation, e.g.
10394 fontsize=56:fontcolor=green:text='Hello World'
10397 Full filter invocation with sendcmd would look like this:
10400 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10404 If the entire argument can't be parsed or applied as valid values then the filter will
10405 continue with its existing parameters.
10407 @subsection Examples
10411 Draw "Test Text" with font FreeSerif, using the default values for the
10412 optional parameters.
10415 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10419 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10420 and y=50 (counting from the top-left corner of the screen), text is
10421 yellow with a red box around it. Both the text and the box have an
10425 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10426 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10429 Note that the double quotes are not necessary if spaces are not used
10430 within the parameter list.
10433 Show the text at the center of the video frame:
10435 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10439 Show the text at a random position, switching to a new position every 30 seconds:
10441 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)"
10445 Show a text line sliding from right to left in the last row of the video
10446 frame. The file @file{LONG_LINE} is assumed to contain a single line
10449 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10453 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10455 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10459 Draw a single green letter "g", at the center of the input video.
10460 The glyph baseline is placed at half screen height.
10462 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10466 Show text for 1 second every 3 seconds:
10468 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10472 Use fontconfig to set the font. Note that the colons need to be escaped.
10474 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10478 Draw "Test Text" with font size dependent on height of the video.
10480 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10484 Print the date of a real-time encoding (see strftime(3)):
10486 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10490 Show text fading in and out (appearing/disappearing):
10493 DS=1.0 # display start
10494 DE=10.0 # display end
10495 FID=1.5 # fade in duration
10496 FOD=5 # fade out duration
10497 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 @}"
10501 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10502 and the @option{fontsize} value are included in the @option{y} offset.
10504 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10505 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10509 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10510 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10511 must have option @option{-export_path_metadata 1} for the special metadata fields
10512 to be available for filters.
10514 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10519 For more information about libfreetype, check:
10520 @url{http://www.freetype.org/}.
10522 For more information about fontconfig, check:
10523 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10525 For more information about libfribidi, check:
10526 @url{http://fribidi.org/}.
10528 @section edgedetect
10530 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10532 The filter accepts the following options:
10537 Set low and high threshold values used by the Canny thresholding
10540 The high threshold selects the "strong" edge pixels, which are then
10541 connected through 8-connectivity with the "weak" edge pixels selected
10542 by the low threshold.
10544 @var{low} and @var{high} threshold values must be chosen in the range
10545 [0,1], and @var{low} should be lesser or equal to @var{high}.
10547 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10551 Define the drawing mode.
10555 Draw white/gray wires on black background.
10558 Mix the colors to create a paint/cartoon effect.
10561 Apply Canny edge detector on all selected planes.
10563 Default value is @var{wires}.
10566 Select planes for filtering. By default all available planes are filtered.
10569 @subsection Examples
10573 Standard edge detection with custom values for the hysteresis thresholding:
10575 edgedetect=low=0.1:high=0.4
10579 Painting effect without thresholding:
10581 edgedetect=mode=colormix:high=0
10587 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10589 For each input image, the filter will compute the optimal mapping from
10590 the input to the output given the codebook length, that is the number
10591 of distinct output colors.
10593 This filter accepts the following options.
10596 @item codebook_length, l
10597 Set codebook length. The value must be a positive integer, and
10598 represents the number of distinct output colors. Default value is 256.
10601 Set the maximum number of iterations to apply for computing the optimal
10602 mapping. The higher the value the better the result and the higher the
10603 computation time. Default value is 1.
10606 Set a random seed, must be an integer included between 0 and
10607 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10608 will try to use a good random seed on a best effort basis.
10611 Set pal8 output pixel format. This option does not work with codebook
10612 length greater than 256.
10617 Measure graylevel entropy in histogram of color channels of video frames.
10619 It accepts the following parameters:
10623 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10625 @var{diff} mode measures entropy of histogram delta values, absolute differences
10626 between neighbour histogram values.
10630 Set brightness, contrast, saturation and approximate gamma adjustment.
10632 The filter accepts the following options:
10636 Set the contrast expression. The value must be a float value in range
10637 @code{-1000.0} to @code{1000.0}. The default value is "1".
10640 Set the brightness expression. The value must be a float value in
10641 range @code{-1.0} to @code{1.0}. The default value is "0".
10644 Set the saturation expression. The value must be a float in
10645 range @code{0.0} to @code{3.0}. The default value is "1".
10648 Set the gamma expression. The value must be a float in range
10649 @code{0.1} to @code{10.0}. The default value is "1".
10652 Set the gamma expression for red. The value must be a float in
10653 range @code{0.1} to @code{10.0}. The default value is "1".
10656 Set the gamma expression for green. The value must be a float in range
10657 @code{0.1} to @code{10.0}. The default value is "1".
10660 Set the gamma expression for blue. The value must be a float in range
10661 @code{0.1} to @code{10.0}. The default value is "1".
10664 Set the gamma weight expression. It can be used to reduce the effect
10665 of a high gamma value on bright image areas, e.g. keep them from
10666 getting overamplified and just plain white. The value must be a float
10667 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10668 gamma correction all the way down while @code{1.0} leaves it at its
10669 full strength. Default is "1".
10672 Set when the expressions for brightness, contrast, saturation and
10673 gamma expressions are evaluated.
10675 It accepts the following values:
10678 only evaluate expressions once during the filter initialization or
10679 when a command is processed
10682 evaluate expressions for each incoming frame
10685 Default value is @samp{init}.
10688 The expressions accept the following parameters:
10691 frame count of the input frame starting from 0
10694 byte position of the corresponding packet in the input file, NAN if
10698 frame rate of the input video, NAN if the input frame rate is unknown
10701 timestamp expressed in seconds, NAN if the input timestamp is unknown
10704 @subsection Commands
10705 The filter supports the following commands:
10709 Set the contrast expression.
10712 Set the brightness expression.
10715 Set the saturation expression.
10718 Set the gamma expression.
10721 Set the gamma_r expression.
10724 Set gamma_g expression.
10727 Set gamma_b expression.
10730 Set gamma_weight expression.
10732 The command accepts the same syntax of the corresponding option.
10734 If the specified expression is not valid, it is kept at its current
10741 Apply erosion effect to the video.
10743 This filter replaces the pixel by the local(3x3) minimum.
10745 It accepts the following options:
10752 Limit the maximum change for each plane, default is 65535.
10753 If 0, plane will remain unchanged.
10756 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10759 Flags to local 3x3 coordinates maps like this:
10766 @subsection Commands
10768 This filter supports the all above options as @ref{commands}.
10770 @section extractplanes
10772 Extract color channel components from input video stream into
10773 separate grayscale video streams.
10775 The filter accepts the following option:
10779 Set plane(s) to extract.
10781 Available values for planes are:
10792 Choosing planes not available in the input will result in an error.
10793 That means you cannot select @code{r}, @code{g}, @code{b} planes
10794 with @code{y}, @code{u}, @code{v} planes at same time.
10797 @subsection Examples
10801 Extract luma, u and v color channel component from input video frame
10802 into 3 grayscale outputs:
10804 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
10810 Apply a fade-in/out effect to the input video.
10812 It accepts the following parameters:
10816 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10818 Default is @code{in}.
10820 @item start_frame, s
10821 Specify the number of the frame to start applying the fade
10822 effect at. Default is 0.
10825 The number of frames that the fade effect lasts. At the end of the
10826 fade-in effect, the output video will have the same intensity as the input video.
10827 At the end of the fade-out transition, the output video will be filled with the
10828 selected @option{color}.
10832 If set to 1, fade only alpha channel, if one exists on the input.
10833 Default value is 0.
10835 @item start_time, st
10836 Specify the timestamp (in seconds) of the frame to start to apply the fade
10837 effect. If both start_frame and start_time are specified, the fade will start at
10838 whichever comes last. Default is 0.
10841 The number of seconds for which the fade effect has to last. At the end of the
10842 fade-in effect the output video will have the same intensity as the input video,
10843 at the end of the fade-out transition the output video will be filled with the
10844 selected @option{color}.
10845 If both duration and nb_frames are specified, duration is used. Default is 0
10846 (nb_frames is used by default).
10849 Specify the color of the fade. Default is "black".
10852 @subsection Examples
10856 Fade in the first 30 frames of video:
10861 The command above is equivalent to:
10867 Fade out the last 45 frames of a 200-frame video:
10870 fade=type=out:start_frame=155:nb_frames=45
10874 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10876 fade=in:0:25, fade=out:975:25
10880 Make the first 5 frames yellow, then fade in from frame 5-24:
10882 fade=in:5:20:color=yellow
10886 Fade in alpha over first 25 frames of video:
10888 fade=in:0:25:alpha=1
10892 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10894 fade=t=in:st=5.5:d=0.5
10900 Denoise frames using 3D FFT (frequency domain filtering).
10902 The filter accepts the following options:
10906 Set the noise sigma constant. This sets denoising strength.
10907 Default value is 1. Allowed range is from 0 to 30.
10908 Using very high sigma with low overlap may give blocking artifacts.
10911 Set amount of denoising. By default all detected noise is reduced.
10912 Default value is 1. Allowed range is from 0 to 1.
10915 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10916 Actual size of block in pixels is 2 to power of @var{block}, so by default
10917 block size in pixels is 2^4 which is 16.
10920 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10923 Set number of previous frames to use for denoising. By default is set to 0.
10926 Set number of next frames to to use for denoising. By default is set to 0.
10929 Set planes which will be filtered, by default are all available filtered
10934 Apply arbitrary expressions to samples in frequency domain
10938 Adjust the dc value (gain) of the luma plane of the image. The filter
10939 accepts an integer value in range @code{0} to @code{1000}. The default
10940 value is set to @code{0}.
10943 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10944 filter accepts an integer value in range @code{0} to @code{1000}. The
10945 default value is set to @code{0}.
10948 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10949 filter accepts an integer value in range @code{0} to @code{1000}. The
10950 default value is set to @code{0}.
10953 Set the frequency domain weight expression for the luma plane.
10956 Set the frequency domain weight expression for the 1st chroma plane.
10959 Set the frequency domain weight expression for the 2nd chroma plane.
10962 Set when the expressions are evaluated.
10964 It accepts the following values:
10967 Only evaluate expressions once during the filter initialization.
10970 Evaluate expressions for each incoming frame.
10973 Default value is @samp{init}.
10975 The filter accepts the following variables:
10978 The coordinates of the current sample.
10982 The width and height of the image.
10985 The number of input frame, starting from 0.
10988 @subsection Examples
10994 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11000 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11006 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11012 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11019 Extract a single field from an interlaced image using stride
11020 arithmetic to avoid wasting CPU time. The output frames are marked as
11023 The filter accepts the following options:
11027 Specify whether to extract the top (if the value is @code{0} or
11028 @code{top}) or the bottom field (if the value is @code{1} or
11034 Create new frames by copying the top and bottom fields from surrounding frames
11035 supplied as numbers by the hint file.
11039 Set file containing hints: absolute/relative frame numbers.
11041 There must be one line for each frame in a clip. Each line must contain two
11042 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11043 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11044 is current frame number for @code{absolute} mode or out of [-1, 1] range
11045 for @code{relative} mode. First number tells from which frame to pick up top
11046 field and second number tells from which frame to pick up bottom field.
11048 If optionally followed by @code{+} output frame will be marked as interlaced,
11049 else if followed by @code{-} output frame will be marked as progressive, else
11050 it will be marked same as input frame.
11051 If optionally followed by @code{t} output frame will use only top field, or in
11052 case of @code{b} it will use only bottom field.
11053 If line starts with @code{#} or @code{;} that line is skipped.
11056 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11059 Example of first several lines of @code{hint} file for @code{relative} mode:
11061 0,0 - # first frame
11062 1,0 - # second frame, use third's frame top field and second's frame bottom field
11063 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11078 @section fieldmatch
11080 Field matching filter for inverse telecine. It is meant to reconstruct the
11081 progressive frames from a telecined stream. The filter does not drop duplicated
11082 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11083 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11085 The separation of the field matching and the decimation is notably motivated by
11086 the possibility of inserting a de-interlacing filter fallback between the two.
11087 If the source has mixed telecined and real interlaced content,
11088 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11089 But these remaining combed frames will be marked as interlaced, and thus can be
11090 de-interlaced by a later filter such as @ref{yadif} before decimation.
11092 In addition to the various configuration options, @code{fieldmatch} can take an
11093 optional second stream, activated through the @option{ppsrc} option. If
11094 enabled, the frames reconstruction will be based on the fields and frames from
11095 this second stream. This allows the first input to be pre-processed in order to
11096 help the various algorithms of the filter, while keeping the output lossless
11097 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11098 or brightness/contrast adjustments can help.
11100 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11101 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11102 which @code{fieldmatch} is based on. While the semantic and usage are very
11103 close, some behaviour and options names can differ.
11105 The @ref{decimate} filter currently only works for constant frame rate input.
11106 If your input has mixed telecined (30fps) and progressive content with a lower
11107 framerate like 24fps use the following filterchain to produce the necessary cfr
11108 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11110 The filter accepts the following options:
11114 Specify the assumed field order of the input stream. Available values are:
11118 Auto detect parity (use FFmpeg's internal parity value).
11120 Assume bottom field first.
11122 Assume top field first.
11125 Note that it is sometimes recommended not to trust the parity announced by the
11128 Default value is @var{auto}.
11131 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11132 sense that it won't risk creating jerkiness due to duplicate frames when
11133 possible, but if there are bad edits or blended fields it will end up
11134 outputting combed frames when a good match might actually exist. On the other
11135 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11136 but will almost always find a good frame if there is one. The other values are
11137 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11138 jerkiness and creating duplicate frames versus finding good matches in sections
11139 with bad edits, orphaned fields, blended fields, etc.
11141 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11143 Available values are:
11147 2-way matching (p/c)
11149 2-way matching, and trying 3rd match if still combed (p/c + n)
11151 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11153 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11154 still combed (p/c + n + u/b)
11156 3-way matching (p/c/n)
11158 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11159 detected as combed (p/c/n + u/b)
11162 The parenthesis at the end indicate the matches that would be used for that
11163 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11166 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11169 Default value is @var{pc_n}.
11172 Mark the main input stream as a pre-processed input, and enable the secondary
11173 input stream as the clean source to pick the fields from. See the filter
11174 introduction for more details. It is similar to the @option{clip2} feature from
11177 Default value is @code{0} (disabled).
11180 Set the field to match from. It is recommended to set this to the same value as
11181 @option{order} unless you experience matching failures with that setting. In
11182 certain circumstances changing the field that is used to match from can have a
11183 large impact on matching performance. Available values are:
11187 Automatic (same value as @option{order}).
11189 Match from the bottom field.
11191 Match from the top field.
11194 Default value is @var{auto}.
11197 Set whether or not chroma is included during the match comparisons. In most
11198 cases it is recommended to leave this enabled. You should set this to @code{0}
11199 only if your clip has bad chroma problems such as heavy rainbowing or other
11200 artifacts. Setting this to @code{0} could also be used to speed things up at
11201 the cost of some accuracy.
11203 Default value is @code{1}.
11207 These define an exclusion band which excludes the lines between @option{y0} and
11208 @option{y1} from being included in the field matching decision. An exclusion
11209 band can be used to ignore subtitles, a logo, or other things that may
11210 interfere with the matching. @option{y0} sets the starting scan line and
11211 @option{y1} sets the ending line; all lines in between @option{y0} and
11212 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11213 @option{y0} and @option{y1} to the same value will disable the feature.
11214 @option{y0} and @option{y1} defaults to @code{0}.
11217 Set the scene change detection threshold as a percentage of maximum change on
11218 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11219 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11220 @option{scthresh} is @code{[0.0, 100.0]}.
11222 Default value is @code{12.0}.
11225 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11226 account the combed scores of matches when deciding what match to use as the
11227 final match. Available values are:
11231 No final matching based on combed scores.
11233 Combed scores are only used when a scene change is detected.
11235 Use combed scores all the time.
11238 Default is @var{sc}.
11241 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11242 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11243 Available values are:
11247 No forced calculation.
11249 Force p/c/n calculations.
11251 Force p/c/n/u/b calculations.
11254 Default value is @var{none}.
11257 This is the area combing threshold used for combed frame detection. This
11258 essentially controls how "strong" or "visible" combing must be to be detected.
11259 Larger values mean combing must be more visible and smaller values mean combing
11260 can be less visible or strong and still be detected. Valid settings are from
11261 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11262 be detected as combed). This is basically a pixel difference value. A good
11263 range is @code{[8, 12]}.
11265 Default value is @code{9}.
11268 Sets whether or not chroma is considered in the combed frame decision. Only
11269 disable this if your source has chroma problems (rainbowing, etc.) that are
11270 causing problems for the combed frame detection with chroma enabled. Actually,
11271 using @option{chroma}=@var{0} is usually more reliable, except for the case
11272 where there is chroma only combing in the source.
11274 Default value is @code{0}.
11278 Respectively set the x-axis and y-axis size of the window used during combed
11279 frame detection. This has to do with the size of the area in which
11280 @option{combpel} pixels are required to be detected as combed for a frame to be
11281 declared combed. See the @option{combpel} parameter description for more info.
11282 Possible values are any number that is a power of 2 starting at 4 and going up
11285 Default value is @code{16}.
11288 The number of combed pixels inside any of the @option{blocky} by
11289 @option{blockx} size blocks on the frame for the frame to be detected as
11290 combed. While @option{cthresh} controls how "visible" the combing must be, this
11291 setting controls "how much" combing there must be in any localized area (a
11292 window defined by the @option{blockx} and @option{blocky} settings) on the
11293 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11294 which point no frames will ever be detected as combed). This setting is known
11295 as @option{MI} in TFM/VFM vocabulary.
11297 Default value is @code{80}.
11300 @anchor{p/c/n/u/b meaning}
11301 @subsection p/c/n/u/b meaning
11303 @subsubsection p/c/n
11305 We assume the following telecined stream:
11308 Top fields: 1 2 2 3 4
11309 Bottom fields: 1 2 3 4 4
11312 The numbers correspond to the progressive frame the fields relate to. Here, the
11313 first two frames are progressive, the 3rd and 4th are combed, and so on.
11315 When @code{fieldmatch} is configured to run a matching from bottom
11316 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11321 B 1 2 3 4 4 <-- matching reference
11330 As a result of the field matching, we can see that some frames get duplicated.
11331 To perform a complete inverse telecine, you need to rely on a decimation filter
11332 after this operation. See for instance the @ref{decimate} filter.
11334 The same operation now matching from top fields (@option{field}=@var{top})
11339 T 1 2 2 3 4 <-- matching reference
11349 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11350 basically, they refer to the frame and field of the opposite parity:
11353 @item @var{p} matches the field of the opposite parity in the previous frame
11354 @item @var{c} matches the field of the opposite parity in the current frame
11355 @item @var{n} matches the field of the opposite parity in the next frame
11360 The @var{u} and @var{b} matching are a bit special in the sense that they match
11361 from the opposite parity flag. In the following examples, we assume that we are
11362 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11363 'x' is placed above and below each matched fields.
11365 With bottom matching (@option{field}=@var{bottom}):
11370 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11371 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11379 With top matching (@option{field}=@var{top}):
11384 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11385 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11393 @subsection Examples
11395 Simple IVTC of a top field first telecined stream:
11397 fieldmatch=order=tff:combmatch=none, decimate
11400 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11402 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11405 @section fieldorder
11407 Transform the field order of the input video.
11409 It accepts the following parameters:
11414 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11415 for bottom field first.
11418 The default value is @samp{tff}.
11420 The transformation is done by shifting the picture content up or down
11421 by one line, and filling the remaining line with appropriate picture content.
11422 This method is consistent with most broadcast field order converters.
11424 If the input video is not flagged as being interlaced, or it is already
11425 flagged as being of the required output field order, then this filter does
11426 not alter the incoming video.
11428 It is very useful when converting to or from PAL DV material,
11429 which is bottom field first.
11433 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11436 @section fifo, afifo
11438 Buffer input images and send them when they are requested.
11440 It is mainly useful when auto-inserted by the libavfilter
11443 It does not take parameters.
11445 @section fillborders
11447 Fill borders of the input video, without changing video stream dimensions.
11448 Sometimes video can have garbage at the four edges and you may not want to
11449 crop video input to keep size multiple of some number.
11451 This filter accepts the following options:
11455 Number of pixels to fill from left border.
11458 Number of pixels to fill from right border.
11461 Number of pixels to fill from top border.
11464 Number of pixels to fill from bottom border.
11469 It accepts the following values:
11472 fill pixels using outermost pixels
11475 fill pixels using mirroring
11478 fill pixels with constant value
11481 Default is @var{smear}.
11484 Set color for pixels in fixed mode. Default is @var{black}.
11487 @subsection Commands
11488 This filter supports same @ref{commands} as options.
11489 The command accepts the same syntax of the corresponding option.
11491 If the specified expression is not valid, it is kept at its current
11496 Find a rectangular object
11498 It accepts the following options:
11502 Filepath of the object image, needs to be in gray8.
11505 Detection threshold, default is 0.5.
11508 Number of mipmaps, default is 3.
11510 @item xmin, ymin, xmax, ymax
11511 Specifies the rectangle in which to search.
11514 @subsection Examples
11518 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11520 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11526 Flood area with values of same pixel components with another values.
11528 It accepts the following options:
11531 Set pixel x coordinate.
11534 Set pixel y coordinate.
11537 Set source #0 component value.
11540 Set source #1 component value.
11543 Set source #2 component value.
11546 Set source #3 component value.
11549 Set destination #0 component value.
11552 Set destination #1 component value.
11555 Set destination #2 component value.
11558 Set destination #3 component value.
11564 Convert the input video to one of the specified pixel formats.
11565 Libavfilter will try to pick one that is suitable as input to
11568 It accepts the following parameters:
11572 A '|'-separated list of pixel format names, such as
11573 "pix_fmts=yuv420p|monow|rgb24".
11577 @subsection Examples
11581 Convert the input video to the @var{yuv420p} format
11583 format=pix_fmts=yuv420p
11586 Convert the input video to any of the formats in the list
11588 format=pix_fmts=yuv420p|yuv444p|yuv410p
11595 Convert the video to specified constant frame rate by duplicating or dropping
11596 frames as necessary.
11598 It accepts the following parameters:
11602 The desired output frame rate. The default is @code{25}.
11605 Assume the first PTS should be the given value, in seconds. This allows for
11606 padding/trimming at the start of stream. By default, no assumption is made
11607 about the first frame's expected PTS, so no padding or trimming is done.
11608 For example, this could be set to 0 to pad the beginning with duplicates of
11609 the first frame if a video stream starts after the audio stream or to trim any
11610 frames with a negative PTS.
11613 Timestamp (PTS) rounding method.
11615 Possible values are:
11622 round towards -infinity
11624 round towards +infinity
11628 The default is @code{near}.
11631 Action performed when reading the last frame.
11633 Possible values are:
11636 Use same timestamp rounding method as used for other frames.
11638 Pass through last frame if input duration has not been reached yet.
11640 The default is @code{round}.
11644 Alternatively, the options can be specified as a flat string:
11645 @var{fps}[:@var{start_time}[:@var{round}]].
11647 See also the @ref{setpts} filter.
11649 @subsection Examples
11653 A typical usage in order to set the fps to 25:
11659 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11661 fps=fps=film:round=near
11667 Pack two different video streams into a stereoscopic video, setting proper
11668 metadata on supported codecs. The two views should have the same size and
11669 framerate and processing will stop when the shorter video ends. Please note
11670 that you may conveniently adjust view properties with the @ref{scale} and
11673 It accepts the following parameters:
11677 The desired packing format. Supported values are:
11682 The views are next to each other (default).
11685 The views are on top of each other.
11688 The views are packed by line.
11691 The views are packed by column.
11694 The views are temporally interleaved.
11703 # Convert left and right views into a frame-sequential video
11704 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11706 # Convert views into a side-by-side video with the same output resolution as the input
11707 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
11712 Change the frame rate by interpolating new video output frames from the source
11715 This filter is not designed to function correctly with interlaced media. If
11716 you wish to change the frame rate of interlaced media then you are required
11717 to deinterlace before this filter and re-interlace after this filter.
11719 A description of the accepted options follows.
11723 Specify the output frames per second. This option can also be specified
11724 as a value alone. The default is @code{50}.
11727 Specify the start of a range where the output frame will be created as a
11728 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11729 the default is @code{15}.
11732 Specify the end of a range where the output frame will be created as a
11733 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11734 the default is @code{240}.
11737 Specify the level at which a scene change is detected as a value between
11738 0 and 100 to indicate a new scene; a low value reflects a low
11739 probability for the current frame to introduce a new scene, while a higher
11740 value means the current frame is more likely to be one.
11741 The default is @code{8.2}.
11744 Specify flags influencing the filter process.
11746 Available value for @var{flags} is:
11749 @item scene_change_detect, scd
11750 Enable scene change detection using the value of the option @var{scene}.
11751 This flag is enabled by default.
11757 Select one frame every N-th frame.
11759 This filter accepts the following option:
11762 Select frame after every @code{step} frames.
11763 Allowed values are positive integers higher than 0. Default value is @code{1}.
11766 @section freezedetect
11768 Detect frozen video.
11770 This filter logs a message and sets frame metadata when it detects that the
11771 input video has no significant change in content during a specified duration.
11772 Video freeze detection calculates the mean average absolute difference of all
11773 the components of video frames and compares it to a noise floor.
11775 The printed times and duration are expressed in seconds. The
11776 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11777 whose timestamp equals or exceeds the detection duration and it contains the
11778 timestamp of the first frame of the freeze. The
11779 @code{lavfi.freezedetect.freeze_duration} and
11780 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11783 The filter accepts the following options:
11787 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11788 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11792 Set freeze duration until notification (default is 2 seconds).
11795 @section freezeframes
11797 Freeze video frames.
11799 This filter freezes video frames using frame from 2nd input.
11801 The filter accepts the following options:
11805 Set number of first frame from which to start freeze.
11808 Set number of last frame from which to end freeze.
11811 Set number of frame from 2nd input which will be used instead of replaced frames.
11817 Apply a frei0r effect to the input video.
11819 To enable the compilation of this filter, you need to install the frei0r
11820 header and configure FFmpeg with @code{--enable-frei0r}.
11822 It accepts the following parameters:
11827 The name of the frei0r effect to load. If the environment variable
11828 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11829 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11830 Otherwise, the standard frei0r paths are searched, in this order:
11831 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11832 @file{/usr/lib/frei0r-1/}.
11834 @item filter_params
11835 A '|'-separated list of parameters to pass to the frei0r effect.
11839 A frei0r effect parameter can be a boolean (its value is either
11840 "y" or "n"), a double, a color (specified as
11841 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11842 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11843 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11844 a position (specified as @var{X}/@var{Y}, where
11845 @var{X} and @var{Y} are floating point numbers) and/or a string.
11847 The number and types of parameters depend on the loaded effect. If an
11848 effect parameter is not specified, the default value is set.
11850 @subsection Examples
11854 Apply the distort0r effect, setting the first two double parameters:
11856 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11860 Apply the colordistance effect, taking a color as the first parameter:
11862 frei0r=colordistance:0.2/0.3/0.4
11863 frei0r=colordistance:violet
11864 frei0r=colordistance:0x112233
11868 Apply the perspective effect, specifying the top left and top right image
11871 frei0r=perspective:0.2/0.2|0.8/0.2
11875 For more information, see
11876 @url{http://frei0r.dyne.org}
11878 @subsection Commands
11880 This filter supports the @option{filter_params} option as @ref{commands}.
11884 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11886 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11887 processing filter, one of them is performed once per block, not per pixel.
11888 This allows for much higher speed.
11890 The filter accepts the following options:
11894 Set quality. This option defines the number of levels for averaging. It accepts
11895 an integer in the range 4-5. Default value is @code{4}.
11898 Force a constant quantization parameter. It accepts an integer in range 0-63.
11899 If not set, the filter will use the QP from the video stream (if available).
11902 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11903 more details but also more artifacts, while higher values make the image smoother
11904 but also blurrier. Default value is @code{0} − PSNR optimal.
11906 @item use_bframe_qp
11907 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11908 option may cause flicker since the B-Frames have often larger QP. Default is
11909 @code{0} (not enabled).
11915 Apply Gaussian blur filter.
11917 The filter accepts the following options:
11921 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11924 Set number of steps for Gaussian approximation. Default is @code{1}.
11927 Set which planes to filter. By default all planes are filtered.
11930 Set vertical sigma, if negative it will be same as @code{sigma}.
11931 Default is @code{-1}.
11934 @subsection Commands
11935 This filter supports same commands as options.
11936 The command accepts the same syntax of the corresponding option.
11938 If the specified expression is not valid, it is kept at its current
11943 Apply generic equation to each pixel.
11945 The filter accepts the following options:
11948 @item lum_expr, lum
11949 Set the luminance expression.
11951 Set the chrominance blue expression.
11953 Set the chrominance red expression.
11954 @item alpha_expr, a
11955 Set the alpha expression.
11957 Set the red expression.
11958 @item green_expr, g
11959 Set the green expression.
11961 Set the blue expression.
11964 The colorspace is selected according to the specified options. If one
11965 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11966 options is specified, the filter will automatically select a YCbCr
11967 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11968 @option{blue_expr} options is specified, it will select an RGB
11971 If one of the chrominance expression is not defined, it falls back on the other
11972 one. If no alpha expression is specified it will evaluate to opaque value.
11973 If none of chrominance expressions are specified, they will evaluate
11974 to the luminance expression.
11976 The expressions can use the following variables and functions:
11980 The sequential number of the filtered frame, starting from @code{0}.
11984 The coordinates of the current sample.
11988 The width and height of the image.
11992 Width and height scale depending on the currently filtered plane. It is the
11993 ratio between the corresponding luma plane number of pixels and the current
11994 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11995 @code{0.5,0.5} for chroma planes.
11998 Time of the current frame, expressed in seconds.
12001 Return the value of the pixel at location (@var{x},@var{y}) of the current
12005 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12009 Return the value of the pixel at location (@var{x},@var{y}) of the
12010 blue-difference chroma plane. Return 0 if there is no such plane.
12013 Return the value of the pixel at location (@var{x},@var{y}) of the
12014 red-difference chroma plane. Return 0 if there is no such plane.
12019 Return the value of the pixel at location (@var{x},@var{y}) of the
12020 red/green/blue component. Return 0 if there is no such component.
12023 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12024 plane. Return 0 if there is no such plane.
12026 @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)
12027 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12028 sums of samples within a rectangle. See the functions without the sum postfix.
12030 @item interpolation
12031 Set one of interpolation methods:
12036 Default is bilinear.
12039 For functions, if @var{x} and @var{y} are outside the area, the value will be
12040 automatically clipped to the closer edge.
12042 Please note that this filter can use multiple threads in which case each slice
12043 will have its own expression state. If you want to use only a single expression
12044 state because your expressions depend on previous state then you should limit
12045 the number of filter threads to 1.
12047 @subsection Examples
12051 Flip the image horizontally:
12057 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12058 wavelength of 100 pixels:
12060 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12064 Generate a fancy enigmatic moving light:
12066 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
12070 Generate a quick emboss effect:
12072 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12076 Modify RGB components depending on pixel position:
12078 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12082 Create a radial gradient that is the same size as the input (also see
12083 the @ref{vignette} filter):
12085 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12091 Fix the banding artifacts that are sometimes introduced into nearly flat
12092 regions by truncation to 8-bit color depth.
12093 Interpolate the gradients that should go where the bands are, and
12096 It is designed for playback only. Do not use it prior to
12097 lossy compression, because compression tends to lose the dither and
12098 bring back the bands.
12100 It accepts the following parameters:
12105 The maximum amount by which the filter will change any one pixel. This is also
12106 the threshold for detecting nearly flat regions. Acceptable values range from
12107 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12111 The neighborhood to fit the gradient to. A larger radius makes for smoother
12112 gradients, but also prevents the filter from modifying the pixels near detailed
12113 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12114 values will be clipped to the valid range.
12118 Alternatively, the options can be specified as a flat string:
12119 @var{strength}[:@var{radius}]
12121 @subsection Examples
12125 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12131 Specify radius, omitting the strength (which will fall-back to the default
12139 @anchor{graphmonitor}
12140 @section graphmonitor
12141 Show various filtergraph stats.
12143 With this filter one can debug complete filtergraph.
12144 Especially issues with links filling with queued frames.
12146 The filter accepts the following options:
12150 Set video output size. Default is @var{hd720}.
12153 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12156 Set output mode, can be @var{fulll} or @var{compact}.
12157 In @var{compact} mode only filters with some queued frames have displayed stats.
12160 Set flags which enable which stats are shown in video.
12162 Available values for flags are:
12165 Display number of queued frames in each link.
12167 @item frame_count_in
12168 Display number of frames taken from filter.
12170 @item frame_count_out
12171 Display number of frames given out from filter.
12174 Display current filtered frame pts.
12177 Display current filtered frame time.
12180 Display time base for filter link.
12183 Display used format for filter link.
12186 Display video size or number of audio channels in case of audio used by filter link.
12189 Display video frame rate or sample rate in case of audio used by filter link.
12192 Display link output status.
12196 Set upper limit for video rate of output stream, Default value is @var{25}.
12197 This guarantee that output video frame rate will not be higher than this value.
12201 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12202 and corrects the scene colors accordingly.
12204 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12206 The filter accepts the following options:
12210 The order of differentiation to be applied on the scene. Must be chosen in the range
12211 [0,2] and default value is 1.
12214 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12215 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12216 max value instead of calculating Minkowski distance.
12219 The standard deviation of Gaussian blur to be applied on the scene. Must be
12220 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12221 can't be equal to 0 if @var{difford} is greater than 0.
12224 @subsection Examples
12230 greyedge=difford=1:minknorm=5:sigma=2
12236 greyedge=difford=1:minknorm=0:sigma=2
12244 Apply a Hald CLUT to a video stream.
12246 First input is the video stream to process, and second one is the Hald CLUT.
12247 The Hald CLUT input can be a simple picture or a complete video stream.
12249 The filter accepts the following options:
12253 Force termination when the shortest input terminates. Default is @code{0}.
12255 Continue applying the last CLUT after the end of the stream. A value of
12256 @code{0} disable the filter after the last frame of the CLUT is reached.
12257 Default is @code{1}.
12260 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12261 filters share the same internals).
12263 This filter also supports the @ref{framesync} options.
12265 More information about the Hald CLUT can be found on Eskil Steenberg's website
12266 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12268 @subsection Workflow examples
12270 @subsubsection Hald CLUT video stream
12272 Generate an identity Hald CLUT stream altered with various effects:
12274 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
12277 Note: make sure you use a lossless codec.
12279 Then use it with @code{haldclut} to apply it on some random stream:
12281 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12284 The Hald CLUT will be applied to the 10 first seconds (duration of
12285 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12286 to the remaining frames of the @code{mandelbrot} stream.
12288 @subsubsection Hald CLUT with preview
12290 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12291 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12292 biggest possible square starting at the top left of the picture. The remaining
12293 padding pixels (bottom or right) will be ignored. This area can be used to add
12294 a preview of the Hald CLUT.
12296 Typically, the following generated Hald CLUT will be supported by the
12297 @code{haldclut} filter:
12300 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12301 pad=iw+320 [padded_clut];
12302 smptebars=s=320x256, split [a][b];
12303 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12304 [main][b] overlay=W-320" -frames:v 1 clut.png
12307 It contains the original and a preview of the effect of the CLUT: SMPTE color
12308 bars are displayed on the right-top, and below the same color bars processed by
12311 Then, the effect of this Hald CLUT can be visualized with:
12313 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12318 Flip the input video horizontally.
