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. Availabe values are:
541 Default is @var{4th}.
548 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
549 each band will be in separate stream:
551 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
555 Same as above, but with higher filter order:
557 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=4th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
561 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
563 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=4th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
569 Reduce audio bit resolution.
571 This filter is bit crusher with enhanced functionality. A bit crusher
572 is used to audibly reduce number of bits an audio signal is sampled
573 with. This doesn't change the bit depth at all, it just produces the
574 effect. Material reduced in bit depth sounds more harsh and "digital".
575 This filter is able to even round to continuous values instead of discrete
577 Additionally it has a D/C offset which results in different crushing of
578 the lower and the upper half of the signal.
579 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
581 Another feature of this filter is the logarithmic mode.
582 This setting switches from linear distances between bits to logarithmic ones.
583 The result is a much more "natural" sounding crusher which doesn't gate low
584 signals for example. The human ear has a logarithmic perception,
585 so this kind of crushing is much more pleasant.
586 Logarithmic crushing is also able to get anti-aliased.
588 The filter accepts the following options:
604 Can be linear: @code{lin} or logarithmic: @code{log}.
613 Set sample reduction.
616 Enable LFO. By default disabled.
627 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
631 Remove impulsive noise from input audio.
633 Samples detected as impulsive noise are replaced by interpolated samples using
634 autoregressive modelling.
638 Set window size, in milliseconds. Allowed range is from @code{10} to
639 @code{100}. Default value is @code{55} milliseconds.
640 This sets size of window which will be processed at once.
643 Set window overlap, in percentage of window size. Allowed range is from
644 @code{50} to @code{95}. Default value is @code{75} percent.
645 Setting this to a very high value increases impulsive noise removal but makes
646 whole process much slower.
649 Set autoregression order, in percentage of window size. Allowed range is from
650 @code{0} to @code{25}. Default value is @code{2} percent. This option also
651 controls quality of interpolated samples using neighbour good samples.
654 Set threshold value. Allowed range is from @code{1} to @code{100}.
655 Default value is @code{2}.
656 This controls the strength of impulsive noise which is going to be removed.
657 The lower value, the more samples will be detected as impulsive noise.
660 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
661 @code{10}. Default value is @code{2}.
662 If any two samples detected as noise are spaced less than this value then any
663 sample between those two samples will be also detected as noise.
668 It accepts the following values:
671 Select overlap-add method. Even not interpolated samples are slightly
672 changed with this method.
675 Select overlap-save method. Not interpolated samples remain unchanged.
678 Default value is @code{a}.
682 Remove clipped samples from input audio.
684 Samples detected as clipped are replaced by interpolated samples using
685 autoregressive modelling.
689 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
690 Default value is @code{55} milliseconds.
691 This sets size of window which will be processed at once.
694 Set window overlap, in percentage of window size. Allowed range is from @code{50}
695 to @code{95}. Default value is @code{75} percent.
698 Set autoregression order, in percentage of window size. Allowed range is from
699 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
700 quality of interpolated samples using neighbour good samples.
703 Set threshold value. Allowed range is from @code{1} to @code{100}.
704 Default value is @code{10}. Higher values make clip detection less aggressive.
707 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
708 Default value is @code{1000}. Higher values make clip detection less aggressive.
713 It accepts the following values:
716 Select overlap-add method. Even not interpolated samples are slightly changed
720 Select overlap-save method. Not interpolated samples remain unchanged.
723 Default value is @code{a}.
728 Delay one or more audio channels.
730 Samples in delayed channel are filled with silence.
732 The filter accepts the following option:
736 Set list of delays in milliseconds for each channel separated by '|'.
737 Unused delays will be silently ignored. If number of given delays is
738 smaller than number of channels all remaining channels will not be delayed.
739 If you want to delay exact number of samples, append 'S' to number.
740 If you want instead to delay in seconds, append 's' to number.
743 Use last set delay for all remaining channels. By default is disabled.
744 This option if enabled changes how option @code{delays} is interpreted.
751 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
752 the second channel (and any other channels that may be present) unchanged.
758 Delay second channel by 500 samples, the third channel by 700 samples and leave
759 the first channel (and any other channels that may be present) unchanged.
765 Delay all channels by same number of samples:
767 adelay=delays=64S:all=1
772 Remedy denormals in audio by adding extremely low-level noise.
774 A description of the accepted parameters follows.
778 Set level of added noise in dB. Default is @code{-351}.
779 Allowed range is from -451 to -90.
782 Set type of added noise.
795 Default is @code{dc}.
798 @section aderivative, aintegral
800 Compute derivative/integral of audio stream.
802 Applying both filters one after another produces original audio.
806 Apply echoing to the input audio.
808 Echoes are reflected sound and can occur naturally amongst mountains
809 (and sometimes large buildings) when talking or shouting; digital echo
810 effects emulate this behaviour and are often used to help fill out the
811 sound of a single instrument or vocal. The time difference between the
812 original signal and the reflection is the @code{delay}, and the
813 loudness of the reflected signal is the @code{decay}.
814 Multiple echoes can have different delays and decays.
816 A description of the accepted parameters follows.
820 Set input gain of reflected signal. Default is @code{0.6}.
823 Set output gain of reflected signal. Default is @code{0.3}.
826 Set list of time intervals in milliseconds between original signal and reflections
827 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
828 Default is @code{1000}.
831 Set list of loudness of reflected signals separated by '|'.
832 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
833 Default is @code{0.5}.
840 Make it sound as if there are twice as many instruments as are actually playing:
842 aecho=0.8:0.88:60:0.4
846 If delay is very short, then it sounds like a (metallic) robot playing music:
852 A longer delay will sound like an open air concert in the mountains:
854 aecho=0.8:0.9:1000:0.3
858 Same as above but with one more mountain:
860 aecho=0.8:0.9:1000|1800:0.3|0.25
865 Audio emphasis filter creates or restores material directly taken from LPs or
866 emphased CDs with different filter curves. E.g. to store music on vinyl the
867 signal has to be altered by a filter first to even out the disadvantages of
868 this recording medium.
869 Once the material is played back the inverse filter has to be applied to
870 restore the distortion of the frequency response.
872 The filter accepts the following options:
882 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
883 use @code{production} mode. Default is @code{reproduction} mode.
886 Set filter type. Selects medium. Can be one of the following:
898 select Compact Disc (CD).
904 select 50µs (FM-KF).
906 select 75µs (FM-KF).
912 Modify an audio signal according to the specified expressions.
914 This filter accepts one or more expressions (one for each channel),
915 which are evaluated and used to modify a corresponding audio signal.
917 It accepts the following parameters:
921 Set the '|'-separated expressions list for each separate channel. If
922 the number of input channels is greater than the number of
923 expressions, the last specified expression is used for the remaining
926 @item channel_layout, c
927 Set output channel layout. If not specified, the channel layout is
928 specified by the number of expressions. If set to @samp{same}, it will
929 use by default the same input channel layout.
932 Each expression in @var{exprs} can contain the following constants and functions:
936 channel number of the current expression
939 number of the evaluated sample, starting from 0
945 time of the evaluated sample expressed in seconds
948 @item nb_out_channels
949 input and output number of channels
952 the value of input channel with number @var{CH}
955 Note: this filter is slow. For faster processing you should use a
964 aeval=val(ch)/2:c=same
968 Invert phase of the second channel:
977 Apply fade-in/out effect to input audio.
979 A description of the accepted parameters follows.
983 Specify the effect type, can be either @code{in} for fade-in, or
984 @code{out} for a fade-out effect. Default is @code{in}.
986 @item start_sample, ss
987 Specify the number of the start sample for starting to apply the fade
988 effect. Default is 0.
991 Specify the number of samples for which the fade effect has to last. At
992 the end of the fade-in effect the output audio will have the same
993 volume as the input audio, at the end of the fade-out transition
994 the output audio will be silence. Default is 44100.
997 Specify the start time of the fade effect. Default is 0.
998 The value must be specified as a time duration; see
999 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1000 for the accepted syntax.
1001 If set this option is used instead of @var{start_sample}.
1004 Specify the duration of the fade effect. See
1005 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1006 for the accepted syntax.
1007 At the end of the fade-in effect the output audio will have the same
1008 volume as the input audio, at the end of the fade-out transition
1009 the output audio will be silence.
1010 By default the duration is determined by @var{nb_samples}.
1011 If set this option is used instead of @var{nb_samples}.
1014 Set curve for fade transition.
1016 It accepts the following values:
1019 select triangular, linear slope (default)
1021 select quarter of sine wave
1023 select half of sine wave
1025 select exponential sine wave
1029 select inverted parabola
1043 select inverted quarter of sine wave
1045 select inverted half of sine wave
1047 select double-exponential seat
1049 select double-exponential sigmoid
1051 select logistic sigmoid
1053 select sine cardinal function
1055 select inverted sine cardinal function
1061 @subsection Examples
1065 Fade in first 15 seconds of audio:
1067 afade=t=in:ss=0:d=15
1071 Fade out last 25 seconds of a 900 seconds audio:
1073 afade=t=out:st=875:d=25
1078 Denoise audio samples with FFT.
1080 A description of the accepted parameters follows.
1084 Set the noise reduction in dB, allowed range is 0.01 to 97.
1085 Default value is 12 dB.
1088 Set the noise floor in dB, allowed range is -80 to -20.
1089 Default value is -50 dB.
1094 It accepts the following values:
1103 Select shellac noise.
1106 Select custom noise, defined in @code{bn} option.
1108 Default value is white noise.
1112 Set custom band noise for every one of 15 bands.
1113 Bands are separated by ' ' or '|'.
1116 Set the residual floor in dB, allowed range is -80 to -20.
1117 Default value is -38 dB.
1120 Enable noise tracking. By default is disabled.
1121 With this enabled, noise floor is automatically adjusted.
1124 Enable residual tracking. By default is disabled.
1127 Set the output mode.
1129 It accepts the following values:
1132 Pass input unchanged.
1135 Pass noise filtered out.
1140 Default value is @var{o}.
1144 @subsection Commands
1146 This filter supports the following commands:
1148 @item sample_noise, sn
1149 Start or stop measuring noise profile.
1150 Syntax for the command is : "start" or "stop" string.
1151 After measuring noise profile is stopped it will be
1152 automatically applied in filtering.
1154 @item noise_reduction, nr
1155 Change noise reduction. Argument is single float number.
1156 Syntax for the command is : "@var{noise_reduction}"
1158 @item noise_floor, nf
1159 Change noise floor. Argument is single float number.
1160 Syntax for the command is : "@var{noise_floor}"
1162 @item output_mode, om
1163 Change output mode operation.
1164 Syntax for the command is : "i", "o" or "n" string.
1168 Apply arbitrary expressions to samples in frequency domain.
1172 Set frequency domain real expression for each separate channel separated
1173 by '|'. Default is "re".
1174 If the number of input channels is greater than the number of
1175 expressions, the last specified expression is used for the remaining
1179 Set frequency domain imaginary expression for each separate channel
1180 separated by '|'. Default is "im".
1182 Each expression in @var{real} and @var{imag} can contain the following
1183 constants and functions:
1190 current frequency bin number
1193 number of available bins
1196 channel number of the current expression
1205 current real part of frequency bin of current channel
1208 current imaginary part of frequency bin of current channel
1211 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1214 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1218 Set window size. Allowed range is from 16 to 131072.
1219 Default is @code{4096}
1222 Set window function. Default is @code{hann}.
1225 Set window overlap. If set to 1, the recommended overlap for selected
1226 window function will be picked. Default is @code{0.75}.
1229 @subsection Examples
1233 Leave almost only low frequencies in audio:
1235 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1239 Apply robotize effect:
1241 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1245 Apply whisper effect:
1247 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"
1254 Apply an arbitrary Finite Impulse Response filter.
1256 This filter is designed for applying long FIR filters,
1257 up to 60 seconds long.
1259 It can be used as component for digital crossover filters,
1260 room equalization, cross talk cancellation, wavefield synthesis,
1261 auralization, ambiophonics, ambisonics and spatialization.
1263 This filter uses the streams higher than first one as FIR coefficients.
1264 If the non-first stream holds a single channel, it will be used
1265 for all input channels in the first stream, otherwise
1266 the number of channels in the non-first stream must be same as
1267 the number of channels in the first stream.
1269 It accepts the following parameters:
1273 Set dry gain. This sets input gain.
1276 Set wet gain. This sets final output gain.
1279 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1282 Enable applying gain measured from power of IR.
1284 Set which approach to use for auto gain measurement.
1288 Do not apply any gain.
1291 select peak gain, very conservative approach. This is default value.
1294 select DC gain, limited application.
1297 select gain to noise approach, this is most popular one.
1301 Set gain to be applied to IR coefficients before filtering.
1302 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1305 Set format of IR stream. Can be @code{mono} or @code{input}.
1306 Default is @code{input}.
1309 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1310 Allowed range is 0.1 to 60 seconds.
1313 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1314 By default it is disabled.
1317 Set for which IR channel to display frequency response. By default is first channel
1318 displayed. This option is used only when @var{response} is enabled.
1321 Set video stream size. This option is used only when @var{response} is enabled.
1324 Set video stream frame rate. This option is used only when @var{response} is enabled.
1327 Set minimal partition size used for convolution. Default is @var{8192}.
1328 Allowed range is from @var{1} to @var{32768}.
1329 Lower values decreases latency at cost of higher CPU usage.
1332 Set maximal partition size used for convolution. Default is @var{8192}.
1333 Allowed range is from @var{8} to @var{32768}.
1334 Lower values may increase CPU usage.
1337 Set number of input impulse responses streams which will be switchable at runtime.
1338 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1341 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1342 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1343 This option can be changed at runtime via @ref{commands}.
1346 @subsection Examples
1350 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1352 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1359 Set output format constraints for the input audio. The framework will
1360 negotiate the most appropriate format to minimize conversions.
1362 It accepts the following parameters:
1365 @item sample_fmts, f
1366 A '|'-separated list of requested sample formats.
1368 @item sample_rates, r
1369 A '|'-separated list of requested sample rates.
1371 @item channel_layouts, cl
1372 A '|'-separated list of requested channel layouts.
1374 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1375 for the required syntax.
1378 If a parameter is omitted, all values are allowed.
1380 Force the output to either unsigned 8-bit or signed 16-bit stereo
1382 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1386 Apply frequency shift to input audio samples.
1388 The filter accepts the following options:
1392 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1393 Default value is 0.0.
1396 @subsection Commands
1398 This filter supports the above option as @ref{commands}.
1402 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1403 processing reduces disturbing noise between useful signals.
1405 Gating is done by detecting the volume below a chosen level @var{threshold}
1406 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1407 floor is set via @var{range}. Because an exact manipulation of the signal
1408 would cause distortion of the waveform the reduction can be levelled over
1409 time. This is done by setting @var{attack} and @var{release}.
1411 @var{attack} determines how long the signal has to fall below the threshold
1412 before any reduction will occur and @var{release} sets the time the signal
1413 has to rise above the threshold to reduce the reduction again.
1414 Shorter signals than the chosen attack time will be left untouched.
1418 Set input level before filtering.
1419 Default is 1. Allowed range is from 0.015625 to 64.
1422 Set the mode of operation. Can be @code{upward} or @code{downward}.
1423 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1424 will be amplified, expanding dynamic range in upward direction.
1425 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1428 Set the level of gain reduction when the signal is below the threshold.
1429 Default is 0.06125. Allowed range is from 0 to 1.
1430 Setting this to 0 disables reduction and then filter behaves like expander.
1433 If a signal rises above this level the gain reduction is released.
1434 Default is 0.125. Allowed range is from 0 to 1.
1437 Set a ratio by which the signal is reduced.
1438 Default is 2. Allowed range is from 1 to 9000.
1441 Amount of milliseconds the signal has to rise above the threshold before gain
1443 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1446 Amount of milliseconds the signal has to fall below the threshold before the
1447 reduction is increased again. Default is 250 milliseconds.
1448 Allowed range is from 0.01 to 9000.
1451 Set amount of amplification of signal after processing.
1452 Default is 1. Allowed range is from 1 to 64.
1455 Curve the sharp knee around the threshold to enter gain reduction more softly.
1456 Default is 2.828427125. Allowed range is from 1 to 8.
1459 Choose if exact signal should be taken for detection or an RMS like one.
1460 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1463 Choose if the average level between all channels or the louder channel affects
1465 Default is @code{average}. Can be @code{average} or @code{maximum}.
1470 Apply an arbitrary Infinite Impulse Response filter.
1472 It accepts the following parameters:
1476 Set B/numerator/zeros/reflection coefficients.
1479 Set A/denominator/poles/ladder coefficients.
1491 Set coefficients format.
1495 lattice-ladder function
1497 analog transfer function
1499 digital transfer function
1501 Z-plane zeros/poles, cartesian (default)
1503 Z-plane zeros/poles, polar radians
1505 Z-plane zeros/poles, polar degrees
1511 Set type of processing.
1523 Set filtering precision.
1527 double-precision floating-point (default)
1529 single-precision floating-point
1537 Normalize filter coefficients, by default is enabled.
1538 Enabling it will normalize magnitude response at DC to 0dB.
1541 How much to use filtered signal in output. Default is 1.
1542 Range is between 0 and 1.
1545 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1546 By default it is disabled.
1549 Set for which IR channel to display frequency response. By default is first channel
1550 displayed. This option is used only when @var{response} is enabled.
1553 Set video stream size. This option is used only when @var{response} is enabled.
1556 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1559 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1560 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1563 Different coefficients and gains can be provided for every channel, in such case
1564 use '|' to separate coefficients or gains. Last provided coefficients will be
1565 used for all remaining channels.
1567 @subsection Examples
1571 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1573 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
1577 Same as above but in @code{zp} format:
1579 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
1583 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1585 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1591 The limiter prevents an input signal from rising over a desired threshold.
1592 This limiter uses lookahead technology to prevent your signal from distorting.
1593 It means that there is a small delay after the signal is processed. Keep in mind
1594 that the delay it produces is the attack time you set.