12320 For example, to horizontally flip the input video with @command{ffmpeg}:
12322 ffmpeg -i in.avi -vf "hflip" out.avi
12326 This filter applies a global color histogram equalization on a
12329 It can be used to correct video that has a compressed range of pixel
12330 intensities. The filter redistributes the pixel intensities to
12331 equalize their distribution across the intensity range. It may be
12332 viewed as an "automatically adjusting contrast filter". This filter is
12333 useful only for correcting degraded or poorly captured source
12336 The filter accepts the following options:
12340 Determine the amount of equalization to be applied. As the strength
12341 is reduced, the distribution of pixel intensities more-and-more
12342 approaches that of the input frame. The value must be a float number
12343 in the range [0,1] and defaults to 0.200.
12346 Set the maximum intensity that can generated and scale the output
12347 values appropriately. The strength should be set as desired and then
12348 the intensity can be limited if needed to avoid washing-out. The value
12349 must be a float number in the range [0,1] and defaults to 0.210.
12352 Set the antibanding level. If enabled the filter will randomly vary
12353 the luminance of output pixels by a small amount to avoid banding of
12354 the histogram. Possible values are @code{none}, @code{weak} or
12355 @code{strong}. It defaults to @code{none}.
12361 Compute and draw a color distribution histogram for the input video.
12363 The computed histogram is a representation of the color component
12364 distribution in an image.
12366 Standard histogram displays the color components distribution in an image.
12367 Displays color graph for each color component. Shows distribution of
12368 the Y, U, V, A or R, G, B components, depending on input format, in the
12369 current frame. Below each graph a color component scale meter is shown.
12371 The filter accepts the following options:
12375 Set height of level. Default value is @code{200}.
12376 Allowed range is [50, 2048].
12379 Set height of color scale. Default value is @code{12}.
12380 Allowed range is [0, 40].
12384 It accepts the following values:
12387 Per color component graphs are placed below each other.
12390 Per color component graphs are placed side by side.
12393 Presents information identical to that in the @code{parade}, except
12394 that the graphs representing color components are superimposed directly
12397 Default is @code{stack}.
12400 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12401 Default is @code{linear}.
12404 Set what color components to display.
12405 Default is @code{7}.
12408 Set foreground opacity. Default is @code{0.7}.
12411 Set background opacity. Default is @code{0.5}.
12414 @subsection Examples
12419 Calculate and draw histogram:
12421 ffplay -i input -vf histogram
12429 This is a high precision/quality 3d denoise filter. It aims to reduce
12430 image noise, producing smooth images and making still images really
12431 still. It should enhance compressibility.
12433 It accepts the following optional parameters:
12437 A non-negative floating point number which specifies spatial luma strength.
12438 It defaults to 4.0.
12440 @item chroma_spatial
12441 A non-negative floating point number which specifies spatial chroma strength.
12442 It defaults to 3.0*@var{luma_spatial}/4.0.
12445 A floating point number which specifies luma temporal strength. It defaults to
12446 6.0*@var{luma_spatial}/4.0.
12449 A floating point number which specifies chroma temporal strength. It defaults to
12450 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12453 @subsection Commands
12454 This filter supports same @ref{commands} as options.
12455 The command accepts the same syntax of the corresponding option.
12457 If the specified expression is not valid, it is kept at its current
12460 @anchor{hwdownload}
12461 @section hwdownload
12463 Download hardware frames to system memory.
12465 The input must be in hardware frames, and the output a non-hardware format.
12466 Not all formats will be supported on the output - it may be necessary to insert
12467 an additional @option{format} filter immediately following in the graph to get
12468 the output in a supported format.
12472 Map hardware frames to system memory or to another device.
12474 This filter has several different modes of operation; which one is used depends
12475 on the input and output formats:
12478 Hardware frame input, normal frame output
12480 Map the input frames to system memory and pass them to the output. If the
12481 original hardware frame is later required (for example, after overlaying
12482 something else on part of it), the @option{hwmap} filter can be used again
12483 in the next mode to retrieve it.
12485 Normal frame input, hardware frame output
12487 If the input is actually a software-mapped hardware frame, then unmap it -
12488 that is, return the original hardware frame.
12490 Otherwise, a device must be provided. Create new hardware surfaces on that
12491 device for the output, then map them back to the software format at the input
12492 and give those frames to the preceding filter. This will then act like the
12493 @option{hwupload} filter, but may be able to avoid an additional copy when
12494 the input is already in a compatible format.
12496 Hardware frame input and output
12498 A device must be supplied for the output, either directly or with the
12499 @option{derive_device} option. The input and output devices must be of
12500 different types and compatible - the exact meaning of this is
12501 system-dependent, but typically it means that they must refer to the same
12502 underlying hardware context (for example, refer to the same graphics card).
12504 If the input frames were originally created on the output device, then unmap
12505 to retrieve the original frames.
12507 Otherwise, map the frames to the output device - create new hardware frames
12508 on the output corresponding to the frames on the input.
12511 The following additional parameters are accepted:
12515 Set the frame mapping mode. Some combination of:
12518 The mapped frame should be readable.
12520 The mapped frame should be writeable.
12522 The mapping will always overwrite the entire frame.
12524 This may improve performance in some cases, as the original contents of the
12525 frame need not be loaded.
12527 The mapping must not involve any copying.
12529 Indirect mappings to copies of frames are created in some cases where either
12530 direct mapping is not possible or it would have unexpected properties.
12531 Setting this flag ensures that the mapping is direct and will fail if that is
12534 Defaults to @var{read+write} if not specified.
12536 @item derive_device @var{type}
12537 Rather than using the device supplied at initialisation, instead derive a new
12538 device of type @var{type} from the device the input frames exist on.
12541 In a hardware to hardware mapping, map in reverse - create frames in the sink
12542 and map them back to the source. This may be necessary in some cases where
12543 a mapping in one direction is required but only the opposite direction is
12544 supported by the devices being used.
12546 This option is dangerous - it may break the preceding filter in undefined
12547 ways if there are any additional constraints on that filter's output.
12548 Do not use it without fully understanding the implications of its use.
12554 Upload system memory frames to hardware surfaces.
12556 The device to upload to must be supplied when the filter is initialised. If
12557 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12558 option or with the @option{derive_device} option. The input and output devices
12559 must be of different types and compatible - the exact meaning of this is
12560 system-dependent, but typically it means that they must refer to the same
12561 underlying hardware context (for example, refer to the same graphics card).
12563 The following additional parameters are accepted:
12566 @item derive_device @var{type}
12567 Rather than using the device supplied at initialisation, instead derive a new
12568 device of type @var{type} from the device the input frames exist on.
12571 @anchor{hwupload_cuda}
12572 @section hwupload_cuda
12574 Upload system memory frames to a CUDA device.
12576 It accepts the following optional parameters:
12580 The number of the CUDA device to use
12585 Apply a high-quality magnification filter designed for pixel art. This filter
12586 was originally created by Maxim Stepin.
12588 It accepts the following option:
12592 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12593 @code{hq3x} and @code{4} for @code{hq4x}.
12594 Default is @code{3}.
12598 Stack input videos horizontally.
12600 All streams must be of same pixel format and of same height.
12602 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12603 to create same output.
12605 The filter accepts the following option:
12609 Set number of input streams. Default is 2.
12612 If set to 1, force the output to terminate when the shortest input
12613 terminates. Default value is 0.
12618 Modify the hue and/or the saturation of the input.
12620 It accepts the following parameters:
12624 Specify the hue angle as a number of degrees. It accepts an expression,
12625 and defaults to "0".
12628 Specify the saturation in the [-10,10] range. It accepts an expression and
12632 Specify the hue angle as a number of radians. It accepts an
12633 expression, and defaults to "0".
12636 Specify the brightness in the [-10,10] range. It accepts an expression and
12640 @option{h} and @option{H} are mutually exclusive, and can't be
12641 specified at the same time.
12643 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12644 expressions containing the following constants:
12648 frame count of the input frame starting from 0
12651 presentation timestamp of the input frame expressed in time base units
12654 frame rate of the input video, NAN if the input frame rate is unknown
12657 timestamp expressed in seconds, NAN if the input timestamp is unknown
12660 time base of the input video
12663 @subsection Examples
12667 Set the hue to 90 degrees and the saturation to 1.0:
12673 Same command but expressing the hue in radians:
12679 Rotate hue and make the saturation swing between 0
12680 and 2 over a period of 1 second:
12682 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12686 Apply a 3 seconds saturation fade-in effect starting at 0:
12688 hue="s=min(t/3\,1)"
12691 The general fade-in expression can be written as:
12693 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12697 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12699 hue="s=max(0\, min(1\, (8-t)/3))"
12702 The general fade-out expression can be written as:
12704 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12709 @subsection Commands
12711 This filter supports the following commands:
12717 Modify the hue and/or the saturation and/or brightness of the input video.
12718 The command accepts the same syntax of the corresponding option.
12720 If the specified expression is not valid, it is kept at its current
12724 @section hysteresis
12726 Grow first stream into second stream by connecting components.
12727 This makes it possible to build more robust edge masks.
12729 This filter accepts the following options:
12733 Set which planes will be processed as bitmap, unprocessed planes will be
12734 copied from first stream.
12735 By default value 0xf, all planes will be processed.
12738 Set threshold which is used in filtering. If pixel component value is higher than
12739 this value filter algorithm for connecting components is activated.
12740 By default value is 0.
12743 The @code{hysteresis} filter also supports the @ref{framesync} options.
12747 Detect video interlacing type.
12749 This filter tries to detect if the input frames are interlaced, progressive,
12750 top or bottom field first. It will also try to detect fields that are
12751 repeated between adjacent frames (a sign of telecine).
12753 Single frame detection considers only immediately adjacent frames when classifying each frame.
12754 Multiple frame detection incorporates the classification history of previous frames.
12756 The filter will log these metadata values:
12759 @item single.current_frame
12760 Detected type of current frame using single-frame detection. One of:
12761 ``tff'' (top field first), ``bff'' (bottom field first),
12762 ``progressive'', or ``undetermined''
12765 Cumulative number of frames detected as top field first using single-frame detection.
12768 Cumulative number of frames detected as top field first using multiple-frame detection.
12771 Cumulative number of frames detected as bottom field first using single-frame detection.
12773 @item multiple.current_frame
12774 Detected type of current frame using multiple-frame detection. One of:
12775 ``tff'' (top field first), ``bff'' (bottom field first),
12776 ``progressive'', or ``undetermined''
12779 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12781 @item single.progressive
12782 Cumulative number of frames detected as progressive using single-frame detection.
12784 @item multiple.progressive
12785 Cumulative number of frames detected as progressive using multiple-frame detection.
12787 @item single.undetermined
12788 Cumulative number of frames that could not be classified using single-frame detection.
12790 @item multiple.undetermined
12791 Cumulative number of frames that could not be classified using multiple-frame detection.
12793 @item repeated.current_frame
12794 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12796 @item repeated.neither
12797 Cumulative number of frames with no repeated field.
12800 Cumulative number of frames with the top field repeated from the previous frame's top field.
12802 @item repeated.bottom
12803 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12806 The filter accepts the following options:
12810 Set interlacing threshold.
12812 Set progressive threshold.
12814 Threshold for repeated field detection.
12816 Number of frames after which a given frame's contribution to the
12817 statistics is halved (i.e., it contributes only 0.5 to its
12818 classification). The default of 0 means that all frames seen are given
12819 full weight of 1.0 forever.
12820 @item analyze_interlaced_flag
12821 When this is not 0 then idet will use the specified number of frames to determine
12822 if the interlaced flag is accurate, it will not count undetermined frames.
12823 If the flag is found to be accurate it will be used without any further
12824 computations, if it is found to be inaccurate it will be cleared without any
12825 further computations. This allows inserting the idet filter as a low computational
12826 method to clean up the interlaced flag
12831 Deinterleave or interleave fields.
12833 This filter allows one to process interlaced images fields without
12834 deinterlacing them. Deinterleaving splits the input frame into 2
12835 fields (so called half pictures). Odd lines are moved to the top
12836 half of the output image, even lines to the bottom half.
12837 You can process (filter) them independently and then re-interleave them.
12839 The filter accepts the following options:
12843 @item chroma_mode, c
12844 @item alpha_mode, a
12845 Available values for @var{luma_mode}, @var{chroma_mode} and
12846 @var{alpha_mode} are:
12852 @item deinterleave, d
12853 Deinterleave fields, placing one above the other.
12855 @item interleave, i
12856 Interleave fields. Reverse the effect of deinterleaving.
12858 Default value is @code{none}.
12860 @item luma_swap, ls
12861 @item chroma_swap, cs
12862 @item alpha_swap, as
12863 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12866 @subsection Commands
12868 This filter supports the all above options as @ref{commands}.
12872 Apply inflate effect to the video.
12874 This filter replaces the pixel by the local(3x3) average by taking into account
12875 only values higher than the pixel.
12877 It accepts the following options:
12884 Limit the maximum change for each plane, default is 65535.
12885 If 0, plane will remain unchanged.
12888 @subsection Commands
12890 This filter supports the all above options as @ref{commands}.
12894 Simple interlacing filter from progressive contents. This interleaves upper (or
12895 lower) lines from odd frames with lower (or upper) lines from even frames,
12896 halving the frame rate and preserving image height.
12899 Original Original New Frame
12900 Frame 'j' Frame 'j+1' (tff)
12901 ========== =========== ==================
12902 Line 0 --------------------> Frame 'j' Line 0
12903 Line 1 Line 1 ----> Frame 'j+1' Line 1
12904 Line 2 ---------------------> Frame 'j' Line 2
12905 Line 3 Line 3 ----> Frame 'j+1' Line 3
12907 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12910 It accepts the following optional parameters:
12914 This determines whether the interlaced frame is taken from the even
12915 (tff - default) or odd (bff) lines of the progressive frame.
12918 Vertical lowpass filter to avoid twitter interlacing and
12919 reduce moire patterns.
12923 Disable vertical lowpass filter
12926 Enable linear filter (default)
12929 Enable complex filter. This will slightly less reduce twitter and moire
12930 but better retain detail and subjective sharpness impression.
12937 Deinterlace input video by applying Donald Graft's adaptive kernel
12938 deinterling. Work on interlaced parts of a video to produce
12939 progressive frames.
12941 The description of the accepted parameters follows.
12945 Set the threshold which affects the filter's tolerance when
12946 determining if a pixel line must be processed. It must be an integer
12947 in the range [0,255] and defaults to 10. A value of 0 will result in
12948 applying the process on every pixels.
12951 Paint pixels exceeding the threshold value to white if set to 1.
12955 Set the fields order. Swap fields if set to 1, leave fields alone if
12959 Enable additional sharpening if set to 1. Default is 0.
12962 Enable twoway sharpening if set to 1. Default is 0.
12965 @subsection Examples
12969 Apply default values:
12971 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12975 Enable additional sharpening:
12981 Paint processed pixels in white:
12989 Slowly update darker pixels.
12991 This filter makes short flashes of light appear longer.
12992 This filter accepts the following options:
12996 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
12999 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13002 @section lenscorrection
13004 Correct radial lens distortion
13006 This filter can be used to correct for radial distortion as can result from the use
13007 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13008 one can use tools available for example as part of opencv or simply trial-and-error.
13009 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13010 and extract the k1 and k2 coefficients from the resulting matrix.
13012 Note that effectively the same filter is available in the open-source tools Krita and
13013 Digikam from the KDE project.
13015 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13016 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13017 brightness distribution, so you may want to use both filters together in certain
13018 cases, though you will have to take care of ordering, i.e. whether vignetting should
13019 be applied before or after lens correction.
13021 @subsection Options
13023 The filter accepts the following options:
13027 Relative x-coordinate of the focal point of the image, and thereby the center of the
13028 distortion. This value has a range [0,1] and is expressed as fractions of the image
13029 width. Default is 0.5.
13031 Relative y-coordinate of the focal point of the image, and thereby the center of the
13032 distortion. This value has a range [0,1] and is expressed as fractions of the image
13033 height. Default is 0.5.
13035 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13036 no correction. Default is 0.
13038 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13039 0 means no correction. Default is 0.
13042 The formula that generates the correction is:
13044 @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)
13046 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13047 distances from the focal point in the source and target images, respectively.
13051 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13053 The @code{lensfun} filter requires the camera make, camera model, and lens model
13054 to apply the lens correction. The filter will load the lensfun database and
13055 query it to find the corresponding camera and lens entries in the database. As
13056 long as these entries can be found with the given options, the filter can
13057 perform corrections on frames. Note that incomplete strings will result in the
13058 filter choosing the best match with the given options, and the filter will
13059 output the chosen camera and lens models (logged with level "info"). You must
13060 provide the make, camera model, and lens model as they are required.
13062 The filter accepts the following options:
13066 The make of the camera (for example, "Canon"). This option is required.
13069 The model of the camera (for example, "Canon EOS 100D"). This option is
13073 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13074 option is required.
13077 The type of correction to apply. The following values are valid options:
13081 Enables fixing lens vignetting.
13084 Enables fixing lens geometry. This is the default.
13087 Enables fixing chromatic aberrations.
13090 Enables fixing lens vignetting and lens geometry.
13093 Enables fixing lens vignetting and chromatic aberrations.
13096 Enables fixing both lens geometry and chromatic aberrations.
13099 Enables all possible corrections.
13103 The focal length of the image/video (zoom; expected constant for video). For
13104 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13105 range should be chosen when using that lens. Default 18.
13108 The aperture of the image/video (expected constant for video). Note that
13109 aperture is only used for vignetting correction. Default 3.5.
13111 @item focus_distance
13112 The focus distance of the image/video (expected constant for video). Note that
13113 focus distance is only used for vignetting and only slightly affects the
13114 vignetting correction process. If unknown, leave it at the default value (which
13118 The scale factor which is applied after transformation. After correction the
13119 video is no longer necessarily rectangular. This parameter controls how much of
13120 the resulting image is visible. The value 0 means that a value will be chosen
13121 automatically such that there is little or no unmapped area in the output
13122 image. 1.0 means that no additional scaling is done. Lower values may result
13123 in more of the corrected image being visible, while higher values may avoid
13124 unmapped areas in the output.
13126 @item target_geometry
13127 The target geometry of the output image/video. The following values are valid
13131 @item rectilinear (default)
13134 @item equirectangular
13135 @item fisheye_orthographic
13136 @item fisheye_stereographic
13137 @item fisheye_equisolid
13138 @item fisheye_thoby
13141 Apply the reverse of image correction (instead of correcting distortion, apply
13144 @item interpolation
13145 The type of interpolation used when correcting distortion. The following values
13150 @item linear (default)
13155 @subsection Examples
13159 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13160 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13164 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
13168 Apply the same as before, but only for the first 5 seconds of video.
13171 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
13178 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13179 score between two input videos.
13181 The obtained VMAF score is printed through the logging system.
13183 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13184 After installing the library it can be enabled using:
13185 @code{./configure --enable-libvmaf}.
13186 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13188 The filter has following options:
13192 Set the model path which is to be used for SVM.
13193 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13196 Set the file path to be used to store logs.
13199 Set the format of the log file (csv, json or xml).
13201 @item enable_transform
13202 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13203 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13204 Default value: @code{false}
13207 Invokes the phone model which will generate VMAF scores higher than in the
13208 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13209 Default value: @code{false}
13212 Enables computing psnr along with vmaf.
13213 Default value: @code{false}
13216 Enables computing ssim along with vmaf.
13217 Default value: @code{false}
13220 Enables computing ms_ssim along with vmaf.
13221 Default value: @code{false}
13224 Set the pool method to be used for computing vmaf.
13225 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13228 Set number of threads to be used when computing vmaf.
13229 Default value: @code{0}, which makes use of all available logical processors.
13232 Set interval for frame subsampling used when computing vmaf.
13233 Default value: @code{1}
13235 @item enable_conf_interval
13236 Enables confidence interval.
13237 Default value: @code{false}
13240 This filter also supports the @ref{framesync} options.
13242 @subsection Examples
13245 On the below examples the input file @file{main.mpg} being processed is
13246 compared with the reference file @file{ref.mpg}.
13249 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13253 Example with options:
13255 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13259 Example with options and different containers:
13261 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 -
13267 Limits the pixel components values to the specified range [min, max].
13269 The filter accepts the following options:
13273 Lower bound. Defaults to the lowest allowed value for the input.
13276 Upper bound. Defaults to the highest allowed value for the input.
13279 Specify which planes will be processed. Defaults to all available.
13286 The filter accepts the following options:
13290 Set the number of loops. Setting this value to -1 will result in infinite loops.
13294 Set maximal size in number of frames. Default is 0.
13297 Set first frame of loop. Default is 0.
13300 @subsection Examples
13304 Loop single first frame infinitely:
13306 loop=loop=-1:size=1:start=0
13310 Loop single first frame 10 times:
13312 loop=loop=10:size=1:start=0
13316 Loop 10 first frames 5 times:
13318 loop=loop=5:size=10:start=0
13324 Apply a 1D LUT to an input video.
13326 The filter accepts the following options:
13330 Set the 1D LUT file name.
13332 Currently supported formats:
13341 Select interpolation mode.
13343 Available values are:
13347 Use values from the nearest defined point.
13349 Interpolate values using the linear interpolation.
13351 Interpolate values using the cosine interpolation.
13353 Interpolate values using the cubic interpolation.
13355 Interpolate values using the spline interpolation.
13362 Apply a 3D LUT to an input video.
13364 The filter accepts the following options:
13368 Set the 3D LUT file name.
13370 Currently supported formats:
13384 Select interpolation mode.
13386 Available values are:
13390 Use values from the nearest defined point.
13392 Interpolate values using the 8 points defining a cube.
13394 Interpolate values using a tetrahedron.
13400 Turn certain luma values into transparency.
13402 The filter accepts the following options:
13406 Set the luma which will be used as base for transparency.
13407 Default value is @code{0}.
13410 Set the range of luma values to be keyed out.
13411 Default value is @code{0.01}.
13414 Set the range of softness. Default value is @code{0}.
13415 Use this to control gradual transition from zero to full transparency.
13418 @subsection Commands
13419 This filter supports same @ref{commands} as options.
13420 The command accepts the same syntax of the corresponding option.
13422 If the specified expression is not valid, it is kept at its current
13425 @section lut, lutrgb, lutyuv
13427 Compute a look-up table for binding each pixel component input value
13428 to an output value, and apply it to the input video.
13430 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13431 to an RGB input video.
13433 These filters accept the following parameters:
13436 set first pixel component expression
13438 set second pixel component expression
13440 set third pixel component expression
13442 set fourth pixel component expression, corresponds to the alpha component
13445 set red component expression
13447 set green component expression
13449 set blue component expression
13451 alpha component expression
13454 set Y/luminance component expression
13456 set U/Cb component expression
13458 set V/Cr component expression
13461 Each of them specifies the expression to use for computing the lookup table for
13462 the corresponding pixel component values.
13464 The exact component associated to each of the @var{c*} options depends on the
13467 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13468 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13470 The expressions can contain the following constants and functions:
13475 The input width and height.
13478 The input value for the pixel component.
13481 The input value, clipped to the @var{minval}-@var{maxval} range.
13484 The maximum value for the pixel component.
13487 The minimum value for the pixel component.
13490 The negated value for the pixel component value, clipped to the
13491 @var{minval}-@var{maxval} range; it corresponds to the expression
13492 "maxval-clipval+minval".
13495 The computed value in @var{val}, clipped to the
13496 @var{minval}-@var{maxval} range.
13498 @item gammaval(gamma)
13499 The computed gamma correction value of the pixel component value,
13500 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13502 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13506 All expressions default to "val".
13508 @subsection Examples
13512 Negate input video:
13514 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13515 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13518 The above is the same as:
13520 lutrgb="r=negval:g=negval:b=negval"
13521 lutyuv="y=negval:u=negval:v=negval"
13531 Remove chroma components, turning the video into a graytone image:
13533 lutyuv="u=128:v=128"
13537 Apply a luma burning effect:
13543 Remove green and blue components:
13549 Set a constant alpha channel value on input:
13551 format=rgba,lutrgb=a="maxval-minval/2"
13555 Correct luminance gamma by a factor of 0.5:
13557 lutyuv=y=gammaval(0.5)
13561 Discard least significant bits of luma:
13563 lutyuv=y='bitand(val, 128+64+32)'
13567 Technicolor like effect:
13569 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13573 @section lut2, tlut2
13575 The @code{lut2} filter takes two input streams and outputs one
13578 The @code{tlut2} (time lut2) filter takes two consecutive frames
13579 from one single stream.
13581 This filter accepts the following parameters:
13584 set first pixel component expression
13586 set second pixel component expression
13588 set third pixel component expression
13590 set fourth pixel component expression, corresponds to the alpha component
13593 set output bit depth, only available for @code{lut2} filter. By default is 0,
13594 which means bit depth is automatically picked from first input format.
13597 The @code{lut2} filter also supports the @ref{framesync} options.
13599 Each of them specifies the expression to use for computing the lookup table for
13600 the corresponding pixel component values.
13602 The exact component associated to each of the @var{c*} options depends on the
13605 The expressions can contain the following constants:
13610 The input width and height.
13613 The first input value for the pixel component.
13616 The second input value for the pixel component.
13619 The first input video bit depth.
13622 The second input video bit depth.
13625 All expressions default to "x".
13627 @subsection Examples
13631 Highlight differences between two RGB video streams:
13633 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)'
13637 Highlight differences between two YUV video streams:
13639 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)'
13643 Show max difference between two video streams:
13645 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)))'
13649 @section maskedclamp
13651 Clamp the first input stream with the second input and third input stream.
13653 Returns the value of first stream to be between second input
13654 stream - @code{undershoot} and third input stream + @code{overshoot}.
13656 This filter accepts the following options:
13659 Default value is @code{0}.
13662 Default value is @code{0}.
13665 Set which planes will be processed as bitmap, unprocessed planes will be
13666 copied from first stream.
13667 By default value 0xf, all planes will be processed.
13672 Merge the second and third input stream into output stream using absolute differences
13673 between second input stream and first input stream and absolute difference between
13674 third input stream and first input stream. The picked value will be from second input
13675 stream if second absolute difference is greater than first one or from third input stream
13678 This filter accepts the following options:
13681 Set which planes will be processed as bitmap, unprocessed planes will be
13682 copied from first stream.
13683 By default value 0xf, all planes will be processed.
13686 @section maskedmerge
13688 Merge the first input stream with the second input stream using per pixel
13689 weights in the third input stream.
13691 A value of 0 in the third stream pixel component means that pixel component
13692 from first stream is returned unchanged, while maximum value (eg. 255 for
13693 8-bit videos) means that pixel component from second stream is returned
13694 unchanged. Intermediate values define the amount of merging between both
13695 input stream's pixel components.
13697 This filter accepts the following options:
13700 Set which planes will be processed as bitmap, unprocessed planes will be
13701 copied from first stream.
13702 By default value 0xf, all planes will be processed.
13707 Merge the second and third input stream into output stream using absolute differences
13708 between second input stream and first input stream and absolute difference between
13709 third input stream and first input stream. The picked value will be from second input
13710 stream if second absolute difference is less than first one or from third input stream
13713 This filter accepts the following options:
13716 Set which planes will be processed as bitmap, unprocessed planes will be
13717 copied from first stream.
13718 By default value 0xf, all planes will be processed.
13721 @section maskedthreshold
13722 Pick pixels comparing absolute difference of two video streams with fixed
13725 If absolute difference between pixel component of first and second video
13726 stream is equal or lower than user supplied threshold than pixel component
13727 from first video stream is picked, otherwise pixel component from second
13728 video stream is picked.
13730 This filter accepts the following options:
13733 Set threshold used when picking pixels from absolute difference from two input
13737 Set which planes will be processed as bitmap, unprocessed planes will be
13738 copied from second stream.
13739 By default value 0xf, all planes will be processed.
13743 Create mask from input video.
13745 For example it is useful to create motion masks after @code{tblend} filter.
13747 This filter accepts the following options:
13751 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13754 Set high threshold. Any pixel component higher than this value will be set to max value
13755 allowed for current pixel format.
13758 Set planes to filter, by default all available planes are filtered.
13761 Fill all frame pixels with this value.
13764 Set max average pixel value for frame. If sum of all pixel components is higher that this
13765 average, output frame will be completely filled with value set by @var{fill} option.
13766 Typically useful for scene changes when used in combination with @code{tblend} filter.
13771 Apply motion-compensation deinterlacing.
13773 It needs one field per frame as input and must thus be used together
13774 with yadif=1/3 or equivalent.
13776 This filter accepts the following options:
13779 Set the deinterlacing mode.
13781 It accepts one of the following values:
13786 use iterative motion estimation
13788 like @samp{slow}, but use multiple reference frames.
13790 Default value is @samp{fast}.
13793 Set the picture field parity assumed for the input video. It must be
13794 one of the following values:
13798 assume top field first
13800 assume bottom field first
13803 Default value is @samp{bff}.
13806 Set per-block quantization parameter (QP) used by the internal
13809 Higher values should result in a smoother motion vector field but less
13810 optimal individual vectors. Default value is 1.
13815 Pick median pixel from certain rectangle defined by radius.
13817 This filter accepts the following options:
13821 Set horizontal radius size. Default value is @code{1}.
13822 Allowed range is integer from 1 to 127.
13825 Set which planes to process. Default is @code{15}, which is all available planes.
13828 Set vertical radius size. Default value is @code{0}.
13829 Allowed range is integer from 0 to 127.
13830 If it is 0, value will be picked from horizontal @code{radius} option.
13833 Set median percentile. Default value is @code{0.5}.
13834 Default value of @code{0.5} will pick always median values, while @code{0} will pick
13835 minimum values, and @code{1} maximum values.
13838 @subsection Commands
13839 This filter supports same @ref{commands} as options.
13840 The command accepts the same syntax of the corresponding option.
13842 If the specified expression is not valid, it is kept at its current
13845 @section mergeplanes
13847 Merge color channel components from several video streams.
13849 The filter accepts up to 4 input streams, and merge selected input
13850 planes to the output video.
13852 This filter accepts the following options:
13855 Set input to output plane mapping. Default is @code{0}.
13857 The mappings is specified as a bitmap. It should be specified as a
13858 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13859 mapping for the first plane of the output stream. 'A' sets the number of
13860 the input stream to use (from 0 to 3), and 'a' the plane number of the
13861 corresponding input to use (from 0 to 3). The rest of the mappings is
13862 similar, 'Bb' describes the mapping for the output stream second
13863 plane, 'Cc' describes the mapping for the output stream third plane and
13864 'Dd' describes the mapping for the output stream fourth plane.
13867 Set output pixel format. Default is @code{yuva444p}.
13870 @subsection Examples
13874 Merge three gray video streams of same width and height into single video stream:
13876 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13880 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13882 [a0][a1]mergeplanes=0x00010210:yuva444p
13886 Swap Y and A plane in yuva444p stream:
13888 format=yuva444p,mergeplanes=0x03010200:yuva444p
13892 Swap U and V plane in yuv420p stream:
13894 format=yuv420p,mergeplanes=0x000201:yuv420p
13898 Cast a rgb24 clip to yuv444p:
13900 format=rgb24,mergeplanes=0x000102:yuv444p
13906 Estimate and export motion vectors using block matching algorithms.
13907 Motion vectors are stored in frame side data to be used by other filters.
13909 This filter accepts the following options:
13912 Specify the motion estimation method. Accepts one of the following values:
13916 Exhaustive search algorithm.
13918 Three step search algorithm.
13920 Two dimensional logarithmic search algorithm.
13922 New three step search algorithm.
13924 Four step search algorithm.
13926 Diamond search algorithm.
13928 Hexagon-based search algorithm.
13930 Enhanced predictive zonal search algorithm.
13932 Uneven multi-hexagon search algorithm.
13934 Default value is @samp{esa}.
13937 Macroblock size. Default @code{16}.
13940 Search parameter. Default @code{7}.
13943 @section midequalizer
13945 Apply Midway Image Equalization effect using two video streams.
13947 Midway Image Equalization adjusts a pair of images to have the same
13948 histogram, while maintaining their dynamics as much as possible. It's
13949 useful for e.g. matching exposures from a pair of stereo cameras.
13951 This filter has two inputs and one output, which must be of same pixel format, but
13952 may be of different sizes. The output of filter is first input adjusted with
13953 midway histogram of both inputs.
13955 This filter accepts the following option:
13959 Set which planes to process. Default is @code{15}, which is all available planes.
13962 @section minterpolate
13964 Convert the video to specified frame rate using motion interpolation.
13966 This filter accepts the following options:
13969 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}.
13972 Motion interpolation mode. Following values are accepted:
13975 Duplicate previous or next frame for interpolating new ones.
13977 Blend source frames. Interpolated frame is mean of previous and next frames.
13979 Motion compensated interpolation. Following options are effective when this mode is selected:
13983 Motion compensation mode. Following values are accepted:
13986 Overlapped block motion compensation.
13988 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13990 Default mode is @samp{obmc}.
13993 Motion estimation mode. Following values are accepted:
13996 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
13998 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14000 Default mode is @samp{bilat}.
14003 The algorithm to be used for motion estimation. Following values are accepted:
14006 Exhaustive search algorithm.
14008 Three step search algorithm.
14010 Two dimensional logarithmic search algorithm.
14012 New three step search algorithm.
14014 Four step search algorithm.
14016 Diamond search algorithm.
14018 Hexagon-based search algorithm.
14020 Enhanced predictive zonal search algorithm.
14022 Uneven multi-hexagon search algorithm.
14024 Default algorithm is @samp{epzs}.
14027 Macroblock size. Default @code{16}.
14030 Motion estimation search parameter. Default @code{32}.
14033 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).
14038 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:
14041 Disable scene change detection.
14043 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14045 Default method is @samp{fdiff}.
14047 @item scd_threshold
14048 Scene change detection threshold. Default is @code{10.}.
14053 Mix several video input streams into one video stream.
14055 A description of the accepted options follows.
14059 The number of inputs. If unspecified, it defaults to 2.
14062 Specify weight of each input video stream as sequence.
14063 Each weight is separated by space. If number of weights
14064 is smaller than number of @var{frames} last specified
14065 weight will be used for all remaining unset weights.
14068 Specify scale, if it is set it will be multiplied with sum
14069 of each weight multiplied with pixel values to give final destination
14070 pixel value. By default @var{scale} is auto scaled to sum of weights.
14073 Specify how end of stream is determined.
14076 The duration of the longest input. (default)
14079 The duration of the shortest input.
14082 The duration of the first input.
14086 @section mpdecimate
14088 Drop frames that do not differ greatly from the previous frame in
14089 order to reduce frame rate.
14091 The main use of this filter is for very-low-bitrate encoding
14092 (e.g. streaming over dialup modem), but it could in theory be used for
14093 fixing movies that were inverse-telecined incorrectly.
14095 A description of the accepted options follows.
14099 Set the maximum number of consecutive frames which can be dropped (if
14100 positive), or the minimum interval between dropped frames (if
14101 negative). If the value is 0, the frame is dropped disregarding the
14102 number of previous sequentially dropped frames.
14104 Default value is 0.
14109 Set the dropping threshold values.
14111 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14112 represent actual pixel value differences, so a threshold of 64
14113 corresponds to 1 unit of difference for each pixel, or the same spread
14114 out differently over the block.