1596 The filter accepts the following options:
1600 Set input gain. Default is 1.
1603 Set output gain. Default is 1.
1606 Don't let signals above this level pass the limiter. Default is 1.
1609 The limiter will reach its attenuation level in this amount of time in
1610 milliseconds. Default is 5 milliseconds.
1613 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1614 Default is 50 milliseconds.
1617 When gain reduction is always needed ASC takes care of releasing to an
1618 average reduction level rather than reaching a reduction of 0 in the release
1622 Select how much the release time is affected by ASC, 0 means nearly no changes
1623 in release time while 1 produces higher release times.
1626 Auto level output signal. Default is enabled.
1627 This normalizes audio back to 0dB if enabled.
1630 Depending on picked setting it is recommended to upsample input 2x or 4x times
1631 with @ref{aresample} before applying this filter.
1635 Apply a two-pole all-pass filter with central frequency (in Hz)
1636 @var{frequency}, and filter-width @var{width}.
1637 An all-pass filter changes the audio's frequency to phase relationship
1638 without changing its frequency to amplitude relationship.
1640 The filter accepts the following options:
1644 Set frequency in Hz.
1647 Set method to specify band-width of filter.
1662 Specify the band-width of a filter in width_type units.
1665 How much to use filtered signal in output. Default is 1.
1666 Range is between 0 and 1.
1669 Specify which channels to filter, by default all available are filtered.
1672 Normalize biquad coefficients, by default is disabled.
1673 Enabling it will normalize magnitude response at DC to 0dB.
1676 Set the filter order, can be 1 or 2. Default is 2.
1679 Set transform type of IIR filter.
1688 @subsection Commands
1690 This filter supports the following commands:
1693 Change allpass frequency.
1694 Syntax for the command is : "@var{frequency}"
1697 Change allpass width_type.
1698 Syntax for the command is : "@var{width_type}"
1701 Change allpass width.
1702 Syntax for the command is : "@var{width}"
1706 Syntax for the command is : "@var{mix}"
1713 The filter accepts the following options:
1717 Set the number of loops. Setting this value to -1 will result in infinite loops.
1721 Set maximal number of samples. Default is 0.
1724 Set first sample of loop. Default is 0.
1730 Merge two or more audio streams into a single multi-channel stream.
1732 The filter accepts the following options:
1737 Set the number of inputs. Default is 2.
1741 If the channel layouts of the inputs are disjoint, and therefore compatible,
1742 the channel layout of the output will be set accordingly and the channels
1743 will be reordered as necessary. If the channel layouts of the inputs are not
1744 disjoint, the output will have all the channels of the first input then all
1745 the channels of the second input, in that order, and the channel layout of
1746 the output will be the default value corresponding to the total number of
1749 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1750 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1751 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1752 first input, b1 is the first channel of the second input).
1754 On the other hand, if both input are in stereo, the output channels will be
1755 in the default order: a1, a2, b1, b2, and the channel layout will be
1756 arbitrarily set to 4.0, which may or may not be the expected value.
1758 All inputs must have the same sample rate, and format.
1760 If inputs do not have the same duration, the output will stop with the
1763 @subsection Examples
1767 Merge two mono files into a stereo stream:
1769 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1773 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1775 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
1781 Mixes multiple audio inputs into a single output.
1783 Note that this filter only supports float samples (the @var{amerge}
1784 and @var{pan} audio filters support many formats). If the @var{amix}
1785 input has integer samples then @ref{aresample} will be automatically
1786 inserted to perform the conversion to float samples.
1790 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1792 will mix 3 input audio streams to a single output with the same duration as the
1793 first input and a dropout transition time of 3 seconds.
1795 It accepts the following parameters:
1799 The number of inputs. If unspecified, it defaults to 2.
1802 How to determine the end-of-stream.
1806 The duration of the longest input. (default)
1809 The duration of the shortest input.
1812 The duration of the first input.
1816 @item dropout_transition
1817 The transition time, in seconds, for volume renormalization when an input
1818 stream ends. The default value is 2 seconds.
1821 Specify weight of each input audio stream as sequence.
1822 Each weight is separated by space. By default all inputs have same weight.
1825 @subsection Commands
1827 This filter supports the following commands:
1830 Syntax is same as option with same name.
1835 Multiply first audio stream with second audio stream and store result
1836 in output audio stream. Multiplication is done by multiplying each
1837 sample from first stream with sample at same position from second stream.
1839 With this element-wise multiplication one can create amplitude fades and
1840 amplitude modulations.
1842 @section anequalizer
1844 High-order parametric multiband equalizer for each channel.
1846 It accepts the following parameters:
1850 This option string is in format:
1851 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1852 Each equalizer band is separated by '|'.
1856 Set channel number to which equalization will be applied.
1857 If input doesn't have that channel the entry is ignored.
1860 Set central frequency for band.
1861 If input doesn't have that frequency the entry is ignored.
1864 Set band width in Hertz.
1867 Set band gain in dB.
1870 Set filter type for band, optional, can be:
1874 Butterworth, this is default.
1885 With this option activated frequency response of anequalizer is displayed
1889 Set video stream size. Only useful if curves option is activated.
1892 Set max gain that will be displayed. Only useful if curves option is activated.
1893 Setting this to a reasonable value makes it possible to display gain which is derived from
1894 neighbour bands which are too close to each other and thus produce higher gain
1895 when both are activated.
1898 Set frequency scale used to draw frequency response in video output.
1899 Can be linear or logarithmic. Default is logarithmic.
1902 Set color for each channel curve which is going to be displayed in video stream.
1903 This is list of color names separated by space or by '|'.
1904 Unrecognised or missing colors will be replaced by white color.
1907 @subsection Examples
1911 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1912 for first 2 channels using Chebyshev type 1 filter:
1914 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1918 @subsection Commands
1920 This filter supports the following commands:
1923 Alter existing filter parameters.
1924 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1926 @var{fN} is existing filter number, starting from 0, if no such filter is available
1928 @var{freq} set new frequency parameter.
1929 @var{width} set new width parameter in Hertz.
1930 @var{gain} set new gain parameter in dB.
1932 Full filter invocation with asendcmd may look like this:
1933 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1938 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1940 Each sample is adjusted by looking for other samples with similar contexts. This
1941 context similarity is defined by comparing their surrounding patches of size
1942 @option{p}. Patches are searched in an area of @option{r} around the sample.
1944 The filter accepts the following options:
1948 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1951 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1952 Default value is 2 milliseconds.
1955 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1956 Default value is 6 milliseconds.
1959 Set the output mode.
1961 It accepts the following values:
1964 Pass input unchanged.
1967 Pass noise filtered out.
1972 Default value is @var{o}.
1976 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
1979 @subsection Commands
1981 This filter supports the all above options as @ref{commands}.
1984 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
1986 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
1987 relate to producing the least mean square of the error signal (difference between the desired,
1988 2nd input audio stream and the actual signal, the 1st input audio stream).
1990 A description of the accepted options follows.
2003 Set the filter leakage.
2006 It accepts the following values:
2015 Pass filtered samples.
2018 Pass difference between desired and filtered samples.
2020 Default value is @var{o}.
2024 @subsection Examples
2028 One of many usages of this filter is noise reduction, input audio is filtered
2029 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2031 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2035 @subsection Commands
2037 This filter supports the same commands as options, excluding option @code{order}.
2041 Pass the audio source unchanged to the output.
2045 Pad the end of an audio stream with silence.
2047 This can be used together with @command{ffmpeg} @option{-shortest} to
2048 extend audio streams to the same length as the video stream.
2050 A description of the accepted options follows.
2054 Set silence packet size. Default value is 4096.
2057 Set the number of samples of silence to add to the end. After the
2058 value is reached, the stream is terminated. This option is mutually
2059 exclusive with @option{whole_len}.
2062 Set the minimum total number of samples in the output audio stream. If
2063 the value is longer than the input audio length, silence is added to
2064 the end, until the value is reached. This option is mutually exclusive
2065 with @option{pad_len}.
2068 Specify the duration of samples of silence to add. See
2069 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2070 for the accepted syntax. Used only if set to non-zero value.
2073 Specify the minimum total duration in the output audio stream. See
2074 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2075 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2076 the input audio length, silence is added to the end, until the value is reached.
2077 This option is mutually exclusive with @option{pad_dur}
2080 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2081 nor @option{whole_dur} option is set, the filter will add silence to the end of
2082 the input stream indefinitely.
2084 @subsection Examples
2088 Add 1024 samples of silence to the end of the input:
2094 Make sure the audio output will contain at least 10000 samples, pad
2095 the input with silence if required:
2097 apad=whole_len=10000
2101 Use @command{ffmpeg} to pad the audio input with silence, so that the
2102 video stream will always result the shortest and will be converted
2103 until the end in the output file when using the @option{shortest}
2106 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2111 Add a phasing effect to the input audio.
2113 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2114 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2116 A description of the accepted parameters follows.
2120 Set input gain. Default is 0.4.
2123 Set output gain. Default is 0.74
2126 Set delay in milliseconds. Default is 3.0.
2129 Set decay. Default is 0.4.
2132 Set modulation speed in Hz. Default is 0.5.
2135 Set modulation type. Default is triangular.
2137 It accepts the following values:
2144 @section aphaseshift
2145 Apply phase shift to input audio samples.
2147 The filter accepts the following options:
2151 Specify phase shift. Allowed range is from -1.0 to 1.0.
2152 Default value is 0.0.
2155 @subsection Commands
2157 This filter supports the above option as @ref{commands}.
2161 Audio pulsator is something between an autopanner and a tremolo.
2162 But it can produce funny stereo effects as well. Pulsator changes the volume
2163 of the left and right channel based on a LFO (low frequency oscillator) with
2164 different waveforms and shifted phases.
2165 This filter have the ability to define an offset between left and right
2166 channel. An offset of 0 means that both LFO shapes match each other.
2167 The left and right channel are altered equally - a conventional tremolo.
2168 An offset of 50% means that the shape of the right channel is exactly shifted
2169 in phase (or moved backwards about half of the frequency) - pulsator acts as
2170 an autopanner. At 1 both curves match again. Every setting in between moves the
2171 phase shift gapless between all stages and produces some "bypassing" sounds with
2172 sine and triangle waveforms. The more you set the offset near 1 (starting from
2173 the 0.5) the faster the signal passes from the left to the right speaker.
2175 The filter accepts the following options:
2179 Set input gain. By default it is 1. Range is [0.015625 - 64].
2182 Set output gain. By default it is 1. Range is [0.015625 - 64].
2185 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2186 sawup or sawdown. Default is sine.
2189 Set modulation. Define how much of original signal is affected by the LFO.
2192 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2195 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2198 Set pulse width. Default is 1. Allowed range is [0 - 2].
2201 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2204 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2208 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2212 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2213 if timing is set to hz.
2219 Resample the input audio to the specified parameters, using the
2220 libswresample library. If none are specified then the filter will
2221 automatically convert between its input and output.
2223 This filter is also able to stretch/squeeze the audio data to make it match
2224 the timestamps or to inject silence / cut out audio to make it match the
2225 timestamps, do a combination of both or do neither.
2227 The filter accepts the syntax
2228 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2229 expresses a sample rate and @var{resampler_options} is a list of
2230 @var{key}=@var{value} pairs, separated by ":". See the
2231 @ref{Resampler Options,,"Resampler Options" section in the
2232 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2233 for the complete list of supported options.
2235 @subsection Examples
2239 Resample the input audio to 44100Hz:
2245 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2246 samples per second compensation:
2248 aresample=async=1000
2254 Reverse an audio clip.
2256 Warning: This filter requires memory to buffer the entire clip, so trimming
2259 @subsection Examples
2263 Take the first 5 seconds of a clip, and reverse it.
2265 atrim=end=5,areverse
2271 Reduce noise from speech using Recurrent Neural Networks.
2273 This filter accepts the following options:
2277 Set train model file to load. This option is always required.
2280 @section asetnsamples
2282 Set the number of samples per each output audio frame.
2284 The last output packet may contain a different number of samples, as
2285 the filter will flush all the remaining samples when the input audio
2288 The filter accepts the following options:
2292 @item nb_out_samples, n
2293 Set the number of frames per each output audio frame. The number is
2294 intended as the number of samples @emph{per each channel}.
2295 Default value is 1024.
2298 If set to 1, the filter will pad the last audio frame with zeroes, so
2299 that the last frame will contain the same number of samples as the
2300 previous ones. Default value is 1.
2303 For example, to set the number of per-frame samples to 1234 and
2304 disable padding for the last frame, use:
2306 asetnsamples=n=1234:p=0
2311 Set the sample rate without altering the PCM data.
2312 This will result in a change of speed and pitch.
2314 The filter accepts the following options:
2317 @item sample_rate, r
2318 Set the output sample rate. Default is 44100 Hz.
2323 Show a line containing various information for each input audio frame.
2324 The input audio is not modified.
2326 The shown line contains a sequence of key/value pairs of the form
2327 @var{key}:@var{value}.
2329 The following values are shown in the output:
2333 The (sequential) number of the input frame, starting from 0.
2336 The presentation timestamp of the input frame, in time base units; the time base
2337 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2340 The presentation timestamp of the input frame in seconds.
2343 position of the frame in the input stream, -1 if this information in
2344 unavailable and/or meaningless (for example in case of synthetic audio)
2353 The sample rate for the audio frame.
2356 The number of samples (per channel) in the frame.
2359 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2360 audio, the data is treated as if all the planes were concatenated.
2362 @item plane_checksums
2363 A list of Adler-32 checksums for each data plane.
2367 Apply audio soft clipping.
2369 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2370 along a smooth curve, rather than the abrupt shape of hard-clipping.
2372 This filter accepts the following options:
2376 Set type of soft-clipping.
2378 It accepts the following values:
2392 Set additional parameter which controls sigmoid function.
2395 Set oversampling factor.
2398 @subsection Commands
2400 This filter supports the all above options as @ref{commands}.
2403 Automatic Speech Recognition
2405 This filter uses PocketSphinx for speech recognition. To enable
2406 compilation of this filter, you need to configure FFmpeg with
2407 @code{--enable-pocketsphinx}.
2409 It accepts the following options:
2413 Set sampling rate of input audio. Defaults is @code{16000}.
2414 This need to match speech models, otherwise one will get poor results.
2417 Set dictionary containing acoustic model files.
2420 Set pronunciation dictionary.
2423 Set language model file.
2426 Set language model set.
2429 Set which language model to use.
2432 Set output for log messages.
2435 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2440 Display time domain statistical information about the audio channels.
2441 Statistics are calculated and displayed for each audio channel and,
2442 where applicable, an overall figure is also given.
2444 It accepts the following option:
2447 Short window length in seconds, used for peak and trough RMS measurement.
2448 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2452 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2453 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2456 Available keys for each channel are:
2502 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2503 this @code{lavfi.astats.Overall.Peak_count}.
2505 For description what each key means read below.
2508 Set number of frame after which stats are going to be recalculated.
2509 Default is disabled.
2511 @item measure_perchannel
2512 Select the entries which need to be measured per channel. The metadata keys can
2513 be used as flags, default is @option{all} which measures everything.
2514 @option{none} disables all per channel measurement.
2516 @item measure_overall
2517 Select the entries which need to be measured overall. The metadata keys can
2518 be used as flags, default is @option{all} which measures everything.
2519 @option{none} disables all overall measurement.
2523 A description of each shown parameter follows:
2527 Mean amplitude displacement from zero.
2530 Minimal sample level.
2533 Maximal sample level.
2535 @item Min difference
2536 Minimal difference between two consecutive samples.
2538 @item Max difference
2539 Maximal difference between two consecutive samples.
2541 @item Mean difference
2542 Mean difference between two consecutive samples.
2543 The average of each difference between two consecutive samples.
2545 @item RMS difference
2546 Root Mean Square difference between two consecutive samples.
2550 Standard peak and RMS level measured in dBFS.
2554 Peak and trough values for RMS level measured over a short window.
2557 Standard ratio of peak to RMS level (note: not in dB).
2560 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2561 (i.e. either @var{Min level} or @var{Max level}).
2564 Number of occasions (not the number of samples) that the signal attained either
2565 @var{Min level} or @var{Max level}.
2567 @item Noise floor dB
2568 Minimum local peak measured in dBFS over a short window.
2570 @item Noise floor count
2571 Number of occasions (not the number of samples) that the signal attained
2575 Overall bit depth of audio. Number of bits used for each sample.
2578 Measured dynamic range of audio in dB.
2580 @item Zero crossings
2581 Number of points where the waveform crosses the zero level axis.
2583 @item Zero crossings rate
2584 Rate of Zero crossings and number of audio samples.
2588 Boost subwoofer frequencies.
2590 The filter accepts the following options:
2594 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2595 Default value is 0.5.
2598 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2599 Default value is 0.8.
2602 Set delay line decay gain value. Allowed range is from 0 to 1.
2603 Default value is 0.7.
2606 Set delay line feedback gain value. Allowed range is from 0 to 1.
2607 Default value is 0.5.
2610 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2611 Default value is 100.
2614 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2615 Default value is 0.5.
2618 Set delay. Allowed range is from 1 to 100.
2619 Default value is 20.
2622 @subsection Commands
2624 This filter supports the all above options as @ref{commands}.
2627 Cut super frequencies.
2629 The filter accepts the following options:
2633 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2634 Default value is 20000.
2637 @subsection Commands
2639 This filter supports the all above options as @ref{commands}.
2645 The filter accepts exactly one parameter, the audio tempo. If not
2646 specified then the filter will assume nominal 1.0 tempo. Tempo must
2647 be in the [0.5, 100.0] range.
2649 Note that tempo greater than 2 will skip some samples rather than
2650 blend them in. If for any reason this is a concern it is always
2651 possible to daisy-chain several instances of atempo to achieve the
2652 desired product tempo.
2654 @subsection Examples
2658 Slow down audio to 80% tempo:
2664 To speed up audio to 300% tempo:
2670 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2672 atempo=sqrt(3),atempo=sqrt(3)
2676 @subsection Commands
2678 This filter supports the following commands:
2681 Change filter tempo scale factor.
2682 Syntax for the command is : "@var{tempo}"
2687 Trim the input so that the output contains one continuous subpart of the input.
2689 It accepts the following parameters:
2692 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2693 sample with the timestamp @var{start} will be the first sample in the output.
2696 Specify time of the first audio sample that will be dropped, i.e. the
2697 audio sample immediately preceding the one with the timestamp @var{end} will be
2698 the last sample in the output.
2701 Same as @var{start}, except this option sets the start timestamp in samples
2705 Same as @var{end}, except this option sets the end timestamp in samples instead
2709 The maximum duration of the output in seconds.
2712 The number of the first sample that should be output.
2715 The number of the first sample that should be dropped.
2718 @option{start}, @option{end}, and @option{duration} are expressed as time
2719 duration specifications; see
2720 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2722 Note that the first two sets of the start/end options and the @option{duration}
2723 option look at the frame timestamp, while the _sample options simply count the
2724 samples that pass through the filter. So start/end_pts and start/end_sample will
2725 give different results when the timestamps are wrong, inexact or do not start at
2726 zero. Also note that this filter does not modify the timestamps. If you wish
2727 to have the output timestamps start at zero, insert the asetpts filter after the
2730 If multiple start or end options are set, this filter tries to be greedy and
2731 keep all samples that match at least one of the specified constraints. To keep
2732 only the part that matches all the constraints at once, chain multiple atrim
2735 The defaults are such that all the input is kept. So it is possible to set e.g.
2736 just the end values to keep everything before the specified time.
2741 Drop everything except the second minute of input:
2743 ffmpeg -i INPUT -af atrim=60:120
2747 Keep only the first 1000 samples:
2749 ffmpeg -i INPUT -af atrim=end_sample=1000
2754 @section axcorrelate
2755 Calculate normalized cross-correlation between two input audio streams.
2757 Resulted samples are always between -1 and 1 inclusive.
2758 If result is 1 it means two input samples are highly correlated in that selected segment.
2759 Result 0 means they are not correlated at all.
2760 If result is -1 it means two input samples are out of phase, which means they cancel each
2763 The filter accepts the following options:
2767 Set size of segment over which cross-correlation is calculated.
2768 Default is 256. Allowed range is from 2 to 131072.
2771 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2772 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2773 are always zero and thus need much less calculations to make.
2774 This is generally not true, but is valid for typical audio streams.
2777 @subsection Examples
2781 Calculate correlation between channels in stereo audio stream:
2783 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2789 Apply a two-pole Butterworth band-pass filter with central
2790 frequency @var{frequency}, and (3dB-point) band-width width.
2791 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2792 instead of the default: constant 0dB peak gain.
2793 The filter roll off at 6dB per octave (20dB per decade).
2795 The filter accepts the following options:
2799 Set the filter's central frequency. Default is @code{3000}.
2802 Constant skirt gain if set to 1. Defaults to 0.
2805 Set method to specify band-width of filter.
2820 Specify the band-width of a filter in width_type units.
2823 How much to use filtered signal in output. Default is 1.
2824 Range is between 0 and 1.
2827 Specify which channels to filter, by default all available are filtered.
2830 Normalize biquad coefficients, by default is disabled.
2831 Enabling it will normalize magnitude response at DC to 0dB.
2834 Set transform type of IIR filter.
2843 @subsection Commands
2845 This filter supports the following commands:
2848 Change bandpass frequency.
2849 Syntax for the command is : "@var{frequency}"
2852 Change bandpass width_type.
2853 Syntax for the command is : "@var{width_type}"
2856 Change bandpass width.
2857 Syntax for the command is : "@var{width}"
2860 Change bandpass mix.
2861 Syntax for the command is : "@var{mix}"
2866 Apply a two-pole Butterworth band-reject filter with central
2867 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2868 The filter roll off at 6dB per octave (20dB per decade).
2870 The filter accepts the following options:
2874 Set the filter's central frequency. Default is @code{3000}.
2877 Set method to specify band-width of filter.
2892 Specify the band-width of a filter in width_type units.
2895 How much to use filtered signal in output. Default is 1.
2896 Range is between 0 and 1.
2899 Specify which channels to filter, by default all available are filtered.
2902 Normalize biquad coefficients, by default is disabled.
2903 Enabling it will normalize magnitude response at DC to 0dB.
2906 Set transform type of IIR filter.
2915 @subsection Commands
2917 This filter supports the following commands:
2920 Change bandreject frequency.
2921 Syntax for the command is : "@var{frequency}"
2924 Change bandreject width_type.
2925 Syntax for the command is : "@var{width_type}"
2928 Change bandreject width.
2929 Syntax for the command is : "@var{width}"
2932 Change bandreject mix.
2933 Syntax for the command is : "@var{mix}"
2936 @section bass, lowshelf
2938 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2939 shelving filter with a response similar to that of a standard
2940 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2942 The filter accepts the following options:
2946 Give the gain at 0 Hz. Its useful range is about -20
2947 (for a large cut) to +20 (for a large boost).
2948 Beware of clipping when using a positive gain.
2951 Set the filter's central frequency and so can be used
2952 to extend or reduce the frequency range to be boosted or cut.
2953 The default value is @code{100} Hz.
2956 Set method to specify band-width of filter.
2971 Determine how steep is the filter's shelf transition.
2974 How much to use filtered signal in output. Default is 1.
2975 Range is between 0 and 1.
2978 Specify which channels to filter, by default all available are filtered.
2981 Normalize biquad coefficients, by default is disabled.
2982 Enabling it will normalize magnitude response at DC to 0dB.
2985 Set transform type of IIR filter.
2994 @subsection Commands
2996 This filter supports the following commands:
2999 Change bass frequency.
3000 Syntax for the command is : "@var{frequency}"
3003 Change bass width_type.
3004 Syntax for the command is : "@var{width_type}"
3008 Syntax for the command is : "@var{width}"
3012 Syntax for the command is : "@var{gain}"
3016 Syntax for the command is : "@var{mix}"
3021 Apply a biquad IIR filter with the given coefficients.
3022 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3023 are the numerator and denominator coefficients respectively.
3024 and @var{channels}, @var{c} specify which channels to filter, by default all
3025 available are filtered.
3027 @subsection Commands
3029 This filter supports the following commands:
3037 Change biquad parameter.
3038 Syntax for the command is : "@var{value}"
3041 How much to use filtered signal in output. Default is 1.
3042 Range is between 0 and 1.
3045 Specify which channels to filter, by default all available are filtered.
3048 Normalize biquad coefficients, by default is disabled.
3049 Enabling it will normalize magnitude response at DC to 0dB.
3052 Set transform type of IIR filter.
3062 Bauer stereo to binaural transformation, which improves headphone listening of
3063 stereo audio records.
3065 To enable compilation of this filter you need to configure FFmpeg with
3066 @code{--enable-libbs2b}.
3068 It accepts the following parameters:
3072 Pre-defined crossfeed level.
3076 Default level (fcut=700, feed=50).
3079 Chu Moy circuit (fcut=700, feed=60).
3082 Jan Meier circuit (fcut=650, feed=95).
3087 Cut frequency (in Hz).
3096 Remap input channels to new locations.
3098 It accepts the following parameters:
3101 Map channels from input to output. The argument is a '|'-separated list of
3102 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3103 @var{in_channel} form. @var{in_channel} can be either the name of the input
3104 channel (e.g. FL for front left) or its index in the input channel layout.
3105 @var{out_channel} is the name of the output channel or its index in the output
3106 channel layout. If @var{out_channel} is not given then it is implicitly an
3107 index, starting with zero and increasing by one for each mapping.
3109 @item channel_layout
3110 The channel layout of the output stream.
3113 If no mapping is present, the filter will implicitly map input channels to
3114 output channels, preserving indices.
3116 @subsection Examples
3120 For example, assuming a 5.1+downmix input MOV file,
3122 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3124 will create an output WAV file tagged as stereo from the downmix channels of
3128 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3130 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3134 @section channelsplit
3136 Split each channel from an input audio stream into a separate output stream.
3138 It accepts the following parameters:
3140 @item channel_layout
3141 The channel layout of the input stream. The default is "stereo".
3143 A channel layout describing the channels to be extracted as separate output streams
3144 or "all" to extract each input channel as a separate stream. The default is "all".
3146 Choosing channels not present in channel layout in the input will result in an error.
3149 @subsection Examples
3153 For example, assuming a stereo input MP3 file,
3155 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3157 will create an output Matroska file with two audio streams, one containing only
3158 the left channel and the other the right channel.
3161 Split a 5.1 WAV file into per-channel files:
3163 ffmpeg -i in.wav -filter_complex
3164 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3165 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3166 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3171 Extract only LFE from a 5.1 WAV file:
3173 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3174 -map '[LFE]' lfe.wav
3179 Add a chorus effect to the audio.
3181 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3183 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3184 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3185 The modulation depth defines the range the modulated delay is played before or after
3186 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3187 sound tuned around the original one, like in a chorus where some vocals are slightly
3190 It accepts the following parameters:
3193 Set input gain. Default is 0.4.
3196 Set output gain. Default is 0.4.
3199 Set delays. A typical delay is around 40ms to 60ms.
3211 @subsection Examples
3217 chorus=0.7:0.9:55:0.4:0.25:2
3223 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3227 Fuller sounding chorus with three delays:
3229 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
3234 Compress or expand the audio's dynamic range.
3236 It accepts the following parameters:
3242 A list of times in seconds for each channel over which the instantaneous level
3243 of the input signal is averaged to determine its volume. @var{attacks} refers to
3244 increase of volume and @var{decays} refers to decrease of volume. For most
3245 situations, the attack time (response to the audio getting louder) should be
3246 shorter than the decay time, because the human ear is more sensitive to sudden
3247 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3248 a typical value for decay is 0.8 seconds.
3249 If specified number of attacks & decays is lower than number of channels, the last
3250 set attack/decay will be used for all remaining channels.
3253 A list of points for the transfer function, specified in dB relative to the
3254 maximum possible signal amplitude. Each key points list must be defined using
3255 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3256 @code{x0/y0 x1/y1 x2/y2 ....}
3258 The input values must be in strictly increasing order but the transfer function
3259 does not have to be monotonically rising. The point @code{0/0} is assumed but
3260 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3261 function are @code{-70/-70|-60/-20|1/0}.
3264 Set the curve radius in dB for all joints. It defaults to 0.01.
3267 Set the additional gain in dB to be applied at all points on the transfer
3268 function. This allows for easy adjustment of the overall gain.
3272 Set an initial volume, in dB, to be assumed for each channel when filtering
3273 starts. This permits the user to supply a nominal level initially, so that, for
3274 example, a very large gain is not applied to initial signal levels before the
3275 companding has begun to operate. A typical value for audio which is initially
3276 quiet is -90 dB. It defaults to 0.
3279 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3280 delayed before being fed to the volume adjuster. Specifying a delay
3281 approximately equal to the attack/decay times allows the filter to effectively
3282 operate in predictive rather than reactive mode. It defaults to 0.
3286 @subsection Examples
3290 Make music with both quiet and loud passages suitable for listening to in a
3293 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3296 Another example for audio with whisper and explosion parts:
3298 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3302 A noise gate for when the noise is at a lower level than the signal:
3304 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3308 Here is another noise gate, this time for when the noise is at a higher level
3309 than the signal (making it, in some ways, similar to squelch):
3311 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3315 2:1 compression starting at -6dB:
3317 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3321 2:1 compression starting at -9dB:
3323 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3327 2:1 compression starting at -12dB:
3329 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3333 2:1 compression starting at -18dB:
3335 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3339 3:1 compression starting at -15dB:
3341 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3347 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3353 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
3357 Hard limiter at -6dB:
3359 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3363 Hard limiter at -12dB:
3365 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3369 Hard noise gate at -35 dB:
3371 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3377 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3381 @section compensationdelay
3383 Compensation Delay Line is a metric based delay to compensate differing
3384 positions of microphones or speakers.
3386 For example, you have recorded guitar with two microphones placed in
3387 different locations. Because the front of sound wave has fixed speed in
3388 normal conditions, the phasing of microphones can vary and depends on
3389 their location and interposition. The best sound mix can be achieved when
3390 these microphones are in phase (synchronized). Note that a distance of
3391 ~30 cm between microphones makes one microphone capture the signal in
3392 antiphase to the other microphone. That makes the final mix sound moody.
3393 This filter helps to solve phasing problems by adding different delays
3394 to each microphone track and make them synchronized.
3396 The best result can be reached when you take one track as base and
3397 synchronize other tracks one by one with it.
3398 Remember that synchronization/delay tolerance depends on sample rate, too.
3399 Higher sample rates will give more tolerance.
3401 The filter accepts the following parameters:
3405 Set millimeters distance. This is compensation distance for fine tuning.
3409 Set cm distance. This is compensation distance for tightening distance setup.
3413 Set meters distance. This is compensation distance for hard distance setup.
3417 Set dry amount. Amount of unprocessed (dry) signal.
3421 Set wet amount. Amount of processed (wet) signal.
3425 Set temperature in degrees Celsius. This is the temperature of the environment.
3430 Apply headphone crossfeed filter.
3432 Crossfeed is the process of blending the left and right channels of stereo
3434 It is mainly used to reduce extreme stereo separation of low frequencies.
3436 The intent is to produce more speaker like sound to the listener.
3438 The filter accepts the following options:
3442 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3443 This sets gain of low shelf filter for side part of stereo image.
3444 Default is -6dB. Max allowed is -30db when strength is set to 1.
3447 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3448 This sets cut off frequency of low shelf filter. Default is cut off near
3449 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3452 Set curve slope of low shelf filter. Default is 0.5.
3453 Allowed range is from 0.01 to 1.
3456 Set input gain. Default is 0.9.
3459 Set output gain. Default is 1.
3462 @subsection Commands
3464 This filter supports the all above options as @ref{commands}.
3466 @section crystalizer
3467 Simple algorithm to expand audio dynamic range.
3469 The filter accepts the following options:
3473 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3474 (unchanged sound) to 10.0 (maximum effect).
3477 Enable clipping. By default is enabled.
3480 @subsection Commands
3482 This filter supports the all above options as @ref{commands}.
3485 Apply a DC shift to the audio.
3487 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3488 in the recording chain) from the audio. The effect of a DC offset is reduced
3489 headroom and hence volume. The @ref{astats} filter can be used to determine if
3490 a signal has a DC offset.
3494 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3498 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3499 used to prevent clipping.
3504 Apply de-essing to the audio samples.
3508 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3512 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3516 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3520 Set the output mode.
3522 It accepts the following values:
3525 Pass input unchanged.
3528 Pass ess filtered out.
3533 Default value is @var{o}.
3539 Measure audio dynamic range.
3541 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3542 is found in transition material. And anything less that 8 have very poor dynamics
3543 and is very compressed.
3545 The filter accepts the following options:
3549 Set window length in seconds used to split audio into segments of equal length.
3550 Default is 3 seconds.
3554 Dynamic Audio Normalizer.
3556 This filter applies a certain amount of gain to the input audio in order
3557 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3558 contrast to more "simple" normalization algorithms, the Dynamic Audio
3559 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3560 This allows for applying extra gain to the "quiet" sections of the audio
3561 while avoiding distortions or clipping the "loud" sections. In other words:
3562 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3563 sections, in the sense that the volume of each section is brought to the
3564 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3565 this goal *without* applying "dynamic range compressing". It will retain 100%
3566 of the dynamic range *within* each section of the audio file.
3570 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3571 Default is 500 milliseconds.
3572 The Dynamic Audio Normalizer processes the input audio in small chunks,
3573 referred to as frames. This is required, because a peak magnitude has no
3574 meaning for just a single sample value. Instead, we need to determine the
3575 peak magnitude for a contiguous sequence of sample values. While a "standard"
3576 normalizer would simply use the peak magnitude of the complete file, the
3577 Dynamic Audio Normalizer determines the peak magnitude individually for each
3578 frame. The length of a frame is specified in milliseconds. By default, the
3579 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3580 been found to give good results with most files.
3581 Note that the exact frame length, in number of samples, will be determined
3582 automatically, based on the sampling rate of the individual input audio file.
3585 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3586 number. Default is 31.
3587 Probably the most important parameter of the Dynamic Audio Normalizer is the
3588 @code{window size} of the Gaussian smoothing filter. The filter's window size
3589 is specified in frames, centered around the current frame. For the sake of
3590 simplicity, this must be an odd number. Consequently, the default value of 31
3591 takes into account the current frame, as well as the 15 preceding frames and
3592 the 15 subsequent frames. Using a larger window results in a stronger
3593 smoothing effect and thus in less gain variation, i.e. slower gain
3594 adaptation. Conversely, using a smaller window results in a weaker smoothing
3595 effect and thus in more gain variation, i.e. faster gain adaptation.
3596 In other words, the more you increase this value, the more the Dynamic Audio
3597 Normalizer will behave like a "traditional" normalization filter. On the
3598 contrary, the more you decrease this value, the more the Dynamic Audio
3599 Normalizer will behave like a dynamic range compressor.
3602 Set the target peak value. This specifies the highest permissible magnitude
3603 level for the normalized audio input. This filter will try to approach the
3604 target peak magnitude as closely as possible, but at the same time it also
3605 makes sure that the normalized signal will never exceed the peak magnitude.
3606 A frame's maximum local gain factor is imposed directly by the target peak
3607 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3608 It is not recommended to go above this value.