14116 A frame is a candidate for dropping if no 8x8 blocks differ by more
14117 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14118 meaning the whole image) differ by more than a threshold of @option{lo}.
14120 Default value for @option{hi} is 64*12, default value for @option{lo} is
14121 64*5, and default value for @option{frac} is 0.33.
14127 Negate (invert) the input video.
14129 It accepts the following option:
14134 With value 1, it negates the alpha component, if present. Default value is 0.
14140 Denoise frames using Non-Local Means algorithm.
14142 Each pixel is adjusted by looking for other pixels with similar contexts. This
14143 context similarity is defined by comparing their surrounding patches of size
14144 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14147 Note that the research area defines centers for patches, which means some
14148 patches will be made of pixels outside that research area.
14150 The filter accepts the following options.
14154 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14157 Set patch size. Default is 7. Must be odd number in range [0, 99].
14160 Same as @option{p} but for chroma planes.
14162 The default value is @var{0} and means automatic.
14165 Set research size. Default is 15. Must be odd number in range [0, 99].
14168 Same as @option{r} but for chroma planes.
14170 The default value is @var{0} and means automatic.
14175 Deinterlace video using neural network edge directed interpolation.
14177 This filter accepts the following options:
14181 Mandatory option, without binary file filter can not work.
14182 Currently file can be found here:
14183 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14186 Set which frames to deinterlace, by default it is @code{all}.
14187 Can be @code{all} or @code{interlaced}.
14190 Set mode of operation.
14192 Can be one of the following:
14196 Use frame flags, both fields.
14198 Use frame flags, single field.
14200 Use top field only.
14202 Use bottom field only.
14204 Use both fields, top first.
14206 Use both fields, bottom first.
14210 Set which planes to process, by default filter process all frames.
14213 Set size of local neighborhood around each pixel, used by the predictor neural
14216 Can be one of the following:
14229 Set the number of neurons in predictor neural network.
14230 Can be one of the following:
14241 Controls the number of different neural network predictions that are blended
14242 together to compute the final output value. Can be @code{fast}, default or
14246 Set which set of weights to use in the predictor.
14247 Can be one of the following:
14251 weights trained to minimize absolute error
14253 weights trained to minimize squared error
14257 Controls whether or not the prescreener neural network is used to decide
14258 which pixels should be processed by the predictor neural network and which
14259 can be handled by simple cubic interpolation.
14260 The prescreener is trained to know whether cubic interpolation will be
14261 sufficient for a pixel or whether it should be predicted by the predictor nn.
14262 The computational complexity of the prescreener nn is much less than that of
14263 the predictor nn. Since most pixels can be handled by cubic interpolation,
14264 using the prescreener generally results in much faster processing.
14265 The prescreener is pretty accurate, so the difference between using it and not
14266 using it is almost always unnoticeable.
14268 Can be one of the following:
14276 Default is @code{new}.
14279 Set various debugging flags.
14284 Force libavfilter not to use any of the specified pixel formats for the
14285 input to the next filter.
14287 It accepts the following parameters:
14291 A '|'-separated list of pixel format names, such as
14292 pix_fmts=yuv420p|monow|rgb24".
14296 @subsection Examples
14300 Force libavfilter to use a format different from @var{yuv420p} for the
14301 input to the vflip filter:
14303 noformat=pix_fmts=yuv420p,vflip
14307 Convert the input video to any of the formats not contained in the list:
14309 noformat=yuv420p|yuv444p|yuv410p
14315 Add noise on video input frame.
14317 The filter accepts the following options:
14325 Set noise seed for specific pixel component or all pixel components in case
14326 of @var{all_seed}. Default value is @code{123457}.
14328 @item all_strength, alls
14329 @item c0_strength, c0s
14330 @item c1_strength, c1s
14331 @item c2_strength, c2s
14332 @item c3_strength, c3s
14333 Set noise strength for specific pixel component or all pixel components in case
14334 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14336 @item all_flags, allf
14337 @item c0_flags, c0f
14338 @item c1_flags, c1f
14339 @item c2_flags, c2f
14340 @item c3_flags, c3f
14341 Set pixel component flags or set flags for all components if @var{all_flags}.
14342 Available values for component flags are:
14345 averaged temporal noise (smoother)
14347 mix random noise with a (semi)regular pattern
14349 temporal noise (noise pattern changes between frames)
14351 uniform noise (gaussian otherwise)
14355 @subsection Examples
14357 Add temporal and uniform noise to input video:
14359 noise=alls=20:allf=t+u
14364 Normalize RGB video (aka histogram stretching, contrast stretching).
14365 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14367 For each channel of each frame, the filter computes the input range and maps
14368 it linearly to the user-specified output range. The output range defaults
14369 to the full dynamic range from pure black to pure white.
14371 Temporal smoothing can be used on the input range to reduce flickering (rapid
14372 changes in brightness) caused when small dark or bright objects enter or leave
14373 the scene. This is similar to the auto-exposure (automatic gain control) on a
14374 video camera, and, like a video camera, it may cause a period of over- or
14375 under-exposure of the video.
14377 The R,G,B channels can be normalized independently, which may cause some
14378 color shifting, or linked together as a single channel, which prevents
14379 color shifting. Linked normalization preserves hue. Independent normalization
14380 does not, so it can be used to remove some color casts. Independent and linked
14381 normalization can be combined in any ratio.
14383 The normalize filter accepts the following options:
14388 Colors which define the output range. The minimum input value is mapped to
14389 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14390 The defaults are black and white respectively. Specifying white for
14391 @var{blackpt} and black for @var{whitept} will give color-inverted,
14392 normalized video. Shades of grey can be used to reduce the dynamic range
14393 (contrast). Specifying saturated colors here can create some interesting
14397 The number of previous frames to use for temporal smoothing. The input range
14398 of each channel is smoothed using a rolling average over the current frame
14399 and the @var{smoothing} previous frames. The default is 0 (no temporal
14403 Controls the ratio of independent (color shifting) channel normalization to
14404 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14405 independent. Defaults to 1.0 (fully independent).
14408 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14409 expensive no-op. Defaults to 1.0 (full strength).
14413 @subsection Commands
14414 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14415 The command accepts the same syntax of the corresponding option.
14417 If the specified expression is not valid, it is kept at its current
14420 @subsection Examples
14422 Stretch video contrast to use the full dynamic range, with no temporal
14423 smoothing; may flicker depending on the source content:
14425 normalize=blackpt=black:whitept=white:smoothing=0
14428 As above, but with 50 frames of temporal smoothing; flicker should be
14429 reduced, depending on the source content:
14431 normalize=blackpt=black:whitept=white:smoothing=50
14434 As above, but with hue-preserving linked channel normalization:
14436 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14439 As above, but with half strength:
14441 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14444 Map the darkest input color to red, the brightest input color to cyan:
14446 normalize=blackpt=red:whitept=cyan
14451 Pass the video source unchanged to the output.
14454 Optical Character Recognition
14456 This filter uses Tesseract for optical character recognition. To enable
14457 compilation of this filter, you need to configure FFmpeg with
14458 @code{--enable-libtesseract}.
14460 It accepts the following options:
14464 Set datapath to tesseract data. Default is to use whatever was
14465 set at installation.
14468 Set language, default is "eng".
14471 Set character whitelist.
14474 Set character blacklist.
14477 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14478 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14482 Apply a video transform using libopencv.
14484 To enable this filter, install the libopencv library and headers and
14485 configure FFmpeg with @code{--enable-libopencv}.
14487 It accepts the following parameters:
14492 The name of the libopencv filter to apply.
14494 @item filter_params
14495 The parameters to pass to the libopencv filter. If not specified, the default
14496 values are assumed.
14500 Refer to the official libopencv documentation for more precise
14502 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14504 Several libopencv filters are supported; see the following subsections.
14509 Dilate an image by using a specific structuring element.
14510 It corresponds to the libopencv function @code{cvDilate}.
14512 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14514 @var{struct_el} represents a structuring element, and has the syntax:
14515 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14517 @var{cols} and @var{rows} represent the number of columns and rows of
14518 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14519 point, and @var{shape} the shape for the structuring element. @var{shape}
14520 must be "rect", "cross", "ellipse", or "custom".
14522 If the value for @var{shape} is "custom", it must be followed by a
14523 string of the form "=@var{filename}". The file with name
14524 @var{filename} is assumed to represent a binary image, with each
14525 printable character corresponding to a bright pixel. When a custom
14526 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14527 or columns and rows of the read file are assumed instead.
14529 The default value for @var{struct_el} is "3x3+0x0/rect".
14531 @var{nb_iterations} specifies the number of times the transform is
14532 applied to the image, and defaults to 1.
14536 # Use the default values
14539 # Dilate using a structuring element with a 5x5 cross, iterating two times
14540 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14542 # Read the shape from the file diamond.shape, iterating two times.
14543 # The file diamond.shape may contain a pattern of characters like this
14549 # The specified columns and rows are ignored
14550 # but the anchor point coordinates are not
14551 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14556 Erode an image by using a specific structuring element.
14557 It corresponds to the libopencv function @code{cvErode}.
14559 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14560 with the same syntax and semantics as the @ref{dilate} filter.
14564 Smooth the input video.
14566 The filter takes the following parameters:
14567 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14569 @var{type} is the type of smooth filter to apply, and must be one of
14570 the following values: "blur", "blur_no_scale", "median", "gaussian",
14571 or "bilateral". The default value is "gaussian".
14573 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14574 depends on the smooth type. @var{param1} and
14575 @var{param2} accept integer positive values or 0. @var{param3} and
14576 @var{param4} accept floating point values.
14578 The default value for @var{param1} is 3. The default value for the
14579 other parameters is 0.
14581 These parameters correspond to the parameters assigned to the
14582 libopencv function @code{cvSmooth}.
14584 @section oscilloscope
14586 2D Video Oscilloscope.
14588 Useful to measure spatial impulse, step responses, chroma delays, etc.
14590 It accepts the following parameters:
14594 Set scope center x position.
14597 Set scope center y position.
14600 Set scope size, relative to frame diagonal.
14603 Set scope tilt/rotation.
14609 Set trace center x position.
14612 Set trace center y position.
14615 Set trace width, relative to width of frame.
14618 Set trace height, relative to height of frame.
14621 Set which components to trace. By default it traces first three components.
14624 Draw trace grid. By default is enabled.
14627 Draw some statistics. By default is enabled.
14630 Draw scope. By default is enabled.
14633 @subsection Commands
14634 This filter supports same @ref{commands} as options.
14635 The command accepts the same syntax of the corresponding option.
14637 If the specified expression is not valid, it is kept at its current
14640 @subsection Examples
14644 Inspect full first row of video frame.
14646 oscilloscope=x=0.5:y=0:s=1
14650 Inspect full last row of video frame.
14652 oscilloscope=x=0.5:y=1:s=1
14656 Inspect full 5th line of video frame of height 1080.
14658 oscilloscope=x=0.5:y=5/1080:s=1
14662 Inspect full last column of video frame.
14664 oscilloscope=x=1:y=0.5:s=1:t=1
14672 Overlay one video on top of another.
14674 It takes two inputs and has one output. The first input is the "main"
14675 video on which the second input is overlaid.
14677 It accepts the following parameters:
14679 A description of the accepted options follows.
14684 Set the expression for the x and y coordinates of the overlaid video
14685 on the main video. Default value is "0" for both expressions. In case
14686 the expression is invalid, it is set to a huge value (meaning that the
14687 overlay will not be displayed within the output visible area).
14690 See @ref{framesync}.
14693 Set when the expressions for @option{x}, and @option{y} are evaluated.
14695 It accepts the following values:
14698 only evaluate expressions once during the filter initialization or
14699 when a command is processed
14702 evaluate expressions for each incoming frame
14705 Default value is @samp{frame}.
14708 See @ref{framesync}.
14711 Set the format for the output video.
14713 It accepts the following values:
14716 force YUV420 output
14719 force YUV420p10 output
14722 force YUV422 output
14725 force YUV422p10 output
14728 force YUV444 output
14731 force packed RGB output
14734 force planar RGB output
14737 automatically pick format
14740 Default value is @samp{yuv420}.
14743 See @ref{framesync}.
14746 Set format of alpha of the overlaid video, it can be @var{straight} or
14747 @var{premultiplied}. Default is @var{straight}.
14750 The @option{x}, and @option{y} expressions can contain the following
14756 The main input width and height.
14760 The overlay input width and height.
14764 The computed values for @var{x} and @var{y}. They are evaluated for
14769 horizontal and vertical chroma subsample values of the output
14770 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14774 the number of input frame, starting from 0
14777 the position in the file of the input frame, NAN if unknown
14780 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14784 This filter also supports the @ref{framesync} options.
14786 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14787 when evaluation is done @emph{per frame}, and will evaluate to NAN
14788 when @option{eval} is set to @samp{init}.
14790 Be aware that frames are taken from each input video in timestamp
14791 order, hence, if their initial timestamps differ, it is a good idea
14792 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14793 have them begin in the same zero timestamp, as the example for
14794 the @var{movie} filter does.
14796 You can chain together more overlays but you should test the
14797 efficiency of such approach.
14799 @subsection Commands
14801 This filter supports the following commands:
14805 Modify the x and y of the overlay input.
14806 The command accepts the same syntax of the corresponding option.
14808 If the specified expression is not valid, it is kept at its current
14812 @subsection Examples
14816 Draw the overlay at 10 pixels from the bottom right corner of the main
14819 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14822 Using named options the example above becomes:
14824 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14828 Insert a transparent PNG logo in the bottom left corner of the input,
14829 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14831 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14835 Insert 2 different transparent PNG logos (second logo on bottom
14836 right corner) using the @command{ffmpeg} tool:
14838 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
14842 Add a transparent color layer on top of the main video; @code{WxH}
14843 must specify the size of the main input to the overlay filter:
14845 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14849 Play an original video and a filtered version (here with the deshake
14850 filter) side by side using the @command{ffplay} tool:
14852 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14855 The above command is the same as:
14857 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14861 Make a sliding overlay appearing from the left to the right top part of the
14862 screen starting since time 2:
14864 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14868 Compose output by putting two input videos side to side:
14870 ffmpeg -i left.avi -i right.avi -filter_complex "
14871 nullsrc=size=200x100 [background];
14872 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14873 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14874 [background][left] overlay=shortest=1 [background+left];
14875 [background+left][right] overlay=shortest=1:x=100 [left+right]
14880 Mask 10-20 seconds of a video by applying the delogo filter to a section
14882 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14883 -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]'
14888 Chain several overlays in cascade:
14890 nullsrc=s=200x200 [bg];
14891 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14892 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14893 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14894 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14895 [in3] null, [mid2] overlay=100:100 [out0]
14900 @anchor{overlay_cuda}
14901 @section overlay_cuda
14903 Overlay one video on top of another.
14905 This is the CUDA variant of the @ref{overlay} filter.
14906 It only accepts CUDA frames. The underlying input pixel formats have to match.
14908 It takes two inputs and has one output. The first input is the "main"
14909 video on which the second input is overlaid.
14911 It accepts the following parameters:
14916 Set the x and y coordinates of the overlaid video on the main video.
14917 Default value is "0" for both expressions.
14920 See @ref{framesync}.
14923 See @ref{framesync}.
14926 See @ref{framesync}.
14930 This filter also supports the @ref{framesync} options.
14934 Apply Overcomplete Wavelet denoiser.
14936 The filter accepts the following options:
14942 Larger depth values will denoise lower frequency components more, but
14943 slow down filtering.
14945 Must be an int in the range 8-16, default is @code{8}.
14947 @item luma_strength, ls
14950 Must be a double value in the range 0-1000, default is @code{1.0}.
14952 @item chroma_strength, cs
14953 Set chroma strength.
14955 Must be a double value in the range 0-1000, default is @code{1.0}.
14961 Add paddings to the input image, and place the original input at the
14962 provided @var{x}, @var{y} coordinates.
14964 It accepts the following parameters:
14969 Specify an expression for the size of the output image with the
14970 paddings added. If the value for @var{width} or @var{height} is 0, the
14971 corresponding input size is used for the output.
14973 The @var{width} expression can reference the value set by the
14974 @var{height} expression, and vice versa.
14976 The default value of @var{width} and @var{height} is 0.
14980 Specify the offsets to place the input image at within the padded area,
14981 with respect to the top/left border of the output image.
14983 The @var{x} expression can reference the value set by the @var{y}
14984 expression, and vice versa.
14986 The default value of @var{x} and @var{y} is 0.
14988 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14989 so the input image is centered on the padded area.
14992 Specify the color of the padded area. For the syntax of this option,
14993 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14994 manual,ffmpeg-utils}.
14996 The default value of @var{color} is "black".
14999 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15001 It accepts the following values:
15005 Only evaluate expressions once during the filter initialization or when
15006 a command is processed.
15009 Evaluate expressions for each incoming frame.
15013 Default value is @samp{init}.
15016 Pad to aspect instead to a resolution.
15020 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15021 options are expressions containing the following constants:
15026 The input video width and height.
15030 These are the same as @var{in_w} and @var{in_h}.
15034 The output width and height (the size of the padded area), as
15035 specified by the @var{width} and @var{height} expressions.
15039 These are the same as @var{out_w} and @var{out_h}.
15043 The x and y offsets as specified by the @var{x} and @var{y}
15044 expressions, or NAN if not yet specified.
15047 same as @var{iw} / @var{ih}
15050 input sample aspect ratio
15053 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15057 The horizontal and vertical chroma subsample values. For example for the
15058 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15061 @subsection Examples
15065 Add paddings with the color "violet" to the input video. The output video
15066 size is 640x480, and the top-left corner of the input video is placed at
15069 pad=640:480:0:40:violet
15072 The example above is equivalent to the following command:
15074 pad=width=640:height=480:x=0:y=40:color=violet
15078 Pad the input to get an output with dimensions increased by 3/2,
15079 and put the input video at the center of the padded area:
15081 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15085 Pad the input to get a squared output with size equal to the maximum
15086 value between the input width and height, and put the input video at
15087 the center of the padded area:
15089 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15093 Pad the input to get a final w/h ratio of 16:9:
15095 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15099 In case of anamorphic video, in order to set the output display aspect
15100 correctly, it is necessary to use @var{sar} in the expression,
15101 according to the relation:
15103 (ih * X / ih) * sar = output_dar
15104 X = output_dar / sar
15107 Thus the previous example needs to be modified to:
15109 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15113 Double the output size and put the input video in the bottom-right
15114 corner of the output padded area:
15116 pad="2*iw:2*ih:ow-iw:oh-ih"
15120 @anchor{palettegen}
15121 @section palettegen
15123 Generate one palette for a whole video stream.
15125 It accepts the following options:
15129 Set the maximum number of colors to quantize in the palette.
15130 Note: the palette will still contain 256 colors; the unused palette entries
15133 @item reserve_transparent
15134 Create a palette of 255 colors maximum and reserve the last one for
15135 transparency. Reserving the transparency color is useful for GIF optimization.
15136 If not set, the maximum of colors in the palette will be 256. You probably want
15137 to disable this option for a standalone image.
15140 @item transparency_color
15141 Set the color that will be used as background for transparency.
15144 Set statistics mode.
15146 It accepts the following values:
15149 Compute full frame histograms.
15151 Compute histograms only for the part that differs from previous frame. This
15152 might be relevant to give more importance to the moving part of your input if
15153 the background is static.
15155 Compute new histogram for each frame.
15158 Default value is @var{full}.
15161 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15162 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15163 color quantization of the palette. This information is also visible at
15164 @var{info} logging level.
15166 @subsection Examples
15170 Generate a representative palette of a given video using @command{ffmpeg}:
15172 ffmpeg -i input.mkv -vf palettegen palette.png
15176 @section paletteuse
15178 Use a palette to downsample an input video stream.
15180 The filter takes two inputs: one video stream and a palette. The palette must
15181 be a 256 pixels image.
15183 It accepts the following options:
15187 Select dithering mode. Available algorithms are:
15190 Ordered 8x8 bayer dithering (deterministic)
15192 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15193 Note: this dithering is sometimes considered "wrong" and is included as a
15195 @item floyd_steinberg
15196 Floyd and Steingberg dithering (error diffusion)
15198 Frankie Sierra dithering v2 (error diffusion)
15200 Frankie Sierra dithering v2 "Lite" (error diffusion)
15203 Default is @var{sierra2_4a}.
15206 When @var{bayer} dithering is selected, this option defines the scale of the
15207 pattern (how much the crosshatch pattern is visible). A low value means more
15208 visible pattern for less banding, and higher value means less visible pattern
15209 at the cost of more banding.
15211 The option must be an integer value in the range [0,5]. Default is @var{2}.
15214 If set, define the zone to process
15218 Only the changing rectangle will be reprocessed. This is similar to GIF
15219 cropping/offsetting compression mechanism. This option can be useful for speed
15220 if only a part of the image is changing, and has use cases such as limiting the
15221 scope of the error diffusal @option{dither} to the rectangle that bounds the
15222 moving scene (it leads to more deterministic output if the scene doesn't change
15223 much, and as a result less moving noise and better GIF compression).
15226 Default is @var{none}.
15229 Take new palette for each output frame.
15231 @item alpha_threshold
15232 Sets the alpha threshold for transparency. Alpha values above this threshold
15233 will be treated as completely opaque, and values below this threshold will be
15234 treated as completely transparent.
15236 The option must be an integer value in the range [0,255]. Default is @var{128}.
15239 @subsection Examples
15243 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15244 using @command{ffmpeg}:
15246 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15250 @section perspective
15252 Correct perspective of video not recorded perpendicular to the screen.
15254 A description of the accepted parameters follows.
15265 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15266 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15267 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15268 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15269 then the corners of the source will be sent to the specified coordinates.
15271 The expressions can use the following variables:
15276 the width and height of video frame.
15280 Output frame count.
15283 @item interpolation
15284 Set interpolation for perspective correction.
15286 It accepts the following values:
15292 Default value is @samp{linear}.
15295 Set interpretation of coordinate options.
15297 It accepts the following values:
15301 Send point in the source specified by the given coordinates to
15302 the corners of the destination.
15304 @item 1, destination
15306 Send the corners of the source to the point in the destination specified
15307 by the given coordinates.
15309 Default value is @samp{source}.
15313 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15315 It accepts the following values:
15318 only evaluate expressions once during the filter initialization or
15319 when a command is processed
15322 evaluate expressions for each incoming frame
15325 Default value is @samp{init}.
15330 Delay interlaced video by one field time so that the field order changes.
15332 The intended use is to fix PAL movies that have been captured with the
15333 opposite field order to the film-to-video transfer.
15335 A description of the accepted parameters follows.
15341 It accepts the following values:
15344 Capture field order top-first, transfer bottom-first.
15345 Filter will delay the bottom field.
15348 Capture field order bottom-first, transfer top-first.
15349 Filter will delay the top field.
15352 Capture and transfer with the same field order. This mode only exists
15353 for the documentation of the other options to refer to, but if you
15354 actually select it, the filter will faithfully do nothing.
15357 Capture field order determined automatically by field flags, transfer
15359 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15360 basis using field flags. If no field information is available,
15361 then this works just like @samp{u}.
15364 Capture unknown or varying, transfer opposite.
15365 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15366 analyzing the images and selecting the alternative that produces best
15367 match between the fields.
15370 Capture top-first, transfer unknown or varying.
15371 Filter selects among @samp{t} and @samp{p} using image analysis.
15374 Capture bottom-first, transfer unknown or varying.
15375 Filter selects among @samp{b} and @samp{p} using image analysis.
15378 Capture determined by field flags, transfer unknown or varying.
15379 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15380 image analysis. If no field information is available, then this works just
15381 like @samp{U}. This is the default mode.
15384 Both capture and transfer unknown or varying.
15385 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15389 @section photosensitivity
15390 Reduce various flashes in video, so to help users with epilepsy.
15392 It accepts the following options:
15395 Set how many frames to use when filtering. Default is 30.
15398 Set detection threshold factor. Default is 1.
15402 Set how many pixels to skip when sampling frames. Default is 1.
15403 Allowed range is from 1 to 1024.
15406 Leave frames unchanged. Default is disabled.
15409 @section pixdesctest
15411 Pixel format descriptor test filter, mainly useful for internal
15412 testing. The output video should be equal to the input video.
15416 format=monow, pixdesctest
15419 can be used to test the monowhite pixel format descriptor definition.
15423 Display sample values of color channels. Mainly useful for checking color
15424 and levels. Minimum supported resolution is 640x480.
15426 The filters accept the following options:
15430 Set scope X position, relative offset on X axis.
15433 Set scope Y position, relative offset on Y axis.
15442 Set window opacity. This window also holds statistics about pixel area.
15445 Set window X position, relative offset on X axis.
15448 Set window Y position, relative offset on Y axis.
15453 Enable the specified chain of postprocessing subfilters using libpostproc. This
15454 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15455 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15456 Each subfilter and some options have a short and a long name that can be used
15457 interchangeably, i.e. dr/dering are the same.
15459 The filters accept the following options:
15463 Set postprocessing subfilters string.
15466 All subfilters share common options to determine their scope:
15470 Honor the quality commands for this subfilter.
15473 Do chrominance filtering, too (default).
15476 Do luminance filtering only (no chrominance).
15479 Do chrominance filtering only (no luminance).
15482 These options can be appended after the subfilter name, separated by a '|'.
15484 Available subfilters are:
15487 @item hb/hdeblock[|difference[|flatness]]
15488 Horizontal deblocking filter
15491 Difference factor where higher values mean more deblocking (default: @code{32}).
15493 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15496 @item vb/vdeblock[|difference[|flatness]]
15497 Vertical deblocking filter
15500 Difference factor where higher values mean more deblocking (default: @code{32}).
15502 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15505 @item ha/hadeblock[|difference[|flatness]]
15506 Accurate horizontal deblocking filter
15509 Difference factor where higher values mean more deblocking (default: @code{32}).
15511 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15514 @item va/vadeblock[|difference[|flatness]]
15515 Accurate vertical deblocking filter
15518 Difference factor where higher values mean more deblocking (default: @code{32}).
15520 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15524 The horizontal and vertical deblocking filters share the difference and
15525 flatness values so you cannot set different horizontal and vertical
15529 @item h1/x1hdeblock
15530 Experimental horizontal deblocking filter
15532 @item v1/x1vdeblock
15533 Experimental vertical deblocking filter
15538 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15541 larger -> stronger filtering
15543 larger -> stronger filtering
15545 larger -> stronger filtering
15548 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15551 Stretch luminance to @code{0-255}.
15554 @item lb/linblenddeint
15555 Linear blend deinterlacing filter that deinterlaces the given block by
15556 filtering all lines with a @code{(1 2 1)} filter.
15558 @item li/linipoldeint
15559 Linear interpolating deinterlacing filter that deinterlaces the given block by
15560 linearly interpolating every second line.
15562 @item ci/cubicipoldeint
15563 Cubic interpolating deinterlacing filter deinterlaces the given block by
15564 cubically interpolating every second line.
15566 @item md/mediandeint
15567 Median deinterlacing filter that deinterlaces the given block by applying a
15568 median filter to every second line.
15570 @item fd/ffmpegdeint
15571 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15572 second line with a @code{(-1 4 2 4 -1)} filter.
15575 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15576 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15578 @item fq/forceQuant[|quantizer]
15579 Overrides the quantizer table from the input with the constant quantizer you
15587 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15590 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15593 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15596 @subsection Examples
15600 Apply horizontal and vertical deblocking, deringing and automatic
15601 brightness/contrast:
15607 Apply default filters without brightness/contrast correction:
15613 Apply default filters and temporal denoiser:
15615 pp=default/tmpnoise|1|2|3
15619 Apply deblocking on luminance only, and switch vertical deblocking on or off
15620 automatically depending on available CPU time:
15627 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15628 similar to spp = 6 with 7 point DCT, where only the center sample is
15631 The filter accepts the following options:
15635 Force a constant quantization parameter. It accepts an integer in range
15636 0 to 63. If not set, the filter will use the QP from the video stream
15640 Set thresholding mode. Available modes are:
15644 Set hard thresholding.
15646 Set soft thresholding (better de-ringing effect, but likely blurrier).
15648 Set medium thresholding (good results, default).
15652 @section premultiply
15653 Apply alpha premultiply effect to input video stream using first plane
15654 of second stream as alpha.
15656 Both streams must have same dimensions and same pixel format.
15658 The filter accepts the following option:
15662 Set which planes will be processed, unprocessed planes will be copied.
15663 By default value 0xf, all planes will be processed.
15666 Do not require 2nd input for processing, instead use alpha plane from input stream.
15670 Apply prewitt operator to input video stream.
15672 The filter accepts the following option:
15676 Set which planes will be processed, unprocessed planes will be copied.
15677 By default value 0xf, all planes will be processed.
15680 Set value which will be multiplied with filtered result.
15683 Set value which will be added to filtered result.
15686 @section pseudocolor
15688 Alter frame colors in video with pseudocolors.
15690 This filter accepts the following options:
15694 set pixel first component expression
15697 set pixel second component expression
15700 set pixel third component expression
15703 set pixel fourth component expression, corresponds to the alpha component
15706 set component to use as base for altering colors
15709 Each of them specifies the expression to use for computing the lookup table for
15710 the corresponding pixel component values.
15712 The expressions can contain the following constants and functions:
15717 The input width and height.
15720 The input value for the pixel component.
15722 @item ymin, umin, vmin, amin
15723 The minimum allowed component value.
15725 @item ymax, umax, vmax, amax
15726 The maximum allowed component value.
15729 All expressions default to "val".
15731 @subsection Examples
15735 Change too high luma values to gradient:
15737 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'"
15743 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15744 Ratio) between two input videos.
15746 This filter takes in input two input videos, the first input is
15747 considered the "main" source and is passed unchanged to the
15748 output. The second input is used as a "reference" video for computing
15751 Both video inputs must have the same resolution and pixel format for
15752 this filter to work correctly. Also it assumes that both inputs
15753 have the same number of frames, which are compared one by one.
15755 The obtained average PSNR is printed through the logging system.
15757 The filter stores the accumulated MSE (mean squared error) of each
15758 frame, and at the end of the processing it is averaged across all frames
15759 equally, and the following formula is applied to obtain the PSNR:
15762 PSNR = 10*log10(MAX^2/MSE)
15765 Where MAX is the average of the maximum values of each component of the
15768 The description of the accepted parameters follows.
15771 @item stats_file, f
15772 If specified the filter will use the named file to save the PSNR of
15773 each individual frame. When filename equals "-" the data is sent to
15776 @item stats_version
15777 Specifies which version of the stats file format to use. Details of
15778 each format are written below.
15779 Default value is 1.
15781 @item stats_add_max
15782 Determines whether the max value is output to the stats log.
15783 Default value is 0.
15784 Requires stats_version >= 2. If this is set and stats_version < 2,
15785 the filter will return an error.
15788 This filter also supports the @ref{framesync} options.
15790 The file printed if @var{stats_file} is selected, contains a sequence of
15791 key/value pairs of the form @var{key}:@var{value} for each compared
15794 If a @var{stats_version} greater than 1 is specified, a header line precedes
15795 the list of per-frame-pair stats, with key value pairs following the frame
15796 format with the following parameters:
15799 @item psnr_log_version
15800 The version of the log file format. Will match @var{stats_version}.
15803 A comma separated list of the per-frame-pair parameters included in
15807 A description of each shown per-frame-pair parameter follows:
15811 sequential number of the input frame, starting from 1
15814 Mean Square Error pixel-by-pixel average difference of the compared
15815 frames, averaged over all the image components.
15817 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15818 Mean Square Error pixel-by-pixel average difference of the compared
15819 frames for the component specified by the suffix.
15821 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15822 Peak Signal to Noise ratio of the compared frames for the component
15823 specified by the suffix.
15825 @item max_avg, max_y, max_u, max_v
15826 Maximum allowed value for each channel, and average over all
15830 @subsection Examples
15835 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15836 [main][ref] psnr="stats_file=stats.log" [out]
15839 On this example the input file being processed is compared with the
15840 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15841 is stored in @file{stats.log}.
15844 Another example with different containers:
15846 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 -
15853 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15854 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15857 The pullup filter is designed to take advantage of future context in making
15858 its decisions. This filter is stateless in the sense that it does not lock
15859 onto a pattern to follow, but it instead looks forward to the following
15860 fields in order to identify matches and rebuild progressive frames.
15862 To produce content with an even framerate, insert the fps filter after
15863 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15864 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15866 The filter accepts the following options:
15873 These options set the amount of "junk" to ignore at the left, right, top, and
15874 bottom of the image, respectively. Left and right are in units of 8 pixels,
15875 while top and bottom are in units of 2 lines.
15876 The default is 8 pixels on each side.
15879 Set the strict breaks. Setting this option to 1 will reduce the chances of
15880 filter generating an occasional mismatched frame, but it may also cause an
15881 excessive number of frames to be dropped during high motion sequences.
15882 Conversely, setting it to -1 will make filter match fields more easily.
15883 This may help processing of video where there is slight blurring between
15884 the fields, but may also cause there to be interlaced frames in the output.
15885 Default value is @code{0}.
15888 Set the metric plane to use. It accepts the following values:
15894 Use chroma blue plane.
15897 Use chroma red plane.
15900 This option may be set to use chroma plane instead of the default luma plane
15901 for doing filter's computations. This may improve accuracy on very clean
15902 source material, but more likely will decrease accuracy, especially if there
15903 is chroma noise (rainbow effect) or any grayscale video.
15904 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15905 load and make pullup usable in realtime on slow machines.
15908 For best results (without duplicated frames in the output file) it is
15909 necessary to change the output frame rate. For example, to inverse
15910 telecine NTSC input:
15912 ffmpeg -i input -vf pullup -r 24000/1001 ...
15917 Change video quantization parameters (QP).
15919 The filter accepts the following option:
15923 Set expression for quantization parameter.
15926 The expression is evaluated through the eval API and can contain, among others,
15927 the following constants:
15931 1 if index is not 129, 0 otherwise.
15934 Sequential index starting from -129 to 128.
15937 @subsection Examples
15941 Some equation like:
15949 Flush video frames from internal cache of frames into a random order.
15950 No frame is discarded.
15951 Inspired by @ref{frei0r} nervous filter.
15955 Set size in number of frames of internal cache, in range from @code{2} to
15956 @code{512}. Default is @code{30}.
15959 Set seed for random number generator, must be an integer included between
15960 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15961 less than @code{0}, the filter will try to use a good random seed on a
15965 @section readeia608
15967 Read closed captioning (EIA-608) information from the top lines of a video frame.
15969 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15970 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15971 with EIA-608 data (starting from 0). A description of each metadata value follows:
15974 @item lavfi.readeia608.X.cc
15975 The two bytes stored as EIA-608 data (printed in hexadecimal).
15977 @item lavfi.readeia608.X.line
15978 The number of the line on which the EIA-608 data was identified and read.
15981 This filter accepts the following options:
15985 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15988 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15991 Set the ratio of width reserved for sync code detection.
15992 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
15995 Enable checking the parity bit. In the event of a parity error, the filter will output
15996 @code{0x00} for that character. Default is false.
15999 Lowpass lines prior to further processing. Default is enabled.
16002 @subsection Commands
16004 This filter supports the all above options as @ref{commands}.
16006 @subsection Examples
16010 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16012 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
16018 Read vertical interval timecode (VITC) information from the top lines of a
16021 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16022 timecode value, if a valid timecode has been detected. Further metadata key
16023 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16024 timecode data has been found or not.