3611 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3612 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3613 factor for each input frame, i.e. the maximum gain factor that does not
3614 result in clipping or distortion. The maximum gain factor is determined by
3615 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3616 additionally bounds the frame's maximum gain factor by a predetermined
3617 (global) maximum gain factor. This is done in order to avoid excessive gain
3618 factors in "silent" or almost silent frames. By default, the maximum gain
3619 factor is 10.0, For most inputs the default value should be sufficient and
3620 it usually is not recommended to increase this value. Though, for input
3621 with an extremely low overall volume level, it may be necessary to allow even
3622 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3623 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3624 Instead, a "sigmoid" threshold function will be applied. This way, the
3625 gain factors will smoothly approach the threshold value, but never exceed that
3629 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3630 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3631 This means that the maximum local gain factor for each frame is defined
3632 (only) by the frame's highest magnitude sample. This way, the samples can
3633 be amplified as much as possible without exceeding the maximum signal
3634 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3635 Normalizer can also take into account the frame's root mean square,
3636 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3637 determine the power of a time-varying signal. It is therefore considered
3638 that the RMS is a better approximation of the "perceived loudness" than
3639 just looking at the signal's peak magnitude. Consequently, by adjusting all
3640 frames to a constant RMS value, a uniform "perceived loudness" can be
3641 established. If a target RMS value has been specified, a frame's local gain
3642 factor is defined as the factor that would result in exactly that RMS value.
3643 Note, however, that the maximum local gain factor is still restricted by the
3644 frame's highest magnitude sample, in order to prevent clipping.
3647 Enable channels coupling. By default is enabled.
3648 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3649 amount. This means the same gain factor will be applied to all channels, i.e.
3650 the maximum possible gain factor is determined by the "loudest" channel.
3651 However, in some recordings, it may happen that the volume of the different
3652 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3653 In this case, this option can be used to disable the channel coupling. This way,
3654 the gain factor will be determined independently for each channel, depending
3655 only on the individual channel's highest magnitude sample. This allows for
3656 harmonizing the volume of the different channels.
3659 Enable DC bias correction. By default is disabled.
3660 An audio signal (in the time domain) is a sequence of sample values.
3661 In the Dynamic Audio Normalizer these sample values are represented in the
3662 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3663 audio signal, or "waveform", should be centered around the zero point.
3664 That means if we calculate the mean value of all samples in a file, or in a
3665 single frame, then the result should be 0.0 or at least very close to that
3666 value. If, however, there is a significant deviation of the mean value from
3667 0.0, in either positive or negative direction, this is referred to as a
3668 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3669 Audio Normalizer provides optional DC bias correction.
3670 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3671 the mean value, or "DC correction" offset, of each input frame and subtract
3672 that value from all of the frame's sample values which ensures those samples
3673 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3674 boundaries, the DC correction offset values will be interpolated smoothly
3675 between neighbouring frames.
3677 @item altboundary, b
3678 Enable alternative boundary mode. By default is disabled.
3679 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3680 around each frame. This includes the preceding frames as well as the
3681 subsequent frames. However, for the "boundary" frames, located at the very
3682 beginning and at the very end of the audio file, not all neighbouring
3683 frames are available. In particular, for the first few frames in the audio
3684 file, the preceding frames are not known. And, similarly, for the last few
3685 frames in the audio file, the subsequent frames are not known. Thus, the
3686 question arises which gain factors should be assumed for the missing frames
3687 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3688 to deal with this situation. The default boundary mode assumes a gain factor
3689 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3690 "fade out" at the beginning and at the end of the input, respectively.
3693 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3694 By default, the Dynamic Audio Normalizer does not apply "traditional"
3695 compression. This means that signal peaks will not be pruned and thus the
3696 full dynamic range will be retained within each local neighbourhood. However,
3697 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3698 normalization algorithm with a more "traditional" compression.
3699 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3700 (thresholding) function. If (and only if) the compression feature is enabled,
3701 all input frames will be processed by a soft knee thresholding function prior
3702 to the actual normalization process. Put simply, the thresholding function is
3703 going to prune all samples whose magnitude exceeds a certain threshold value.
3704 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3705 value. Instead, the threshold value will be adjusted for each individual
3707 In general, smaller parameters result in stronger compression, and vice versa.
3708 Values below 3.0 are not recommended, because audible distortion may appear.
3711 Set the target threshold value. This specifies the lowest permissible
3712 magnitude level for the audio input which will be normalized.
3713 If input frame volume is above this value frame will be normalized.
3714 Otherwise frame may not be normalized at all. The default value is set
3715 to 0, which means all input frames will be normalized.
3716 This option is mostly useful if digital noise is not wanted to be amplified.
3719 @subsection Commands
3721 This filter supports the all above options as @ref{commands}.
3725 Make audio easier to listen to on headphones.
3727 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3728 so that when listened to on headphones the stereo image is moved from
3729 inside your head (standard for headphones) to outside and in front of
3730 the listener (standard for speakers).
3736 Apply a two-pole peaking equalisation (EQ) filter. With this
3737 filter, the signal-level at and around a selected frequency can
3738 be increased or decreased, whilst (unlike bandpass and bandreject
3739 filters) that at all other frequencies is unchanged.
3741 In order to produce complex equalisation curves, this filter can
3742 be given several times, each with a different central frequency.
3744 The filter accepts the following options:
3748 Set the filter's central frequency in Hz.
3751 Set method to specify band-width of filter.
3766 Specify the band-width of a filter in width_type units.
3769 Set the required gain or attenuation in dB.
3770 Beware of clipping when using a positive gain.
3773 How much to use filtered signal in output. Default is 1.
3774 Range is between 0 and 1.
3777 Specify which channels to filter, by default all available are filtered.
3780 Normalize biquad coefficients, by default is disabled.
3781 Enabling it will normalize magnitude response at DC to 0dB.
3784 Set transform type of IIR filter.
3793 @subsection Examples
3796 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3798 equalizer=f=1000:t=h:width=200:g=-10
3802 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3804 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3808 @subsection Commands
3810 This filter supports the following commands:
3813 Change equalizer frequency.
3814 Syntax for the command is : "@var{frequency}"
3817 Change equalizer width_type.
3818 Syntax for the command is : "@var{width_type}"
3821 Change equalizer width.
3822 Syntax for the command is : "@var{width}"
3825 Change equalizer gain.
3826 Syntax for the command is : "@var{gain}"
3829 Change equalizer mix.
3830 Syntax for the command is : "@var{mix}"
3833 @section extrastereo
3835 Linearly increases the difference between left and right channels which
3836 adds some sort of "live" effect to playback.
3838 The filter accepts the following options:
3842 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3843 (average of both channels), with 1.0 sound will be unchanged, with
3844 -1.0 left and right channels will be swapped.
3847 Enable clipping. By default is enabled.
3850 @subsection Commands
3852 This filter supports the all above options as @ref{commands}.
3854 @section firequalizer
3855 Apply FIR Equalization using arbitrary frequency response.
3857 The filter accepts the following option:
3861 Set gain curve equation (in dB). The expression can contain variables:
3864 the evaluated frequency
3868 channel number, set to 0 when multichannels evaluation is disabled
3870 channel id, see libavutil/channel_layout.h, set to the first channel id when
3871 multichannels evaluation is disabled
3875 channel_layout, see libavutil/channel_layout.h
3880 @item gain_interpolate(f)
3881 interpolate gain on frequency f based on gain_entry
3882 @item cubic_interpolate(f)
3883 same as gain_interpolate, but smoother
3885 This option is also available as command. Default is @code{gain_interpolate(f)}.
3888 Set gain entry for gain_interpolate function. The expression can
3892 store gain entry at frequency f with value g
3894 This option is also available as command.
3897 Set filter delay in seconds. Higher value means more accurate.
3898 Default is @code{0.01}.
3901 Set filter accuracy in Hz. Lower value means more accurate.
3902 Default is @code{5}.
3905 Set window function. Acceptable values are:
3908 rectangular window, useful when gain curve is already smooth
3910 hann window (default)
3916 3-terms continuous 1st derivative nuttall window
3918 minimum 3-terms discontinuous nuttall window
3920 4-terms continuous 1st derivative nuttall window
3922 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3924 blackman-harris window
3930 If enabled, use fixed number of audio samples. This improves speed when
3931 filtering with large delay. Default is disabled.
3934 Enable multichannels evaluation on gain. Default is disabled.
3937 Enable zero phase mode by subtracting timestamp to compensate delay.
3938 Default is disabled.
3941 Set scale used by gain. Acceptable values are:
3944 linear frequency, linear gain
3946 linear frequency, logarithmic (in dB) gain (default)
3948 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3950 logarithmic frequency, logarithmic gain
3954 Set file for dumping, suitable for gnuplot.
3957 Set scale for dumpfile. Acceptable values are same with scale option.
3961 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3962 Default is disabled.
3965 Enable minimum phase impulse response. Default is disabled.
3968 @subsection Examples
3973 firequalizer=gain='if(lt(f,1000), 0, -INF)'
3976 lowpass at 1000 Hz with gain_entry:
3978 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
3981 custom equalization:
3983 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
3986 higher delay with zero phase to compensate delay:
3988 firequalizer=delay=0.1:fixed=on:zero_phase=on
3991 lowpass on left channel, highpass on right channel:
3993 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
3994 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
3999 Apply a flanging effect to the audio.
4001 The filter accepts the following options:
4005 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4008 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4011 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4015 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4016 Default value is 71.
4019 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4022 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4023 Default value is @var{sinusoidal}.
4026 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4027 Default value is 25.
4030 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4031 Default is @var{linear}.
4035 Apply Haas effect to audio.
4037 Note that this makes most sense to apply on mono signals.
4038 With this filter applied to mono signals it give some directionality and
4039 stretches its stereo image.
4041 The filter accepts the following options:
4045 Set input level. By default is @var{1}, or 0dB
4048 Set output level. By default is @var{1}, or 0dB.
4051 Set gain applied to side part of signal. By default is @var{1}.
4054 Set kind of middle source. Can be one of the following:
4064 Pick middle part signal of stereo image.
4067 Pick side part signal of stereo image.
4071 Change middle phase. By default is disabled.
4074 Set left channel delay. By default is @var{2.05} milliseconds.
4077 Set left channel balance. By default is @var{-1}.
4080 Set left channel gain. By default is @var{1}.
4083 Change left phase. By default is disabled.
4086 Set right channel delay. By defaults is @var{2.12} milliseconds.
4089 Set right channel balance. By default is @var{1}.
4092 Set right channel gain. By default is @var{1}.
4095 Change right phase. By default is enabled.
4100 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4101 embedded HDCD codes is expanded into a 20-bit PCM stream.
4103 The filter supports the Peak Extend and Low-level Gain Adjustment features
4104 of HDCD, and detects the Transient Filter flag.
4107 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4110 When using the filter with wav, note the default encoding for wav is 16-bit,
4111 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4112 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4114 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4115 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4118 The filter accepts the following options:
4121 @item disable_autoconvert
4122 Disable any automatic format conversion or resampling in the filter graph.
4124 @item process_stereo
4125 Process the stereo channels together. If target_gain does not match between
4126 channels, consider it invalid and use the last valid target_gain.
4129 Set the code detect timer period in ms.
4132 Always extend peaks above -3dBFS even if PE isn't signaled.
4135 Replace audio with a solid tone and adjust the amplitude to signal some
4136 specific aspect of the decoding process. The output file can be loaded in
4137 an audio editor alongside the original to aid analysis.
4139 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4146 Gain adjustment level at each sample
4148 Samples where peak extend occurs
4150 Samples where the code detect timer is active
4152 Samples where the target gain does not match between channels
4158 Apply head-related transfer functions (HRTFs) to create virtual
4159 loudspeakers around the user for binaural listening via headphones.
4160 The HRIRs are provided via additional streams, for each channel
4161 one stereo input stream is needed.
4163 The filter accepts the following options:
4167 Set mapping of input streams for convolution.
4168 The argument is a '|'-separated list of channel names in order as they
4169 are given as additional stream inputs for filter.
4170 This also specify number of input streams. Number of input streams
4171 must be not less than number of channels in first stream plus one.
4174 Set gain applied to audio. Value is in dB. Default is 0.
4177 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4178 processing audio in time domain which is slow.
4179 @var{freq} is processing audio in frequency domain which is fast.
4180 Default is @var{freq}.
4183 Set custom gain for LFE channels. Value is in dB. Default is 0.
4186 Set size of frame in number of samples which will be processed at once.
4187 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4190 Set format of hrir stream.
4191 Default value is @var{stereo}. Alternative value is @var{multich}.
4192 If value is set to @var{stereo}, number of additional streams should
4193 be greater or equal to number of input channels in first input stream.
4194 Also each additional stream should have stereo number of channels.
4195 If value is set to @var{multich}, number of additional streams should
4196 be exactly one. Also number of input channels of additional stream
4197 should be equal or greater than twice number of channels of first input
4201 @subsection Examples
4205 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4206 each amovie filter use stereo file with IR coefficients as input.
4207 The files give coefficients for each position of virtual loudspeaker:
4210 -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"
4215 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4216 but now in @var{multich} @var{hrir} format.
4218 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"
4225 Apply a high-pass filter with 3dB point frequency.
4226 The filter can be either single-pole, or double-pole (the default).
4227 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4229 The filter accepts the following options:
4233 Set frequency in Hz. Default is 3000.
4236 Set number of poles. Default is 2.
4239 Set method to specify band-width of filter.
4254 Specify the band-width of a filter in width_type units.
4255 Applies only to double-pole filter.
4256 The default is 0.707q and gives a Butterworth response.
4259 How much to use filtered signal in output. Default is 1.
4260 Range is between 0 and 1.
4263 Specify which channels to filter, by default all available are filtered.
4266 Normalize biquad coefficients, by default is disabled.
4267 Enabling it will normalize magnitude response at DC to 0dB.
4270 Set transform type of IIR filter.
4279 @subsection Commands
4281 This filter supports the following commands:
4284 Change highpass frequency.
4285 Syntax for the command is : "@var{frequency}"
4288 Change highpass width_type.
4289 Syntax for the command is : "@var{width_type}"
4292 Change highpass width.
4293 Syntax for the command is : "@var{width}"
4296 Change highpass mix.
4297 Syntax for the command is : "@var{mix}"
4302 Join multiple input streams into one multi-channel stream.
4304 It accepts the following parameters:
4308 The number of input streams. It defaults to 2.
4310 @item channel_layout
4311 The desired output channel layout. It defaults to stereo.
4314 Map channels from inputs to output. The argument is a '|'-separated list of
4315 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4316 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4317 can be either the name of the input channel (e.g. FL for front left) or its
4318 index in the specified input stream. @var{out_channel} is the name of the output
4322 The filter will attempt to guess the mappings when they are not specified
4323 explicitly. It does so by first trying to find an unused matching input channel
4324 and if that fails it picks the first unused input channel.
4326 Join 3 inputs (with properly set channel layouts):
4328 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4331 Build a 5.1 output from 6 single-channel streams:
4333 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4334 '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'
4340 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4342 To enable compilation of this filter you need to configure FFmpeg with
4343 @code{--enable-ladspa}.
4347 Specifies the name of LADSPA plugin library to load. If the environment
4348 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4349 each one of the directories specified by the colon separated list in
4350 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4351 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4352 @file{/usr/lib/ladspa/}.
4355 Specifies the plugin within the library. Some libraries contain only
4356 one plugin, but others contain many of them. If this is not set filter
4357 will list all available plugins within the specified library.
4360 Set the '|' separated list of controls which are zero or more floating point
4361 values that determine the behavior of the loaded plugin (for example delay,
4363 Controls need to be defined using the following syntax:
4364 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4365 @var{valuei} is the value set on the @var{i}-th control.
4366 Alternatively they can be also defined using the following syntax:
4367 @var{value0}|@var{value1}|@var{value2}|..., where
4368 @var{valuei} is the value set on the @var{i}-th control.
4369 If @option{controls} is set to @code{help}, all available controls and
4370 their valid ranges are printed.
4372 @item sample_rate, s
4373 Specify the sample rate, default to 44100. Only used if plugin have
4377 Set the number of samples per channel per each output frame, default
4378 is 1024. Only used if plugin have zero inputs.
4381 Set the minimum duration of the sourced audio. See
4382 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4383 for the accepted syntax.
4384 Note that the resulting duration may be greater than the specified duration,
4385 as the generated audio is always cut at the end of a complete frame.
4386 If not specified, or the expressed duration is negative, the audio is
4387 supposed to be generated forever.
4388 Only used if plugin have zero inputs.
4391 Enable latency compensation, by default is disabled.
4392 Only used if plugin have inputs.
4395 @subsection Examples
4399 List all available plugins within amp (LADSPA example plugin) library:
4405 List all available controls and their valid ranges for @code{vcf_notch}
4406 plugin from @code{VCF} library:
4408 ladspa=f=vcf:p=vcf_notch:c=help
4412 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4415 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4419 Add reverberation to the audio using TAP-plugins
4420 (Tom's Audio Processing plugins):
4422 ladspa=file=tap_reverb:tap_reverb
4426 Generate white noise, with 0.2 amplitude:
4428 ladspa=file=cmt:noise_source_white:c=c0=.2
4432 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4433 @code{C* Audio Plugin Suite} (CAPS) library:
4435 ladspa=file=caps:Click:c=c1=20'
4439 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4441 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4445 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4446 @code{SWH Plugins} collection:
4448 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4452 Attenuate low frequencies using Multiband EQ from Steve Harris
4453 @code{SWH Plugins} collection:
4455 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4459 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4462 ladspa=caps:Narrower
4466 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4468 ladspa=caps:White:.2
4472 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4474 ladspa=caps:Fractal:c=c1=1
4478 Dynamic volume normalization using @code{VLevel} plugin:
4480 ladspa=vlevel-ladspa:vlevel_mono
4484 @subsection Commands
4486 This filter supports the following commands:
4489 Modify the @var{N}-th control value.
4491 If the specified value is not valid, it is ignored and prior one is kept.
4496 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4497 Support for both single pass (livestreams, files) and double pass (files) modes.
4498 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4499 detect true peaks, the audio stream will be upsampled to 192 kHz.