16026 This filter accepts the following options:
16030 Set the maximum number of lines to scan for VITC data. If the value is set to
16031 @code{-1} the full video frame is scanned. Default is @code{45}.
16034 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16035 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16038 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16039 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16042 @subsection Examples
16046 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16047 draw @code{--:--:--:--} as a placeholder:
16049 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16055 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16057 Destination pixel at position (X, Y) will be picked from source (x, y) position
16058 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16059 value for pixel will be used for destination pixel.
16061 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16062 will have Xmap/Ymap video stream dimensions.
16063 Xmap and Ymap input video streams are 16bit depth, single channel.
16067 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16068 Default is @code{color}.
16071 Specify the color of the unmapped pixels. For the syntax of this option,
16072 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16073 manual,ffmpeg-utils}. Default color is @code{black}.
16076 @section removegrain
16078 The removegrain filter is a spatial denoiser for progressive video.
16082 Set mode for the first plane.
16085 Set mode for the second plane.
16088 Set mode for the third plane.
16091 Set mode for the fourth plane.
16094 Range of mode is from 0 to 24. Description of each mode follows:
16098 Leave input plane unchanged. Default.
16101 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16104 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16107 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16110 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16111 This is equivalent to a median filter.
16114 Line-sensitive clipping giving the minimal change.
16117 Line-sensitive clipping, intermediate.
16120 Line-sensitive clipping, intermediate.
16123 Line-sensitive clipping, intermediate.
16126 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16129 Replaces the target pixel with the closest neighbour.
16132 [1 2 1] horizontal and vertical kernel blur.
16138 Bob mode, interpolates top field from the line where the neighbours
16139 pixels are the closest.
16142 Bob mode, interpolates bottom field from the line where the neighbours
16143 pixels are the closest.
16146 Bob mode, interpolates top field. Same as 13 but with a more complicated
16147 interpolation formula.
16150 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16151 interpolation formula.
16154 Clips the pixel with the minimum and maximum of respectively the maximum and
16155 minimum of each pair of opposite neighbour pixels.
16158 Line-sensitive clipping using opposite neighbours whose greatest distance from
16159 the current pixel is minimal.
16162 Replaces the pixel with the average of its 8 neighbours.
16165 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16168 Clips pixels using the averages of opposite neighbour.
16171 Same as mode 21 but simpler and faster.
16174 Small edge and halo removal, but reputed useless.
16180 @section removelogo
16182 Suppress a TV station logo, using an image file to determine which
16183 pixels comprise the logo. It works by filling in the pixels that
16184 comprise the logo with neighboring pixels.
16186 The filter accepts the following options:
16190 Set the filter bitmap file, which can be any image format supported by
16191 libavformat. The width and height of the image file must match those of the
16192 video stream being processed.
16195 Pixels in the provided bitmap image with a value of zero are not
16196 considered part of the logo, non-zero pixels are considered part of
16197 the logo. If you use white (255) for the logo and black (0) for the
16198 rest, you will be safe. For making the filter bitmap, it is
16199 recommended to take a screen capture of a black frame with the logo
16200 visible, and then using a threshold filter followed by the erode
16201 filter once or twice.
16203 If needed, little splotches can be fixed manually. Remember that if
16204 logo pixels are not covered, the filter quality will be much
16205 reduced. Marking too many pixels as part of the logo does not hurt as
16206 much, but it will increase the amount of blurring needed to cover over
16207 the image and will destroy more information than necessary, and extra
16208 pixels will slow things down on a large logo.
16210 @section repeatfields
16212 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16213 fields based on its value.
16217 Reverse a video clip.
16219 Warning: This filter requires memory to buffer the entire clip, so trimming
16222 @subsection Examples
16226 Take the first 5 seconds of a clip, and reverse it.
16233 Shift R/G/B/A pixels horizontally and/or vertically.
16235 The filter accepts the following options:
16238 Set amount to shift red horizontally.
16240 Set amount to shift red vertically.
16242 Set amount to shift green horizontally.
16244 Set amount to shift green vertically.
16246 Set amount to shift blue horizontally.
16248 Set amount to shift blue vertically.
16250 Set amount to shift alpha horizontally.
16252 Set amount to shift alpha vertically.
16254 Set edge mode, can be @var{smear}, default, or @var{warp}.
16257 @subsection Commands
16259 This filter supports the all above options as @ref{commands}.
16262 Apply roberts cross operator to input video stream.
16264 The filter accepts the following option:
16268 Set which planes will be processed, unprocessed planes will be copied.
16269 By default value 0xf, all planes will be processed.
16272 Set value which will be multiplied with filtered result.
16275 Set value which will be added to filtered result.
16280 Rotate video by an arbitrary angle expressed in radians.
16282 The filter accepts the following options:
16284 A description of the optional parameters follows.
16287 Set an expression for the angle by which to rotate the input video
16288 clockwise, expressed as a number of radians. A negative value will
16289 result in a counter-clockwise rotation. By default it is set to "0".
16291 This expression is evaluated for each frame.
16294 Set the output width expression, default value is "iw".
16295 This expression is evaluated just once during configuration.
16298 Set the output height expression, default value is "ih".
16299 This expression is evaluated just once during configuration.
16302 Enable bilinear interpolation if set to 1, a value of 0 disables
16303 it. Default value is 1.
16306 Set the color used to fill the output area not covered by the rotated
16307 image. For the general syntax of this option, check the
16308 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16309 If the special value "none" is selected then no
16310 background is printed (useful for example if the background is never shown).
16312 Default value is "black".
16315 The expressions for the angle and the output size can contain the
16316 following constants and functions:
16320 sequential number of the input frame, starting from 0. It is always NAN
16321 before the first frame is filtered.
16324 time in seconds of the input frame, it is set to 0 when the filter is
16325 configured. It is always NAN before the first frame is filtered.
16329 horizontal and vertical chroma subsample values. For example for the
16330 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16334 the input video width and height
16338 the output width and height, that is the size of the padded area as
16339 specified by the @var{width} and @var{height} expressions
16343 the minimal width/height required for completely containing the input
16344 video rotated by @var{a} radians.
16346 These are only available when computing the @option{out_w} and
16347 @option{out_h} expressions.
16350 @subsection Examples
16354 Rotate the input by PI/6 radians clockwise:
16360 Rotate the input by PI/6 radians counter-clockwise:
16366 Rotate the input by 45 degrees clockwise:
16372 Apply a constant rotation with period T, starting from an angle of PI/3:
16374 rotate=PI/3+2*PI*t/T
16378 Make the input video rotation oscillating with a period of T
16379 seconds and an amplitude of A radians:
16381 rotate=A*sin(2*PI/T*t)
16385 Rotate the video, output size is chosen so that the whole rotating
16386 input video is always completely contained in the output:
16388 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16392 Rotate the video, reduce the output size so that no background is ever
16395 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16399 @subsection Commands
16401 The filter supports the following commands:
16405 Set the angle expression.
16406 The command accepts the same syntax of the corresponding option.
16408 If the specified expression is not valid, it is kept at its current
16414 Apply Shape Adaptive Blur.
16416 The filter accepts the following options:
16419 @item luma_radius, lr
16420 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16421 value is 1.0. A greater value will result in a more blurred image, and
16422 in slower processing.
16424 @item luma_pre_filter_radius, lpfr
16425 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16428 @item luma_strength, ls
16429 Set luma maximum difference between pixels to still be considered, must
16430 be a value in the 0.1-100.0 range, default value is 1.0.
16432 @item chroma_radius, cr
16433 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16434 greater value will result in a more blurred image, and in slower
16437 @item chroma_pre_filter_radius, cpfr
16438 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16440 @item chroma_strength, cs
16441 Set chroma maximum difference between pixels to still be considered,
16442 must be a value in the -0.9-100.0 range.
16445 Each chroma option value, if not explicitly specified, is set to the
16446 corresponding luma option value.
16451 Scale (resize) the input video, using the libswscale library.
16453 The scale filter forces the output display aspect ratio to be the same
16454 of the input, by changing the output sample aspect ratio.
16456 If the input image format is different from the format requested by
16457 the next filter, the scale filter will convert the input to the
16460 @subsection Options
16461 The filter accepts the following options, or any of the options
16462 supported by the libswscale scaler.
16464 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16465 the complete list of scaler options.
16470 Set the output video dimension expression. Default value is the input
16473 If the @var{width} or @var{w} value is 0, the input width is used for
16474 the output. If the @var{height} or @var{h} value is 0, the input height
16475 is used for the output.
16477 If one and only one of the values is -n with n >= 1, the scale filter
16478 will use a value that maintains the aspect ratio of the input image,
16479 calculated from the other specified dimension. After that it will,
16480 however, make sure that the calculated dimension is divisible by n and
16481 adjust the value if necessary.
16483 If both values are -n with n >= 1, the behavior will be identical to
16484 both values being set to 0 as previously detailed.
16486 See below for the list of accepted constants for use in the dimension
16490 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16494 Only evaluate expressions once during the filter initialization or when a command is processed.
16497 Evaluate expressions for each incoming frame.
16501 Default value is @samp{init}.
16505 Set the interlacing mode. It accepts the following values:
16509 Force interlaced aware scaling.
16512 Do not apply interlaced scaling.
16515 Select interlaced aware scaling depending on whether the source frames
16516 are flagged as interlaced or not.
16519 Default value is @samp{0}.
16522 Set libswscale scaling flags. See
16523 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16524 complete list of values. If not explicitly specified the filter applies
16528 @item param0, param1
16529 Set libswscale input parameters for scaling algorithms that need them. See
16530 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16531 complete documentation. If not explicitly specified the filter applies
16537 Set the video size. For the syntax of this option, check the
16538 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16540 @item in_color_matrix
16541 @item out_color_matrix
16542 Set in/output YCbCr color space type.
16544 This allows the autodetected value to be overridden as well as allows forcing
16545 a specific value used for the output and encoder.
16547 If not specified, the color space type depends on the pixel format.
16553 Choose automatically.
16556 Format conforming to International Telecommunication Union (ITU)
16557 Recommendation BT.709.
16560 Set color space conforming to the United States Federal Communications
16561 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16566 Set color space conforming to:
16570 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16573 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16576 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16581 Set color space conforming to SMPTE ST 240:1999.
16584 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16589 Set in/output YCbCr sample range.
16591 This allows the autodetected value to be overridden as well as allows forcing
16592 a specific value used for the output and encoder. If not specified, the
16593 range depends on the pixel format. Possible values:
16597 Choose automatically.
16600 Set full range (0-255 in case of 8-bit luma).
16602 @item mpeg/limited/tv
16603 Set "MPEG" range (16-235 in case of 8-bit luma).
16606 @item force_original_aspect_ratio
16607 Enable decreasing or increasing output video width or height if necessary to
16608 keep the original aspect ratio. Possible values:
16612 Scale the video as specified and disable this feature.
16615 The output video dimensions will automatically be decreased if needed.
16618 The output video dimensions will automatically be increased if needed.
16622 One useful instance of this option is that when you know a specific device's
16623 maximum allowed resolution, you can use this to limit the output video to
16624 that, while retaining the aspect ratio. For example, device A allows
16625 1280x720 playback, and your video is 1920x800. Using this option (set it to
16626 decrease) and specifying 1280x720 to the command line makes the output
16629 Please note that this is a different thing than specifying -1 for @option{w}
16630 or @option{h}, you still need to specify the output resolution for this option
16633 @item force_divisible_by
16634 Ensures that both the output dimensions, width and height, are divisible by the
16635 given integer when used together with @option{force_original_aspect_ratio}. This
16636 works similar to using @code{-n} in the @option{w} and @option{h} options.
16638 This option respects the value set for @option{force_original_aspect_ratio},
16639 increasing or decreasing the resolution accordingly. The video's aspect ratio
16640 may be slightly modified.
16642 This option can be handy if you need to have a video fit within or exceed
16643 a defined resolution using @option{force_original_aspect_ratio} but also have
16644 encoder restrictions on width or height divisibility.
16648 The values of the @option{w} and @option{h} options are expressions
16649 containing the following constants:
16654 The input width and height
16658 These are the same as @var{in_w} and @var{in_h}.
16662 The output (scaled) width and height
16666 These are the same as @var{out_w} and @var{out_h}
16669 The same as @var{iw} / @var{ih}
16672 input sample aspect ratio
16675 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16679 horizontal and vertical input chroma subsample values. For example for the
16680 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16684 horizontal and vertical output chroma subsample values. For example for the
16685 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16688 The (sequential) number of the input frame, starting from 0.
16689 Only available with @code{eval=frame}.
16692 The presentation timestamp of the input frame, expressed as a number of
16693 seconds. Only available with @code{eval=frame}.
16696 The position (byte offset) of the frame in the input stream, or NaN if
16697 this information is unavailable and/or meaningless (for example in case of synthetic video).
16698 Only available with @code{eval=frame}.
16701 @subsection Examples
16705 Scale the input video to a size of 200x100
16710 This is equivalent to:
16721 Specify a size abbreviation for the output size:
16726 which can also be written as:
16732 Scale the input to 2x:
16734 scale=w=2*iw:h=2*ih
16738 The above is the same as:
16740 scale=2*in_w:2*in_h
16744 Scale the input to 2x with forced interlaced scaling:
16746 scale=2*iw:2*ih:interl=1
16750 Scale the input to half size:
16752 scale=w=iw/2:h=ih/2
16756 Increase the width, and set the height to the same size:
16762 Seek Greek harmony:
16769 Increase the height, and set the width to 3/2 of the height:
16771 scale=w=3/2*oh:h=3/5*ih
16775 Increase the size, making the size a multiple of the chroma
16778 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16782 Increase the width to a maximum of 500 pixels,
16783 keeping the same aspect ratio as the input:
16785 scale=w='min(500\, iw*3/2):h=-1'
16789 Make pixels square by combining scale and setsar:
16791 scale='trunc(ih*dar):ih',setsar=1/1
16795 Make pixels square by combining scale and setsar,
16796 making sure the resulting resolution is even (required by some codecs):
16798 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16802 @subsection Commands
16804 This filter supports the following commands:
16808 Set the output video dimension expression.
16809 The command accepts the same syntax of the corresponding option.
16811 If the specified expression is not valid, it is kept at its current
16817 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16818 format conversion on CUDA video frames. Setting the output width and height
16819 works in the same way as for the @var{scale} filter.
16821 The following additional options are accepted:
16824 The pixel format of the output CUDA frames. If set to the string "same" (the
16825 default), the input format will be kept. Note that automatic format negotiation
16826 and conversion is not yet supported for hardware frames
16829 The interpolation algorithm used for resizing. One of the following:
16836 @item cubic2p_bspline
16837 2-parameter cubic (B=1, C=0)
16839 @item cubic2p_catmullrom
16840 2-parameter cubic (B=0, C=1/2)
16842 @item cubic2p_b05c03
16843 2-parameter cubic (B=1/2, C=3/10)
16851 @item force_original_aspect_ratio
16852 Enable decreasing or increasing output video width or height if necessary to
16853 keep the original aspect ratio. Possible values:
16857 Scale the video as specified and disable this feature.
16860 The output video dimensions will automatically be decreased if needed.
16863 The output video dimensions will automatically be increased if needed.
16867 One useful instance of this option is that when you know a specific device's
16868 maximum allowed resolution, you can use this to limit the output video to
16869 that, while retaining the aspect ratio. For example, device A allows
16870 1280x720 playback, and your video is 1920x800. Using this option (set it to
16871 decrease) and specifying 1280x720 to the command line makes the output
16874 Please note that this is a different thing than specifying -1 for @option{w}
16875 or @option{h}, you still need to specify the output resolution for this option
16878 @item force_divisible_by
16879 Ensures that both the output dimensions, width and height, are divisible by the
16880 given integer when used together with @option{force_original_aspect_ratio}. This
16881 works similar to using @code{-n} in the @option{w} and @option{h} options.
16883 This option respects the value set for @option{force_original_aspect_ratio},
16884 increasing or decreasing the resolution accordingly. The video's aspect ratio
16885 may be slightly modified.
16887 This option can be handy if you need to have a video fit within or exceed
16888 a defined resolution using @option{force_original_aspect_ratio} but also have
16889 encoder restrictions on width or height divisibility.
16895 Scale (resize) the input video, based on a reference video.
16897 See the scale filter for available options, scale2ref supports the same but
16898 uses the reference video instead of the main input as basis. scale2ref also
16899 supports the following additional constants for the @option{w} and
16900 @option{h} options:
16905 The main input video's width and height
16908 The same as @var{main_w} / @var{main_h}
16911 The main input video's sample aspect ratio
16913 @item main_dar, mdar
16914 The main input video's display aspect ratio. Calculated from
16915 @code{(main_w / main_h) * main_sar}.
16919 The main input video's horizontal and vertical chroma subsample values.
16920 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16924 The (sequential) number of the main input frame, starting from 0.
16925 Only available with @code{eval=frame}.
16928 The presentation timestamp of the main input frame, expressed as a number of
16929 seconds. Only available with @code{eval=frame}.
16932 The position (byte offset) of the frame in the main input stream, or NaN if
16933 this information is unavailable and/or meaningless (for example in case of synthetic video).
16934 Only available with @code{eval=frame}.
16937 @subsection Examples
16941 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16943 'scale2ref[b][a];[a][b]overlay'
16947 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16949 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16953 @subsection Commands
16955 This filter supports the following commands:
16959 Set the output video dimension expression.
16960 The command accepts the same syntax of the corresponding option.
16962 If the specified expression is not valid, it is kept at its current
16967 Scroll input video horizontally and/or vertically by constant speed.
16969 The filter accepts the following options:
16971 @item horizontal, h
16972 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16973 Negative values changes scrolling direction.
16976 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16977 Negative values changes scrolling direction.
16980 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16983 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16986 @subsection Commands
16988 This filter supports the following @ref{commands}:
16990 @item horizontal, h
16991 Set the horizontal scrolling speed.
16993 Set the vertical scrolling speed.
16999 Detect video scene change.
17001 This filter sets frame metadata with mafd between frame, the scene score, and
17002 forward the frame to the next filter, so they can use these metadata to detect
17003 scene change or others.
17005 In addition, this filter logs a message and sets frame metadata when it detects
17006 a scene change by @option{threshold}.
17008 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17010 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17011 to detect scene change.
17013 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17014 detect scene change with @option{threshold}.
17016 The filter accepts the following options:
17020 Set the scene change detection threshold as a percentage of maximum change. Good
17021 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17024 Default value is @code{10.}.
17027 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17028 You can enable it if you want to get snapshot of scene change frames only.
17031 @anchor{selectivecolor}
17032 @section selectivecolor
17034 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17035 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17036 by the "purity" of the color (that is, how saturated it already is).
17038 This filter is similar to the Adobe Photoshop Selective Color tool.
17040 The filter accepts the following options:
17043 @item correction_method
17044 Select color correction method.
17046 Available values are:
17049 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17052 Specified adjustments are relative to the original component value.
17054 Default is @code{absolute}.
17056 Adjustments for red pixels (pixels where the red component is the maximum)
17058 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17060 Adjustments for green pixels (pixels where the green component is the maximum)
17062 Adjustments for cyan pixels (pixels where the red component is the minimum)
17064 Adjustments for blue pixels (pixels where the blue component is the maximum)
17066 Adjustments for magenta pixels (pixels where the green component is the minimum)
17068 Adjustments for white pixels (pixels where all components are greater than 128)
17070 Adjustments for all pixels except pure black and pure white
17072 Adjustments for black pixels (pixels where all components are lesser than 128)
17074 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17077 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17078 4 space separated floating point adjustment values in the [-1,1] range,
17079 respectively to adjust the amount of cyan, magenta, yellow and black for the
17080 pixels of its range.
17082 @subsection Examples
17086 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17087 increase magenta by 27% in blue areas:
17089 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17093 Use a Photoshop selective color preset:
17095 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17099 @anchor{separatefields}
17100 @section separatefields
17102 The @code{separatefields} takes a frame-based video input and splits
17103 each frame into its components fields, producing a new half height clip
17104 with twice the frame rate and twice the frame count.
17106 This filter use field-dominance information in frame to decide which
17107 of each pair of fields to place first in the output.
17108 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17110 @section setdar, setsar
17112 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17115 This is done by changing the specified Sample (aka Pixel) Aspect
17116 Ratio, according to the following equation:
17118 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17121 Keep in mind that the @code{setdar} filter does not modify the pixel
17122 dimensions of the video frame. Also, the display aspect ratio set by
17123 this filter may be changed by later filters in the filterchain,
17124 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17127 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17128 the filter output video.
17130 Note that as a consequence of the application of this filter, the
17131 output display aspect ratio will change according to the equation
17134 Keep in mind that the sample aspect ratio set by the @code{setsar}
17135 filter may be changed by later filters in the filterchain, e.g. if
17136 another "setsar" or a "setdar" filter is applied.
17138 It accepts the following parameters:
17141 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17142 Set the aspect ratio used by the filter.
17144 The parameter can be a floating point number string, an expression, or
17145 a string of the form @var{num}:@var{den}, where @var{num} and
17146 @var{den} are the numerator and denominator of the aspect ratio. If
17147 the parameter is not specified, it is assumed the value "0".
17148 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17152 Set the maximum integer value to use for expressing numerator and
17153 denominator when reducing the expressed aspect ratio to a rational.
17154 Default value is @code{100}.
17158 The parameter @var{sar} is an expression containing
17159 the following constants:
17163 These are approximated values for the mathematical constants e
17164 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17167 The input width and height.
17170 These are the same as @var{w} / @var{h}.
17173 The input sample aspect ratio.
17176 The input display aspect ratio. It is the same as
17177 (@var{w} / @var{h}) * @var{sar}.
17180 Horizontal and vertical chroma subsample values. For example, for the
17181 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17184 @subsection Examples
17189 To change the display aspect ratio to 16:9, specify one of the following:
17196 To change the sample aspect ratio to 10:11, specify:
17202 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17203 1000 in the aspect ratio reduction, use the command:
17205 setdar=ratio=16/9:max=1000
17213 Force field for the output video frame.
17215 The @code{setfield} filter marks the interlace type field for the
17216 output frames. It does not change the input frame, but only sets the
17217 corresponding property, which affects how the frame is treated by
17218 following filters (e.g. @code{fieldorder} or @code{yadif}).
17220 The filter accepts the following options:
17225 Available values are:
17229 Keep the same field property.
17232 Mark the frame as bottom-field-first.
17235 Mark the frame as top-field-first.
17238 Mark the frame as progressive.
17245 Force frame parameter for the output video frame.
17247 The @code{setparams} filter marks interlace and color range for the
17248 output frames. It does not change the input frame, but only sets the
17249 corresponding property, which affects how the frame is treated by
17254 Available values are:
17258 Keep the same field property (default).
17261 Mark the frame as bottom-field-first.
17264 Mark the frame as top-field-first.
17267 Mark the frame as progressive.
17271 Available values are:
17275 Keep the same color range property (default).
17277 @item unspecified, unknown
17278 Mark the frame as unspecified color range.
17280 @item limited, tv, mpeg
17281 Mark the frame as limited range.
17283 @item full, pc, jpeg
17284 Mark the frame as full range.
17287 @item color_primaries
17288 Set the color primaries.
17289 Available values are:
17293 Keep the same color primaries property (default).
17310 Set the color transfer.
17311 Available values are:
17315 Keep the same color trc property (default).
17337 Set the colorspace.
17338 Available values are:
17342 Keep the same colorspace property (default).
17355 @item chroma-derived-nc
17356 @item chroma-derived-c
17363 Show a line containing various information for each input video frame.
17364 The input video is not modified.
17366 This filter supports the following options:
17370 Calculate checksums of each plane. By default enabled.
17373 The shown line contains a sequence of key/value pairs of the form
17374 @var{key}:@var{value}.
17376 The following values are shown in the output:
17380 The (sequential) number of the input frame, starting from 0.
17383 The Presentation TimeStamp of the input frame, expressed as a number of
17384 time base units. The time base unit depends on the filter input pad.
17387 The Presentation TimeStamp of the input frame, expressed as a number of
17391 The position of the frame in the input stream, or -1 if this information is
17392 unavailable and/or meaningless (for example in case of synthetic video).
17395 The pixel format name.
17398 The sample aspect ratio of the input frame, expressed in the form
17399 @var{num}/@var{den}.
17402 The size of the input frame. For the syntax of this option, check the
17403 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17406 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17407 for bottom field first).
17410 This is 1 if the frame is a key frame, 0 otherwise.
17413 The picture type of the input frame ("I" for an I-frame, "P" for a
17414 P-frame, "B" for a B-frame, or "?" for an unknown type).
17415 Also refer to the documentation of the @code{AVPictureType} enum and of
17416 the @code{av_get_picture_type_char} function defined in
17417 @file{libavutil/avutil.h}.
17420 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17422 @item plane_checksum
17423 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17424 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17427 The mean value of pixels in each plane of the input frame, expressed in the form
17428 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17431 The standard deviation of pixel values in each plane of the input frame, expressed
17432 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17436 @section showpalette
17438 Displays the 256 colors palette of each frame. This filter is only relevant for
17439 @var{pal8} pixel format frames.
17441 It accepts the following option:
17445 Set the size of the box used to represent one palette color entry. Default is
17446 @code{30} (for a @code{30x30} pixel box).
17449 @section shuffleframes
17451 Reorder and/or duplicate and/or drop video frames.
17453 It accepts the following parameters:
17457 Set the destination indexes of input frames.
17458 This is space or '|' separated list of indexes that maps input frames to output
17459 frames. Number of indexes also sets maximal value that each index may have.
17460 '-1' index have special meaning and that is to drop frame.
17463 The first frame has the index 0. The default is to keep the input unchanged.
17465 @subsection Examples
17469 Swap second and third frame of every three frames of the input:
17471 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17475 Swap 10th and 1st frame of every ten frames of the input:
17477 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17481 @section shuffleplanes
17483 Reorder and/or duplicate video planes.
17485 It accepts the following parameters:
17490 The index of the input plane to be used as the first output plane.
17493 The index of the input plane to be used as the second output plane.
17496 The index of the input plane to be used as the third output plane.
17499 The index of the input plane to be used as the fourth output plane.
17503 The first plane has the index 0. The default is to keep the input unchanged.
17505 @subsection Examples
17509 Swap the second and third planes of the input:
17511 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17515 @anchor{signalstats}
17516 @section signalstats
17517 Evaluate various visual metrics that assist in determining issues associated
17518 with the digitization of analog video media.
17520 By default the filter will log these metadata values:
17524 Display the minimal Y value contained within the input frame. Expressed in
17528 Display the Y value at the 10% percentile within the input frame. Expressed in
17532 Display the average Y value within the input frame. Expressed in range of
17536 Display the Y value at the 90% percentile within the input frame. Expressed in
17540 Display the maximum Y value contained within the input frame. Expressed in
17544 Display the minimal U value contained within the input frame. Expressed in
17548 Display the U value at the 10% percentile within the input frame. Expressed in
17552 Display the average U value within the input frame. Expressed in range of
17556 Display the U value at the 90% percentile within the input frame. Expressed in
17560 Display the maximum U value contained within the input frame. Expressed in
17564 Display the minimal V value contained within the input frame. Expressed in
17568 Display the V value at the 10% percentile within the input frame. Expressed in
17572 Display the average V value within the input frame. Expressed in range of
17576 Display the V value at the 90% percentile within the input frame. Expressed in
17580 Display the maximum V value contained within the input frame. Expressed in
17584 Display the minimal saturation value contained within the input frame.
17585 Expressed in range of [0-~181.02].
17588 Display the saturation value at the 10% percentile within the input frame.
17589 Expressed in range of [0-~181.02].
17592 Display the average saturation value within the input frame. Expressed in range
17596 Display the saturation value at the 90% percentile within the input frame.
17597 Expressed in range of [0-~181.02].
17600 Display the maximum saturation value contained within the input frame.
17601 Expressed in range of [0-~181.02].
17604 Display the median value for hue within the input frame. Expressed in range of
17608 Display the average value for hue within the input frame. Expressed in range of
17612 Display the average of sample value difference between all values of the Y
17613 plane in the current frame and corresponding values of the previous input frame.
17614 Expressed in range of [0-255].
17617 Display the average of sample value difference between all values of the U
17618 plane in the current frame and corresponding values of the previous input frame.
17619 Expressed in range of [0-255].
17622 Display the average of sample value difference between all values of the V
17623 plane in the current frame and corresponding values of the previous input frame.
17624 Expressed in range of [0-255].
17627 Display bit depth of Y plane in current frame.
17628 Expressed in range of [0-16].
17631 Display bit depth of U plane in current frame.
17632 Expressed in range of [0-16].
17635 Display bit depth of V plane in current frame.
17636 Expressed in range of [0-16].
17639 The filter accepts the following options:
17645 @option{stat} specify an additional form of image analysis.
17646 @option{out} output video with the specified type of pixel highlighted.
17648 Both options accept the following values:
17652 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17653 unlike the neighboring pixels of the same field. Examples of temporal outliers
17654 include the results of video dropouts, head clogs, or tape tracking issues.
17657 Identify @var{vertical line repetition}. Vertical line repetition includes
17658 similar rows of pixels within a frame. In born-digital video vertical line
17659 repetition is common, but this pattern is uncommon in video digitized from an
17660 analog source. When it occurs in video that results from the digitization of an
17661 analog source it can indicate concealment from a dropout compensator.
17664 Identify pixels that fall outside of legal broadcast range.
17668 Set the highlight color for the @option{out} option. The default color is
17672 @subsection Examples
17676 Output data of various video metrics:
17678 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17682 Output specific data about the minimum and maximum values of the Y plane per frame:
17684 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17688 Playback video while highlighting pixels that are outside of broadcast range in red.
17690 ffplay example.mov -vf signalstats="out=brng:color=red"
17694 Playback video with signalstats metadata drawn over the frame.
17696 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17699 The contents of signalstat_drawtext.txt used in the command are:
17702 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
17703 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
17704 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
17705 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
17713 Calculates the MPEG-7 Video Signature. The filter can handle more than one
17714 input. In this case the matching between the inputs can be calculated additionally.
17715 The filter always passes through the first input. The signature of each stream can
17716 be written into a file.
17718 It accepts the following options:
17722 Enable or disable the matching process.
17724 Available values are:
17728 Disable the calculation of a matching (default).
17730 Calculate the matching for the whole video and output whether the whole video
17731 matches or only parts.
17733 Calculate only until a matching is found or the video ends. Should be faster in
17738 Set the number of inputs. The option value must be a non negative integer.
17739 Default value is 1.
17742 Set the path to which the output is written. If there is more than one input,
17743 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
17744 integer), that will be replaced with the input number. If no filename is
17745 specified, no output will be written. This is the default.
17748 Choose the output format.
17750 Available values are:
17754 Use the specified binary representation (default).
17756 Use the specified xml representation.
17760 Set threshold to detect one word as similar. The option value must be an integer
17761 greater than zero. The default value is 9000.
17764 Set threshold to detect all words as similar. The option value must be an integer
17765 greater than zero. The default value is 60000.
17768 Set threshold to detect frames as similar. The option value must be an integer
17769 greater than zero. The default value is 116.
17772 Set the minimum length of a sequence in frames to recognize it as matching
17773 sequence. The option value must be a non negative integer value.
17774 The default value is 0.
17777 Set the minimum relation, that matching frames to all frames must have.
17778 The option value must be a double value between 0 and 1. The default value is 0.5.
17781 @subsection Examples
17785 To calculate the signature of an input video and store it in signature.bin:
17787 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17791 To detect whether two videos match and store the signatures in XML format in
17792 signature0.xml and signature1.xml:
17794 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 -
17802 Blur the input video without impacting the outlines.
17804 It accepts the following options:
17807 @item luma_radius, lr
17808 Set the luma radius. The option value must be a float number in
17809 the range [0.1,5.0] that specifies the variance of the gaussian filter
17810 used to blur the image (slower if larger). Default value is 1.0.
17812 @item luma_strength, ls
17813 Set the luma strength. The option value must be a float number
17814 in the range [-1.0,1.0] that configures the blurring. A value included
17815 in [0.0,1.0] will blur the image whereas a value included in
17816 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17818 @item luma_threshold, lt
17819 Set the luma threshold used as a coefficient to determine
17820 whether a pixel should be blurred or not. The option value must be an
17821 integer in the range [-30,30]. A value of 0 will filter all the image,
17822 a value included in [0,30] will filter flat areas and a value included
17823 in [-30,0] will filter edges. Default value is 0.
17825 @item chroma_radius, cr
17826 Set the chroma radius. The option value must be a float number in
17827 the range [0.1,5.0] that specifies the variance of the gaussian filter
17828 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17830 @item chroma_strength, cs
17831 Set the chroma strength. The option value must be a float number
17832 in the range [-1.0,1.0] that configures the blurring. A value included
17833 in [0.0,1.0] will blur the image whereas a value included in
17834 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17836 @item chroma_threshold, ct
17837 Set the chroma threshold used as a coefficient to determine
17838 whether a pixel should be blurred or not. The option value must be an
17839 integer in the range [-30,30]. A value of 0 will filter all the image,
17840 a value included in [0,30] will filter flat areas and a value included
17841 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17844 If a chroma option is not explicitly set, the corresponding luma value
17848 Apply sobel operator to input video stream.
17850 The filter accepts the following option:
17854 Set which planes will be processed, unprocessed planes will be copied.
17855 By default value 0xf, all planes will be processed.
17858 Set value which will be multiplied with filtered result.
17861 Set value which will be added to filtered result.
17867 Apply a simple postprocessing filter that compresses and decompresses the image
17868 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17869 and average the results.
17871 The filter accepts the following options:
17875 Set quality. This option defines the number of levels for averaging. It accepts
17876 an integer in the range 0-6. If set to @code{0}, the filter will have no
17877 effect. A value of @code{6} means the higher quality. For each increment of
17878 that value the speed drops by a factor of approximately 2. Default value is
17882 Force a constant quantization parameter. If not set, the filter will use the QP
17883 from the video stream (if available).
17886 Set thresholding mode. Available modes are:
17890 Set hard thresholding (default).
17892 Set soft thresholding (better de-ringing effect, but likely blurrier).
17895 @item use_bframe_qp
17896 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17897 option may cause flicker since the B-Frames have often larger QP. Default is
17898 @code{0} (not enabled).
17901 @subsection Commands
17903 This filter supports the following commands:
17905 @item quality, level
17906 Set quality level. The value @code{max} can be used to set the maximum level,
17907 currently @code{6}.
17913 Scale the input by applying one of the super-resolution methods based on
17914 convolutional neural networks. Supported models:
17918 Super-Resolution Convolutional Neural Network model (SRCNN).
17919 See @url{https://arxiv.org/abs/1501.00092}.
17922 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17923 See @url{https://arxiv.org/abs/1609.05158}.
17926 Training scripts as well as scripts for model file (.pb) saving can be found at
17927 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17928 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17930 Native model files (.model) can be generated from TensorFlow model
17931 files (.pb) by using tools/python/convert.py
17933 The filter accepts the following options:
17937 Specify which DNN backend to use for model loading and execution. This option accepts
17938 the following values:
17942 Native implementation of DNN loading and execution.
17945 TensorFlow backend. To enable this backend you
17946 need to install the TensorFlow for C library (see
17947 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17948 @code{--enable-libtensorflow}
17951 Default value is @samp{native}.
17954 Set path to model file specifying network architecture and its parameters.
17955 Note that different backends use different file formats. TensorFlow backend
17956 can load files for both formats, while native backend can load files for only
17960 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17961 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17962 input upscaled using bicubic upscaling with proper scale factor.
17965 This feature can also be finished with @ref{dnn_processing} filter.
17969 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17971 This filter takes in input two input videos, the first input is
17972 considered the "main" source and is passed unchanged to the
17973 output. The second input is used as a "reference" video for computing
17976 Both video inputs must have the same resolution and pixel format for
17977 this filter to work correctly. Also it assumes that both inputs
17978 have the same number of frames, which are compared one by one.
17980 The filter stores the calculated SSIM of each frame.
17982 The description of the accepted parameters follows.
17985 @item stats_file, f
17986 If specified the filter will use the named file to save the SSIM of
17987 each individual frame. When filename equals "-" the data is sent to
17991 The file printed if @var{stats_file} is selected, contains a sequence of
17992 key/value pairs of the form @var{key}:@var{value} for each compared
17995 A description of each shown parameter follows:
17999 sequential number of the input frame, starting from 1
18001 @item Y, U, V, R, G, B
18002 SSIM of the compared frames for the component specified by the suffix.
18005 SSIM of the compared frames for the whole frame.
18008 Same as above but in dB representation.
18011 This filter also supports the @ref{framesync} options.
18013 @subsection Examples
18018 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18019 [main][ref] ssim="stats_file=stats.log" [out]
18022 On this example the input file being processed is compared with the
18023 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18024 is stored in @file{stats.log}.
18027 Another example with both psnr and ssim at same time:
18029 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18033 Another example with different containers:
18035 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 -
18041 Convert between different stereoscopic image formats.
18043 The filters accept the following options:
18047 Set stereoscopic image format of input.
18049 Available values for input image formats are:
18052 side by side parallel (left eye left, right eye right)
18055 side by side crosseye (right eye left, left eye right)
18058 side by side parallel with half width resolution
18059 (left eye left, right eye right)
18062 side by side crosseye with half width resolution
18063 (right eye left, left eye right)
18067 above-below (left eye above, right eye below)
18071 above-below (right eye above, left eye below)
18075 above-below with half height resolution
18076 (left eye above, right eye below)
18080 above-below with half height resolution
18081 (right eye above, left eye below)
18084 alternating frames (left eye first, right eye second)
18087 alternating frames (right eye first, left eye second)
18090 interleaved rows (left eye has top row, right eye starts on next row)
18093 interleaved rows (right eye has top row, left eye starts on next row)
18096 interleaved columns, left eye first
18099 interleaved columns, right eye first
18101 Default value is @samp{sbsl}.
18105 Set stereoscopic image format of output.
18109 side by side parallel (left eye left, right eye right)
18112 side by side crosseye (right eye left, left eye right)
18115 side by side parallel with half width resolution
18116 (left eye left, right eye right)
18119 side by side crosseye with half width resolution
18120 (right eye left, left eye right)
18124 above-below (left eye above, right eye below)
18128 above-below (right eye above, left eye below)
18132 above-below with half height resolution
18133 (left eye above, right eye below)
18137 above-below with half height resolution
18138 (right eye above, left eye below)
18141 alternating frames (left eye first, right eye second)
18144 alternating frames (right eye first, left eye second)
18147 interleaved rows (left eye has top row, right eye starts on next row)
18150 interleaved rows (right eye has top row, left eye starts on next row)
18153 anaglyph red/blue gray
18154 (red filter on left eye, blue filter on right eye)
18157 anaglyph red/green gray
18158 (red filter on left eye, green filter on right eye)
18161 anaglyph red/cyan gray
18162 (red filter on left eye, cyan filter on right eye)
18165 anaglyph red/cyan half colored
18166 (red filter on left eye, cyan filter on right eye)
18169 anaglyph red/cyan color
18170 (red filter on left eye, cyan filter on right eye)
18173 anaglyph red/cyan color optimized with the least squares projection of dubois
18174 (red filter on left eye, cyan filter on right eye)
18177 anaglyph green/magenta gray
18178 (green filter on left eye, magenta filter on right eye)
18181 anaglyph green/magenta half colored
18182 (green filter on left eye, magenta filter on right eye)
18185 anaglyph green/magenta colored
18186 (green filter on left eye, magenta filter on right eye)
18189 anaglyph green/magenta color optimized with the least squares projection of dubois
18190 (green filter on left eye, magenta filter on right eye)
18193 anaglyph yellow/blue gray
18194 (yellow filter on left eye, blue filter on right eye)
18197 anaglyph yellow/blue half colored
18198 (yellow filter on left eye, blue filter on right eye)
18201 anaglyph yellow/blue colored
18202 (yellow filter on left eye, blue filter on right eye)
18205 anaglyph yellow/blue color optimized with the least squares projection of dubois
18206 (yellow filter on left eye, blue filter on right eye)
18209 mono output (left eye only)
18212 mono output (right eye only)
18215 checkerboard, left eye first
18218 checkerboard, right eye first
18221 interleaved columns, left eye first
18224 interleaved columns, right eye first
18230 Default value is @samp{arcd}.
18233 @subsection Examples
18237 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18243 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18249 @section streamselect, astreamselect
18250 Select video or audio streams.
18252 The filter accepts the following options:
18256 Set number of inputs. Default is 2.
18259 Set input indexes to remap to outputs.
18262 @subsection Commands
18264 The @code{streamselect} and @code{astreamselect} filter supports the following
18269 Set input indexes to remap to outputs.
18272 @subsection Examples
18276 Select first 5 seconds 1st stream and rest of time 2nd stream:
18278 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18282 Same as above, but for audio:
18284 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18291 Draw subtitles on top of input video using the libass library.
18293 To enable compilation of this filter you need to configure FFmpeg with
18294 @code{--enable-libass}. This filter also requires a build with libavcodec and
18295 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18296 Alpha) subtitles format.
18298 The filter accepts the following options:
18302 Set the filename of the subtitle file to read. It must be specified.
18304 @item original_size
18305 Specify the size of the original video, the video for which the ASS file
18306 was composed. For the syntax of this option, check the
18307 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18308 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18309 correctly scale the fonts if the aspect ratio has been changed.
18312 Set a directory path containing fonts that can be used by the filter.
18313 These fonts will be used in addition to whatever the font provider uses.
18316 Process alpha channel, by default alpha channel is untouched.
18319 Set subtitles input character encoding. @code{subtitles} filter only. Only
18320 useful if not UTF-8.
18322 @item stream_index, si
18323 Set subtitles stream index. @code{subtitles} filter only.
18326 Override default style or script info parameters of the subtitles. It accepts a
18327 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18330 If the first key is not specified, it is assumed that the first value
18331 specifies the @option{filename}.
18333 For example, to render the file @file{sub.srt} on top of the input
18334 video, use the command:
18339 which is equivalent to:
18341 subtitles=filename=sub.srt
18344 To render the default subtitles stream from file @file{video.mkv}, use:
18346 subtitles=video.mkv
18349 To render the second subtitles stream from that file, use:
18351 subtitles=video.mkv:si=1
18354 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18355 @code{DejaVu Serif}, use:
18357 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18360 @section super2xsai
18362 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18363 Interpolate) pixel art scaling algorithm.
18365 Useful for enlarging pixel art images without reducing sharpness.
18369 Swap two rectangular objects in video.
18371 This filter accepts the following options:
18381 Set 1st rect x coordinate.
18384 Set 1st rect y coordinate.
18387 Set 2nd rect x coordinate.
18390 Set 2nd rect y coordinate.
18392 All expressions are evaluated once for each frame.
18395 The all options are expressions containing the following constants:
18400 The input width and height.
18403 same as @var{w} / @var{h}
18406 input sample aspect ratio
18409 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18412 The number of the input frame, starting from 0.
18415 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18418 the position in the file of the input frame, NAN if unknown
18425 Blend successive video frames.
18431 Apply telecine process to the video.
18433 This filter accepts the following options:
18442 The default value is @code{top}.
18446 A string of numbers representing the pulldown pattern you wish to apply.
18447 The default value is @code{23}.
18451 Some typical patterns:
18456 24p: 2332 (preferred)
18463 24p: 222222222223 ("Euro pulldown")
18468 @section thistogram
18470 Compute and draw a color distribution histogram for the input video across time.
18472 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18473 at certain time, this filter shows also past histograms of number of frames defined
18474 by @code{width} option.
18476 The computed histogram is a representation of the color component
18477 distribution in an image.
18479 The filter accepts the following options:
18483 Set width of single color component output. Default value is @code{0}.
18484 Value of @code{0} means width will be picked from input video.
18485 This also set number of passed histograms to keep.
18486 Allowed range is [0, 8192].
18488 @item display_mode, d
18490 It accepts the following values:
18493 Per color component graphs are placed below each other.
18496 Per color component graphs are placed side by side.
18499 Presents information identical to that in the @code{parade}, except
18500 that the graphs representing color components are superimposed directly
18503 Default is @code{stack}.
18505 @item levels_mode, m
18506 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18507 Default is @code{linear}.
18509 @item components, c
18510 Set what color components to display.
18511 Default is @code{7}.
18514 Set background opacity. Default is @code{0.9}.
18517 Show envelope. Default is disabled.
18520 Set envelope color. Default is @code{gold}.
18525 Available values for slide is:
18528 Draw new frame when right border is reached.
18531 Replace old columns with new ones.
18534 Scroll from right to left.
18537 Scroll from left to right.
18540 Draw single picture.
18543 Default is @code{replace}.
18548 Apply threshold effect to video stream.
18550 This filter needs four video streams to perform thresholding.
18551 First stream is stream we are filtering.
18552 Second stream is holding threshold values, third stream is holding min values,
18553 and last, fourth stream is holding max values.
18555 The filter accepts the following option:
18559 Set which planes will be processed, unprocessed planes will be copied.
18560 By default value 0xf, all planes will be processed.
18563 For example if first stream pixel's component value is less then threshold value
18564 of pixel component from 2nd threshold stream, third stream value will picked,
18565 otherwise fourth stream pixel component value will be picked.
18567 Using color source filter one can perform various types of thresholding:
18569 @subsection Examples
18573 Binary threshold, using gray color as threshold:
18575 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18579 Inverted binary threshold, using gray color as threshold:
18581 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18585 Truncate binary threshold, using gray color as threshold:
18587 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18591 Threshold to zero, using gray color as threshold:
18593 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18597 Inverted threshold to zero, using gray color as threshold:
18599 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18604 Select the most representative frame in a given sequence of consecutive frames.
18606 The filter accepts the following options:
18610 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18611 will pick one of them, and then handle the next batch of @var{n} frames until
18612 the end. Default is @code{100}.
18615 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18616 value will result in a higher memory usage, so a high value is not recommended.
18618 @subsection Examples
18622 Extract one picture each 50 frames:
18628 Complete example of a thumbnail creation with @command{ffmpeg}:
18630 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18637 Tile several successive frames together.
18639 The @ref{untile} filter can do the reverse.
18641 The filter accepts the following options:
18646 Set the grid size (i.e. the number of lines and columns). For the syntax of
18647 this option, check the
18648 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18651 Set the maximum number of frames to render in the given area. It must be less
18652 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18653 the area will be used.
18656 Set the outer border margin in pixels.
18659 Set the inner border thickness (i.e. the number of pixels between frames). For
18660 more advanced padding options (such as having different values for the edges),
18661 refer to the pad video filter.
18664 Specify the color of the unused area. For the syntax of this option, check the
18665 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18666 The default value of @var{color} is "black".
18669 Set the number of frames to overlap when tiling several successive frames together.
18670 The value must be between @code{0} and @var{nb_frames - 1}.
18673 Set the number of frames to initially be empty before displaying first output frame.
18674 This controls how soon will one get first output frame.
18675 The value must be between @code{0} and @var{nb_frames - 1}.
18678 @subsection Examples
18682 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18684 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18686 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18687 duplicating each output frame to accommodate the originally detected frame
18691 Display @code{5} pictures in an area of @code{3x2} frames,
18692 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18693 mixed flat and named options:
18695 tile=3x2:nb_frames=5:padding=7:margin=2
18699 @section tinterlace
18701 Perform various types of temporal field interlacing.
18703 Frames are counted starting from 1, so the first input frame is
18706 The filter accepts the following options:
18711 Specify the mode of the interlacing. This option can also be specified
18712 as a value alone. See below for a list of values for this option.
18714 Available values are:
18718 Move odd frames into the upper field, even into the lower field,
18719 generating a double height frame at half frame rate.
18723 Frame 1 Frame 2 Frame 3 Frame 4
18725 11111 22222 33333 44444
18726 11111 22222 33333 44444
18727 11111 22222 33333 44444
18728 11111 22222 33333 44444
18742 Only output odd frames, even frames are dropped, generating a frame with
18743 unchanged height at half frame rate.
18748 Frame 1 Frame 2 Frame 3 Frame 4
18750 11111 22222 33333 44444
18751 11111 22222 33333 44444
18752 11111 22222 33333 44444
18753 11111 22222 33333 44444
18763 Only output even frames, odd frames are dropped, generating a frame with
18764 unchanged height at half frame rate.
18769 Frame 1 Frame 2 Frame 3 Frame 4
18771 11111 22222 33333 44444
18772 11111 22222 33333 44444
18773 11111 22222 33333 44444
18774 11111 22222 33333 44444
18784 Expand each frame to full height, but pad alternate lines with black,
18785 generating a frame with double height at the same input frame rate.
18790 Frame 1 Frame 2 Frame 3 Frame 4
18792 11111 22222 33333 44444
18793 11111 22222 33333 44444
18794 11111 22222 33333 44444
18795 11111 22222 33333 44444
18798 11111 ..... 33333 .....
18799 ..... 22222 ..... 44444
18800 11111 ..... 33333 .....
18801 ..... 22222 ..... 44444
18802 11111 ..... 33333 .....
18803 ..... 22222 ..... 44444
18804 11111 ..... 33333 .....
18805 ..... 22222 ..... 44444
18809 @item interleave_top, 4
18810 Interleave the upper field from odd frames with the lower field from
18811 even frames, generating a frame with unchanged height at half frame rate.
18816 Frame 1 Frame 2 Frame 3 Frame 4
18818 11111<- 22222 33333<- 44444
18819 11111 22222<- 33333 44444<-
18820 11111<- 22222 33333<- 44444
18821 11111 22222<- 33333 44444<-
18831 @item interleave_bottom, 5
18832 Interleave the lower field from odd frames with the upper field from
18833 even frames, generating a frame with unchanged height at half frame rate.
18838 Frame 1 Frame 2 Frame 3 Frame 4
18840 11111 22222<- 33333 44444<-
18841 11111<- 22222 33333<- 44444
18842 11111 22222<- 33333 44444<-
18843 11111<- 22222 33333<- 44444
18853 @item interlacex2, 6
18854 Double frame rate with unchanged height. Frames are inserted each
18855 containing the second temporal field from the previous input frame and
18856 the first temporal field from the next input frame. This mode relies on
18857 the top_field_first flag. Useful for interlaced video displays with no
18858 field synchronisation.
18863 Frame 1 Frame 2 Frame 3 Frame 4
18865 11111 22222 33333 44444
18866 11111 22222 33333 44444
18867 11111 22222 33333 44444
18868 11111 22222 33333 44444
18871 11111 22222 22222 33333 33333 44444 44444
18872 11111 11111 22222 22222 33333 33333 44444
18873 11111 22222 22222 33333 33333 44444 44444
18874 11111 11111 22222 22222 33333 33333 44444
18879 Move odd frames into the upper field, even into the lower field,
18880 generating a double height frame at same frame rate.
18885 Frame 1 Frame 2 Frame 3 Frame 4
18887 11111 22222 33333 44444
18888 11111 22222 33333 44444
18889 11111 22222 33333 44444
18890 11111 22222 33333 44444
18893 11111 33333 33333 55555
18894 22222 22222 44444 44444
18895 11111 33333 33333 55555
18896 22222 22222 44444 44444
18897 11111 33333 33333 55555
18898 22222 22222 44444 44444
18899 11111 33333 33333 55555
18900 22222 22222 44444 44444
18905 Numeric values are deprecated but are accepted for backward
18906 compatibility reasons.
18908 Default mode is @code{merge}.
18911 Specify flags influencing the filter process.
18913 Available value for @var{flags} is:
18916 @item low_pass_filter, vlpf
18917 Enable linear vertical low-pass filtering in the filter.
18918 Vertical low-pass filtering is required when creating an interlaced
18919 destination from a progressive source which contains high-frequency
18920 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18923 @item complex_filter, cvlpf
18924 Enable complex vertical low-pass filtering.
18925 This will slightly less reduce interlace 'twitter' and Moire
18926 patterning but better retain detail and subjective sharpness impression.
18929 Bypass already interlaced frames, only adjust the frame rate.
18932 Vertical low-pass filtering and bypassing already interlaced frames can only be
18933 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
18938 Pick median pixels from several successive input video frames.
18940 The filter accepts the following options:
18944 Set radius of median filter.
18945 Default is 1. Allowed range is from 1 to 127.
18948 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
18951 Set median percentile. Default value is @code{0.5}.
18952 Default value of @code{0.5} will pick always median values, while @code{0} will pick
18953 minimum values, and @code{1} maximum values.
18958 Mix successive video frames.
18960 A description of the accepted options follows.
18964 The number of successive frames to mix. If unspecified, it defaults to 3.
18967 Specify weight of each input video frame.
18968 Each weight is separated by space. If number of weights is smaller than
18969 number of @var{frames} last specified weight will be used for all remaining
18973 Specify scale, if it is set it will be multiplied with sum
18974 of each weight multiplied with pixel values to give final destination
18975 pixel value. By default @var{scale} is auto scaled to sum of weights.
18978 @subsection Examples
18982 Average 7 successive frames:
18984 tmix=frames=7:weights="1 1 1 1 1 1 1"
18988 Apply simple temporal convolution:
18990 tmix=frames=3:weights="-1 3 -1"
18994 Similar as above but only showing temporal differences:
18996 tmix=frames=3:weights="-1 2 -1":scale=1
19002 Tone map colors from different dynamic ranges.
19004 This filter expects data in single precision floating point, as it needs to
19005 operate on (and can output) out-of-range values. Another filter, such as
19006 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19008 The tonemapping algorithms implemented only work on linear light, so input
19009 data should be linearized beforehand (and possibly correctly tagged).
19012 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19015 @subsection Options
19016 The filter accepts the following options.
19020 Set the tone map algorithm to use.
19022 Possible values are:
19025 Do not apply any tone map, only desaturate overbright pixels.
19028 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19029 in-range values, while distorting out-of-range values.
19032 Stretch the entire reference gamut to a linear multiple of the display.
19035 Fit a logarithmic transfer between the tone curves.
19038 Preserve overall image brightness with a simple curve, using nonlinear
19039 contrast, which results in flattening details and degrading color accuracy.
19042 Preserve both dark and bright details better than @var{reinhard}, at the cost
19043 of slightly darkening everything. Use it when detail preservation is more
19044 important than color and brightness accuracy.
19047 Smoothly map out-of-range values, while retaining contrast and colors for
19048 in-range material as much as possible. Use it when color accuracy is more
19049 important than detail preservation.
19055 Tune the tone mapping algorithm.
19057 This affects the following algorithms:
19063 Specifies the scale factor to use while stretching.
19067 Specifies the exponent of the function.
19071 Specify an extra linear coefficient to multiply into the signal before clipping.
19075 Specify the local contrast coefficient at the display peak.
19076 Default to 0.5, which means that in-gamut values will be about half as bright
19083 Specify the transition point from linear to mobius transform. Every value
19084 below this point is guaranteed to be mapped 1:1. The higher the value, the
19085 more accurate the result will be, at the cost of losing bright details.
19086 Default to 0.3, which due to the steep initial slope still preserves in-range
19087 colors fairly accurately.
19091 Apply desaturation for highlights that exceed this level of brightness. The
19092 higher the parameter, the more color information will be preserved. This
19093 setting helps prevent unnaturally blown-out colors for super-highlights, by
19094 (smoothly) turning into white instead. This makes images feel more natural,
19095 at the cost of reducing information about out-of-range colors.
19097 The default of 2.0 is somewhat conservative and will mostly just apply to
19098 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19100 This option works only if the input frame has a supported color tag.
19103 Override signal/nominal/reference peak with this value. Useful when the
19104 embedded peak information in display metadata is not reliable or when tone
19105 mapping from a lower range to a higher range.
19110 Temporarily pad video frames.
19112 The filter accepts the following options:
19116 Specify number of delay frames before input video stream. Default is 0.
19119 Specify number of padding frames after input video stream.
19120 Set to -1 to pad indefinitely. Default is 0.
19123 Set kind of frames added to beginning of stream.
19124 Can be either @var{add} or @var{clone}.
19125 With @var{add} frames of solid-color are added.
19126 With @var{clone} frames are clones of first frame.
19127 Default is @var{add}.
19130 Set kind of frames added to end of stream.
19131 Can be either @var{add} or @var{clone}.
19132 With @var{add} frames of solid-color are added.
19133 With @var{clone} frames are clones of last frame.
19134 Default is @var{add}.
19136 @item start_duration, stop_duration
19137 Specify the duration of the start/stop delay. See
19138 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19139 for the accepted syntax.
19140 These options override @var{start} and @var{stop}. Default is 0.
19143 Specify the color of the padded area. For the syntax of this option,
19144 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19145 manual,ffmpeg-utils}.
19147 The default value of @var{color} is "black".
19153 Transpose rows with columns in the input video and optionally flip it.
19155 It accepts the following parameters:
19160 Specify the transposition direction.
19162 Can assume the following values:
19164 @item 0, 4, cclock_flip
19165 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19173 Rotate by 90 degrees clockwise, that is:
19181 Rotate by 90 degrees counterclockwise, that is:
19188 @item 3, 7, clock_flip
19189 Rotate by 90 degrees clockwise and vertically flip, that is:
19197 For values between 4-7, the transposition is only done if the input
19198 video geometry is portrait and not landscape. These values are
19199 deprecated, the @code{passthrough} option should be used instead.
19201 Numerical values are deprecated, and should be dropped in favor of
19202 symbolic constants.
19205 Do not apply the transposition if the input geometry matches the one
19206 specified by the specified value. It accepts the following values:
19209 Always apply transposition.
19211 Preserve portrait geometry (when @var{height} >= @var{width}).
19213 Preserve landscape geometry (when @var{width} >= @var{height}).
19216 Default value is @code{none}.
19219 For example to rotate by 90 degrees clockwise and preserve portrait
19222 transpose=dir=1:passthrough=portrait
19225 The command above can also be specified as:
19227 transpose=1:portrait
19230 @section transpose_npp
19232 Transpose rows with columns in the input video and optionally flip it.
19233 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19235 It accepts the following parameters:
19240 Specify the transposition direction.
19242 Can assume the following values:
19245 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19248 Rotate by 90 degrees clockwise.
19251 Rotate by 90 degrees counterclockwise.
19254 Rotate by 90 degrees clockwise and vertically flip.
19258 Do not apply the transposition if the input geometry matches the one
19259 specified by the specified value. It accepts the following values:
19262 Always apply transposition. (default)
19264 Preserve portrait geometry (when @var{height} >= @var{width}).
19266 Preserve landscape geometry (when @var{width} >= @var{height}).
19272 Trim the input so that the output contains one continuous subpart of the input.
19274 It accepts the following parameters:
19277 Specify the time of the start of the kept section, i.e. the frame with the
19278 timestamp @var{start} will be the first frame in the output.
19281 Specify the time of the first frame that will be dropped, i.e. the frame
19282 immediately preceding the one with the timestamp @var{end} will be the last
19283 frame in the output.
19286 This is the same as @var{start}, except this option sets the start timestamp
19287 in timebase units instead of seconds.
19290 This is the same as @var{end}, except this option sets the end timestamp
19291 in timebase units instead of seconds.
19294 The maximum duration of the output in seconds.
19297 The number of the first frame that should be passed to the output.
19300 The number of the first frame that should be dropped.
19303 @option{start}, @option{end}, and @option{duration} are expressed as time
19304 duration specifications; see
19305 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19306 for the accepted syntax.
19308 Note that the first two sets of the start/end options and the @option{duration}
19309 option look at the frame timestamp, while the _frame variants simply count the
19310 frames that pass through the filter. Also note that this filter does not modify
19311 the timestamps. If you wish for the output timestamps to start at zero, insert a
19312 setpts filter after the trim filter.
19314 If multiple start or end options are set, this filter tries to be greedy and
19315 keep all the frames that match at least one of the specified constraints. To keep
19316 only the part that matches all the constraints at once, chain multiple trim
19319 The defaults are such that all the input is kept. So it is possible to set e.g.
19320 just the end values to keep everything before the specified time.
19325 Drop everything except the second minute of input:
19327 ffmpeg -i INPUT -vf trim=60:120
19331 Keep only the first second:
19333 ffmpeg -i INPUT -vf trim=duration=1
19338 @section unpremultiply
19339 Apply alpha unpremultiply effect to input video stream using first plane
19340 of second stream as alpha.
19342 Both streams must have same dimensions and same pixel format.
19344 The filter accepts the following option:
19348 Set which planes will be processed, unprocessed planes will be copied.
19349 By default value 0xf, all planes will be processed.
19351 If the format has 1 or 2 components, then luma is bit 0.
19352 If the format has 3 or 4 components:
19353 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19354 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19355 If present, the alpha channel is always the last bit.
19358 Do not require 2nd input for processing, instead use alpha plane from input stream.
19364 Sharpen or blur the input video.
19366 It accepts the following parameters:
19369 @item luma_msize_x, lx
19370 Set the luma matrix horizontal size. It must be an odd integer between
19371 3 and 23. The default value is 5.
19373 @item luma_msize_y, ly
19374 Set the luma matrix vertical size. It must be an odd integer between 3
19375 and 23. The default value is 5.
19377 @item luma_amount, la
19378 Set the luma effect strength. It must be a floating point number, reasonable
19379 values lay between -1.5 and 1.5.
19381 Negative values will blur the input video, while positive values will
19382 sharpen it, a value of zero will disable the effect.
19384 Default value is 1.0.
19386 @item chroma_msize_x, cx
19387 Set the chroma matrix horizontal size. It must be an odd integer
19388 between 3 and 23. The default value is 5.
19390 @item chroma_msize_y, cy
19391 Set the chroma matrix vertical size. It must be an odd integer
19392 between 3 and 23. The default value is 5.
19394 @item chroma_amount, ca
19395 Set the chroma effect strength. It must be a floating point number, reasonable
19396 values lay between -1.5 and 1.5.
19398 Negative values will blur the input video, while positive values will
19399 sharpen it, a value of zero will disable the effect.
19401 Default value is 0.0.
19405 All parameters are optional and default to the equivalent of the
19406 string '5:5:1.0:5:5:0.0'.
19408 @subsection Examples
19412 Apply strong luma sharpen effect:
19414 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19418 Apply a strong blur of both luma and chroma parameters:
19420 unsharp=7:7:-2:7:7:-2
19427 Decompose a video made of tiled images into the individual images.
19429 The frame rate of the output video is the frame rate of the input video
19430 multiplied by the number of tiles.
19432 This filter does the reverse of @ref{tile}.
19434 The filter accepts the following options:
19439 Set the grid size (i.e. the number of lines and columns). For the syntax of
19440 this option, check the
19441 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19444 @subsection Examples
19448 Produce a 1-second video from a still image file made of 25 frames stacked
19449 vertically, like an analogic film reel:
19451 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19457 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19458 the image at several (or - in the case of @option{quality} level @code{8} - all)
19459 shifts and average the results.
19461 The way this differs from the behavior of spp is that uspp actually encodes &
19462 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19463 DCT similar to MJPEG.
19465 The filter accepts the following options:
19469 Set quality. This option defines the number of levels for averaging. It accepts
19470 an integer in the range 0-8. If set to @code{0}, the filter will have no
19471 effect. A value of @code{8} means the higher quality. For each increment of
19472 that value the speed drops by a factor of approximately 2. Default value is
19476 Force a constant quantization parameter. If not set, the filter will use the QP
19477 from the video stream (if available).
19482 Convert 360 videos between various formats.
19484 The filter accepts the following options:
19490 Set format of the input/output video.
19498 Equirectangular projection.
19503 Cubemap with 3x2/6x1/1x6 layout.
19505 Format specific options:
19510 Set padding proportion for the input/output cubemap. Values in decimals.
19517 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)
19520 Default value is @b{@samp{0}}.
19521 Maximum value is @b{@samp{0.1}}.
19525 Set fixed padding for the input/output cubemap. Values in pixels.
19527 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19531 Set order of faces for the input/output cubemap. Choose one direction for each position.
19533 Designation of directions:
19549 Default value is @b{@samp{rludfb}}.
19553 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19555 Designation of angles:
19558 0 degrees clockwise
19560 90 degrees clockwise
19562 180 degrees clockwise
19564 270 degrees clockwise
19567 Default value is @b{@samp{000000}}.
19571 Equi-Angular Cubemap.
19578 Format specific options:
19583 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19585 If diagonal field of view is set it overrides horizontal and vertical field of view.
19590 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19592 If diagonal field of view is set it overrides horizontal and vertical field of view.
19598 Format specific options:
19603 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19605 If diagonal field of view is set it overrides horizontal and vertical field of view.
19610 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19612 If diagonal field of view is set it overrides horizontal and vertical field of view.
19618 Facebook's 360 formats.
19621 Stereographic format.
19623 Format specific options:
19628 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19630 If diagonal field of view is set it overrides horizontal and vertical field of view.
19635 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19637 If diagonal field of view is set it overrides horizontal and vertical field of view.
19644 Ball format, gives significant distortion toward the back.
19647 Hammer-Aitoff map projection format.
19650 Sinusoidal map projection format.
19653 Fisheye projection.
19655 Format specific options:
19660 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19662 If diagonal field of view is set it overrides horizontal and vertical field of view.
19667 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19669 If diagonal field of view is set it overrides horizontal and vertical field of view.
19673 Pannini projection.
19675 Format specific options:
19678 Set output pannini parameter.
19681 Set input pannini parameter.
19685 Cylindrical projection.
19687 Format specific options:
19692 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19694 If diagonal field of view is set it overrides horizontal and vertical field of view.
19699 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19701 If diagonal field of view is set it overrides horizontal and vertical field of view.
19705 Perspective projection. @i{(output only)}
19707 Format specific options:
19710 Set perspective parameter.
19714 Tetrahedron projection.
19717 Truncated square pyramid projection.
19721 Half equirectangular projection.
19726 Format specific options:
19731 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19733 If diagonal field of view is set it overrides horizontal and vertical field of view.
19738 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19740 If diagonal field of view is set it overrides horizontal and vertical field of view.
19744 Orthographic format.
19746 Format specific options:
19751 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19753 If diagonal field of view is set it overrides horizontal and vertical field of view.
19758 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19760 If diagonal field of view is set it overrides horizontal and vertical field of view.
19764 Octahedron projection.
19768 Set interpolation method.@*
19769 @i{Note: more complex interpolation methods require much more memory to run.}
19779 Bilinear interpolation.
19781 Lagrange9 interpolation.
19784 Bicubic interpolation.
19787 Lanczos interpolation.
19790 Spline16 interpolation.
19793 Gaussian interpolation.
19795 Mitchell interpolation.
19798 Default value is @b{@samp{line}}.
19802 Set the output video resolution.
19804 Default resolution depends on formats.
19808 Set the input/output stereo format.
19819 Default value is @b{@samp{2d}} for input and output format.
19824 Set rotation for the output video. Values in degrees.
19827 Set rotation order for the output video. Choose one item for each position.
19838 Default value is @b{@samp{ypr}}.
19843 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
19847 Set if input video is flipped horizontally/vertically. Boolean values.
19850 Set if input video is transposed. Boolean value, by default disabled.
19853 Set if output video needs to be transposed. Boolean value, by default disabled.
19856 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
19859 @subsection Examples
19863 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
19865 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19868 Extract back view of Equi-Angular Cubemap:
19870 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19873 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
19875 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19879 @subsection Commands
19881 This filter supports subset of above options as @ref{commands}.
19883 @section vaguedenoiser
19885 Apply a wavelet based denoiser.
19887 It transforms each frame from the video input into the wavelet domain,
19888 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19889 the obtained coefficients. It does an inverse wavelet transform after.
19890 Due to wavelet properties, it should give a nice smoothed result, and
19891 reduced noise, without blurring picture features.
19893 This filter accepts the following options:
19897 The filtering strength. The higher, the more filtered the video will be.
19898 Hard thresholding can use a higher threshold than soft thresholding
19899 before the video looks overfiltered. Default value is 2.
19902 The filtering method the filter will use.
19904 It accepts the following values:
19907 All values under the threshold will be zeroed.
19910 All values under the threshold will be zeroed. All values above will be
19911 reduced by the threshold.
19914 Scales or nullifies coefficients - intermediary between (more) soft and
19915 (less) hard thresholding.
19918 Default is garrote.
19921 Number of times, the wavelet will decompose the picture. Picture can't
19922 be decomposed beyond a particular point (typically, 8 for a 640x480
19923 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
19926 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
19929 A list of the planes to process. By default all planes are processed.
19932 The threshold type the filter will use.
19934 It accepts the following values:
19937 Threshold used is same for all decompositions.
19940 Threshold used depends also on each decomposition coefficients.
19943 Default is universal.
19946 @section vectorscope
19948 Display 2 color component values in the two dimensional graph (which is called
19951 This filter accepts the following options:
19955 Set vectorscope mode.
19957 It accepts the following values:
19961 Gray values are displayed on graph, higher brightness means more pixels have
19962 same component color value on location in graph. This is the default mode.
19965 Gray values are displayed on graph. Surrounding pixels values which are not
19966 present in video frame are drawn in gradient of 2 color components which are
19967 set by option @code{x} and @code{y}. The 3rd color component is static.
19970 Actual color components values present in video frame are displayed on graph.
19973 Similar as color2 but higher frequency of same values @code{x} and @code{y}
19974 on graph increases value of another color component, which is luminance by
19975 default values of @code{x} and @code{y}.
19978 Actual colors present in video frame are displayed on graph. If two different
19979 colors map to same position on graph then color with higher value of component
19980 not present in graph is picked.
19983 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
19984 component picked from radial gradient.
19988 Set which color component will be represented on X-axis. Default is @code{1}.
19991 Set which color component will be represented on Y-axis. Default is @code{2}.
19994 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
19995 of color component which represents frequency of (X, Y) location in graph.
20000 No envelope, this is default.
20003 Instant envelope, even darkest single pixel will be clearly highlighted.
20006 Hold maximum and minimum values presented in graph over time. This way you
20007 can still spot out of range values without constantly looking at vectorscope.
20010 Peak and instant envelope combined together.
20014 Set what kind of graticule to draw.
20023 Set graticule opacity.
20026 Set graticule flags.
20030 Draw graticule for white point.
20033 Draw graticule for black point.
20036 Draw color points short names.
20040 Set background opacity.
20042 @item lthreshold, l
20043 Set low threshold for color component not represented on X or Y axis.
20044 Values lower than this value will be ignored. Default is 0.
20045 Note this value is multiplied with actual max possible value one pixel component
20046 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20049 @item hthreshold, h
20050 Set high threshold for color component not represented on X or Y axis.
20051 Values higher than this value will be ignored. Default is 1.
20052 Note this value is multiplied with actual max possible value one pixel component
20053 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20054 is 0.9 * 255 = 230.