4500 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4502 The filter accepts the following options:
4506 Set integrated loudness target.
4507 Range is -70.0 - -5.0. Default value is -24.0.
4510 Set loudness range target.
4511 Range is 1.0 - 20.0. Default value is 7.0.
4514 Set maximum true peak.
4515 Range is -9.0 - +0.0. Default value is -2.0.
4517 @item measured_I, measured_i
4518 Measured IL of input file.
4519 Range is -99.0 - +0.0.
4521 @item measured_LRA, measured_lra
4522 Measured LRA of input file.
4523 Range is 0.0 - 99.0.
4525 @item measured_TP, measured_tp
4526 Measured true peak of input file.
4527 Range is -99.0 - +99.0.
4529 @item measured_thresh
4530 Measured threshold of input file.
4531 Range is -99.0 - +0.0.
4534 Set offset gain. Gain is applied before the true-peak limiter.
4535 Range is -99.0 - +99.0. Default is +0.0.
4538 Normalize by linearly scaling the source audio.
4539 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4540 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4541 be lower than source LRA and the change in integrated loudness shouldn't
4542 result in a true peak which exceeds the target TP. If any of these
4543 conditions aren't met, normalization mode will revert to @var{dynamic}.
4544 Options are @code{true} or @code{false}. Default is @code{true}.
4547 Treat mono input files as "dual-mono". If a mono file is intended for playback
4548 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4549 If set to @code{true}, this option will compensate for this effect.
4550 Multi-channel input files are not affected by this option.
4551 Options are true or false. Default is false.
4554 Set print format for stats. Options are summary, json, or none.
4555 Default value is none.
4560 Apply a low-pass filter with 3dB point frequency.
4561 The filter can be either single-pole or double-pole (the default).
4562 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4564 The filter accepts the following options:
4568 Set frequency in Hz. Default is 500.
4571 Set number of poles. Default is 2.
4574 Set method to specify band-width of filter.
4589 Specify the band-width of a filter in width_type units.
4590 Applies only to double-pole filter.
4591 The default is 0.707q and gives a Butterworth response.
4594 How much to use filtered signal in output. Default is 1.
4595 Range is between 0 and 1.
4598 Specify which channels to filter, by default all available are filtered.
4601 Normalize biquad coefficients, by default is disabled.
4602 Enabling it will normalize magnitude response at DC to 0dB.
4605 Set transform type of IIR filter.
4614 @subsection Examples
4617 Lowpass only LFE channel, it LFE is not present it does nothing:
4623 @subsection Commands
4625 This filter supports the following commands:
4628 Change lowpass frequency.
4629 Syntax for the command is : "@var{frequency}"
4632 Change lowpass width_type.
4633 Syntax for the command is : "@var{width_type}"
4636 Change lowpass width.
4637 Syntax for the command is : "@var{width}"
4641 Syntax for the command is : "@var{mix}"
4646 Load a LV2 (LADSPA Version 2) plugin.
4648 To enable compilation of this filter you need to configure FFmpeg with
4649 @code{--enable-lv2}.
4653 Specifies the plugin URI. You may need to escape ':'.
4656 Set the '|' separated list of controls which are zero or more floating point
4657 values that determine the behavior of the loaded plugin (for example delay,
4659 If @option{controls} is set to @code{help}, all available controls and
4660 their valid ranges are printed.
4662 @item sample_rate, s
4663 Specify the sample rate, default to 44100. Only used if plugin have
4667 Set the number of samples per channel per each output frame, default
4668 is 1024. Only used if plugin have zero inputs.
4671 Set the minimum duration of the sourced audio. See
4672 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4673 for the accepted syntax.
4674 Note that the resulting duration may be greater than the specified duration,
4675 as the generated audio is always cut at the end of a complete frame.
4676 If not specified, or the expressed duration is negative, the audio is
4677 supposed to be generated forever.
4678 Only used if plugin have zero inputs.
4681 @subsection Examples
4685 Apply bass enhancer plugin from Calf:
4687 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4691 Apply vinyl plugin from Calf:
4693 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4697 Apply bit crusher plugin from ArtyFX:
4699 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4704 Multiband Compress or expand the audio's dynamic range.
4706 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4707 This is akin to the crossover of a loudspeaker, and results in flat frequency
4708 response when absent compander action.
4710 It accepts the following parameters:
4714 This option syntax is:
4715 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4716 For explanation of each item refer to compand filter documentation.
4722 Mix channels with specific gain levels. The filter accepts the output
4723 channel layout followed by a set of channels definitions.
4725 This filter is also designed to efficiently remap the channels of an audio
4728 The filter accepts parameters of the form:
4729 "@var{l}|@var{outdef}|@var{outdef}|..."
4733 output channel layout or number of channels
4736 output channel specification, of the form:
4737 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4740 output channel to define, either a channel name (FL, FR, etc.) or a channel
4741 number (c0, c1, etc.)
4744 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4747 input channel to use, see out_name for details; it is not possible to mix
4748 named and numbered input channels
4751 If the `=' in a channel specification is replaced by `<', then the gains for
4752 that specification will be renormalized so that the total is 1, thus
4753 avoiding clipping noise.
4755 @subsection Mixing examples
4757 For example, if you want to down-mix from stereo to mono, but with a bigger
4758 factor for the left channel:
4760 pan=1c|c0=0.9*c0+0.1*c1
4763 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4764 7-channels surround:
4766 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4769 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4770 that should be preferred (see "-ac" option) unless you have very specific
4773 @subsection Remapping examples
4775 The channel remapping will be effective if, and only if:
4778 @item gain coefficients are zeroes or ones,
4779 @item only one input per channel output,
4782 If all these conditions are satisfied, the filter will notify the user ("Pure
4783 channel mapping detected"), and use an optimized and lossless method to do the
4786 For example, if you have a 5.1 source and want a stereo audio stream by
4787 dropping the extra channels:
4789 pan="stereo| c0=FL | c1=FR"
4792 Given the same source, you can also switch front left and front right channels
4793 and keep the input channel layout:
4795 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4798 If the input is a stereo audio stream, you can mute the front left channel (and
4799 still keep the stereo channel layout) with:
4804 Still with a stereo audio stream input, you can copy the right channel in both
4805 front left and right:
4807 pan="stereo| c0=FR | c1=FR"
4812 ReplayGain scanner filter. This filter takes an audio stream as an input and
4813 outputs it unchanged.
4814 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4818 Convert the audio sample format, sample rate and channel layout. It is
4819 not meant to be used directly.
4822 Apply time-stretching and pitch-shifting with librubberband.
4824 To enable compilation of this filter, you need to configure FFmpeg with
4825 @code{--enable-librubberband}.
4827 The filter accepts the following options:
4831 Set tempo scale factor.
4834 Set pitch scale factor.
4837 Set transients detector.
4838 Possible values are:
4847 Possible values are:
4856 Possible values are:
4863 Set processing window size.
4864 Possible values are:
4873 Possible values are:
4880 Enable formant preservation when shift pitching.
4881 Possible values are:
4889 Possible values are:
4898 Possible values are:
4905 @subsection Commands
4907 This filter supports the following commands:
4910 Change filter tempo scale factor.
4911 Syntax for the command is : "@var{tempo}"
4914 Change filter pitch scale factor.
4915 Syntax for the command is : "@var{pitch}"
4918 @section sidechaincompress
4920 This filter acts like normal compressor but has the ability to compress
4921 detected signal using second input signal.
4922 It needs two input streams and returns one output stream.
4923 First input stream will be processed depending on second stream signal.
4924 The filtered signal then can be filtered with other filters in later stages of
4925 processing. See @ref{pan} and @ref{amerge} filter.
4927 The filter accepts the following options:
4931 Set input gain. Default is 1. Range is between 0.015625 and 64.
4934 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4935 Default is @code{downward}.
4938 If a signal of second stream raises above this level it will affect the gain
4939 reduction of first stream.
4940 By default is 0.125. Range is between 0.00097563 and 1.
4943 Set a ratio about which the signal is reduced. 1:2 means that if the level
4944 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4945 Default is 2. Range is between 1 and 20.
4948 Amount of milliseconds the signal has to rise above the threshold before gain
4949 reduction starts. Default is 20. Range is between 0.01 and 2000.
4952 Amount of milliseconds the signal has to fall below the threshold before
4953 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4956 Set the amount by how much signal will be amplified after processing.
4957 Default is 1. Range is from 1 to 64.
4960 Curve the sharp knee around the threshold to enter gain reduction more softly.
4961 Default is 2.82843. Range is between 1 and 8.
4964 Choose if the @code{average} level between all channels of side-chain stream
4965 or the louder(@code{maximum}) channel of side-chain stream affects the
4966 reduction. Default is @code{average}.
4969 Should the exact signal be taken in case of @code{peak} or an RMS one in case
4970 of @code{rms}. Default is @code{rms} which is mainly smoother.
4973 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
4976 How much to use compressed signal in output. Default is 1.
4977 Range is between 0 and 1.
4980 @subsection Commands
4982 This filter supports the all above options as @ref{commands}.
4984 @subsection Examples
4988 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
4989 depending on the signal of 2nd input and later compressed signal to be
4990 merged with 2nd input:
4992 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
4996 @section sidechaingate
4998 A sidechain gate acts like a normal (wideband) gate but has the ability to
4999 filter the detected signal before sending it to the gain reduction stage.
5000 Normally a gate uses the full range signal to detect a level above the
5002 For example: If you cut all lower frequencies from your sidechain signal
5003 the gate will decrease the volume of your track only if not enough highs
5004 appear. With this technique you are able to reduce the resonation of a
5005 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5007 It needs two input streams and returns one output stream.
5008 First input stream will be processed depending on second stream signal.
5010 The filter accepts the following options:
5014 Set input level before filtering.
5015 Default is 1. Allowed range is from 0.015625 to 64.
5018 Set the mode of operation. Can be @code{upward} or @code{downward}.
5019 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5020 will be amplified, expanding dynamic range in upward direction.
5021 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5024 Set the level of gain reduction when the signal is below the threshold.
5025 Default is 0.06125. Allowed range is from 0 to 1.
5026 Setting this to 0 disables reduction and then filter behaves like expander.
5029 If a signal rises above this level the gain reduction is released.
5030 Default is 0.125. Allowed range is from 0 to 1.
5033 Set a ratio about which the signal is reduced.
5034 Default is 2. Allowed range is from 1 to 9000.
5037 Amount of milliseconds the signal has to rise above the threshold before gain
5039 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5042 Amount of milliseconds the signal has to fall below the threshold before the
5043 reduction is increased again. Default is 250 milliseconds.
5044 Allowed range is from 0.01 to 9000.
5047 Set amount of amplification of signal after processing.
5048 Default is 1. Allowed range is from 1 to 64.
5051 Curve the sharp knee around the threshold to enter gain reduction more softly.
5052 Default is 2.828427125. Allowed range is from 1 to 8.
5055 Choose if exact signal should be taken for detection or an RMS like one.
5056 Default is rms. Can be peak or rms.
5059 Choose if the average level between all channels or the louder channel affects
5061 Default is average. Can be average or maximum.
5064 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5067 @section silencedetect
5069 Detect silence in an audio stream.
5071 This filter logs a message when it detects that the input audio volume is less
5072 or equal to a noise tolerance value for a duration greater or equal to the
5073 minimum detected noise duration.
5075 The printed times and duration are expressed in seconds. The
5076 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5077 is set on the first frame whose timestamp equals or exceeds the detection
5078 duration and it contains the timestamp of the first frame of the silence.
5080 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5081 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5082 keys are set on the first frame after the silence. If @option{mono} is
5083 enabled, and each channel is evaluated separately, the @code{.X}
5084 suffixed keys are used, and @code{X} corresponds to the channel number.
5086 The filter accepts the following options:
5090 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5091 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5094 Set silence duration until notification (default is 2 seconds). See
5095 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5096 for the accepted syntax.
5099 Process each channel separately, instead of combined. By default is disabled.
5102 @subsection Examples
5106 Detect 5 seconds of silence with -50dB noise tolerance:
5108 silencedetect=n=-50dB:d=5
5112 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5113 tolerance in @file{silence.mp3}:
5115 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5119 @section silenceremove
5121 Remove silence from the beginning, middle or end of the audio.
5123 The filter accepts the following options:
5127 This value is used to indicate if audio should be trimmed at beginning of
5128 the audio. A value of zero indicates no silence should be trimmed from the
5129 beginning. When specifying a non-zero value, it trims audio up until it
5130 finds non-silence. Normally, when trimming silence from beginning of audio
5131 the @var{start_periods} will be @code{1} but it can be increased to higher
5132 values to trim all audio up to specific count of non-silence periods.
5133 Default value is @code{0}.
5135 @item start_duration
5136 Specify the amount of time that non-silence must be detected before it stops
5137 trimming audio. By increasing the duration, bursts of noises can be treated
5138 as silence and trimmed off. Default value is @code{0}.
5140 @item start_threshold
5141 This indicates what sample value should be treated as silence. For digital
5142 audio, a value of @code{0} may be fine but for audio recorded from analog,
5143 you may wish to increase the value to account for background noise.
5144 Can be specified in dB (in case "dB" is appended to the specified value)
5145 or amplitude ratio. Default value is @code{0}.
5148 Specify max duration of silence at beginning that will be kept after
5149 trimming. Default is 0, which is equal to trimming all samples detected
5153 Specify mode of detection of silence end in start of multi-channel audio.
5154 Can be @var{any} or @var{all}. Default is @var{any}.
5155 With @var{any}, any sample that is detected as non-silence will cause
5156 stopped trimming of silence.
5157 With @var{all}, only if all channels are detected as non-silence will cause
5158 stopped trimming of silence.
5161 Set the count for trimming silence from the end of audio.
5162 To remove silence from the middle of a file, specify a @var{stop_periods}
5163 that is negative. This value is then treated as a positive value and is
5164 used to indicate the effect should restart processing as specified by
5165 @var{start_periods}, making it suitable for removing periods of silence
5166 in the middle of the audio.
5167 Default value is @code{0}.
5170 Specify a duration of silence that must exist before audio is not copied any
5171 more. By specifying a higher duration, silence that is wanted can be left in
5173 Default value is @code{0}.
5175 @item stop_threshold
5176 This is the same as @option{start_threshold} but for trimming silence from
5178 Can be specified in dB (in case "dB" is appended to the specified value)
5179 or amplitude ratio. Default value is @code{0}.
5182 Specify max duration of silence at end that will be kept after
5183 trimming. Default is 0, which is equal to trimming all samples detected
5187 Specify mode of detection of silence start in end of multi-channel audio.
5188 Can be @var{any} or @var{all}. Default is @var{any}.
5189 With @var{any}, any sample that is detected as non-silence will cause
5190 stopped trimming of silence.
5191 With @var{all}, only if all channels are detected as non-silence will cause
5192 stopped trimming of silence.
5195 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5196 and works better with digital silence which is exactly 0.
5197 Default value is @code{rms}.
5200 Set duration in number of seconds used to calculate size of window in number
5201 of samples for detecting silence.
5202 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5205 @subsection Examples
5209 The following example shows how this filter can be used to start a recording
5210 that does not contain the delay at the start which usually occurs between
5211 pressing the record button and the start of the performance:
5213 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5217 Trim all silence encountered from beginning to end where there is more than 1
5218 second of silence in audio:
5220 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5224 Trim all digital silence samples, using peak detection, from beginning to end
5225 where there is more than 0 samples of digital silence in audio and digital
5226 silence is detected in all channels at same positions in stream:
5228 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5234 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5235 loudspeakers around the user for binaural listening via headphones (audio
5236 formats up to 9 channels supported).
5237 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5238 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5239 Austrian Academy of Sciences.
5241 To enable compilation of this filter you need to configure FFmpeg with
5242 @code{--enable-libmysofa}.
5244 The filter accepts the following options:
5248 Set the SOFA file used for rendering.
5251 Set gain applied to audio. Value is in dB. Default is 0.
5254 Set rotation of virtual loudspeakers in deg. Default is 0.
5257 Set elevation of virtual speakers in deg. Default is 0.
5260 Set distance in meters between loudspeakers and the listener with near-field
5261 HRTFs. Default is 1.
5264 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5265 processing audio in time domain which is slow.
5266 @var{freq} is processing audio in frequency domain which is fast.
5267 Default is @var{freq}.
5270 Set custom positions of virtual loudspeakers. Syntax for this option is:
5271 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5272 Each virtual loudspeaker is described with short channel name following with
5273 azimuth and elevation in degrees.
5274 Each virtual loudspeaker description is separated by '|'.
5275 For example to override front left and front right channel positions use:
5276 'speakers=FL 45 15|FR 345 15'.
5277 Descriptions with unrecognised channel names are ignored.
5280 Set custom gain for LFE channels. Value is in dB. Default is 0.
5283 Set custom frame size in number of samples. Default is 1024.
5284 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5285 is set to @var{freq}.
5288 Should all IRs be normalized upon importing SOFA file.
5289 By default is enabled.
5292 Should nearest IRs be interpolated with neighbor IRs if exact position
5293 does not match. By default is disabled.
5296 Minphase all IRs upon loading of SOFA file. By default is disabled.
5299 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5302 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5305 @subsection Examples
5309 Using ClubFritz6 sofa file:
5311 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5315 Using ClubFritz12 sofa file and bigger radius with small rotation:
5317 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5321 Similar as above but with custom speaker positions for front left, front right, back left and back right
5322 and also with custom gain:
5324 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5331 This filter expands or compresses each half-cycle of audio samples
5332 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5333 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5335 The filter accepts the following options:
5339 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5340 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5343 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5344 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5345 would be such that local peak value reaches target peak value but never to surpass it and that
5346 ratio between new and previous peak value does not surpass this option value.
5348 @item compression, c
5349 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5350 This option controls maximum local half-cycle of samples compression. This option is used
5351 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5352 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5353 that peak's half-cycle will be compressed by current compression factor.
5356 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5357 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5358 Any half-cycle samples with their local peak value below or same as this option value will be
5359 compressed by current compression factor, otherwise, if greater than threshold value they will be
5360 expanded with expansion factor so that it could reach peak target value but never surpass it.
5363 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5364 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5365 each new half-cycle until it reaches @option{expansion} value.
5366 Setting this options too high may lead to distortions.
5369 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5370 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5371 each new half-cycle until it reaches @option{compression} value.
5374 Specify which channels to filter, by default all available channels are filtered.
5377 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5378 option. When enabled any half-cycle of samples with their local peak value below or same as
5379 @option{threshold} option will be expanded otherwise it will be compressed.