20056 @item colorspace, c
20057 Set what kind of colorspace to use when drawing graticule.
20067 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20068 This means no tint, and output will remain gray.
20071 @anchor{vidstabdetect}
20072 @section vidstabdetect
20074 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20075 @ref{vidstabtransform} for pass 2.
20077 This filter generates a file with relative translation and rotation
20078 transform information about subsequent frames, which is then used by
20079 the @ref{vidstabtransform} filter.
20081 To enable compilation of this filter you need to configure FFmpeg with
20082 @code{--enable-libvidstab}.
20084 This filter accepts the following options:
20088 Set the path to the file used to write the transforms information.
20089 Default value is @file{transforms.trf}.
20092 Set how shaky the video is and how quick the camera is. It accepts an
20093 integer in the range 1-10, a value of 1 means little shakiness, a
20094 value of 10 means strong shakiness. Default value is 5.
20097 Set the accuracy of the detection process. It must be a value in the
20098 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20099 accuracy. Default value is 15.
20102 Set stepsize of the search process. The region around minimum is
20103 scanned with 1 pixel resolution. Default value is 6.
20106 Set minimum contrast. Below this value a local measurement field is
20107 discarded. Must be a floating point value in the range 0-1. Default
20111 Set reference frame number for tripod mode.
20113 If enabled, the motion of the frames is compared to a reference frame
20114 in the filtered stream, identified by the specified number. The idea
20115 is to compensate all movements in a more-or-less static scene and keep
20116 the camera view absolutely still.
20118 If set to 0, it is disabled. The frames are counted starting from 1.
20121 Show fields and transforms in the resulting frames. It accepts an
20122 integer in the range 0-2. Default value is 0, which disables any
20126 @subsection Examples
20130 Use default values:
20136 Analyze strongly shaky movie and put the results in file
20137 @file{mytransforms.trf}:
20139 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20143 Visualize the result of internal transformations in the resulting
20146 vidstabdetect=show=1
20150 Analyze a video with medium shakiness using @command{ffmpeg}:
20152 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20156 @anchor{vidstabtransform}
20157 @section vidstabtransform
20159 Video stabilization/deshaking: pass 2 of 2,
20160 see @ref{vidstabdetect} for pass 1.
20162 Read a file with transform information for each frame and
20163 apply/compensate them. Together with the @ref{vidstabdetect}
20164 filter this can be used to deshake videos. See also
20165 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20166 the @ref{unsharp} filter, see below.
20168 To enable compilation of this filter you need to configure FFmpeg with
20169 @code{--enable-libvidstab}.
20171 @subsection Options
20175 Set path to the file used to read the transforms. Default value is
20176 @file{transforms.trf}.
20179 Set the number of frames (value*2 + 1) used for lowpass filtering the
20180 camera movements. Default value is 10.
20182 For example a number of 10 means that 21 frames are used (10 in the
20183 past and 10 in the future) to smoothen the motion in the video. A
20184 larger value leads to a smoother video, but limits the acceleration of
20185 the camera (pan/tilt movements). 0 is a special case where a static
20186 camera is simulated.
20189 Set the camera path optimization algorithm.
20191 Accepted values are:
20194 gaussian kernel low-pass filter on camera motion (default)
20196 averaging on transformations
20200 Set maximal number of pixels to translate frames. Default value is -1,
20204 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20205 value is -1, meaning no limit.
20208 Specify how to deal with borders that may be visible due to movement
20211 Available values are:
20214 keep image information from previous frame (default)
20216 fill the border black
20220 Invert transforms if set to 1. Default value is 0.
20223 Consider transforms as relative to previous frame if set to 1,
20224 absolute if set to 0. Default value is 0.
20227 Set percentage to zoom. A positive value will result in a zoom-in
20228 effect, a negative value in a zoom-out effect. Default value is 0 (no
20232 Set optimal zooming to avoid borders.
20234 Accepted values are:
20239 optimal static zoom value is determined (only very strong movements
20240 will lead to visible borders) (default)
20242 optimal adaptive zoom value is determined (no borders will be
20243 visible), see @option{zoomspeed}
20246 Note that the value given at zoom is added to the one calculated here.
20249 Set percent to zoom maximally each frame (enabled when
20250 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20254 Specify type of interpolation.
20256 Available values are:
20261 linear only horizontal
20263 linear in both directions (default)
20265 cubic in both directions (slow)
20269 Enable virtual tripod mode if set to 1, which is equivalent to
20270 @code{relative=0:smoothing=0}. Default value is 0.
20272 Use also @code{tripod} option of @ref{vidstabdetect}.
20275 Increase log verbosity if set to 1. Also the detected global motions
20276 are written to the temporary file @file{global_motions.trf}. Default
20280 @subsection Examples
20284 Use @command{ffmpeg} for a typical stabilization with default values:
20286 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20289 Note the use of the @ref{unsharp} filter which is always recommended.
20292 Zoom in a bit more and load transform data from a given file:
20294 vidstabtransform=zoom=5:input="mytransforms.trf"
20298 Smoothen the video even more:
20300 vidstabtransform=smoothing=30
20306 Flip the input video vertically.
20308 For example, to vertically flip a video with @command{ffmpeg}:
20310 ffmpeg -i in.avi -vf "vflip" out.avi
20315 Detect variable frame rate video.
20317 This filter tries to detect if the input is variable or constant frame rate.
20319 At end it will output number of frames detected as having variable delta pts,
20320 and ones with constant delta pts.
20321 If there was frames with variable delta, than it will also show min, max and
20322 average delta encountered.
20326 Boost or alter saturation.
20328 The filter accepts the following options:
20331 Set strength of boost if positive value or strength of alter if negative value.
20332 Default is 0. Allowed range is from -2 to 2.
20335 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20338 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20341 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20344 Set the red luma coefficient.
20347 Set the green luma coefficient.
20350 Set the blue luma coefficient.
20353 If @code{intensity} is negative and this is set to 1, colors will change,
20354 otherwise colors will be less saturated, more towards gray.
20357 @subsection Commands
20359 This filter supports the all above options as @ref{commands}.
20364 Make or reverse a natural vignetting effect.
20366 The filter accepts the following options:
20370 Set lens angle expression as a number of radians.
20372 The value is clipped in the @code{[0,PI/2]} range.
20374 Default value: @code{"PI/5"}
20378 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20382 Set forward/backward mode.
20384 Available modes are:
20387 The larger the distance from the central point, the darker the image becomes.
20390 The larger the distance from the central point, the brighter the image becomes.
20391 This can be used to reverse a vignette effect, though there is no automatic
20392 detection to extract the lens @option{angle} and other settings (yet). It can
20393 also be used to create a burning effect.
20396 Default value is @samp{forward}.
20399 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20401 It accepts the following values:
20404 Evaluate expressions only once during the filter initialization.
20407 Evaluate expressions for each incoming frame. This is way slower than the
20408 @samp{init} mode since it requires all the scalers to be re-computed, but it
20409 allows advanced dynamic expressions.
20412 Default value is @samp{init}.
20415 Set dithering to reduce the circular banding effects. Default is @code{1}
20419 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20420 Setting this value to the SAR of the input will make a rectangular vignetting
20421 following the dimensions of the video.
20423 Default is @code{1/1}.
20426 @subsection Expressions
20428 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20429 following parameters.
20434 input width and height
20437 the number of input frame, starting from 0
20440 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20441 @var{TB} units, NAN if undefined
20444 frame rate of the input video, NAN if the input frame rate is unknown
20447 the PTS (Presentation TimeStamp) of the filtered video frame,
20448 expressed in seconds, NAN if undefined
20451 time base of the input video
20455 @subsection Examples
20459 Apply simple strong vignetting effect:
20465 Make a flickering vignetting:
20467 vignette='PI/4+random(1)*PI/50':eval=frame
20472 @section vmafmotion
20474 Obtain the average VMAF motion score of a video.
20475 It is one of the component metrics of VMAF.
20477 The obtained average motion score is printed through the logging system.
20479 The filter accepts the following options:
20483 If specified, the filter will use the named file to save the motion score of
20484 each frame with respect to the previous frame.
20485 When filename equals "-" the data is sent to standard output.
20490 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20494 Stack input videos vertically.
20496 All streams must be of same pixel format and of same width.
20498 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20499 to create same output.
20501 The filter accepts the following options:
20505 Set number of input streams. Default is 2.
20508 If set to 1, force the output to terminate when the shortest input
20509 terminates. Default value is 0.
20514 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20515 Deinterlacing Filter").
20517 Based on the process described by Martin Weston for BBC R&D, and
20518 implemented based on the de-interlace algorithm written by Jim
20519 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20520 uses filter coefficients calculated by BBC R&D.
20522 This filter uses field-dominance information in frame to decide which
20523 of each pair of fields to place first in the output.
20524 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20526 There are two sets of filter coefficients, so called "simple"
20527 and "complex". Which set of filter coefficients is used can
20528 be set by passing an optional parameter:
20532 Set the interlacing filter coefficients. Accepts one of the following values:
20536 Simple filter coefficient set.
20538 More-complex filter coefficient set.
20540 Default value is @samp{complex}.
20543 Specify which frames to deinterlace. Accepts one of the following values:
20547 Deinterlace all frames,
20549 Only deinterlace frames marked as interlaced.
20552 Default value is @samp{all}.
20556 Video waveform monitor.
20558 The waveform monitor plots color component intensity. By default luminance
20559 only. Each column of the waveform corresponds to a column of pixels in the
20562 It accepts the following options:
20566 Can be either @code{row}, or @code{column}. Default is @code{column}.
20567 In row mode, the graph on the left side represents color component value 0 and
20568 the right side represents value = 255. In column mode, the top side represents
20569 color component value = 0 and bottom side represents value = 255.
20572 Set intensity. Smaller values are useful to find out how many values of the same
20573 luminance are distributed across input rows/columns.
20574 Default value is @code{0.04}. Allowed range is [0, 1].
20577 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20578 In mirrored mode, higher values will be represented on the left
20579 side for @code{row} mode and at the top for @code{column} mode. Default is
20580 @code{1} (mirrored).
20584 It accepts the following values:
20587 Presents information identical to that in the @code{parade}, except
20588 that the graphs representing color components are superimposed directly
20591 This display mode makes it easier to spot relative differences or similarities
20592 in overlapping areas of the color components that are supposed to be identical,
20593 such as neutral whites, grays, or blacks.
20596 Display separate graph for the color components side by side in
20597 @code{row} mode or one below the other in @code{column} mode.
20600 Display separate graph for the color components side by side in
20601 @code{column} mode or one below the other in @code{row} mode.
20603 Using this display mode makes it easy to spot color casts in the highlights
20604 and shadows of an image, by comparing the contours of the top and the bottom
20605 graphs of each waveform. Since whites, grays, and blacks are characterized
20606 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20607 should display three waveforms of roughly equal width/height. If not, the
20608 correction is easy to perform by making level adjustments the three waveforms.
20610 Default is @code{stack}.
20612 @item components, c
20613 Set which color components to display. Default is 1, which means only luminance
20614 or red color component if input is in RGB colorspace. If is set for example to
20615 7 it will display all 3 (if) available color components.
20620 No envelope, this is default.
20623 Instant envelope, minimum and maximum values presented in graph will be easily
20624 visible even with small @code{step} value.
20627 Hold minimum and maximum values presented in graph across time. This way you
20628 can still spot out of range values without constantly looking at waveforms.
20631 Peak and instant envelope combined together.
20637 No filtering, this is default.
20640 Luma and chroma combined together.
20643 Similar as above, but shows difference between blue and red chroma.
20646 Similar as above, but use different colors.
20649 Similar as above, but again with different colors.
20652 Displays only chroma.
20655 Displays actual color value on waveform.
20658 Similar as above, but with luma showing frequency of chroma values.
20662 Set which graticule to display.
20666 Do not display graticule.
20669 Display green graticule showing legal broadcast ranges.
20672 Display orange graticule showing legal broadcast ranges.
20675 Display invert graticule showing legal broadcast ranges.
20679 Set graticule opacity.
20682 Set graticule flags.
20686 Draw numbers above lines. By default enabled.
20689 Draw dots instead of lines.
20693 Set scale used for displaying graticule.
20700 Default is digital.
20703 Set background opacity.
20707 Set tint for output.
20708 Only used with lowpass filter and when display is not overlay and input
20709 pixel formats are not RGB.
20712 @section weave, doubleweave
20714 The @code{weave} takes a field-based video input and join
20715 each two sequential fields into single frame, producing a new double
20716 height clip with half the frame rate and half the frame count.
20718 The @code{doubleweave} works same as @code{weave} but without
20719 halving frame rate and frame count.
20721 It accepts the following option:
20725 Set first field. Available values are:
20729 Set the frame as top-field-first.
20732 Set the frame as bottom-field-first.
20736 @subsection Examples
20740 Interlace video using @ref{select} and @ref{separatefields} filter:
20742 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20747 Apply the xBR high-quality magnification filter which is designed for pixel
20748 art. It follows a set of edge-detection rules, see
20749 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
20751 It accepts the following option:
20755 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
20756 @code{3xBR} and @code{4} for @code{4xBR}.
20757 Default is @code{3}.
20762 Apply cross fade from one input video stream to another input video stream.
20763 The cross fade is applied for specified duration.
20765 The filter accepts the following options:
20769 Set one of available transition effects:
20817 Default transition effect is fade.
20820 Set cross fade duration in seconds.
20821 Default duration is 1 second.
20824 Set cross fade start relative to first input stream in seconds.
20825 Default offset is 0.
20828 Set expression for custom transition effect.
20830 The expressions can use the following variables and functions:
20835 The coordinates of the current sample.
20839 The width and height of the image.
20842 Progress of transition effect.
20845 Currently processed plane.
20848 Return value of first input at current location and plane.
20851 Return value of second input at current location and plane.
20857 Return the value of the pixel at location (@var{x},@var{y}) of the
20858 first/second/third/fourth component of first input.
20864 Return the value of the pixel at location (@var{x},@var{y}) of the
20865 first/second/third/fourth component of second input.
20869 @subsection Examples
20873 Cross fade from one input video to another input video, with fade transition and duration of transition
20874 of 2 seconds starting at offset of 5 seconds:
20876 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20881 Pick median pixels from several input videos.
20883 The filter accepts the following options:
20887 Set number of inputs.
20888 Default is 3. Allowed range is from 3 to 255.
20889 If number of inputs is even number, than result will be mean value between two median values.
20892 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
20895 Set median percentile. Default value is @code{0.5}.
20896 Default value of @code{0.5} will pick always median values, while @code{0} will pick
20897 minimum values, and @code{1} maximum values.
20901 Stack video inputs into custom layout.
20903 All streams must be of same pixel format.
20905 The filter accepts the following options:
20909 Set number of input streams. Default is 2.
20912 Specify layout of inputs.
20913 This option requires the desired layout configuration to be explicitly set by the user.
20914 This sets position of each video input in output. Each input
20915 is separated by '|'.
20916 The first number represents the column, and the second number represents the row.
20917 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
20918 where X is video input from which to take width or height.
20919 Multiple values can be used when separated by '+'. In such
20920 case values are summed together.
20922 Note that if inputs are of different sizes gaps may appear, as not all of
20923 the output video frame will be filled. Similarly, videos can overlap each
20924 other if their position doesn't leave enough space for the full frame of
20927 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
20928 a layout must be set by the user.
20931 If set to 1, force the output to terminate when the shortest input
20932 terminates. Default value is 0.
20935 If set to valid color, all unused pixels will be filled with that color.
20936 By default fill is set to none, so it is disabled.
20939 @subsection Examples
20943 Display 4 inputs into 2x2 grid.
20947 input1(0, 0) | input3(w0, 0)
20948 input2(0, h0) | input4(w0, h0)
20952 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20955 Note that if inputs are of different sizes, gaps or overlaps may occur.
20958 Display 4 inputs into 1x4 grid.
20965 input4(0, h0+h1+h2)
20969 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20972 Note that if inputs are of different widths, unused space will appear.
20975 Display 9 inputs into 3x3 grid.
20979 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20980 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20981 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20985 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
20988 Note that if inputs are of different sizes, gaps or overlaps may occur.
20991 Display 16 inputs into 4x4 grid.
20995 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
20996 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
20997 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
20998 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21002 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|
21003 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
21006 Note that if inputs are of different sizes, gaps or overlaps may occur.
21013 Deinterlace the input video ("yadif" means "yet another deinterlacing
21016 It accepts the following parameters:
21022 The interlacing mode to adopt. It accepts one of the following values:
21025 @item 0, send_frame
21026 Output one frame for each frame.
21027 @item 1, send_field
21028 Output one frame for each field.
21029 @item 2, send_frame_nospatial
21030 Like @code{send_frame}, but it skips the spatial interlacing check.
21031 @item 3, send_field_nospatial
21032 Like @code{send_field}, but it skips the spatial interlacing check.
21035 The default value is @code{send_frame}.
21038 The picture field parity assumed for the input interlaced video. It accepts one
21039 of the following values:
21043 Assume the top field is first.
21045 Assume the bottom field is first.
21047 Enable automatic detection of field parity.
21050 The default value is @code{auto}.
21051 If the interlacing is unknown or the decoder does not export this information,
21052 top field first will be assumed.
21055 Specify which frames to deinterlace. Accepts one of the following
21060 Deinterlace all frames.
21061 @item 1, interlaced
21062 Only deinterlace frames marked as interlaced.
21065 The default value is @code{all}.
21068 @section yadif_cuda
21070 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21071 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21074 It accepts the following parameters:
21080 The interlacing mode to adopt. It accepts one of the following values:
21083 @item 0, send_frame
21084 Output one frame for each frame.
21085 @item 1, send_field
21086 Output one frame for each field.
21087 @item 2, send_frame_nospatial
21088 Like @code{send_frame}, but it skips the spatial interlacing check.
21089 @item 3, send_field_nospatial
21090 Like @code{send_field}, but it skips the spatial interlacing check.
21093 The default value is @code{send_frame}.
21096 The picture field parity assumed for the input interlaced video. It accepts one
21097 of the following values:
21101 Assume the top field is first.
21103 Assume the bottom field is first.
21105 Enable automatic detection of field parity.
21108 The default value is @code{auto}.
21109 If the interlacing is unknown or the decoder does not export this information,
21110 top field first will be assumed.
21113 Specify which frames to deinterlace. Accepts one of the following
21118 Deinterlace all frames.
21119 @item 1, interlaced
21120 Only deinterlace frames marked as interlaced.
21123 The default value is @code{all}.
21128 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21129 The algorithm is described in
21130 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21132 It accepts the following parameters:
21136 Set the window radius. Default value is 3.
21139 Set which planes to filter. Default is only the first plane.
21142 Set blur strength. Default value is 128.
21145 @subsection Commands
21146 This filter supports same @ref{commands} as options.
21150 Apply Zoom & Pan effect.
21152 This filter accepts the following options:
21156 Set the zoom expression. Range is 1-10. Default is 1.
21160 Set the x and y expression. Default is 0.
21163 Set the duration expression in number of frames.
21164 This sets for how many number of frames effect will last for
21165 single input image.
21168 Set the output image size, default is 'hd720'.
21171 Set the output frame rate, default is '25'.
21174 Each expression can contain the following constants:
21193 Output frame count.
21196 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21198 @item out_time, time, ot
21199 The output timestamp expressed in seconds.
21203 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21204 for current input frame.
21208 'x' and 'y' of last output frame of previous input frame or 0 when there was
21209 not yet such frame (first input frame).
21212 Last calculated zoom from 'z' expression for current input frame.
21215 Last calculated zoom of last output frame of previous input frame.
21218 Number of output frames for current input frame. Calculated from 'd' expression
21219 for each input frame.
21222 number of output frames created for previous input frame
21225 Rational number: input width / input height
21228 sample aspect ratio
21231 display aspect ratio
21235 @subsection Examples
21239 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21241 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
21245 Zoom in up to 1.5x and pan always at center of picture:
21247 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21251 Same as above but without pausing:
21253 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21257 Zoom in 2x into center of picture only for the first second of the input video:
21259 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21266 Scale (resize) the input video, using the z.lib library:
21267 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21268 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21270 The zscale filter forces the output display aspect ratio to be the same
21271 as the input, by changing the output sample aspect ratio.
21273 If the input image format is different from the format requested by
21274 the next filter, the zscale filter will convert the input to the
21277 @subsection Options
21278 The filter accepts the following options.
21283 Set the output video dimension expression. Default value is the input
21286 If the @var{width} or @var{w} value is 0, the input width is used for
21287 the output. If the @var{height} or @var{h} value is 0, the input height
21288 is used for the output.
21290 If one and only one of the values is -n with n >= 1, the zscale filter
21291 will use a value that maintains the aspect ratio of the input image,
21292 calculated from the other specified dimension. After that it will,
21293 however, make sure that the calculated dimension is divisible by n and
21294 adjust the value if necessary.
21296 If both values are -n with n >= 1, the behavior will be identical to
21297 both values being set to 0 as previously detailed.
21299 See below for the list of accepted constants for use in the dimension
21303 Set the video size. For the syntax of this option, check the
21304 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21307 Set the dither type.
21309 Possible values are:
21314 @item error_diffusion
21320 Set the resize filter type.
21322 Possible values are:
21332 Default is bilinear.
21335 Set the color range.
21337 Possible values are:
21344 Default is same as input.
21347 Set the color primaries.
21349 Possible values are:
21359 Default is same as input.
21362 Set the transfer characteristics.
21364 Possible values are:
21378 Default is same as input.
21381 Set the colorspace matrix.
21383 Possible value are:
21394 Default is same as input.
21397 Set the input color range.
21399 Possible values are:
21406 Default is same as input.
21408 @item primariesin, pin
21409 Set the input color primaries.
21411 Possible values are:
21421 Default is same as input.
21423 @item transferin, tin
21424 Set the input transfer characteristics.
21426 Possible values are:
21437 Default is same as input.
21439 @item matrixin, min
21440 Set the input colorspace matrix.
21442 Possible value are:
21454 Set the output chroma location.
21456 Possible values are:
21467 @item chromalin, cin
21468 Set the input chroma location.
21470 Possible values are:
21482 Set the nominal peak luminance.
21485 The values of the @option{w} and @option{h} options are expressions
21486 containing the following constants:
21491 The input width and height
21495 These are the same as @var{in_w} and @var{in_h}.
21499 The output (scaled) width and height
21503 These are the same as @var{out_w} and @var{out_h}
21506 The same as @var{iw} / @var{ih}
21509 input sample aspect ratio
21512 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21516 horizontal and vertical input chroma subsample values. For example for the
21517 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21521 horizontal and vertical output chroma subsample values. For example for the
21522 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21525 @subsection Commands
21527 This filter supports the following commands:
21531 Set the output video dimension expression.
21532 The command accepts the same syntax of the corresponding option.
21534 If the specified expression is not valid, it is kept at its current
21538 @c man end VIDEO FILTERS
21540 @chapter OpenCL Video Filters
21541 @c man begin OPENCL VIDEO FILTERS
21543 Below is a description of the currently available OpenCL video filters.
21545 To enable compilation of these filters you need to configure FFmpeg with
21546 @code{--enable-opencl}.
21548 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21551 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21552 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21553 given device parameters.
21555 @item -filter_hw_device @var{name}
21556 Pass the hardware device called @var{name} to all filters in any filter graph.
21560 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21564 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21566 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21570 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.
21572 @section avgblur_opencl
21574 Apply average blur filter.
21576 The filter accepts the following options:
21580 Set horizontal radius size.
21581 Range is @code{[1, 1024]} and default value is @code{1}.
21584 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21587 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21590 @subsection Example
21594 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.
21596 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21600 @section boxblur_opencl
21602 Apply a boxblur algorithm to the input video.
21604 It accepts the following parameters:
21608 @item luma_radius, lr
21609 @item luma_power, lp
21610 @item chroma_radius, cr
21611 @item chroma_power, cp
21612 @item alpha_radius, ar
21613 @item alpha_power, ap
21617 A description of the accepted options follows.
21620 @item luma_radius, lr
21621 @item chroma_radius, cr
21622 @item alpha_radius, ar
21623 Set an expression for the box radius in pixels used for blurring the
21624 corresponding input plane.
21626 The radius value must be a non-negative number, and must not be
21627 greater than the value of the expression @code{min(w,h)/2} for the
21628 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21631 Default value for @option{luma_radius} is "2". If not specified,
21632 @option{chroma_radius} and @option{alpha_radius} default to the
21633 corresponding value set for @option{luma_radius}.
21635 The expressions can contain the following constants:
21639 The input width and height in pixels.
21643 The input chroma image width and height in pixels.
21647 The horizontal and vertical chroma subsample values. For example, for the
21648 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21651 @item luma_power, lp
21652 @item chroma_power, cp
21653 @item alpha_power, ap
21654 Specify how many times the boxblur filter is applied to the
21655 corresponding plane.
21657 Default value for @option{luma_power} is 2. If not specified,
21658 @option{chroma_power} and @option{alpha_power} default to the
21659 corresponding value set for @option{luma_power}.
21661 A value of 0 will disable the effect.
21664 @subsection Examples
21666 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.
21670 Apply a boxblur filter with the luma, chroma, and alpha radius
21671 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.
21673 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21674 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21678 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.
21680 For the luma plane, a 2x2 box radius will be run once.
21682 For the chroma plane, a 4x4 box radius will be run 5 times.
21684 For the alpha plane, a 3x3 box radius will be run 7 times.
21686 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21690 @section colorkey_opencl
21691 RGB colorspace color keying.
21693 The filter accepts the following options:
21697 The color which will be replaced with transparency.
21700 Similarity percentage with the key color.
21702 0.01 matches only the exact key color, while 1.0 matches everything.
21707 0.0 makes pixels either fully transparent, or not transparent at all.
21709 Higher values result in semi-transparent pixels, with a higher transparency
21710 the more similar the pixels color is to the key color.
21713 @subsection Examples
21717 Make every semi-green pixel in the input transparent with some slight blending:
21719 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
21723 @section convolution_opencl
21725 Apply convolution of 3x3, 5x5, 7x7 matrix.
21727 The filter accepts the following options:
21734 Set matrix for each plane.
21735 Matrix is sequence of 9, 25 or 49 signed numbers.
21736 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
21742 Set multiplier for calculated value for each plane.
21743 If unset or 0, it will be sum of all matrix elements.
21744 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
21750 Set bias for each plane. This value is added to the result of the multiplication.
21751 Useful for making the overall image brighter or darker.
21752 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
21756 @subsection Examples
21762 -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
21768 -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
21772 Apply edge enhance:
21774 -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
21780 -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
21784 Apply laplacian edge detector which includes diagonals:
21786 -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
21792 -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
21796 @section erosion_opencl
21798 Apply erosion effect to the video.
21800 This filter replaces the pixel by the local(3x3) minimum.
21802 It accepts the following options:
21809 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21810 If @code{0}, plane will remain unchanged.
21813 Flag which specifies the pixel to refer to.
21814 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21816 Flags to local 3x3 coordinates region centered on @code{x}:
21825 @subsection Example
21829 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.
21831 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21835 @section deshake_opencl
21836 Feature-point based video stabilization filter.
21838 The filter accepts the following options:
21842 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
21845 Whether or not additional debug info should be displayed, both in the processed output and in the console.
21847 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
21849 Viewing point matches in the output video is only supported for RGB input.
21851 Defaults to @code{0}.
21853 @item adaptive_crop
21854 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
21856 Defaults to @code{1}.
21858 @item refine_features
21859 Whether or not feature points should be refined at a sub-pixel level.
21861 This can be turned off for a slight performance gain at the cost of precision.
21863 Defaults to @code{1}.
21865 @item smooth_strength
21866 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
21868 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
21870 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
21872 Defaults to @code{0.0}.
21874 @item smooth_window_multiplier
21875 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
21877 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
21879 Acceptable values range from @code{0.1} to @code{10.0}.
21881 Larger values increase the amount of motion data available for determining how to smooth the camera path,
21882 potentially improving smoothness, but also increase latency and memory usage.
21884 Defaults to @code{2.0}.
21888 @subsection Examples
21892 Stabilize a video with a fixed, medium smoothing strength:
21894 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21898 Stabilize a video with debugging (both in console and in rendered video):
21900 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21904 @section dilation_opencl
21906 Apply dilation effect to the video.
21908 This filter replaces the pixel by the local(3x3) maximum.
21910 It accepts the following options:
21917 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21918 If @code{0}, plane will remain unchanged.
21921 Flag which specifies the pixel to refer to.
21922 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21924 Flags to local 3x3 coordinates region centered on @code{x}:
21933 @subsection Example
21937 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.
21939 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21943 @section nlmeans_opencl
21945 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
21947 @section overlay_opencl
21949 Overlay one video on top of another.
21951 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
21952 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
21954 The filter accepts the following options:
21959 Set the x coordinate of the overlaid video on the main video.
21960 Default value is @code{0}.
21963 Set the y coordinate of the overlaid video on the main video.
21964 Default value is @code{0}.
21968 @subsection Examples
21972 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
21974 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21977 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
21979 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21984 @section pad_opencl
21986 Add paddings to the input image, and place the original input at the
21987 provided @var{x}, @var{y} coordinates.
21989 It accepts the following options:
21994 Specify an expression for the size of the output image with the
21995 paddings added. If the value for @var{width} or @var{height} is 0, the
21996 corresponding input size is used for the output.
21998 The @var{width} expression can reference the value set by the
21999 @var{height} expression, and vice versa.
22001 The default value of @var{width} and @var{height} is 0.
22005 Specify the offsets to place the input image at within the padded area,
22006 with respect to the top/left border of the output image.
22008 The @var{x} expression can reference the value set by the @var{y}
22009 expression, and vice versa.
22011 The default value of @var{x} and @var{y} is 0.
22013 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22014 so the input image is centered on the padded area.
22017 Specify the color of the padded area. For the syntax of this option,
22018 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22019 manual,ffmpeg-utils}.
22022 Pad to an aspect instead to a resolution.
22025 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22026 options are expressions containing the following constants:
22031 The input video width and height.
22035 These are the same as @var{in_w} and @var{in_h}.
22039 The output width and height (the size of the padded area), as
22040 specified by the @var{width} and @var{height} expressions.
22044 These are the same as @var{out_w} and @var{out_h}.
22048 The x and y offsets as specified by the @var{x} and @var{y}
22049 expressions, or NAN if not yet specified.
22052 same as @var{iw} / @var{ih}
22055 input sample aspect ratio
22058 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22061 @section prewitt_opencl
22063 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22065 The filter accepts the following option:
22069 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22072 Set value which will be multiplied with filtered result.
22073 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22076 Set value which will be added to filtered result.
22077 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22080 @subsection Example
22084 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22086 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22090 @anchor{program_opencl}
22091 @section program_opencl
22093 Filter video using an OpenCL program.
22098 OpenCL program source file.
22101 Kernel name in program.
22104 Number of inputs to the filter. Defaults to 1.
22107 Size of output frames. Defaults to the same as the first input.
22111 The @code{program_opencl} filter also supports the @ref{framesync} options.
22113 The program source file must contain a kernel function with the given name,
22114 which will be run once for each plane of the output. Each run on a plane
22115 gets enqueued as a separate 2D global NDRange with one work-item for each
22116 pixel to be generated. The global ID offset for each work-item is therefore
22117 the coordinates of a pixel in the destination image.
22119 The kernel function needs to take the following arguments:
22122 Destination image, @var{__write_only image2d_t}.
22124 This image will become the output; the kernel should write all of it.
22126 Frame index, @var{unsigned int}.
22128 This is a counter starting from zero and increasing by one for each frame.
22130 Source images, @var{__read_only image2d_t}.
22132 These are the most recent images on each input. The kernel may read from
22133 them to generate the output, but they can't be written to.
22140 Copy the input to the output (output must be the same size as the input).
22142 __kernel void copy(__write_only image2d_t destination,
22143 unsigned int index,
22144 __read_only image2d_t source)
22146 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22148 int2 location = (int2)(get_global_id(0), get_global_id(1));
22150 float4 value = read_imagef(source, sampler, location);
22152 write_imagef(destination, location, value);
22157 Apply a simple transformation, rotating the input by an amount increasing
22158 with the index counter. Pixel values are linearly interpolated by the
22159 sampler, and the output need not have the same dimensions as the input.
22161 __kernel void rotate_image(__write_only image2d_t dst,
22162 unsigned int index,
22163 __read_only image2d_t src)
22165 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22166 CLK_FILTER_LINEAR);
22168 float angle = (float)index / 100.0f;
22170 float2 dst_dim = convert_float2(get_image_dim(dst));
22171 float2 src_dim = convert_float2(get_image_dim(src));
22173 float2 dst_cen = dst_dim / 2.0f;
22174 float2 src_cen = src_dim / 2.0f;
22176 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22178 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22180 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22181 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22183 src_pos = src_pos * src_dim / dst_dim;
22185 float2 src_loc = src_pos + src_cen;
22187 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22188 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22189 write_imagef(dst, dst_loc, 0.5f);
22191 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22196 Blend two inputs together, with the amount of each input used varying
22197 with the index counter.
22199 __kernel void blend_images(__write_only image2d_t dst,
22200 unsigned int index,
22201 __read_only image2d_t src1,
22202 __read_only image2d_t src2)
22204 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22205 CLK_FILTER_LINEAR);
22207 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22209 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22210 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22211 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22213 float4 val1 = read_imagef(src1, sampler, src1_loc);
22214 float4 val2 = read_imagef(src2, sampler, src2_loc);
22216 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22222 @section roberts_opencl
22223 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22225 The filter accepts the following option:
22229 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22232 Set value which will be multiplied with filtered result.
22233 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22236 Set value which will be added to filtered result.
22237 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22240 @subsection Example
22244 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22246 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22250 @section sobel_opencl
22252 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22254 The filter accepts the following option:
22258 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22261 Set value which will be multiplied with filtered result.
22262 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22265 Set value which will be added to filtered result.
22266 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22269 @subsection Example
22273 Apply sobel operator with scale set to 2 and delta set to 10
22275 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22279 @section tonemap_opencl
22281 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22283 It accepts the following parameters:
22287 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22290 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22293 Apply desaturation for highlights that exceed this level of brightness. The
22294 higher the parameter, the more color information will be preserved. This
22295 setting helps prevent unnaturally blown-out colors for super-highlights, by
22296 (smoothly) turning into white instead. This makes images feel more natural,
22297 at the cost of reducing information about out-of-range colors.
22299 The default value is 0.5, and the algorithm here is a little different from
22300 the cpu version tonemap currently. A setting of 0.0 disables this option.
22303 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22304 is used to detect whether the scene has changed or not. If the distance between
22305 the current frame average brightness and the current running average exceeds
22306 a threshold value, we would re-calculate scene average and peak brightness.
22307 The default value is 0.2.
22310 Specify the output pixel format.
22312 Currently supported formats are:
22319 Set the output color range.
22321 Possible values are:
22327 Default is same as input.
22330 Set the output color primaries.
22332 Possible values are:
22338 Default is same as input.
22341 Set the output transfer characteristics.
22343 Possible values are:
22352 Set the output colorspace matrix.
22354 Possible value are:
22360 Default is same as input.
22364 @subsection Example
22368 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22370 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22374 @section unsharp_opencl
22376 Sharpen or blur the input video.
22378 It accepts the following parameters:
22381 @item luma_msize_x, lx
22382 Set the luma matrix horizontal size.
22383 Range is @code{[1, 23]} and default value is @code{5}.
22385 @item luma_msize_y, ly
22386 Set the luma matrix vertical size.
22387 Range is @code{[1, 23]} and default value is @code{5}.
22389 @item luma_amount, la
22390 Set the luma effect strength.
22391 Range is @code{[-10, 10]} and default value is @code{1.0}.
22393 Negative values will blur the input video, while positive values will
22394 sharpen it, a value of zero will disable the effect.
22396 @item chroma_msize_x, cx
22397 Set the chroma matrix horizontal size.
22398 Range is @code{[1, 23]} and default value is @code{5}.
22400 @item chroma_msize_y, cy
22401 Set the chroma matrix vertical size.
22402 Range is @code{[1, 23]} and default value is @code{5}.
22404 @item chroma_amount, ca
22405 Set the chroma effect strength.
22406 Range is @code{[-10, 10]} and default value is @code{0.0}.
22408 Negative values will blur the input video, while positive values will
22409 sharpen it, a value of zero will disable the effect.
22413 All parameters are optional and default to the equivalent of the
22414 string '5:5:1.0:5:5:0.0'.
22416 @subsection Examples
22420 Apply strong luma sharpen effect:
22422 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22426 Apply a strong blur of both luma and chroma parameters:
22428 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22432 @section xfade_opencl
22434 Cross fade two videos with custom transition effect by using OpenCL.
22436 It accepts the following options:
22440 Set one of possible transition effects.
22444 Select custom transition effect, the actual transition description
22445 will be picked from source and kernel options.
22457 Default transition is fade.
22461 OpenCL program source file for custom transition.
22464 Set name of kernel to use for custom transition from program source file.
22467 Set duration of video transition.
22470 Set time of start of transition relative to first video.
22473 The program source file must contain a kernel function with the given name,
22474 which will be run once for each plane of the output. Each run on a plane
22475 gets enqueued as a separate 2D global NDRange with one work-item for each
22476 pixel to be generated. The global ID offset for each work-item is therefore
22477 the coordinates of a pixel in the destination image.
22479 The kernel function needs to take the following arguments:
22482 Destination image, @var{__write_only image2d_t}.
22484 This image will become the output; the kernel should write all of it.
22487 First Source image, @var{__read_only image2d_t}.
22488 Second Source image, @var{__read_only image2d_t}.
22490 These are the most recent images on each input. The kernel may read from
22491 them to generate the output, but they can't be written to.
22494 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22501 Apply dots curtain transition effect:
22503 __kernel void blend_images(__write_only image2d_t dst,
22504 __read_only image2d_t src1,
22505 __read_only image2d_t src2,
22508 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22509 CLK_FILTER_LINEAR);
22510 int2 p = (int2)(get_global_id(0), get_global_id(1));
22511 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22512 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22515 float2 dots = (float2)(20.0, 20.0);
22516 float2 center = (float2)(0,0);
22519 float4 val1 = read_imagef(src1, sampler, p);
22520 float4 val2 = read_imagef(src2, sampler, p);
22521 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22523 write_imagef(dst, p, next ? val1 : val2);
22529 @c man end OPENCL VIDEO FILTERS
22531 @chapter VAAPI Video Filters
22532 @c man begin VAAPI VIDEO FILTERS
22534 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22536 To enable compilation of these filters you need to configure FFmpeg with
22537 @code{--enable-vaapi}.
22539 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}
22541 @section tonemap_vaapi
22543 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22544 It maps the dynamic range of HDR10 content to the SDR content.
22545 It currently only accepts HDR10 as input.
22547 It accepts the following parameters:
22551 Specify the output pixel format.
22553 Currently supported formats are:
22562 Set the output color primaries.
22564 Default is same as input.
22567 Set the output transfer characteristics.
22572 Set the output colorspace matrix.
22574 Default is same as input.
22578 @subsection Example
22582 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22584 tonemap_vaapi=format=p010:t=bt2020-10
22588 @c man end VAAPI VIDEO FILTERS
22590 @chapter Video Sources
22591 @c man begin VIDEO SOURCES
22593 Below is a description of the currently available video sources.
22597 Buffer video frames, and make them available to the filter chain.
22599 This source is mainly intended for a programmatic use, in particular
22600 through the interface defined in @file{libavfilter/buffersrc.h}.
22602 It accepts the following parameters:
22607 Specify the size (width and height) of the buffered video frames. For the
22608 syntax of this option, check the
22609 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22612 The input video width.
22615 The input video height.
22618 A string representing the pixel format of the buffered video frames.
22619 It may be a number corresponding to a pixel format, or a pixel format
22623 Specify the timebase assumed by the timestamps of the buffered frames.
22626 Specify the frame rate expected for the video stream.
22628 @item pixel_aspect, sar
22629 The sample (pixel) aspect ratio of the input video.
22632 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22633 to the filtergraph description to specify swscale flags for automatically
22634 inserted scalers. See @ref{Filtergraph syntax}.
22636 @item hw_frames_ctx
22637 When using a hardware pixel format, this should be a reference to an
22638 AVHWFramesContext describing input frames.
22643 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22646 will instruct the source to accept video frames with size 320x240 and
22647 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22648 square pixels (1:1 sample aspect ratio).
22649 Since the pixel format with name "yuv410p" corresponds to the number 6
22650 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22651 this example corresponds to:
22653 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22656 Alternatively, the options can be specified as a flat string, but this
22657 syntax is deprecated:
22659 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22663 Create a pattern generated by an elementary cellular automaton.
22665 The initial state of the cellular automaton can be defined through the
22666 @option{filename} and @option{pattern} options. If such options are
22667 not specified an initial state is created randomly.
22669 At each new frame a new row in the video is filled with the result of
22670 the cellular automaton next generation. The behavior when the whole
22671 frame is filled is defined by the @option{scroll} option.
22673 This source accepts the following options:
22677 Read the initial cellular automaton state, i.e. the starting row, from
22678 the specified file.
22679 In the file, each non-whitespace character is considered an alive
22680 cell, a newline will terminate the row, and further characters in the
22681 file will be ignored.
22684 Read the initial cellular automaton state, i.e. the starting row, from
22685 the specified string.
22687 Each non-whitespace character in the string is considered an alive
22688 cell, a newline will terminate the row, and further characters in the
22689 string will be ignored.
22692 Set the video rate, that is the number of frames generated per second.
22695 @item random_fill_ratio, ratio
22696 Set the random fill ratio for the initial cellular automaton row. It
22697 is a floating point number value ranging from 0 to 1, defaults to
22700 This option is ignored when a file or a pattern is specified.
22702 @item random_seed, seed
22703 Set the seed for filling randomly the initial row, must be an integer
22704 included between 0 and UINT32_MAX. If not specified, or if explicitly
22705 set to -1, the filter will try to use a good random seed on a best
22709 Set the cellular automaton rule, it is a number ranging from 0 to 255.
22710 Default value is 110.
22713 Set the size of the output video. For the syntax of this option, check the
22714 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22716 If @option{filename} or @option{pattern} is specified, the size is set
22717 by default to the width of the specified initial state row, and the
22718 height is set to @var{width} * PHI.
22720 If @option{size} is set, it must contain the width of the specified
22721 pattern string, and the specified pattern will be centered in the
22724 If a filename or a pattern string is not specified, the size value
22725 defaults to "320x518" (used for a randomly generated initial state).
22728 If set to 1, scroll the output upward when all the rows in the output
22729 have been already filled. If set to 0, the new generated row will be
22730 written over the top row just after the bottom row is filled.
22733 @item start_full, full
22734 If set to 1, completely fill the output with generated rows before
22735 outputting the first frame.
22736 This is the default behavior, for disabling set the value to 0.
22739 If set to 1, stitch the left and right row edges together.
22740 This is the default behavior, for disabling set the value to 0.
22743 @subsection Examples
22747 Read the initial state from @file{pattern}, and specify an output of
22750 cellauto=f=pattern:s=200x400
22754 Generate a random initial row with a width of 200 cells, with a fill
22757 cellauto=ratio=2/3:s=200x200
22761 Create a pattern generated by rule 18 starting by a single alive cell
22762 centered on an initial row with width 100:
22764 cellauto=p=@@:s=100x400:full=0:rule=18
22768 Specify a more elaborated initial pattern:
22770 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
22775 @anchor{coreimagesrc}
22776 @section coreimagesrc
22777 Video source generated on GPU using Apple's CoreImage API on OSX.
22779 This video source is a specialized version of the @ref{coreimage} video filter.
22780 Use a core image generator at the beginning of the applied filterchain to
22781 generate the content.
22783 The coreimagesrc video source accepts the following options:
22785 @item list_generators
22786 List all available generators along with all their respective options as well as
22787 possible minimum and maximum values along with the default values.
22789 list_generators=true
22793 Specify the size of the sourced video. For the syntax of this option, check the
22794 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22795 The default value is @code{320x240}.
22798 Specify the frame rate of the sourced video, as the number of frames
22799 generated per second. It has to be a string in the format
22800 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22801 number or a valid video frame rate abbreviation. The default value is
22805 Set the sample aspect ratio of the sourced video.
22808 Set the duration of the sourced video. See
22809 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22810 for the accepted syntax.
22812 If not specified, or the expressed duration is negative, the video is
22813 supposed to be generated forever.
22816 Additionally, all options of the @ref{coreimage} video filter are accepted.
22817 A complete filterchain can be used for further processing of the
22818 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
22819 and examples for details.
22821 @subsection Examples
22826 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22827 given as complete and escaped command-line for Apple's standard bash shell:
22829 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
22831 This example is equivalent to the QRCode example of @ref{coreimage} without the
22832 need for a nullsrc video source.
22837 Generate several gradients.
22841 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22842 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22845 Set frame rate, expressed as number of frames per second. Default
22848 @item c0, c1, c2, c3, c4, c5, c6, c7
22849 Set 8 colors. Default values for colors is to pick random one.
22851 @item x0, y0, y0, y1
22852 Set gradient line source and destination points. If negative or out of range, random ones
22856 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
22859 Set seed for picking gradient line points.
22862 Set the duration of the sourced video. See
22863 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22864 for the accepted syntax.
22866 If not specified, or the expressed duration is negative, the video is
22867 supposed to be generated forever.
22870 Set speed of gradients rotation.
22874 @section mandelbrot
22876 Generate a Mandelbrot set fractal, and progressively zoom towards the
22877 point specified with @var{start_x} and @var{start_y}.
22879 This source accepts the following options:
22884 Set the terminal pts value. Default value is 400.
22887 Set the terminal scale value.
22888 Must be a floating point value. Default value is 0.3.
22891 Set the inner coloring mode, that is the algorithm used to draw the
22892 Mandelbrot fractal internal region.
22894 It shall assume one of the following values:
22899 Show time until convergence.
22901 Set color based on point closest to the origin of the iterations.
22906 Default value is @var{mincol}.
22909 Set the bailout value. Default value is 10.0.
22912 Set the maximum of iterations performed by the rendering
22913 algorithm. Default value is 7189.
22916 Set outer coloring mode.
22917 It shall assume one of following values:
22919 @item iteration_count
22920 Set iteration count mode.
22921 @item normalized_iteration_count
22922 set normalized iteration count mode.
22924 Default value is @var{normalized_iteration_count}.
22927 Set frame rate, expressed as number of frames per second. Default
22931 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22932 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22935 Set the initial scale value. Default value is 3.0.
22938 Set the initial x position. Must be a floating point value between
22939 -100 and 100. Default value is -0.743643887037158704752191506114774.
22942 Set the initial y position. Must be a floating point value between
22943 -100 and 100. Default value is -0.131825904205311970493132056385139.
22948 Generate various test patterns, as generated by the MPlayer test filter.
22950 The size of the generated video is fixed, and is 256x256.
22951 This source is useful in particular for testing encoding features.
22953 This source accepts the following options:
22958 Specify the frame rate of the sourced video, as the number of frames
22959 generated per second. It has to be a string in the format
22960 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22961 number or a valid video frame rate abbreviation. The default value is
22965 Set the duration of the sourced video. See
22966 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22967 for the accepted syntax.
22969 If not specified, or the expressed duration is negative, the video is
22970 supposed to be generated forever.
22974 Set the number or the name of the test to perform. Supported tests are:
22988 @item max_frames, m
22989 Set the maximum number of frames generated for each test, default value is 30.
22993 Default value is "all", which will cycle through the list of all tests.
22998 mptestsrc=t=dc_luma
23001 will generate a "dc_luma" test pattern.
23003 @section frei0r_src
23005 Provide a frei0r source.
23007 To enable compilation of this filter you need to install the frei0r
23008 header and configure FFmpeg with @code{--enable-frei0r}.
23010 This source accepts the following parameters:
23015 The size of the video to generate. For the syntax of this option, check the
23016 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23019 The framerate of the generated video. It may be a string of the form
23020 @var{num}/@var{den} or a frame rate abbreviation.
23023 The name to the frei0r source to load. For more information regarding frei0r and
23024 how to set the parameters, read the @ref{frei0r} section in the video filters
23027 @item filter_params
23028 A '|'-separated list of parameters to pass to the frei0r source.
23032 For example, to generate a frei0r partik0l source with size 200x200
23033 and frame rate 10 which is overlaid on the overlay filter main input:
23035 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23040 Generate a life pattern.
23042 This source is based on a generalization of John Conway's life game.
23044 The sourced input represents a life grid, each pixel represents a cell
23045 which can be in one of two possible states, alive or dead. Every cell
23046 interacts with its eight neighbours, which are the cells that are
23047 horizontally, vertically, or diagonally adjacent.
23049 At each interaction the grid evolves according to the adopted rule,
23050 which specifies the number of neighbor alive cells which will make a
23051 cell stay alive or born. The @option{rule} option allows one to specify
23054 This source accepts the following options:
23058 Set the file from which to read the initial grid state. In the file,
23059 each non-whitespace character is considered an alive cell, and newline
23060 is used to delimit the end of each row.
23062 If this option is not specified, the initial grid is generated
23066 Set the video rate, that is the number of frames generated per second.
23069 @item random_fill_ratio, ratio
23070 Set the random fill ratio for the initial random grid. It is a
23071 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23072 It is ignored when a file is specified.
23074 @item random_seed, seed
23075 Set the seed for filling the initial random grid, must be an integer
23076 included between 0 and UINT32_MAX. If not specified, or if explicitly
23077 set to -1, the filter will try to use a good random seed on a best
23083 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23084 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23085 @var{NS} specifies the number of alive neighbor cells which make a
23086 live cell stay alive, and @var{NB} the number of alive neighbor cells
23087 which make a dead cell to become alive (i.e. to "born").
23088 "s" and "b" can be used in place of "S" and "B", respectively.
23090 Alternatively a rule can be specified by an 18-bits integer. The 9
23091 high order bits are used to encode the next cell state if it is alive
23092 for each number of neighbor alive cells, the low order bits specify
23093 the rule for "borning" new cells. Higher order bits encode for an
23094 higher number of neighbor cells.
23095 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23096 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23098 Default value is "S23/B3", which is the original Conway's game of life
23099 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23100 cells, and will born a new cell if there are three alive cells around
23104 Set the size of the output video. For the syntax of this option, check the
23105 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23107 If @option{filename} is specified, the size is set by default to the
23108 same size of the input file. If @option{size} is set, it must contain
23109 the size specified in the input file, and the initial grid defined in
23110 that file is centered in the larger resulting area.
23112 If a filename is not specified, the size value defaults to "320x240"
23113 (used for a randomly generated initial grid).
23116 If set to 1, stitch the left and right grid edges together, and the
23117 top and bottom edges also. Defaults to 1.
23120 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23121 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23122 value from 0 to 255.
23125 Set the color of living (or new born) cells.
23128 Set the color of dead cells. If @option{mold} is set, this is the first color
23129 used to represent a dead cell.
23132 Set mold color, for definitely dead and moldy cells.
23134 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23135 ffmpeg-utils manual,ffmpeg-utils}.
23138 @subsection Examples
23142 Read a grid from @file{pattern}, and center it on a grid of size
23145 life=f=pattern:s=300x300
23149 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23151 life=ratio=2/3:s=200x200
23155 Specify a custom rule for evolving a randomly generated grid:
23161 Full example with slow death effect (mold) using @command{ffplay}:
23163 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23170 @anchor{haldclutsrc}
23173 @anchor{pal100bars}
23174 @anchor{rgbtestsrc}
23176 @anchor{smptehdbars}
23179 @anchor{yuvtestsrc}
23180 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23182 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23184 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23186 The @code{color} source provides an uniformly colored input.
23188 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23189 @ref{haldclut} filter.
23191 The @code{nullsrc} source returns unprocessed video frames. It is
23192 mainly useful to be employed in analysis / debugging tools, or as the
23193 source for filters which ignore the input data.
23195 The @code{pal75bars} source generates a color bars pattern, based on
23196 EBU PAL recommendations with 75% color levels.
23198 The @code{pal100bars} source generates a color bars pattern, based on
23199 EBU PAL recommendations with 100% color levels.
23201 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23202 detecting RGB vs BGR issues. You should see a red, green and blue
23203 stripe from top to bottom.
23205 The @code{smptebars} source generates a color bars pattern, based on
23206 the SMPTE Engineering Guideline EG 1-1990.
23208 The @code{smptehdbars} source generates a color bars pattern, based on
23209 the SMPTE RP 219-2002.
23211 The @code{testsrc} source generates a test video pattern, showing a
23212 color pattern, a scrolling gradient and a timestamp. This is mainly
23213 intended for testing purposes.
23215 The @code{testsrc2} source is similar to testsrc, but supports more
23216 pixel formats instead of just @code{rgb24}. This allows using it as an
23217 input for other tests without requiring a format conversion.
23219 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23220 see a y, cb and cr stripe from top to bottom.
23222 The sources accept the following parameters:
23227 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23228 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23229 pixels to be used as identity matrix for 3D lookup tables. Each component is
23230 coded on a @code{1/(N*N)} scale.
23233 Specify the color of the source, only available in the @code{color}
23234 source. For the syntax of this option, check the
23235 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23238 Specify the size of the sourced video. For the syntax of this option, check the
23239 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23240 The default value is @code{320x240}.
23242 This option is not available with the @code{allrgb}, @code{allyuv}, and
23243 @code{haldclutsrc} filters.
23246 Specify the frame rate of the sourced video, as the number of frames
23247 generated per second. It has to be a string in the format
23248 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23249 number or a valid video frame rate abbreviation. The default value is
23253 Set the duration of the sourced video. See
23254 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23255 for the accepted syntax.
23257 If not specified, or the expressed duration is negative, the video is
23258 supposed to be generated forever.
23260 Since the frame rate is used as time base, all frames including the last one
23261 will have their full duration. If the specified duration is not a multiple
23262 of the frame duration, it will be rounded up.
23265 Set the sample aspect ratio of the sourced video.
23268 Specify the alpha (opacity) of the background, only available in the
23269 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23270 255 (fully opaque, the default).
23273 Set the number of decimals to show in the timestamp, only available in the
23274 @code{testsrc} source.
23276 The displayed timestamp value will correspond to the original
23277 timestamp value multiplied by the power of 10 of the specified
23278 value. Default value is 0.
23281 @subsection Examples
23285 Generate a video with a duration of 5.3 seconds, with size
23286 176x144 and a frame rate of 10 frames per second:
23288 testsrc=duration=5.3:size=qcif:rate=10
23292 The following graph description will generate a red source
23293 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23296 color=c=red@@0.2:s=qcif:r=10
23300 If the input content is to be ignored, @code{nullsrc} can be used. The
23301 following command generates noise in the luminance plane by employing
23302 the @code{geq} filter:
23304 nullsrc=s=256x256, geq=random(1)*255:128:128
23308 @subsection Commands
23310 The @code{color} source supports the following commands:
23314 Set the color of the created image. Accepts the same syntax of the
23315 corresponding @option{color} option.
23320 Generate video using an OpenCL program.
23325 OpenCL program source file.
23328 Kernel name in program.
23331 Size of frames to generate. This must be set.
23334 Pixel format to use for the generated frames. This must be set.
23337 Number of frames generated every second. Default value is '25'.
23341 For details of how the program loading works, see the @ref{program_opencl}
23348 Generate a colour ramp by setting pixel values from the position of the pixel
23349 in the output image. (Note that this will work with all pixel formats, but
23350 the generated output will not be the same.)
23352 __kernel void ramp(__write_only image2d_t dst,
23353 unsigned int index)
23355 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23358 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23360 write_imagef(dst, loc, val);
23365 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23367 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23368 unsigned int index)
23370 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23372 float4 value = 0.0f;
23373 int x = loc.x + index;
23374 int y = loc.y + index;
23375 while (x > 0 || y > 0) {
23376 if (x % 3 == 1 && y % 3 == 1) {
23384 write_imagef(dst, loc, value);
23390 @section sierpinski
23392 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23394 This source accepts the following options:
23398 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23399 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23402 Set frame rate, expressed as number of frames per second. Default
23406 Set seed which is used for random panning.
23409 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23412 Set fractal type, can be default @code{carpet} or @code{triangle}.
23415 @c man end VIDEO SOURCES
23417 @chapter Video Sinks
23418 @c man begin VIDEO SINKS
23420 Below is a description of the currently available video sinks.
23422 @section buffersink
23424 Buffer video frames, and make them available to the end of the filter
23427 This sink is mainly intended for programmatic use, in particular
23428 through the interface defined in @file{libavfilter/buffersink.h}
23429 or the options system.
23431 It accepts a pointer to an AVBufferSinkContext structure, which
23432 defines the incoming buffers' formats, to be passed as the opaque
23433 parameter to @code{avfilter_init_filter} for initialization.
23437 Null video sink: do absolutely nothing with the input video. It is
23438 mainly useful as a template and for use in analysis / debugging
23441 @c man end VIDEO SINKS
23443 @chapter Multimedia Filters
23444 @c man begin MULTIMEDIA FILTERS
23446 Below is a description of the currently available multimedia filters.
23450 Convert input audio to a video output, displaying the audio bit scope.
23452 The filter accepts the following options:
23456 Set frame rate, expressed as number of frames per second. Default
23460 Specify the video size for the output. For the syntax of this option, check the
23461 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23462 Default value is @code{1024x256}.
23465 Specify list of colors separated by space or by '|' which will be used to
23466 draw channels. Unrecognized or missing colors will be replaced
23470 @section adrawgraph
23471 Draw a graph using input audio metadata.
23473 See @ref{drawgraph}
23475 @section agraphmonitor
23477 See @ref{graphmonitor}.
23479 @section ahistogram
23481 Convert input audio to a video output, displaying the volume histogram.
23483 The filter accepts the following options:
23487 Specify how histogram is calculated.
23489 It accepts the following values:
23492 Use single histogram for all channels.
23494 Use separate histogram for each channel.
23496 Default is @code{single}.
23499 Set frame rate, expressed as number of frames per second. Default
23503 Specify the video size for the output. For the syntax of this option, check the
23504 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23505 Default value is @code{hd720}.
23510 It accepts the following values:
23521 reverse logarithmic
23523 Default is @code{log}.
23526 Set amplitude scale.
23528 It accepts the following values:
23535 Default is @code{log}.
23538 Set how much frames to accumulate in histogram.
23539 Default is 1. Setting this to -1 accumulates all frames.
23542 Set histogram ratio of window height.
23545 Set sonogram sliding.
23547 It accepts the following values:
23550 replace old rows with new ones.
23552 scroll from top to bottom.
23554 Default is @code{replace}.
23557 @section aphasemeter
23559 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23560 representing mean phase of current audio frame. A video output can also be produced and is
23561 enabled by default. The audio is passed through as first output.
23563 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23564 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23565 and @code{1} means channels are in phase.
23567 The filter accepts the following options, all related to its video output:
23571 Set the output frame rate. Default value is @code{25}.
23574 Set the video size for the output. For the syntax of this option, check the
23575 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23576 Default value is @code{800x400}.
23581 Specify the red, green, blue contrast. Default values are @code{2},
23582 @code{7} and @code{1}.
23583 Allowed range is @code{[0, 255]}.
23586 Set color which will be used for drawing median phase. If color is
23587 @code{none} which is default, no median phase value will be drawn.
23590 Enable video output. Default is enabled.
23593 @subsection phasing detection
23595 The filter also detects out of phase and mono sequences in stereo streams.
23596 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23598 The filter accepts the following options for this detection:
23602 Enable mono and out of phase detection. Default is disabled.
23605 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23606 Allowed range is @code{[0, 1]}.
23609 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23610 Allowed range is @code{[90, 180]}.
23613 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23616 @subsection Examples
23620 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23622 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23626 @section avectorscope
23628 Convert input audio to a video output, representing the audio vector
23631 The filter is used to measure the difference between channels of stereo
23632 audio stream. A monaural signal, consisting of identical left and right
23633 signal, results in straight vertical line. Any stereo separation is visible
23634 as a deviation from this line, creating a Lissajous figure.
23635 If the straight (or deviation from it) but horizontal line appears this
23636 indicates that the left and right channels are out of phase.
23638 The filter accepts the following options:
23642 Set the vectorscope mode.
23644 Available values are:
23647 Lissajous rotated by 45 degrees.
23650 Same as above but not rotated.
23653 Shape resembling half of circle.
23656 Default value is @samp{lissajous}.
23659 Set the video size for the output. For the syntax of this option, check the
23660 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23661 Default value is @code{400x400}.
23664 Set the output frame rate. Default value is @code{25}.
23670 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23671 @code{160}, @code{80} and @code{255}.
23672 Allowed range is @code{[0, 255]}.
23678 Specify the red, green, blue and alpha fade. Default values are @code{15},
23679 @code{10}, @code{5} and @code{5}.
23680 Allowed range is @code{[0, 255]}.
23683 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23684 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23687 Set the vectorscope drawing mode.
23689 Available values are:
23692 Draw dot for each sample.
23695 Draw line between previous and current sample.
23698 Default value is @samp{dot}.
23701 Specify amplitude scale of audio samples.
23703 Available values are:
23719 Swap left channel axis with right channel axis.
23729 Mirror only x axis.
23732 Mirror only y axis.
23740 @subsection Examples
23744 Complete example using @command{ffplay}:
23746 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23747 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
23751 @section bench, abench
23753 Benchmark part of a filtergraph.
23755 The filter accepts the following options:
23759 Start or stop a timer.
23761 Available values are:
23764 Get the current time, set it as frame metadata (using the key
23765 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
23768 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
23769 the input frame metadata to get the time difference. Time difference, average,
23770 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
23771 @code{min}) are then printed. The timestamps are expressed in seconds.
23775 @subsection Examples
23779 Benchmark @ref{selectivecolor} filter:
23781 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
23787 Concatenate audio and video streams, joining them together one after the
23790 The filter works on segments of synchronized video and audio streams. All
23791 segments must have the same number of streams of each type, and that will
23792 also be the number of streams at output.
23794 The filter accepts the following options:
23799 Set the number of segments. Default is 2.
23802 Set the number of output video streams, that is also the number of video
23803 streams in each segment. Default is 1.
23806 Set the number of output audio streams, that is also the number of audio
23807 streams in each segment. Default is 0.
23810 Activate unsafe mode: do not fail if segments have a different format.
23814 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
23815 @var{a} audio outputs.
23817 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
23818 segment, in the same order as the outputs, then the inputs for the second
23821 Related streams do not always have exactly the same duration, for various
23822 reasons including codec frame size or sloppy authoring. For that reason,
23823 related synchronized streams (e.g. a video and its audio track) should be
23824 concatenated at once. The concat filter will use the duration of the longest
23825 stream in each segment (except the last one), and if necessary pad shorter
23826 audio streams with silence.
23828 For this filter to work correctly, all segments must start at timestamp 0.
23830 All corresponding streams must have the same parameters in all segments; the
23831 filtering system will automatically select a common pixel format for video
23832 streams, and a common sample format, sample rate and channel layout for
23833 audio streams, but other settings, such as resolution, must be converted
23834 explicitly by the user.
23836 Different frame rates are acceptable but will result in variable frame rate
23837 at output; be sure to configure the output file to handle it.
23839 @subsection Examples
23843 Concatenate an opening, an episode and an ending, all in bilingual version
23844 (video in stream 0, audio in streams 1 and 2):
23846 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23847 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23848 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23849 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23853 Concatenate two parts, handling audio and video separately, using the
23854 (a)movie sources, and adjusting the resolution:
23856 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23857 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23858 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23860 Note that a desync will happen at the stitch if the audio and video streams
23861 do not have exactly the same duration in the first file.
23865 @subsection Commands
23867 This filter supports the following commands:
23870 Close the current segment and step to the next one
23876 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
23877 level. By default, it logs a message at a frequency of 10Hz with the
23878 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
23879 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
23881 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
23882 sample format is double-precision floating point. The input stream will be converted to
23883 this specification, if needed. Users may need to insert aformat and/or aresample filters
23884 after this filter to obtain the original parameters.
23886 The filter also has a video output (see the @var{video} option) with a real
23887 time graph to observe the loudness evolution. The graphic contains the logged
23888 message mentioned above, so it is not printed anymore when this option is set,
23889 unless the verbose logging is set. The main graphing area contains the
23890 short-term loudness (3 seconds of analysis), and the gauge on the right is for
23891 the momentary loudness (400 milliseconds), but can optionally be configured
23892 to instead display short-term loudness (see @var{gauge}).
23894 The green area marks a +/- 1LU target range around the target loudness
23895 (-23LUFS by default, unless modified through @var{target}).
23897 More information about the Loudness Recommendation EBU R128 on
23898 @url{http://tech.ebu.ch/loudness}.
23900 The filter accepts the following options:
23905 Activate the video output. The audio stream is passed unchanged whether this
23906 option is set or no. The video stream will be the first output stream if
23907 activated. Default is @code{0}.
23910 Set the video size. This option is for video only. For the syntax of this
23912 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23913 Default and minimum resolution is @code{640x480}.
23916 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
23917 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
23918 other integer value between this range is allowed.
23921 Set metadata injection. If set to @code{1}, the audio input will be segmented
23922 into 100ms output frames, each of them containing various loudness information
23923 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
23925 Default is @code{0}.
23928 Force the frame logging level.
23930 Available values are:
23933 information logging level
23935 verbose logging level
23938 By default, the logging level is set to @var{info}. If the @option{video} or
23939 the @option{metadata} options are set, it switches to @var{verbose}.
23944 Available modes can be cumulated (the option is a @code{flag} type). Possible
23948 Disable any peak mode (default).
23950 Enable sample-peak mode.
23952 Simple peak mode looking for the higher sample value. It logs a message
23953 for sample-peak (identified by @code{SPK}).
23955 Enable true-peak mode.
23957 If enabled, the peak lookup is done on an over-sampled version of the input
23958 stream for better peak accuracy. It logs a message for true-peak.
23959 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
23960 This mode requires a build with @code{libswresample}.
23964 Treat mono input files as "dual mono". If a mono file is intended for playback
23965 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
23966 If set to @code{true}, this option will compensate for this effect.
23967 Multi-channel input files are not affected by this option.
23970 Set a specific pan law to be used for the measurement of dual mono files.
23971 This parameter is optional, and has a default value of -3.01dB.
23974 Set a specific target level (in LUFS) used as relative zero in the visualization.
23975 This parameter is optional and has a default value of -23LUFS as specified
23976 by EBU R128. However, material published online may prefer a level of -16LUFS
23977 (e.g. for use with podcasts or video platforms).
23980 Set the value displayed by the gauge. Valid values are @code{momentary} and s
23981 @code{shortterm}. By default the momentary value will be used, but in certain
23982 scenarios it may be more useful to observe the short term value instead (e.g.
23986 Sets the display scale for the loudness. Valid parameters are @code{absolute}
23987 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
23988 video output, not the summary or continuous log output.
23991 @subsection Examples
23995 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
23997 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24001 Run an analysis with @command{ffmpeg}:
24003 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24007 @section interleave, ainterleave
24009 Temporally interleave frames from several inputs.
24011 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24013 These filters read frames from several inputs and send the oldest
24014 queued frame to the output.
24016 Input streams must have well defined, monotonically increasing frame
24019 In order to submit one frame to output, these filters need to enqueue
24020 at least one frame for each input, so they cannot work in case one
24021 input is not yet terminated and will not receive incoming frames.
24023 For example consider the case when one input is a @code{select} filter
24024 which always drops input frames. The @code{interleave} filter will keep
24025 reading from that input, but it will never be able to send new frames
24026 to output until the input sends an end-of-stream signal.
24028 Also, depending on inputs synchronization, the filters will drop
24029 frames in case one input receives more frames than the other ones, and
24030 the queue is already filled.
24032 These filters accept the following options:
24036 Set the number of different inputs, it is 2 by default.
24039 How to determine the end-of-stream.
24043 The duration of the longest input. (default)
24046 The duration of the shortest input.
24049 The duration of the first input.
24054 @subsection Examples
24058 Interleave frames belonging to different streams using @command{ffmpeg}:
24060 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24064 Add flickering blur effect:
24066 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24070 @section metadata, ametadata
24072 Manipulate frame metadata.
24074 This filter accepts the following options:
24078 Set mode of operation of the filter.
24080 Can be one of the following:
24084 If both @code{value} and @code{key} is set, select frames
24085 which have such metadata. If only @code{key} is set, select
24086 every frame that has such key in metadata.
24089 Add new metadata @code{key} and @code{value}. If key is already available
24093 Modify value of already present key.
24096 If @code{value} is set, delete only keys that have such value.
24097 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24101 Print key and its value if metadata was found. If @code{key} is not set print all
24102 metadata values available in frame.
24106 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24109 Set metadata value which will be used. This option is mandatory for
24110 @code{modify} and @code{add} mode.
24113 Which function to use when comparing metadata value and @code{value}.
24115 Can be one of following:
24119 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24122 Values are interpreted as strings, returns true if metadata value starts with
24123 the @code{value} option string.
24126 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24129 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24132 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24135 Values are interpreted as floats, returns true if expression from option @code{expr}
24139 Values are interpreted as strings, returns true if metadata value ends with
24140 the @code{value} option string.
24144 Set expression which is used when @code{function} is set to @code{expr}.
24145 The expression is evaluated through the eval API and can contain the following
24150 Float representation of @code{value} from metadata key.
24153 Float representation of @code{value} as supplied by user in @code{value} option.
24157 If specified in @code{print} mode, output is written to the named file. Instead of
24158 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24159 for standard output. If @code{file} option is not set, output is written to the log
24160 with AV_LOG_INFO loglevel.
24163 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24167 @subsection Examples
24171 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24174 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24177 Print silencedetect output to file @file{metadata.txt}.
24179 silencedetect,ametadata=mode=print:file=metadata.txt
24182 Direct all metadata to a pipe with file descriptor 4.
24184 metadata=mode=print:file='pipe\:4'
24188 @section perms, aperms
24190 Set read/write permissions for the output frames.
24192 These filters are mainly aimed at developers to test direct path in the
24193 following filter in the filtergraph.
24195 The filters accept the following options:
24199 Select the permissions mode.
24201 It accepts the following values:
24204 Do nothing. This is the default.
24206 Set all the output frames read-only.
24208 Set all the output frames directly writable.
24210 Make the frame read-only if writable, and writable if read-only.
24212 Set each output frame read-only or writable randomly.
24216 Set the seed for the @var{random} mode, must be an integer included between
24217 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24218 @code{-1}, the filter will try to use a good random seed on a best effort
24222 Note: in case of auto-inserted filter between the permission filter and the
24223 following one, the permission might not be received as expected in that
24224 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24225 perms/aperms filter can avoid this problem.