5382 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5383 When disabled each filtered channel gain calculation is independent, otherwise when this option
5384 is enabled the minimum of all possible gains for each filtered channel is used.
5387 @subsection Commands
5389 This filter supports the all above options as @ref{commands}.
5391 @section stereotools
5393 This filter has some handy utilities to manage stereo signals, for converting
5394 M/S stereo recordings to L/R signal while having control over the parameters
5395 or spreading the stereo image of master track.
5397 The filter accepts the following options:
5401 Set input level before filtering for both channels. Defaults is 1.
5402 Allowed range is from 0.015625 to 64.
5405 Set output level after filtering for both channels. Defaults is 1.
5406 Allowed range is from 0.015625 to 64.
5409 Set input balance between both channels. Default is 0.
5410 Allowed range is from -1 to 1.
5413 Set output balance between both channels. Default is 0.
5414 Allowed range is from -1 to 1.
5417 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5418 clipping. Disabled by default.
5421 Mute the left channel. Disabled by default.
5424 Mute the right channel. Disabled by default.
5427 Change the phase of the left channel. Disabled by default.
5430 Change the phase of the right channel. Disabled by default.
5433 Set stereo mode. Available values are:
5437 Left/Right to Left/Right, this is default.
5440 Left/Right to Mid/Side.
5443 Mid/Side to Left/Right.
5446 Left/Right to Left/Left.
5449 Left/Right to Right/Right.
5452 Left/Right to Left + Right.
5455 Left/Right to Right/Left.
5458 Mid/Side to Left/Left.
5461 Mid/Side to Right/Right.
5465 Set level of side signal. Default is 1.
5466 Allowed range is from 0.015625 to 64.
5469 Set balance of side signal. Default is 0.
5470 Allowed range is from -1 to 1.
5473 Set level of the middle signal. Default is 1.
5474 Allowed range is from 0.015625 to 64.
5477 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5480 Set stereo base between mono and inversed channels. Default is 0.
5481 Allowed range is from -1 to 1.
5484 Set delay in milliseconds how much to delay left from right channel and
5485 vice versa. Default is 0. Allowed range is from -20 to 20.
5488 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5491 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5493 @item bmode_in, bmode_out
5494 Set balance mode for balance_in/balance_out option.
5496 Can be one of the following:
5500 Classic balance mode. Attenuate one channel at time.
5501 Gain is raised up to 1.
5504 Similar as classic mode above but gain is raised up to 2.
5507 Equal power distribution, from -6dB to +6dB range.
5511 @subsection Examples
5515 Apply karaoke like effect:
5517 stereotools=mlev=0.015625
5521 Convert M/S signal to L/R:
5523 "stereotools=mode=ms>lr"
5527 @section stereowiden
5529 This filter enhance the stereo effect by suppressing signal common to both
5530 channels and by delaying the signal of left into right and vice versa,
5531 thereby widening the stereo effect.
5533 The filter accepts the following options:
5537 Time in milliseconds of the delay of left signal into right and vice versa.
5538 Default is 20 milliseconds.
5541 Amount of gain in delayed signal into right and vice versa. Gives a delay
5542 effect of left signal in right output and vice versa which gives widening
5543 effect. Default is 0.3.
5546 Cross feed of left into right with inverted phase. This helps in suppressing
5547 the mono. If the value is 1 it will cancel all the signal common to both
5548 channels. Default is 0.3.
5551 Set level of input signal of original channel. Default is 0.8.
5554 @subsection Commands
5556 This filter supports the all above options except @code{delay} as @ref{commands}.
5558 @section superequalizer
5559 Apply 18 band equalizer.
5561 The filter accepts the following options:
5568 Set 131Hz band gain.
5570 Set 185Hz band gain.
5572 Set 262Hz band gain.
5574 Set 370Hz band gain.
5576 Set 523Hz band gain.
5578 Set 740Hz band gain.
5580 Set 1047Hz band gain.
5582 Set 1480Hz band gain.
5584 Set 2093Hz band gain.
5586 Set 2960Hz band gain.
5588 Set 4186Hz band gain.
5590 Set 5920Hz band gain.
5592 Set 8372Hz band gain.
5594 Set 11840Hz band gain.
5596 Set 16744Hz band gain.
5598 Set 20000Hz band gain.
5602 Apply audio surround upmix filter.
5604 This filter allows to produce multichannel output from audio stream.
5606 The filter accepts the following options:
5610 Set output channel layout. By default, this is @var{5.1}.
5612 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5613 for the required syntax.
5616 Set input channel layout. By default, this is @var{stereo}.
5618 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5619 for the required syntax.
5622 Set input volume level. By default, this is @var{1}.
5625 Set output volume level. By default, this is @var{1}.
5628 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5631 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5634 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5637 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5638 In @var{add} mode, LFE channel is created from input audio and added to output.
5639 In @var{sub} mode, LFE channel is created from input audio and added to output but
5640 also all non-LFE output channels are subtracted with output LFE channel.
5643 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5644 Default is @var{90}.
5647 Set front center input volume. By default, this is @var{1}.
5650 Set front center output volume. By default, this is @var{1}.
5653 Set front left input volume. By default, this is @var{1}.
5656 Set front left output volume. By default, this is @var{1}.
5659 Set front right input volume. By default, this is @var{1}.
5662 Set front right output volume. By default, this is @var{1}.
5665 Set side left input volume. By default, this is @var{1}.
5668 Set side left output volume. By default, this is @var{1}.
5671 Set side right input volume. By default, this is @var{1}.
5674 Set side right output volume. By default, this is @var{1}.
5677 Set back left input volume. By default, this is @var{1}.
5680 Set back left output volume. By default, this is @var{1}.
5683 Set back right input volume. By default, this is @var{1}.
5686 Set back right output volume. By default, this is @var{1}.
5689 Set back center input volume. By default, this is @var{1}.
5692 Set back center output volume. By default, this is @var{1}.
5695 Set LFE input volume. By default, this is @var{1}.
5698 Set LFE output volume. By default, this is @var{1}.
5701 Set spread usage of stereo image across X axis for all channels.
5704 Set spread usage of stereo image across Y axis for all channels.
5706 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5707 Set spread usage of stereo image across X axis for each channel.
5709 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5710 Set spread usage of stereo image across Y axis for each channel.
5713 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5716 Set window function.
5718 It accepts the following values:
5741 Default is @code{hann}.
5744 Set window overlap. If set to 1, the recommended overlap for selected
5745 window function will be picked. Default is @code{0.5}.
5748 @section treble, highshelf
5750 Boost or cut treble (upper) frequencies of the audio using a two-pole
5751 shelving filter with a response similar to that of a standard
5752 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5754 The filter accepts the following options:
5758 Give the gain at whichever is the lower of ~22 kHz and the
5759 Nyquist frequency. Its useful range is about -20 (for a large cut)
5760 to +20 (for a large boost). Beware of clipping when using a positive gain.
5763 Set the filter's central frequency and so can be used
5764 to extend or reduce the frequency range to be boosted or cut.
5765 The default value is @code{3000} Hz.
5768 Set method to specify band-width of filter.
5783 Determine how steep is the filter's shelf transition.
5786 How much to use filtered signal in output. Default is 1.
5787 Range is between 0 and 1.
5790 Specify which channels to filter, by default all available are filtered.
5793 Normalize biquad coefficients, by default is disabled.
5794 Enabling it will normalize magnitude response at DC to 0dB.
5797 Set transform type of IIR filter.
5806 @subsection Commands
5808 This filter supports the following commands:
5811 Change treble frequency.
5812 Syntax for the command is : "@var{frequency}"
5815 Change treble width_type.
5816 Syntax for the command is : "@var{width_type}"
5819 Change treble width.
5820 Syntax for the command is : "@var{width}"
5824 Syntax for the command is : "@var{gain}"
5828 Syntax for the command is : "@var{mix}"
5833 Sinusoidal amplitude modulation.
5835 The filter accepts the following options:
5839 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5840 (20 Hz or lower) will result in a tremolo effect.
5841 This filter may also be used as a ring modulator by specifying
5842 a modulation frequency higher than 20 Hz.
5843 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5846 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5847 Default value is 0.5.
5852 Sinusoidal phase modulation.
5854 The filter accepts the following options:
5858 Modulation frequency in Hertz.
5859 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5862 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5863 Default value is 0.5.
5868 Adjust the input audio volume.
5870 It accepts the following parameters:
5874 Set audio volume expression.
5876 Output values are clipped to the maximum value.
5878 The output audio volume is given by the relation:
5880 @var{output_volume} = @var{volume} * @var{input_volume}
5883 The default value for @var{volume} is "1.0".
5886 This parameter represents the mathematical precision.
5888 It determines which input sample formats will be allowed, which affects the
5889 precision of the volume scaling.
5893 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5895 32-bit floating-point; this limits input sample format to FLT. (default)
5897 64-bit floating-point; this limits input sample format to DBL.
5901 Choose the behaviour on encountering ReplayGain side data in input frames.
5905 Remove ReplayGain side data, ignoring its contents (the default).
5908 Ignore ReplayGain side data, but leave it in the frame.
5911 Prefer the track gain, if present.
5914 Prefer the album gain, if present.
5917 @item replaygain_preamp
5918 Pre-amplification gain in dB to apply to the selected replaygain gain.
5920 Default value for @var{replaygain_preamp} is 0.0.
5922 @item replaygain_noclip
5923 Prevent clipping by limiting the gain applied.
5925 Default value for @var{replaygain_noclip} is 1.
5928 Set when the volume expression is evaluated.
5930 It accepts the following values:
5933 only evaluate expression once during the filter initialization, or
5934 when the @samp{volume} command is sent
5937 evaluate expression for each incoming frame
5940 Default value is @samp{once}.
5943 The volume expression can contain the following parameters.
5947 frame number (starting at zero)
5950 @item nb_consumed_samples
5951 number of samples consumed by the filter
5953 number of samples in the current frame
5955 original frame position in the file
5961 PTS at start of stream
5963 time at start of stream
5969 last set volume value
5972 Note that when @option{eval} is set to @samp{once} only the
5973 @var{sample_rate} and @var{tb} variables are available, all other
5974 variables will evaluate to NAN.
5976 @subsection Commands
5978 This filter supports the following commands:
5981 Modify the volume expression.
5982 The command accepts the same syntax of the corresponding option.
5984 If the specified expression is not valid, it is kept at its current
5988 @subsection Examples
5992 Halve the input audio volume:
5996 volume=volume=-6.0206dB
5999 In all the above example the named key for @option{volume} can be
6000 omitted, for example like in:
6006 Increase input audio power by 6 decibels using fixed-point precision:
6008 volume=volume=6dB:precision=fixed
6012 Fade volume after time 10 with an annihilation period of 5 seconds:
6014 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6018 @section volumedetect
6020 Detect the volume of the input video.
6022 The filter has no parameters. The input is not modified. Statistics about
6023 the volume will be printed in the log when the input stream end is reached.
6025 In particular it will show the mean volume (root mean square), maximum
6026 volume (on a per-sample basis), and the beginning of a histogram of the
6027 registered volume values (from the maximum value to a cumulated 1/1000 of
6030 All volumes are in decibels relative to the maximum PCM value.
6032 @subsection Examples
6034 Here is an excerpt of the output:
6036 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6037 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6038 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6039 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6040 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6041 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6042 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6043 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6044 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6050 The mean square energy is approximately -27 dB, or 10^-2.7.
6052 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6054 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6057 In other words, raising the volume by +4 dB does not cause any clipping,
6058 raising it by +5 dB causes clipping for 6 samples, etc.
6060 @c man end AUDIO FILTERS
6062 @chapter Audio Sources
6063 @c man begin AUDIO SOURCES
6065 Below is a description of the currently available audio sources.
6069 Buffer audio frames, and make them available to the filter chain.
6071 This source is mainly intended for a programmatic use, in particular
6072 through the interface defined in @file{libavfilter/buffersrc.h}.
6074 It accepts the following parameters:
6078 The timebase which will be used for timestamps of submitted frames. It must be
6079 either a floating-point number or in @var{numerator}/@var{denominator} form.
6082 The sample rate of the incoming audio buffers.
6085 The sample format of the incoming audio buffers.
6086 Either a sample format name or its corresponding integer representation from
6087 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6089 @item channel_layout
6090 The channel layout of the incoming audio buffers.
6091 Either a channel layout name from channel_layout_map in
6092 @file{libavutil/channel_layout.c} or its corresponding integer representation
6093 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6096 The number of channels of the incoming audio buffers.
6097 If both @var{channels} and @var{channel_layout} are specified, then they
6102 @subsection Examples
6105 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6108 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6109 Since the sample format with name "s16p" corresponds to the number
6110 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6113 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6118 Generate an audio signal specified by an expression.
6120 This source accepts in input one or more expressions (one for each
6121 channel), which are evaluated and used to generate a corresponding
6124 This source accepts the following options:
6128 Set the '|'-separated expressions list for each separate channel. In case the
6129 @option{channel_layout} option is not specified, the selected channel layout
6130 depends on the number of provided expressions. Otherwise the last
6131 specified expression is applied to the remaining output channels.
6133 @item channel_layout, c
6134 Set the channel layout. The number of channels in the specified layout
6135 must be equal to the number of specified expressions.
6138 Set the minimum duration of the sourced audio. See
6139 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6140 for the accepted syntax.
6141 Note that the resulting duration may be greater than the specified
6142 duration, as the generated audio is always cut at the end of a
6145 If not specified, or the expressed duration is negative, the audio is
6146 supposed to be generated forever.
6149 Set the number of samples per channel per each output frame,
6152 @item sample_rate, s
6153 Specify the sample rate, default to 44100.
6156 Each expression in @var{exprs} can contain the following constants:
6160 number of the evaluated sample, starting from 0
6163 time of the evaluated sample expressed in seconds, starting from 0
6170 @subsection Examples
6180 Generate a sin signal with frequency of 440 Hz, set sample rate to
6183 aevalsrc="sin(440*2*PI*t):s=8000"
6187 Generate a two channels signal, specify the channel layout (Front
6188 Center + Back Center) explicitly:
6190 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6194 Generate white noise:
6196 aevalsrc="-2+random(0)"
6200 Generate an amplitude modulated signal:
6202 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6206 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6208 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6215 Generate a FIR coefficients using frequency sampling method.
6217 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6219 The filter accepts the following options:
6223 Set number of filter coefficents in output audio stream.
6224 Default value is 1025.
6227 Set frequency points from where magnitude and phase are set.
6228 This must be in non decreasing order, and first element must be 0, while last element
6229 must be 1. Elements are separated by white spaces.
6232 Set magnitude value for every frequency point set by @option{frequency}.
6233 Number of values must be same as number of frequency points.
6234 Values are separated by white spaces.
6237 Set phase value for every frequency point set by @option{frequency}.
6238 Number of values must be same as number of frequency points.
6239 Values are separated by white spaces.
6241 @item sample_rate, r
6242 Set sample rate, default is 44100.
6245 Set number of samples per each frame. Default is 1024.
6248 Set window function. Default is blackman.
6253 The null audio source, return unprocessed audio frames. It is mainly useful
6254 as a template and to be employed in analysis / debugging tools, or as
6255 the source for filters which ignore the input data (for example the sox
6258 This source accepts the following options:
6262 @item channel_layout, cl
6264 Specifies the channel layout, and can be either an integer or a string
6265 representing a channel layout. The default value of @var{channel_layout}
6268 Check the channel_layout_map definition in
6269 @file{libavutil/channel_layout.c} for the mapping between strings and
6270 channel layout values.
6272 @item sample_rate, r
6273 Specifies the sample rate, and defaults to 44100.
6276 Set the number of samples per requested frames.
6279 Set the duration of the sourced audio. See
6280 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6281 for the accepted syntax.
6283 If not specified, or the expressed duration is negative, the audio is
6284 supposed to be generated forever.
6287 @subsection Examples
6291 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6293 anullsrc=r=48000:cl=4
6297 Do the same operation with a more obvious syntax:
6299 anullsrc=r=48000:cl=mono
6303 All the parameters need to be explicitly defined.
6307 Synthesize a voice utterance using the libflite library.
6309 To enable compilation of this filter you need to configure FFmpeg with
6310 @code{--enable-libflite}.
6312 Note that versions of the flite library prior to 2.0 are not thread-safe.
6314 The filter accepts the following options:
6319 If set to 1, list the names of the available voices and exit
6320 immediately. Default value is 0.
6323 Set the maximum number of samples per frame. Default value is 512.
6326 Set the filename containing the text to speak.
6329 Set the text to speak.
6332 Set the voice to use for the speech synthesis. Default value is
6333 @code{kal}. See also the @var{list_voices} option.
6336 @subsection Examples
6340 Read from file @file{speech.txt}, and synthesize the text using the
6341 standard flite voice:
6343 flite=textfile=speech.txt
6347 Read the specified text selecting the @code{slt} voice:
6349 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6353 Input text to ffmpeg:
6355 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6359 Make @file{ffplay} speak the specified text, using @code{flite} and
6360 the @code{lavfi} device:
6362 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6366 For more information about libflite, check:
6367 @url{http://www.festvox.org/flite/}
6371 Generate a noise audio signal.
6373 The filter accepts the following options:
6376 @item sample_rate, r
6377 Specify the sample rate. Default value is 48000 Hz.
6380 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6384 Specify the duration of the generated audio stream. Not specifying this option
6385 results in noise with an infinite length.
6387 @item color, colour, c
6388 Specify the color of noise. Available noise colors are white, pink, brown,
6389 blue, violet and velvet. Default color is white.
6392 Specify a value used to seed the PRNG.
6395 Set the number of samples per each output frame, default is 1024.
6398 @subsection Examples
6403 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6405 anoisesrc=d=60:c=pink:r=44100:a=0.5
6411 Generate odd-tap Hilbert transform FIR coefficients.
6413 The resulting stream can be used with @ref{afir} filter for phase-shifting
6414 the signal by 90 degrees.