24227 @section realtime, arealtime
24229 Slow down filtering to match real time approximately.
24231 These filters will pause the filtering for a variable amount of time to
24232 match the output rate with the input timestamps.
24233 They are similar to the @option{re} option to @code{ffmpeg}.
24235 They accept the following options:
24239 Time limit for the pauses. Any pause longer than that will be considered
24240 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24242 Speed factor for processing. The value must be a float larger than zero.
24243 Values larger than 1.0 will result in faster than realtime processing,
24244 smaller will slow processing down. The @var{limit} is automatically adapted
24245 accordingly. Default is 1.0.
24247 A processing speed faster than what is possible without these filters cannot
24252 @section select, aselect
24254 Select frames to pass in output.
24256 This filter accepts the following options:
24261 Set expression, which is evaluated for each input frame.
24263 If the expression is evaluated to zero, the frame is discarded.
24265 If the evaluation result is negative or NaN, the frame is sent to the
24266 first output; otherwise it is sent to the output with index
24267 @code{ceil(val)-1}, assuming that the input index starts from 0.
24269 For example a value of @code{1.2} corresponds to the output with index
24270 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24273 Set the number of outputs. The output to which to send the selected
24274 frame is based on the result of the evaluation. Default value is 1.
24277 The expression can contain the following constants:
24281 The (sequential) number of the filtered frame, starting from 0.
24284 The (sequential) number of the selected frame, starting from 0.
24286 @item prev_selected_n
24287 The sequential number of the last selected frame. It's NAN if undefined.
24290 The timebase of the input timestamps.
24293 The PTS (Presentation TimeStamp) of the filtered video frame,
24294 expressed in @var{TB} units. It's NAN if undefined.
24297 The PTS of the filtered video frame,
24298 expressed in seconds. It's NAN if undefined.
24301 The PTS of the previously filtered video frame. It's NAN if undefined.
24303 @item prev_selected_pts
24304 The PTS of the last previously filtered video frame. It's NAN if undefined.
24306 @item prev_selected_t
24307 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24310 The PTS of the first video frame in the video. It's NAN if undefined.
24313 The time of the first video frame in the video. It's NAN if undefined.
24315 @item pict_type @emph{(video only)}
24316 The type of the filtered frame. It can assume one of the following
24328 @item interlace_type @emph{(video only)}
24329 The frame interlace type. It can assume one of the following values:
24332 The frame is progressive (not interlaced).
24334 The frame is top-field-first.
24336 The frame is bottom-field-first.
24339 @item consumed_sample_n @emph{(audio only)}
24340 the number of selected samples before the current frame
24342 @item samples_n @emph{(audio only)}
24343 the number of samples in the current frame
24345 @item sample_rate @emph{(audio only)}
24346 the input sample rate
24349 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24352 the position in the file of the filtered frame, -1 if the information
24353 is not available (e.g. for synthetic video)
24355 @item scene @emph{(video only)}
24356 value between 0 and 1 to indicate a new scene; a low value reflects a low
24357 probability for the current frame to introduce a new scene, while a higher
24358 value means the current frame is more likely to be one (see the example below)
24360 @item concatdec_select
24361 The concat demuxer can select only part of a concat input file by setting an
24362 inpoint and an outpoint, but the output packets may not be entirely contained
24363 in the selected interval. By using this variable, it is possible to skip frames
24364 generated by the concat demuxer which are not exactly contained in the selected
24367 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24368 and the @var{lavf.concat.duration} packet metadata values which are also
24369 present in the decoded frames.
24371 The @var{concatdec_select} variable is -1 if the frame pts is at least
24372 start_time and either the duration metadata is missing or the frame pts is less
24373 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24376 That basically means that an input frame is selected if its pts is within the
24377 interval set by the concat demuxer.
24381 The default value of the select expression is "1".
24383 @subsection Examples
24387 Select all frames in input:
24392 The example above is the same as:
24404 Select only I-frames:
24406 select='eq(pict_type\,I)'
24410 Select one frame every 100:
24412 select='not(mod(n\,100))'
24416 Select only frames contained in the 10-20 time interval:
24418 select=between(t\,10\,20)
24422 Select only I-frames contained in the 10-20 time interval:
24424 select=between(t\,10\,20)*eq(pict_type\,I)
24428 Select frames with a minimum distance of 10 seconds:
24430 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24434 Use aselect to select only audio frames with samples number > 100:
24436 aselect='gt(samples_n\,100)'
24440 Create a mosaic of the first scenes:
24442 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24445 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24449 Send even and odd frames to separate outputs, and compose them:
24451 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24455 Select useful frames from an ffconcat file which is using inpoints and
24456 outpoints but where the source files are not intra frame only.
24458 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24462 @section sendcmd, asendcmd
24464 Send commands to filters in the filtergraph.
24466 These filters read commands to be sent to other filters in the
24469 @code{sendcmd} must be inserted between two video filters,
24470 @code{asendcmd} must be inserted between two audio filters, but apart
24471 from that they act the same way.
24473 The specification of commands can be provided in the filter arguments
24474 with the @var{commands} option, or in a file specified by the
24475 @var{filename} option.
24477 These filters accept the following options:
24480 Set the commands to be read and sent to the other filters.
24482 Set the filename of the commands to be read and sent to the other
24486 @subsection Commands syntax
24488 A commands description consists of a sequence of interval
24489 specifications, comprising a list of commands to be executed when a
24490 particular event related to that interval occurs. The occurring event
24491 is typically the current frame time entering or leaving a given time
24494 An interval is specified by the following syntax:
24496 @var{START}[-@var{END}] @var{COMMANDS};
24499 The time interval is specified by the @var{START} and @var{END} times.
24500 @var{END} is optional and defaults to the maximum time.
24502 The current frame time is considered within the specified interval if
24503 it is included in the interval [@var{START}, @var{END}), that is when
24504 the time is greater or equal to @var{START} and is lesser than
24507 @var{COMMANDS} consists of a sequence of one or more command
24508 specifications, separated by ",", relating to that interval. The
24509 syntax of a command specification is given by:
24511 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24514 @var{FLAGS} is optional and specifies the type of events relating to
24515 the time interval which enable sending the specified command, and must
24516 be a non-null sequence of identifier flags separated by "+" or "|" and
24517 enclosed between "[" and "]".
24519 The following flags are recognized:
24522 The command is sent when the current frame timestamp enters the
24523 specified interval. In other words, the command is sent when the
24524 previous frame timestamp was not in the given interval, and the
24528 The command is sent when the current frame timestamp leaves the
24529 specified interval. In other words, the command is sent when the
24530 previous frame timestamp was in the given interval, and the
24534 The command @var{ARG} is interpreted as expression and result of
24535 expression is passed as @var{ARG}.
24537 The expression is evaluated through the eval API and can contain the following
24542 Original position in the file of the frame, or undefined if undefined
24543 for the current frame.
24546 The presentation timestamp in input.
24549 The count of the input frame for video or audio, starting from 0.
24552 The time in seconds of the current frame.
24555 The start time in seconds of the current command interval.
24558 The end time in seconds of the current command interval.
24561 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24566 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24569 @var{TARGET} specifies the target of the command, usually the name of
24570 the filter class or a specific filter instance name.
24572 @var{COMMAND} specifies the name of the command for the target filter.
24574 @var{ARG} is optional and specifies the optional list of argument for
24575 the given @var{COMMAND}.
24577 Between one interval specification and another, whitespaces, or
24578 sequences of characters starting with @code{#} until the end of line,
24579 are ignored and can be used to annotate comments.
24581 A simplified BNF description of the commands specification syntax
24584 @var{COMMAND_FLAG} ::= "enter" | "leave"
24585 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24586 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24587 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24588 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24589 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24592 @subsection Examples
24596 Specify audio tempo change at second 4:
24598 asendcmd=c='4.0 atempo tempo 1.5',atempo
24602 Target a specific filter instance:
24604 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24608 Specify a list of drawtext and hue commands in a file.
24610 # show text in the interval 5-10
24611 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24612 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24614 # desaturate the image in the interval 15-20
24615 15.0-20.0 [enter] hue s 0,
24616 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24618 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24620 # apply an exponential saturation fade-out effect, starting from time 25
24621 25 [enter] hue s exp(25-t)
24624 A filtergraph allowing to read and process the above command list
24625 stored in a file @file{test.cmd}, can be specified with:
24627 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24632 @section setpts, asetpts
24634 Change the PTS (presentation timestamp) of the input frames.
24636 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24638 This filter accepts the following options:
24643 The expression which is evaluated for each frame to construct its timestamp.
24647 The expression is evaluated through the eval API and can contain the following
24651 @item FRAME_RATE, FR
24652 frame rate, only defined for constant frame-rate video
24655 The presentation timestamp in input
24658 The count of the input frame for video or the number of consumed samples,
24659 not including the current frame for audio, starting from 0.
24661 @item NB_CONSUMED_SAMPLES
24662 The number of consumed samples, not including the current frame (only
24665 @item NB_SAMPLES, S
24666 The number of samples in the current frame (only audio)
24668 @item SAMPLE_RATE, SR
24669 The audio sample rate.
24672 The PTS of the first frame.
24675 the time in seconds of the first frame
24678 State whether the current frame is interlaced.
24681 the time in seconds of the current frame
24684 original position in the file of the frame, or undefined if undefined
24685 for the current frame
24688 The previous input PTS.
24691 previous input time in seconds
24694 The previous output PTS.
24697 previous output time in seconds
24700 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
24704 The wallclock (RTC) time at the start of the movie in microseconds.
24707 The timebase of the input timestamps.
24711 @subsection Examples
24715 Start counting PTS from zero
24717 setpts=PTS-STARTPTS
24721 Apply fast motion effect:
24727 Apply slow motion effect:
24733 Set fixed rate of 25 frames per second:
24739 Set fixed rate 25 fps with some jitter:
24741 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
24745 Apply an offset of 10 seconds to the input PTS:
24751 Generate timestamps from a "live source" and rebase onto the current timebase:
24753 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
24757 Generate timestamps by counting samples:
24766 Force color range for the output video frame.
24768 The @code{setrange} filter marks the color range property for the
24769 output frames. It does not change the input frame, but only sets the
24770 corresponding property, which affects how the frame is treated by
24773 The filter accepts the following options:
24778 Available values are:
24782 Keep the same color range property.
24784 @item unspecified, unknown
24785 Set the color range as unspecified.
24787 @item limited, tv, mpeg
24788 Set the color range as limited.
24790 @item full, pc, jpeg
24791 Set the color range as full.
24795 @section settb, asettb
24797 Set the timebase to use for the output frames timestamps.
24798 It is mainly useful for testing timebase configuration.
24800 It accepts the following parameters:
24805 The expression which is evaluated into the output timebase.
24809 The value for @option{tb} is an arithmetic expression representing a
24810 rational. The expression can contain the constants "AVTB" (the default
24811 timebase), "intb" (the input timebase) and "sr" (the sample rate,
24812 audio only). Default value is "intb".
24814 @subsection Examples
24818 Set the timebase to 1/25:
24824 Set the timebase to 1/10:
24830 Set the timebase to 1001/1000:
24836 Set the timebase to 2*intb:
24842 Set the default timebase value:
24849 Convert input audio to a video output representing frequency spectrum
24850 logarithmically using Brown-Puckette constant Q transform algorithm with
24851 direct frequency domain coefficient calculation (but the transform itself
24852 is not really constant Q, instead the Q factor is actually variable/clamped),
24853 with musical tone scale, from E0 to D#10.
24855 The filter accepts the following options:
24859 Specify the video size for the output. It must be even. For the syntax of this option,
24860 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24861 Default value is @code{1920x1080}.
24864 Set the output frame rate. Default value is @code{25}.
24867 Set the bargraph height. It must be even. Default value is @code{-1} which
24868 computes the bargraph height automatically.
24871 Set the axis height. It must be even. Default value is @code{-1} which computes
24872 the axis height automatically.
24875 Set the sonogram height. It must be even. Default value is @code{-1} which
24876 computes the sonogram height automatically.
24879 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
24880 instead. Default value is @code{1}.
24882 @item sono_v, volume
24883 Specify the sonogram volume expression. It can contain variables:
24886 the @var{bar_v} evaluated expression
24887 @item frequency, freq, f
24888 the frequency where it is evaluated
24889 @item timeclamp, tc
24890 the value of @var{timeclamp} option
24894 @item a_weighting(f)
24895 A-weighting of equal loudness
24896 @item b_weighting(f)
24897 B-weighting of equal loudness
24898 @item c_weighting(f)
24899 C-weighting of equal loudness.
24901 Default value is @code{16}.
24903 @item bar_v, volume2
24904 Specify the bargraph volume expression. It can contain variables:
24907 the @var{sono_v} evaluated expression
24908 @item frequency, freq, f
24909 the frequency where it is evaluated
24910 @item timeclamp, tc
24911 the value of @var{timeclamp} option
24915 @item a_weighting(f)
24916 A-weighting of equal loudness
24917 @item b_weighting(f)
24918 B-weighting of equal loudness
24919 @item c_weighting(f)
24920 C-weighting of equal loudness.
24922 Default value is @code{sono_v}.
24924 @item sono_g, gamma
24925 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
24926 higher gamma makes the spectrum having more range. Default value is @code{3}.
24927 Acceptable range is @code{[1, 7]}.
24929 @item bar_g, gamma2
24930 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
24934 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
24935 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
24937 @item timeclamp, tc
24938 Specify the transform timeclamp. At low frequency, there is trade-off between
24939 accuracy in time domain and frequency domain. If timeclamp is lower,
24940 event in time domain is represented more accurately (such as fast bass drum),
24941 otherwise event in frequency domain is represented more accurately
24942 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
24945 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
24946 limits future samples by applying asymmetric windowing in time domain, useful
24947 when low latency is required. Accepted range is @code{[0, 1]}.
24950 Specify the transform base frequency. Default value is @code{20.01523126408007475},
24951 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
24954 Specify the transform end frequency. Default value is @code{20495.59681441799654},
24955 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
24958 This option is deprecated and ignored.
24961 Specify the transform length in time domain. Use this option to control accuracy
24962 trade-off between time domain and frequency domain at every frequency sample.
24963 It can contain variables:
24965 @item frequency, freq, f
24966 the frequency where it is evaluated
24967 @item timeclamp, tc
24968 the value of @var{timeclamp} option.
24970 Default value is @code{384*tc/(384+tc*f)}.
24973 Specify the transform count for every video frame. Default value is @code{6}.
24974 Acceptable range is @code{[1, 30]}.
24977 Specify the transform count for every single pixel. Default value is @code{0},
24978 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
24981 Specify font file for use with freetype to draw the axis. If not specified,
24982 use embedded font. Note that drawing with font file or embedded font is not
24983 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
24987 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
24988 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
24992 Specify font color expression. This is arithmetic expression that should return
24993 integer value 0xRRGGBB. It can contain variables:
24995 @item frequency, freq, f
24996 the frequency where it is evaluated
24997 @item timeclamp, tc
24998 the value of @var{timeclamp} option
25003 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25004 @item r(x), g(x), b(x)
25005 red, green, and blue value of intensity x.
25007 Default value is @code{st(0, (midi(f)-59.5)/12);
25008 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25009 r(1-ld(1)) + b(ld(1))}.
25012 Specify image file to draw the axis. This option override @var{fontfile} and
25013 @var{fontcolor} option.
25016 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25017 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25018 Default value is @code{1}.
25021 Set colorspace. The accepted values are:
25024 Unspecified (default)
25033 BT.470BG or BT.601-6 625
25036 SMPTE-170M or BT.601-6 525
25042 BT.2020 with non-constant luminance
25047 Set spectrogram color scheme. This is list of floating point values with format
25048 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25049 The default is @code{1|0.5|0|0|0.5|1}.
25053 @subsection Examples
25057 Playing audio while showing the spectrum:
25059 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25063 Same as above, but with frame rate 30 fps:
25065 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25069 Playing at 1280x720:
25071 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25075 Disable sonogram display:
25081 A1 and its harmonics: A1, A2, (near)E3, A3:
25083 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),
25084 asplit[a][out1]; [a] showcqt [out0]'
25088 Same as above, but with more accuracy in frequency domain:
25090 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),
25091 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25097 bar_v=10:sono_v=bar_v*a_weighting(f)
25101 Custom gamma, now spectrum is linear to the amplitude.
25107 Custom tlength equation:
25109 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)))'
25113 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25115 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25119 Custom font using fontconfig:
25121 font='Courier New,Monospace,mono|bold'
25125 Custom frequency range with custom axis using image file:
25127 axisfile=myaxis.png:basefreq=40:endfreq=10000
25133 Convert input audio to video output representing the audio power spectrum.
25134 Audio amplitude is on Y-axis while frequency is on X-axis.
25136 The filter accepts the following options:
25140 Specify size of video. For the syntax of this option, check the
25141 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25142 Default is @code{1024x512}.
25146 This set how each frequency bin will be represented.
25148 It accepts the following values:
25154 Default is @code{bar}.
25157 Set amplitude scale.
25159 It accepts the following values:
25173 Default is @code{log}.
25176 Set frequency scale.
25178 It accepts the following values:
25187 Reverse logarithmic scale.
25189 Default is @code{lin}.
25192 Set window size. Allowed range is from 16 to 65536.
25194 Default is @code{2048}
25197 Set windowing function.
25199 It accepts the following values:
25222 Default is @code{hanning}.
25225 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25226 which means optimal overlap for selected window function will be picked.
25229 Set time averaging. Setting this to 0 will display current maximal peaks.
25230 Default is @code{1}, which means time averaging is disabled.
25233 Specify list of colors separated by space or by '|' which will be used to
25234 draw channel frequencies. Unrecognized or missing colors will be replaced
25238 Set channel display mode.
25240 It accepts the following values:
25245 Default is @code{combined}.
25248 Set minimum amplitude used in @code{log} amplitude scaler.
25251 Set data display mode.
25253 It accepts the following values:
25259 Default is @code{magnitude}.
25262 @section showspatial
25264 Convert stereo input audio to a video output, representing the spatial relationship
25265 between two channels.
25267 The filter accepts the following options:
25271 Specify the video size for the output. For the syntax of this option, check the
25272 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25273 Default value is @code{512x512}.
25276 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25279 Set window function.
25281 It accepts the following values:
25306 Default value is @code{hann}.
25309 Set ratio of overlap window. Default value is @code{0.5}.
25310 When value is @code{1} overlap is set to recommended size for specific
25311 window function currently used.
25314 @anchor{showspectrum}
25315 @section showspectrum
25317 Convert input audio to a video output, representing the audio frequency
25320 The filter accepts the following options:
25324 Specify the video size for the output. For the syntax of this option, check the
25325 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25326 Default value is @code{640x512}.
25329 Specify how the spectrum should slide along the window.
25331 It accepts the following values:
25334 the samples start again on the left when they reach the right
25336 the samples scroll from right to left
25338 frames are only produced when the samples reach the right
25340 the samples scroll from left to right
25343 Default value is @code{replace}.
25346 Specify display mode.
25348 It accepts the following values:
25351 all channels are displayed in the same row
25353 all channels are displayed in separate rows
25356 Default value is @samp{combined}.
25359 Specify display color mode.
25361 It accepts the following values:
25364 each channel is displayed in a separate color
25366 each channel is displayed using the same color scheme
25368 each channel is displayed using the rainbow color scheme
25370 each channel is displayed using the moreland color scheme
25372 each channel is displayed using the nebulae color scheme
25374 each channel is displayed using the fire color scheme
25376 each channel is displayed using the fiery color scheme
25378 each channel is displayed using the fruit color scheme
25380 each channel is displayed using the cool color scheme
25382 each channel is displayed using the magma color scheme
25384 each channel is displayed using the green color scheme
25386 each channel is displayed using the viridis color scheme
25388 each channel is displayed using the plasma color scheme
25390 each channel is displayed using the cividis color scheme
25392 each channel is displayed using the terrain color scheme
25395 Default value is @samp{channel}.
25398 Specify scale used for calculating intensity color values.
25400 It accepts the following values:
25405 square root, default
25416 Default value is @samp{sqrt}.
25419 Specify frequency scale.
25421 It accepts the following values:
25429 Default value is @samp{lin}.
25432 Set saturation modifier for displayed colors. Negative values provide
25433 alternative color scheme. @code{0} is no saturation at all.
25434 Saturation must be in [-10.0, 10.0] range.
25435 Default value is @code{1}.
25438 Set window function.
25440 It accepts the following values:
25465 Default value is @code{hann}.
25468 Set orientation of time vs frequency axis. Can be @code{vertical} or
25469 @code{horizontal}. Default is @code{vertical}.
25472 Set ratio of overlap window. Default value is @code{0}.
25473 When value is @code{1} overlap is set to recommended size for specific
25474 window function currently used.
25477 Set scale gain for calculating intensity color values.
25478 Default value is @code{1}.
25481 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25484 Set color rotation, must be in [-1.0, 1.0] range.
25485 Default value is @code{0}.
25488 Set start frequency from which to display spectrogram. Default is @code{0}.
25491 Set stop frequency to which to display spectrogram. Default is @code{0}.
25494 Set upper frame rate limit. Default is @code{auto}, unlimited.
25497 Draw time and frequency axes and legends. Default is disabled.
25500 The usage is very similar to the showwaves filter; see the examples in that
25503 @subsection Examples
25507 Large window with logarithmic color scaling:
25509 showspectrum=s=1280x480:scale=log
25513 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25515 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25516 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25520 @section showspectrumpic
25522 Convert input audio to a single video frame, representing the audio frequency
25525 The filter accepts the following options:
25529 Specify the video size for the output. For the syntax of this option, check the
25530 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25531 Default value is @code{4096x2048}.
25534 Specify display mode.
25536 It accepts the following values:
25539 all channels are displayed in the same row
25541 all channels are displayed in separate rows
25543 Default value is @samp{combined}.
25546 Specify display color mode.
25548 It accepts the following values:
25551 each channel is displayed in a separate color
25553 each channel is displayed using the same color scheme
25555 each channel is displayed using the rainbow color scheme
25557 each channel is displayed using the moreland color scheme
25559 each channel is displayed using the nebulae color scheme
25561 each channel is displayed using the fire color scheme
25563 each channel is displayed using the fiery color scheme
25565 each channel is displayed using the fruit color scheme
25567 each channel is displayed using the cool color scheme
25569 each channel is displayed using the magma color scheme
25571 each channel is displayed using the green color scheme
25573 each channel is displayed using the viridis color scheme
25575 each channel is displayed using the plasma color scheme
25577 each channel is displayed using the cividis color scheme
25579 each channel is displayed using the terrain color scheme
25581 Default value is @samp{intensity}.
25584 Specify scale used for calculating intensity color values.
25586 It accepts the following values:
25591 square root, default
25601 Default value is @samp{log}.
25604 Specify frequency scale.
25606 It accepts the following values:
25614 Default value is @samp{lin}.
25617 Set saturation modifier for displayed colors. Negative values provide
25618 alternative color scheme. @code{0} is no saturation at all.
25619 Saturation must be in [-10.0, 10.0] range.
25620 Default value is @code{1}.
25623 Set window function.
25625 It accepts the following values:
25649 Default value is @code{hann}.
25652 Set orientation of time vs frequency axis. Can be @code{vertical} or
25653 @code{horizontal}. Default is @code{vertical}.
25656 Set scale gain for calculating intensity color values.
25657 Default value is @code{1}.
25660 Draw time and frequency axes and legends. Default is enabled.
25663 Set color rotation, must be in [-1.0, 1.0] range.
25664 Default value is @code{0}.
25667 Set start frequency from which to display spectrogram. Default is @code{0}.
25670 Set stop frequency to which to display spectrogram. Default is @code{0}.
25673 @subsection Examples
25677 Extract an audio spectrogram of a whole audio track
25678 in a 1024x1024 picture using @command{ffmpeg}:
25680 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25684 @section showvolume
25686 Convert input audio volume to a video output.
25688 The filter accepts the following options:
25695 Set border width, allowed range is [0, 5]. Default is 1.
25698 Set channel width, allowed range is [80, 8192]. Default is 400.
25701 Set channel height, allowed range is [1, 900]. Default is 20.
25704 Set fade, allowed range is [0, 1]. Default is 0.95.
25707 Set volume color expression.
25709 The expression can use the following variables:
25713 Current max volume of channel in dB.
25719 Current channel number, starting from 0.
25723 If set, displays channel names. Default is enabled.
25726 If set, displays volume values. Default is enabled.
25729 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
25730 default is @code{h}.
25733 Set step size, allowed range is [0, 5]. Default is 0, which means
25737 Set background opacity, allowed range is [0, 1]. Default is 0.
25740 Set metering mode, can be peak: @code{p} or rms: @code{r},
25741 default is @code{p}.
25744 Set display scale, can be linear: @code{lin} or log: @code{log},
25745 default is @code{lin}.
25749 If set to > 0., display a line for the max level
25750 in the previous seconds.
25751 default is disabled: @code{0.}
25754 The color of the max line. Use when @code{dm} option is set to > 0.
25755 default is: @code{orange}
25760 Convert input audio to a video output, representing the samples waves.
25762 The filter accepts the following options:
25766 Specify the video size for the output. For the syntax of this option, check the
25767 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25768 Default value is @code{600x240}.
25773 Available values are:
25776 Draw a point for each sample.
25779 Draw a vertical line for each sample.
25782 Draw a point for each sample and a line between them.
25785 Draw a centered vertical line for each sample.
25788 Default value is @code{point}.
25791 Set the number of samples which are printed on the same column. A
25792 larger value will decrease the frame rate. Must be a positive
25793 integer. This option can be set only if the value for @var{rate}
25794 is not explicitly specified.
25797 Set the (approximate) output frame rate. This is done by setting the
25798 option @var{n}. Default value is "25".
25800 @item split_channels
25801 Set if channels should be drawn separately or overlap. Default value is 0.
25804 Set colors separated by '|' which are going to be used for drawing of each channel.
25807 Set amplitude scale.
25809 Available values are:
25827 Set the draw mode. This is mostly useful to set for high @var{n}.
25829 Available values are:
25832 Scale pixel values for each drawn sample.
25835 Draw every sample directly.
25838 Default value is @code{scale}.
25841 @subsection Examples
25845 Output the input file audio and the corresponding video representation
25848 amovie=a.mp3,asplit[out0],showwaves[out1]
25852 Create a synthetic signal and show it with showwaves, forcing a
25853 frame rate of 30 frames per second:
25855 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
25859 @section showwavespic
25861 Convert input audio to a single video frame, representing the samples waves.
25863 The filter accepts the following options:
25867 Specify the video size for the output. For the syntax of this option, check the
25868 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25869 Default value is @code{600x240}.
25871 @item split_channels
25872 Set if channels should be drawn separately or overlap. Default value is 0.
25875 Set colors separated by '|' which are going to be used for drawing of each channel.
25878 Set amplitude scale.
25880 Available values are:
25900 Available values are:
25903 Scale pixel values for each drawn sample.
25906 Draw every sample directly.
25909 Default value is @code{scale}.
25912 Set the filter mode.
25914 Available values are:
25917 Use average samples values for each drawn sample.
25920 Use peak samples values for each drawn sample.
25923 Default value is @code{average}.
25926 @subsection Examples
25930 Extract a channel split representation of the wave form of a whole audio track
25931 in a 1024x800 picture using @command{ffmpeg}:
25933 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25937 @section sidedata, asidedata
25939 Delete frame side data, or select frames based on it.
25941 This filter accepts the following options:
25945 Set mode of operation of the filter.
25947 Can be one of the following:
25951 Select every frame with side data of @code{type}.
25954 Delete side data of @code{type}. If @code{type} is not set, delete all side
25960 Set side data type used with all modes. Must be set for @code{select} mode. For
25961 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
25962 in @file{libavutil/frame.h}. For example, to choose
25963 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
25967 @section spectrumsynth
25969 Synthesize audio from 2 input video spectrums, first input stream represents
25970 magnitude across time and second represents phase across time.
25971 The filter will transform from frequency domain as displayed in videos back
25972 to time domain as presented in audio output.
25974 This filter is primarily created for reversing processed @ref{showspectrum}
25975 filter outputs, but can synthesize sound from other spectrograms too.
25976 But in such case results are going to be poor if the phase data is not
25977 available, because in such cases phase data need to be recreated, usually
25978 it's just recreated from random noise.
25979 For best results use gray only output (@code{channel} color mode in
25980 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
25981 @code{lin} scale for phase video. To produce phase, for 2nd video, use
25982 @code{data} option. Inputs videos should generally use @code{fullframe}
25983 slide mode as that saves resources needed for decoding video.
25985 The filter accepts the following options:
25989 Specify sample rate of output audio, the sample rate of audio from which
25990 spectrum was generated may differ.
25993 Set number of channels represented in input video spectrums.
25996 Set scale which was used when generating magnitude input spectrum.
25997 Can be @code{lin} or @code{log}. Default is @code{log}.
26000 Set slide which was used when generating inputs spectrums.
26001 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26002 Default is @code{fullframe}.
26005 Set window function used for resynthesis.
26008 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26009 which means optimal overlap for selected window function will be picked.
26012 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26013 Default is @code{vertical}.
26016 @subsection Examples
26020 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26021 then resynthesize videos back to audio with spectrumsynth:
26023 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
26024 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
26025 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26029 @section split, asplit
26031 Split input into several identical outputs.
26033 @code{asplit} works with audio input, @code{split} with video.
26035 The filter accepts a single parameter which specifies the number of outputs. If
26036 unspecified, it defaults to 2.
26038 @subsection Examples
26042 Create two separate outputs from the same input:
26044 [in] split [out0][out1]
26048 To create 3 or more outputs, you need to specify the number of
26051 [in] asplit=3 [out0][out1][out2]
26055 Create two separate outputs from the same input, one cropped and
26058 [in] split [splitout1][splitout2];
26059 [splitout1] crop=100:100:0:0 [cropout];
26060 [splitout2] pad=200:200:100:100 [padout];
26064 Create 5 copies of the input audio with @command{ffmpeg}:
26066 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26072 Receive commands sent through a libzmq client, and forward them to
26073 filters in the filtergraph.
26075 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26076 must be inserted between two video filters, @code{azmq} between two
26077 audio filters. Both are capable to send messages to any filter type.
26079 To enable these filters you need to install the libzmq library and
26080 headers and configure FFmpeg with @code{--enable-libzmq}.
26082 For more information about libzmq see:
26083 @url{http://www.zeromq.org/}
26085 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26086 receives messages sent through a network interface defined by the
26087 @option{bind_address} (or the abbreviation "@option{b}") option.
26088 Default value of this option is @file{tcp://localhost:5555}. You may
26089 want to alter this value to your needs, but do not forget to escape any
26090 ':' signs (see @ref{filtergraph escaping}).
26092 The received message must be in the form:
26094 @var{TARGET} @var{COMMAND} [@var{ARG}]
26097 @var{TARGET} specifies the target of the command, usually the name of
26098 the filter class or a specific filter instance name. The default
26099 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26100 but you can override this by using the @samp{filter_name@@id} syntax
26101 (see @ref{Filtergraph syntax}).
26103 @var{COMMAND} specifies the name of the command for the target filter.
26105 @var{ARG} is optional and specifies the optional argument list for the
26106 given @var{COMMAND}.
26108 Upon reception, the message is processed and the corresponding command
26109 is injected into the filtergraph. Depending on the result, the filter
26110 will send a reply to the client, adopting the format:
26112 @var{ERROR_CODE} @var{ERROR_REASON}
26116 @var{MESSAGE} is optional.
26118 @subsection Examples
26120 Look at @file{tools/zmqsend} for an example of a zmq client which can
26121 be used to send commands processed by these filters.
26123 Consider the following filtergraph generated by @command{ffplay}.
26124 In this example the last overlay filter has an instance name. All other
26125 filters will have default instance names.
26128 ffplay -dumpgraph 1 -f lavfi "
26129 color=s=100x100:c=red [l];
26130 color=s=100x100:c=blue [r];
26131 nullsrc=s=200x100, zmq [bg];
26132 [bg][l] overlay [bg+l];
26133 [bg+l][r] overlay@@my=x=100 "
26136 To change the color of the left side of the video, the following
26137 command can be used:
26139 echo Parsed_color_0 c yellow | tools/zmqsend
26142 To change the right side:
26144 echo Parsed_color_1 c pink | tools/zmqsend
26147 To change the position of the right side:
26149 echo overlay@@my x 150 | tools/zmqsend
26153 @c man end MULTIMEDIA FILTERS
26155 @chapter Multimedia Sources
26156 @c man begin MULTIMEDIA SOURCES
26158 Below is a description of the currently available multimedia sources.
26162 This is the same as @ref{movie} source, except it selects an audio
26168 Read audio and/or video stream(s) from a movie container.
26170 It accepts the following parameters:
26174 The name of the resource to read (not necessarily a file; it can also be a
26175 device or a stream accessed through some protocol).
26177 @item format_name, f
26178 Specifies the format assumed for the movie to read, and can be either
26179 the name of a container or an input device. If not specified, the
26180 format is guessed from @var{movie_name} or by probing.
26182 @item seek_point, sp
26183 Specifies the seek point in seconds. The frames will be output
26184 starting from this seek point. The parameter is evaluated with
26185 @code{av_strtod}, so the numerical value may be suffixed by an IS
26186 postfix. The default value is "0".
26189 Specifies the streams to read. Several streams can be specified,
26190 separated by "+". The source will then have as many outputs, in the
26191 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26192 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26193 respectively the default (best suited) video and audio stream. Default
26194 is "dv", or "da" if the filter is called as "amovie".
26196 @item stream_index, si
26197 Specifies the index of the video stream to read. If the value is -1,
26198 the most suitable video stream will be automatically selected. The default
26199 value is "-1". Deprecated. If the filter is called "amovie", it will select
26200 audio instead of video.
26203 Specifies how many times to read the stream in sequence.
26204 If the value is 0, the stream will be looped infinitely.
26205 Default value is "1".
26207 Note that when the movie is looped the source timestamps are not
26208 changed, so it will generate non monotonically increasing timestamps.
26210 @item discontinuity
26211 Specifies the time difference between frames above which the point is
26212 considered a timestamp discontinuity which is removed by adjusting the later
26216 It allows overlaying a second video on top of the main input of
26217 a filtergraph, as shown in this graph:
26219 input -----------> deltapts0 --> overlay --> output
26222 movie --> scale--> deltapts1 -------+
26224 @subsection Examples
26228 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26229 on top of the input labelled "in":
26231 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26232 [in] setpts=PTS-STARTPTS [main];
26233 [main][over] overlay=16:16 [out]
26237 Read from a video4linux2 device, and overlay it on top of the input
26240 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26241 [in] setpts=PTS-STARTPTS [main];
26242 [main][over] overlay=16:16 [out]
26246 Read the first video stream and the audio stream with id 0x81 from
26247 dvd.vob; the video is connected to the pad named "video" and the audio is
26248 connected to the pad named "audio":
26250 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26254 @subsection Commands
26256 Both movie and amovie support the following commands:
26259 Perform seek using "av_seek_frame".
26260 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26263 @var{stream_index}: If stream_index is -1, a default
26264 stream is selected, and @var{timestamp} is automatically converted
26265 from AV_TIME_BASE units to the stream specific time_base.
26267 @var{timestamp}: Timestamp in AVStream.time_base units
26268 or, if no stream is specified, in AV_TIME_BASE units.
26270 @var{flags}: Flags which select direction and seeking mode.
26274 Get movie duration in AV_TIME_BASE units.
26278 @c man end MULTIMEDIA SOURCES