6416 This is used in many matrix coding schemes and for analytic signal generation.
6417 The process is often written as a multiplication by i (or j), the imaginary unit.
6419 The filter accepts the following options:
6423 @item sample_rate, s
6424 Set sample rate, default is 44100.
6427 Set length of FIR filter, default is 22051.
6430 Set number of samples per each frame.
6433 Set window function to be used when generating FIR coefficients.
6438 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6440 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6442 The filter accepts the following options:
6445 @item sample_rate, r
6446 Set sample rate, default is 44100.
6449 Set number of samples per each frame. Default is 1024.
6452 Set high-pass frequency. Default is 0.
6455 Set low-pass frequency. Default is 0.
6456 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6457 is higher than 0 then filter will create band-pass filter coefficients,
6458 otherwise band-reject filter coefficients.
6461 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6464 Set Kaiser window beta.
6467 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6470 Enable rounding, by default is disabled.
6473 Set number of taps for high-pass filter.
6476 Set number of taps for low-pass filter.
6481 Generate an audio signal made of a sine wave with amplitude 1/8.
6483 The audio signal is bit-exact.
6485 The filter accepts the following options:
6490 Set the carrier frequency. Default is 440 Hz.
6492 @item beep_factor, b
6493 Enable a periodic beep every second with frequency @var{beep_factor} times
6494 the carrier frequency. Default is 0, meaning the beep is disabled.
6496 @item sample_rate, r
6497 Specify the sample rate, default is 44100.
6500 Specify the duration of the generated audio stream.
6502 @item samples_per_frame
6503 Set the number of samples per output frame.
6505 The expression can contain the following constants:
6509 The (sequential) number of the output audio frame, starting from 0.
6512 The PTS (Presentation TimeStamp) of the output audio frame,
6513 expressed in @var{TB} units.
6516 The PTS of the output audio frame, expressed in seconds.
6519 The timebase of the output audio frames.
6522 Default is @code{1024}.
6525 @subsection Examples
6530 Generate a simple 440 Hz sine wave:
6536 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6540 sine=frequency=220:beep_factor=4:duration=5
6544 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6547 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6551 @c man end AUDIO SOURCES
6553 @chapter Audio Sinks
6554 @c man begin AUDIO SINKS
6556 Below is a description of the currently available audio sinks.
6558 @section abuffersink
6560 Buffer audio frames, and make them available to the end of filter chain.
6562 This sink is mainly intended for programmatic use, in particular
6563 through the interface defined in @file{libavfilter/buffersink.h}
6564 or the options system.
6566 It accepts a pointer to an AVABufferSinkContext structure, which
6567 defines the incoming buffers' formats, to be passed as the opaque
6568 parameter to @code{avfilter_init_filter} for initialization.
6571 Null audio sink; do absolutely nothing with the input audio. It is
6572 mainly useful as a template and for use in analysis / debugging
6575 @c man end AUDIO SINKS
6577 @chapter Video Filters
6578 @c man begin VIDEO FILTERS
6580 When you configure your FFmpeg build, you can disable any of the
6581 existing filters using @code{--disable-filters}.
6582 The configure output will show the video filters included in your
6585 Below is a description of the currently available video filters.
6589 Mark a region of interest in a video frame.
6591 The frame data is passed through unchanged, but metadata is attached
6592 to the frame indicating regions of interest which can affect the
6593 behaviour of later encoding. Multiple regions can be marked by
6594 applying the filter multiple times.
6598 Region distance in pixels from the left edge of the frame.
6600 Region distance in pixels from the top edge of the frame.
6602 Region width in pixels.
6604 Region height in pixels.
6606 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6607 and may contain the following variables:
6610 Width of the input frame.
6612 Height of the input frame.
6616 Quantisation offset to apply within the region.
6618 This must be a real value in the range -1 to +1. A value of zero
6619 indicates no quality change. A negative value asks for better quality
6620 (less quantisation), while a positive value asks for worse quality
6621 (greater quantisation).
6623 The range is calibrated so that the extreme values indicate the
6624 largest possible offset - if the rest of the frame is encoded with the
6625 worst possible quality, an offset of -1 indicates that this region
6626 should be encoded with the best possible quality anyway. Intermediate
6627 values are then interpolated in some codec-dependent way.
6629 For example, in 10-bit H.264 the quantisation parameter varies between
6630 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6631 this region should be encoded with a QP around one-tenth of the full
6632 range better than the rest of the frame. So, if most of the frame
6633 were to be encoded with a QP of around 30, this region would get a QP
6634 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6635 An extreme value of -1 would indicate that this region should be
6636 encoded with the best possible quality regardless of the treatment of
6637 the rest of the frame - that is, should be encoded at a QP of -12.
6639 If set to true, remove any existing regions of interest marked on the
6640 frame before adding the new one.
6643 @subsection Examples
6647 Mark the centre quarter of the frame as interesting.
6649 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6652 Mark the 100-pixel-wide region on the left edge of the frame as very
6653 uninteresting (to be encoded at much lower quality than the rest of
6656 addroi=0:0:100:ih:+1/5
6660 @section alphaextract
6662 Extract the alpha component from the input as a grayscale video. This
6663 is especially useful with the @var{alphamerge} filter.
6667 Add or replace the alpha component of the primary input with the
6668 grayscale value of a second input. This is intended for use with
6669 @var{alphaextract} to allow the transmission or storage of frame
6670 sequences that have alpha in a format that doesn't support an alpha
6673 For example, to reconstruct full frames from a normal YUV-encoded video
6674 and a separate video created with @var{alphaextract}, you might use:
6676 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6681 Amplify differences between current pixel and pixels of adjacent frames in
6682 same pixel location.
6684 This filter accepts the following options:
6688 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6689 For example radius of 3 will instruct filter to calculate average of 7 frames.
6692 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6695 Set threshold for difference amplification. Any difference greater or equal to
6696 this value will not alter source pixel. Default is 10.
6697 Allowed range is from 0 to 65535.
6700 Set tolerance for difference amplification. Any difference lower to
6701 this value will not alter source pixel. Default is 0.
6702 Allowed range is from 0 to 65535.
6705 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6706 This option controls maximum possible value that will decrease source pixel value.
6709 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6710 This option controls maximum possible value that will increase source pixel value.
6713 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6716 @subsection Commands
6718 This filter supports the following @ref{commands} that corresponds to option of same name:
6730 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6731 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6732 Substation Alpha) subtitles files.
6734 This filter accepts the following option in addition to the common options from
6735 the @ref{subtitles} filter:
6739 Set the shaping engine
6741 Available values are:
6744 The default libass shaping engine, which is the best available.
6746 Fast, font-agnostic shaper that can do only substitutions
6748 Slower shaper using OpenType for substitutions and positioning
6751 The default is @code{auto}.
6755 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6757 The filter accepts the following options:
6761 Set threshold A for 1st plane. Default is 0.02.
6762 Valid range is 0 to 0.3.
6765 Set threshold B for 1st plane. Default is 0.04.
6766 Valid range is 0 to 5.
6769 Set threshold A for 2nd plane. Default is 0.02.
6770 Valid range is 0 to 0.3.
6773 Set threshold B for 2nd plane. Default is 0.04.
6774 Valid range is 0 to 5.
6777 Set threshold A for 3rd plane. Default is 0.02.
6778 Valid range is 0 to 0.3.
6781 Set threshold B for 3rd plane. Default is 0.04.
6782 Valid range is 0 to 5.
6784 Threshold A is designed to react on abrupt changes in the input signal and
6785 threshold B is designed to react on continuous changes in the input signal.
6788 Set number of frames filter will use for averaging. Default is 9. Must be odd
6789 number in range [5, 129].
6792 Set what planes of frame filter will use for averaging. Default is all.
6795 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6796 Alternatively can be set to @code{s} serial.
6798 Parallel can be faster then serial, while other way around is never true.
6799 Parallel will abort early on first change being greater then thresholds, while serial
6800 will continue processing other side of frames if they are equal or bellow thresholds.
6803 @subsection Commands
6804 This filter supports same @ref{commands} as options except option @code{s}.
6805 The command accepts the same syntax of the corresponding option.
6809 Apply average blur filter.
6811 The filter accepts the following options:
6815 Set horizontal radius size.
6818 Set which planes to filter. By default all planes are filtered.
6821 Set vertical radius size, if zero it will be same as @code{sizeX}.
6822 Default is @code{0}.
6825 @subsection Commands
6826 This filter supports same commands as options.
6827 The command accepts the same syntax of the corresponding option.
6829 If the specified expression is not valid, it is kept at its current
6834 Compute the bounding box for the non-black pixels in the input frame
6837 This filter computes the bounding box containing all the pixels with a
6838 luminance value greater than the minimum allowed value.
6839 The parameters describing the bounding box are printed on the filter
6842 The filter accepts the following option:
6846 Set the minimal luminance value. Default is @code{16}.
6850 Apply bilateral filter, spatial smoothing while preserving edges.
6852 The filter accepts the following options:
6855 Set sigma of gaussian function to calculate spatial weight.
6856 Allowed range is 0 to 512. Default is 0.1.
6859 Set sigma of gaussian function to calculate range weight.
6860 Allowed range is 0 to 1. Default is 0.1.
6863 Set planes to filter. Default is first only.
6866 @section bitplanenoise
6868 Show and measure bit plane noise.
6870 The filter accepts the following options:
6874 Set which plane to analyze. Default is @code{1}.
6877 Filter out noisy pixels from @code{bitplane} set above.
6878 Default is disabled.
6881 @section blackdetect
6883 Detect video intervals that are (almost) completely black. Can be
6884 useful to detect chapter transitions, commercials, or invalid
6887 The filter outputs its detection analysis to both the log as well as
6888 frame metadata. If a black segment of at least the specified minimum
6889 duration is found, a line with the start and end timestamps as well
6890 as duration is printed to the log with level @code{info}. In addition,
6891 a log line with level @code{debug} is printed per frame showing the
6892 black amount detected for that frame.
6894 The filter also attaches metadata to the first frame of a black
6895 segment with key @code{lavfi.black_start} and to the first frame
6896 after the black segment ends with key @code{lavfi.black_end}. The
6897 value is the frame's timestamp. This metadata is added regardless
6898 of the minimum duration specified.
6900 The filter accepts the following options:
6903 @item black_min_duration, d
6904 Set the minimum detected black duration expressed in seconds. It must
6905 be a non-negative floating point number.
6907 Default value is 2.0.
6909 @item picture_black_ratio_th, pic_th
6910 Set the threshold for considering a picture "black".
6911 Express the minimum value for the ratio:
6913 @var{nb_black_pixels} / @var{nb_pixels}
6916 for which a picture is considered black.
6917 Default value is 0.98.
6919 @item pixel_black_th, pix_th
6920 Set the threshold for considering a pixel "black".
6922 The threshold expresses the maximum pixel luminance value for which a
6923 pixel is considered "black". The provided value is scaled according to
6924 the following equation:
6926 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6929 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6930 the input video format, the range is [0-255] for YUV full-range
6931 formats and [16-235] for YUV non full-range formats.
6933 Default value is 0.10.
6936 The following example sets the maximum pixel threshold to the minimum
6937 value, and detects only black intervals of 2 or more seconds:
6939 blackdetect=d=2:pix_th=0.00
6944 Detect frames that are (almost) completely black. Can be useful to
6945 detect chapter transitions or commercials. Output lines consist of
6946 the frame number of the detected frame, the percentage of blackness,
6947 the position in the file if known or -1 and the timestamp in seconds.
6949 In order to display the output lines, you need to set the loglevel at
6950 least to the AV_LOG_INFO value.
6952 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6953 The value represents the percentage of pixels in the picture that
6954 are below the threshold value.
6956 It accepts the following parameters:
6961 The percentage of the pixels that have to be below the threshold; it defaults to
6964 @item threshold, thresh
6965 The threshold below which a pixel value is considered black; it defaults to
6973 Blend two video frames into each other.
6975 The @code{blend} filter takes two input streams and outputs one
6976 stream, the first input is the "top" layer and second input is
6977 "bottom" layer. By default, the output terminates when the longest input terminates.
6979 The @code{tblend} (time blend) filter takes two consecutive frames
6980 from one single stream, and outputs the result obtained by blending
6981 the new frame on top of the old frame.
6983 A description of the accepted options follows.
6991 Set blend mode for specific pixel component or all pixel components in case
6992 of @var{all_mode}. Default value is @code{normal}.
6994 Available values for component modes are:
7036 Set blend opacity for specific pixel component or all pixel components in case
7037 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7044 Set blend expression for specific pixel component or all pixel components in case
7045 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7047 The expressions can use the following variables:
7051 The sequential number of the filtered frame, starting from @code{0}.
7055 the coordinates of the current sample
7059 the width and height of currently filtered plane
7063 Width and height scale for the plane being filtered. It is the
7064 ratio between the dimensions of the current plane to the luma plane,
7065 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7066 the luma plane and @code{0.5,0.5} for the chroma planes.
7069 Time of the current frame, expressed in seconds.
7072 Value of pixel component at current location for first video frame (top layer).
7075 Value of pixel component at current location for second video frame (bottom layer).
7079 The @code{blend} filter also supports the @ref{framesync} options.
7081 @subsection Examples
7085 Apply transition from bottom layer to top layer in first 10 seconds:
7087 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7091 Apply linear horizontal transition from top layer to bottom layer:
7093 blend=all_expr='A*(X/W)+B*(1-X/W)'
7097 Apply 1x1 checkerboard effect:
7099 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7103 Apply uncover left effect:
7105 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7109 Apply uncover down effect:
7111 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7115 Apply uncover up-left effect:
7117 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7121 Split diagonally video and shows top and bottom layer on each side:
7123 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7127 Display differences between the current and the previous frame:
7129 tblend=all_mode=grainextract
7135 Denoise frames using Block-Matching 3D algorithm.
7137 The filter accepts the following options.
7141 Set denoising strength. Default value is 1.
7142 Allowed range is from 0 to 999.9.
7143 The denoising algorithm is very sensitive to sigma, so adjust it
7144 according to the source.
7147 Set local patch size. This sets dimensions in 2D.
7150 Set sliding step for processing blocks. Default value is 4.
7151 Allowed range is from 1 to 64.
7152 Smaller values allows processing more reference blocks and is slower.
7155 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7156 When set to 1, no block matching is done. Larger values allows more blocks
7158 Allowed range is from 1 to 256.
7161 Set radius for search block matching. Default is 9.
7162 Allowed range is from 1 to INT32_MAX.
7165 Set step between two search locations for block matching. Default is 1.
7166 Allowed range is from 1 to 64. Smaller is slower.
7169 Set threshold of mean square error for block matching. Valid range is 0 to
7173 Set thresholding parameter for hard thresholding in 3D transformed domain.
7174 Larger values results in stronger hard-thresholding filtering in frequency
7178 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7179 Default is @code{basic}.
7182 If enabled, filter will use 2nd stream for block matching.
7183 Default is disabled for @code{basic} value of @var{estim} option,
7184 and always enabled if value of @var{estim} is @code{final}.
7187 Set planes to filter. Default is all available except alpha.
7190 @subsection Examples
7194 Basic filtering with bm3d:
7196 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7200 Same as above, but filtering only luma:
7202 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7206 Same as above, but with both estimation modes:
7208 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
7212 Same as above, but prefilter with @ref{nlmeans} filter instead:
7214 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
7220 Apply a boxblur algorithm to the input video.
7222 It accepts the following parameters:
7226 @item luma_radius, lr
7227 @item luma_power, lp
7228 @item chroma_radius, cr
7229 @item chroma_power, cp
7230 @item alpha_radius, ar
7231 @item alpha_power, ap
7235 A description of the accepted options follows.
7238 @item luma_radius, lr
7239 @item chroma_radius, cr
7240 @item alpha_radius, ar
7241 Set an expression for the box radius in pixels used for blurring the
7242 corresponding input plane.
7244 The radius value must be a non-negative number, and must not be
7245 greater than the value of the expression @code{min(w,h)/2} for the
7246 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7249 Default value for @option{luma_radius} is "2". If not specified,
7250 @option{chroma_radius} and @option{alpha_radius} default to the
7251 corresponding value set for @option{luma_radius}.
7253 The expressions can contain the following constants:
7257 The input width and height in pixels.
7261 The input chroma image width and height in pixels.
7265 The horizontal and vertical chroma subsample values. For example, for the
7266 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7269 @item luma_power, lp
7270 @item chroma_power, cp
7271 @item alpha_power, ap
7272 Specify how many times the boxblur filter is applied to the
7273 corresponding plane.
7275 Default value for @option{luma_power} is 2. If not specified,
7276 @option{chroma_power} and @option{alpha_power} default to the
7277 corresponding value set for @option{luma_power}.
7279 A value of 0 will disable the effect.
7282 @subsection Examples
7286 Apply a boxblur filter with the luma, chroma, and alpha radii
7289 boxblur=luma_radius=2:luma_power=1
7294 Set the luma radius to 2, and alpha and chroma radius to 0:
7296 boxblur=2:1:cr=0:ar=0
7300 Set the luma and chroma radii to a fraction of the video dimension:
7302 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7308 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7309 Deinterlacing Filter").
7311 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7312 interpolation algorithms.
7313 It accepts the following parameters:
7317 The interlacing mode to adopt. It accepts one of the following values:
7321 Output one frame for each frame.
7323 Output one frame for each field.
7326 The default value is @code{send_field}.
7329 The picture field parity assumed for the input interlaced video. It accepts one
7330 of the following values:
7334 Assume the top field is first.
7336 Assume the bottom field is first.
7338 Enable automatic detection of field parity.
7341 The default value is @code{auto}.
7342 If the interlacing is unknown or the decoder does not export this information,
7343 top field first will be assumed.
7346 Specify which frames to deinterlace. Accepts one of the following
7351 Deinterlace all frames.
7353 Only deinterlace frames marked as interlaced.
7356 The default value is @code{all}.
7361 Apply Contrast Adaptive Sharpen filter to video stream.
7363 The filter accepts the following options:
7367 Set the sharpening strength. Default value is 0.
7370 Set planes to filter. Default value is to filter all
7371 planes except alpha plane.
7375 Remove all color information for all colors except for certain one.
7377 The filter accepts the following options:
7381 The color which will not be replaced with neutral chroma.
7384 Similarity percentage with the above color.
7385 0.01 matches only the exact key color, while 1.0 matches everything.
7389 0.0 makes pixels either fully gray, or not gray at all.
7390 Higher values result in more preserved color.
7393 Signals that the color passed is already in YUV instead of RGB.
7395 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7396 This can be used to pass exact YUV values as hexadecimal numbers.
7399 @subsection Commands
7400 This filter supports same @ref{commands} as options.
7401 The command accepts the same syntax of the corresponding option.
7403 If the specified expression is not valid, it is kept at its current
7407 YUV colorspace color/chroma keying.
7409 The filter accepts the following options:
7413 The color which will be replaced with transparency.
7416 Similarity percentage with the key color.
7418 0.01 matches only the exact key color, while 1.0 matches everything.
7423 0.0 makes pixels either fully transparent, or not transparent at all.
7425 Higher values result in semi-transparent pixels, with a higher transparency
7426 the more similar the pixels color is to the key color.
7429 Signals that the color passed is already in YUV instead of RGB.
7431 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7432 This can be used to pass exact YUV values as hexadecimal numbers.
7435 @subsection Commands
7436 This filter supports same @ref{commands} as options.
7437 The command accepts the same syntax of the corresponding option.
7439 If the specified expression is not valid, it is kept at its current
7442 @subsection Examples
7446 Make every green pixel in the input image transparent:
7448 ffmpeg -i input.png -vf chromakey=green out.png
7452 Overlay a greenscreen-video on top of a static black background.
7454 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
7459 Reduce chrominance noise.
7461 The filter accepts the following options:
7465 Set threshold for averaging chrominance values.
7466 Sum of absolute difference of U and V pixel components or current
7467 pixel and neighbour pixels lower than this threshold will be used in
7468 averaging. Luma component is left unchanged and is copied to output.
7469 Default value is 30. Allowed range is from 1 to 200.
7472 Set horizontal radius of rectangle used for averaging.
7473 Allowed range is from 1 to 100. Default value is 5.
7476 Set vertical radius of rectangle used for averaging.
7477 Allowed range is from 1 to 100. Default value is 5.
7480 Set horizontal step when averaging. Default value is 1.
7481 Allowed range is from 1 to 50.
7482 Mostly useful to speed-up filtering.
7485 Set vertical step when averaging. Default value is 1.
7486 Allowed range is from 1 to 50.
7487 Mostly useful to speed-up filtering.
7490 @subsection Commands
7491 This filter supports same @ref{commands} as options.
7492 The command accepts the same syntax of the corresponding option.
7494 @section chromashift
7495 Shift chroma pixels horizontally and/or vertically.
7497 The filter accepts the following options:
7500 Set amount to shift chroma-blue horizontally.
7502 Set amount to shift chroma-blue vertically.
7504 Set amount to shift chroma-red horizontally.
7506 Set amount to shift chroma-red vertically.
7508 Set edge mode, can be @var{smear}, default, or @var{warp}.
7511 @subsection Commands
7513 This filter supports the all above options as @ref{commands}.
7517 Display CIE color diagram with pixels overlaid onto it.
7519 The filter accepts the following options:
7534 @item uhdtv, rec2020
7548 Set what gamuts to draw.
7550 See @code{system} option for available values.
7553 Set ciescope size, by default set to 512.
7556 Set intensity used to map input pixel values to CIE diagram.
7559 Set contrast used to draw tongue colors that are out of active color system gamut.
7562 Correct gamma displayed on scope, by default enabled.
7565 Show white point on CIE diagram, by default disabled.
7568 Set input gamma. Used only with XYZ input color space.
7573 Visualize information exported by some codecs.
7575 Some codecs can export information through frames using side-data or other
7576 means. For example, some MPEG based codecs export motion vectors through the
7577 @var{export_mvs} flag in the codec @option{flags2} option.
7579 The filter accepts the following option:
7583 Set motion vectors to visualize.
7585 Available flags for @var{mv} are:
7589 forward predicted MVs of P-frames
7591 forward predicted MVs of B-frames
7593 backward predicted MVs of B-frames
7597 Display quantization parameters using the chroma planes.
7600 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7602 Available flags for @var{mv_type} are:
7606 forward predicted MVs
7608 backward predicted MVs
7611 @item frame_type, ft
7612 Set frame type to visualize motion vectors of.
7614 Available flags for @var{frame_type} are:
7618 intra-coded frames (I-frames)
7620 predicted frames (P-frames)
7622 bi-directionally predicted frames (B-frames)
7626 @subsection Examples
7630 Visualize forward predicted MVs of all frames using @command{ffplay}:
7632 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7636 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7638 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7642 @section colorbalance
7643 Modify intensity of primary colors (red, green and blue) of input frames.
7645 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7646 regions for the red-cyan, green-magenta or blue-yellow balance.
7648 A positive adjustment value shifts the balance towards the primary color, a negative
7649 value towards the complementary color.
7651 The filter accepts the following options:
7657 Adjust red, green and blue shadows (darkest pixels).
7662 Adjust red, green and blue midtones (medium pixels).
7667 Adjust red, green and blue highlights (brightest pixels).
7669 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7672 Preserve lightness when changing color balance. Default is disabled.
7675 @subsection Examples
7679 Add red color cast to shadows:
7685 @subsection Commands
7687 This filter supports the all above options as @ref{commands}.
7689 @section colorchannelmixer
7691 Adjust video input frames by re-mixing color channels.
7693 This filter modifies a color channel by adding the values associated to
7694 the other channels of the same pixels. For example if the value to
7695 modify is red, the output value will be:
7697 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7700 The filter accepts the following options:
7707 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7708 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7714 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7715 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7721 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7722 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7728 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7729 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7731 Allowed ranges for options are @code{[-2.0, 2.0]}.
7734 @subsection Examples
7738 Convert source to grayscale:
7740 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7743 Simulate sepia tones:
7745 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7749 @subsection Commands
7751 This filter supports the all above options as @ref{commands}.
7754 RGB colorspace color keying.
7756 The filter accepts the following options:
7760 The color which will be replaced with transparency.
7763 Similarity percentage with the key color.
7765 0.01 matches only the exact key color, while 1.0 matches everything.
7770 0.0 makes pixels either fully transparent, or not transparent at all.
7772 Higher values result in semi-transparent pixels, with a higher transparency
7773 the more similar the pixels color is to the key color.
7776 @subsection Examples
7780 Make every green pixel in the input image transparent:
7782 ffmpeg -i input.png -vf colorkey=green out.png
7786 Overlay a greenscreen-video on top of a static background image.
7788 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
7792 @subsection Commands
7793 This filter supports same @ref{commands} as options.
7794 The command accepts the same syntax of the corresponding option.
7796 If the specified expression is not valid, it is kept at its current
7800 Remove all color information for all RGB colors except for certain one.
7802 The filter accepts the following options:
7806 The color which will not be replaced with neutral gray.
7809 Similarity percentage with the above color.
7810 0.01 matches only the exact key color, while 1.0 matches everything.
7813 Blend percentage. 0.0 makes pixels fully gray.
7814 Higher values result in more preserved color.
7817 @subsection Commands
7818 This filter supports same @ref{commands} as options.
7819 The command accepts the same syntax of the corresponding option.
7821 If the specified expression is not valid, it is kept at its current
7824 @section colorlevels
7826 Adjust video input frames using levels.
7828 The filter accepts the following options:
7835 Adjust red, green, blue and alpha input black point.
7836 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7842 Adjust red, green, blue and alpha input white point.
7843 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7845 Input levels are used to lighten highlights (bright tones), darken shadows
7846 (dark tones), change the balance of bright and dark tones.
7852 Adjust red, green, blue and alpha output black point.
7853 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7859 Adjust red, green, blue and alpha output white point.
7860 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7862 Output levels allows manual selection of a constrained output level range.
7865 @subsection Examples
7869 Make video output darker:
7871 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7877 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7881 Make video output lighter:
7883 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7887 Increase brightness:
7889 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7893 @subsection Commands
7895 This filter supports the all above options as @ref{commands}.
7897 @section colormatrix
7899 Convert color matrix.
7901 The filter accepts the following options:
7906 Specify the source and destination color matrix. Both values must be
7909 The accepted values are:
7937 For example to convert from BT.601 to SMPTE-240M, use the command:
7939 colormatrix=bt601:smpte240m
7944 Convert colorspace, transfer characteristics or color primaries.
7945 Input video needs to have an even size.
7947 The filter accepts the following options:
7952 Specify all color properties at once.
7954 The accepted values are:
7984 Specify output colorspace.
7986 The accepted values are:
7995 BT.470BG or BT.601-6 625
7998 SMPTE-170M or BT.601-6 525
8007 BT.2020 with non-constant luminance
8013 Specify output transfer characteristics.
8015 The accepted values are:
8027 Constant gamma of 2.2
8030 Constant gamma of 2.8
8033 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8051 BT.2020 for 10-bits content
8054 BT.2020 for 12-bits content
8060 Specify output color primaries.
8062 The accepted values are:
8071 BT.470BG or BT.601-6 625
8074 SMPTE-170M or BT.601-6 525
8098 Specify output color range.
8100 The accepted values are:
8103 TV (restricted) range
8106 MPEG (restricted) range
8117 Specify output color format.
8119 The accepted values are:
8122 YUV 4:2:0 planar 8-bits
8125 YUV 4:2:0 planar 10-bits
8128 YUV 4:2:0 planar 12-bits
8131 YUV 4:2:2 planar 8-bits
8134 YUV 4:2:2 planar 10-bits
8137 YUV 4:2:2 planar 12-bits
8140 YUV 4:4:4 planar 8-bits
8143 YUV 4:4:4 planar 10-bits
8146 YUV 4:4:4 planar 12-bits
8151 Do a fast conversion, which skips gamma/primary correction. This will take
8152 significantly less CPU, but will be mathematically incorrect. To get output
8153 compatible with that produced by the colormatrix filter, use fast=1.
8156 Specify dithering mode.
8158 The accepted values are:
8164 Floyd-Steinberg dithering
8168 Whitepoint adaptation mode.
8170 The accepted values are:
8173 Bradford whitepoint adaptation
8176 von Kries whitepoint adaptation
8179 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8183 Override all input properties at once. Same accepted values as @ref{all}.
8186 Override input colorspace. Same accepted values as @ref{space}.
8189 Override input color primaries. Same accepted values as @ref{primaries}.
8192 Override input transfer characteristics. Same accepted values as @ref{trc}.
8195 Override input color range. Same accepted values as @ref{range}.
8199 The filter converts the transfer characteristics, color space and color
8200 primaries to the specified user values. The output value, if not specified,
8201 is set to a default value based on the "all" property. If that property is
8202 also not specified, the filter will log an error. The output color range and
8203 format default to the same value as the input color range and format. The
8204 input transfer characteristics, color space, color primaries and color range
8205 should be set on the input data. If any of these are missing, the filter will
8206 log an error and no conversion will take place.
8208 For example to convert the input to SMPTE-240M, use the command:
8210 colorspace=smpte240m
8213 @section convolution
8215 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8217 The filter accepts the following options:
8224 Set matrix for each plane.
8225 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8226 and from 1 to 49 odd number of signed integers in @var{row} mode.
8232 Set multiplier for calculated value for each plane.
8233 If unset or 0, it will be sum of all matrix elements.
8239 Set bias for each plane. This value is added to the result of the multiplication.
8240 Useful for making the overall image brighter or darker. Default is 0.0.
8246 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8247 Default is @var{square}.
8250 @subsection Examples
8256 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"
8262 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"
8268 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"
8274 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"
8278 Apply laplacian edge detector which includes diagonals:
8280 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"
8286 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"
8292 Apply 2D convolution of video stream in frequency domain using second stream
8295 The filter accepts the following options:
8299 Set which planes to process.
8302 Set which impulse video frames will be processed, can be @var{first}
8303 or @var{all}. Default is @var{all}.
8306 The @code{convolve} filter also supports the @ref{framesync} options.
8310 Copy the input video source unchanged to the output. This is mainly useful for
8315 Video filtering on GPU using Apple's CoreImage API on OSX.
8317 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8318 processed by video hardware. However, software-based OpenGL implementations
8319 exist which means there is no guarantee for hardware processing. It depends on
8322 There are many filters and image generators provided by Apple that come with a
8323 large variety of options. The filter has to be referenced by its name along
8326 The coreimage filter accepts the following options:
8329 List all available filters and generators along with all their respective
8330 options as well as possible minimum and maximum values along with the default
8337 Specify all filters by their respective name and options.
8338 Use @var{list_filters} to determine all valid filter names and options.
8339 Numerical options are specified by a float value and are automatically clamped
8340 to their respective value range. Vector and color options have to be specified
8341 by a list of space separated float values. Character escaping has to be done.
8342 A special option name @code{default} is available to use default options for a
8345 It is required to specify either @code{default} or at least one of the filter options.
8346 All omitted options are used with their default values.
8347 The syntax of the filter string is as follows:
8349 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8353 Specify a rectangle where the output of the filter chain is copied into the
8354 input image. It is given by a list of space separated float values:
8356 output_rect=x\ y\ width\ height
8358 If not given, the output rectangle equals the dimensions of the input image.
8359 The output rectangle is automatically cropped at the borders of the input
8360 image. Negative values are valid for each component.
8362 output_rect=25\ 25\ 100\ 100
8366 Several filters can be chained for successive processing without GPU-HOST
8367 transfers allowing for fast processing of complex filter chains.
8368 Currently, only filters with zero (generators) or exactly one (filters) input
8369 image and one output image are supported. Also, transition filters are not yet
8372 Some filters generate output images with additional padding depending on the
8373 respective filter kernel. The padding is automatically removed to ensure the
8374 filter output has the same size as the input image.
8376 For image generators, the size of the output image is determined by the
8377 previous output image of the filter chain or the input image of the whole
8378 filterchain, respectively. The generators do not use the pixel information of
8379 this image to generate their output. However, the generated output is
8380 blended onto this image, resulting in partial or complete coverage of the
8383 The @ref{coreimagesrc} video source can be used for generating input images
8384 which are directly fed into the filter chain. By using it, providing input
8385 images by another video source or an input video is not required.
8387 @subsection Examples
8392 List all filters available:
8394 coreimage=list_filters=true
8398 Use the CIBoxBlur filter with default options to blur an image:
8400 coreimage=filter=CIBoxBlur@@default
8404 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8405 its center at 100x100 and a radius of 50 pixels:
8407 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8411 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8412 given as complete and escaped command-line for Apple's standard bash shell:
8414 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8420 Cover a rectangular object
8422 It accepts the following options:
8426 Filepath of the optional cover image, needs to be in yuv420.
8431 It accepts the following values:
8434 cover it by the supplied image
8436 cover it by interpolating the surrounding pixels
8439 Default value is @var{blur}.
8442 @subsection Examples
8446 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8448 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8454 Crop the input video to given dimensions.
8456 It accepts the following parameters:
8460 The width of the output video. It defaults to @code{iw}.
8461 This expression is evaluated only once during the filter
8462 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8465 The height of the output video. It defaults to @code{ih}.
8466 This expression is evaluated only once during the filter
8467 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8470 The horizontal position, in the input video, of the left edge of the output
8471 video. It defaults to @code{(in_w-out_w)/2}.
8472 This expression is evaluated per-frame.
8475 The vertical position, in the input video, of the top edge of the output video.
8476 It defaults to @code{(in_h-out_h)/2}.
8477 This expression is evaluated per-frame.
8480 If set to 1 will force the output display aspect ratio
8481 to be the same of the input, by changing the output sample aspect
8482 ratio. It defaults to 0.
8485 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8486 width/height/x/y as specified and will not be rounded to nearest smaller value.
8490 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8491 expressions containing the following constants:
8496 The computed values for @var{x} and @var{y}. They are evaluated for
8501 The input width and height.
8505 These are the same as @var{in_w} and @var{in_h}.
8509 The output (cropped) width and height.
8513 These are the same as @var{out_w} and @var{out_h}.
8516 same as @var{iw} / @var{ih}
8519 input sample aspect ratio
8522 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8526 horizontal and vertical chroma subsample values. For example for the
8527 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8530 The number of the input frame, starting from 0.
8533 the position in the file of the input frame, NAN if unknown
8536 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8540 The expression for @var{out_w} may depend on the value of @var{out_h},
8541 and the expression for @var{out_h} may depend on @var{out_w}, but they
8542 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8543 evaluated after @var{out_w} and @var{out_h}.
8545 The @var{x} and @var{y} parameters specify the expressions for the
8546 position of the top-left corner of the output (non-cropped) area. They
8547 are evaluated for each frame. If the evaluated value is not valid, it
8548 is approximated to the nearest valid value.
8550 The expression for @var{x} may depend on @var{y}, and the expression
8551 for @var{y} may depend on @var{x}.
8553 @subsection Examples
8557 Crop area with size 100x100 at position (12,34).
8562 Using named options, the example above becomes:
8564 crop=w=100:h=100:x=12:y=34
8568 Crop the central input area with size 100x100:
8574 Crop the central input area with size 2/3 of the input video:
8576 crop=2/3*in_w:2/3*in_h
8580 Crop the input video central square:
8587 Delimit the rectangle with the top-left corner placed at position
8588 100:100 and the right-bottom corner corresponding to the right-bottom
8589 corner of the input image.
8591 crop=in_w-100:in_h-100:100:100
8595 Crop 10 pixels from the left and right borders, and 20 pixels from
8596 the top and bottom borders
8598 crop=in_w-2*10:in_h-2*20
8602 Keep only the bottom right quarter of the input image:
8604 crop=in_w/2:in_h/2:in_w/2:in_h/2
8608 Crop height for getting Greek harmony:
8610 crop=in_w:1/PHI*in_w
8614 Apply trembling effect:
8616 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)
8620 Apply erratic camera effect depending on timestamp:
8622 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)"
8626 Set x depending on the value of y:
8628 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8632 @subsection Commands
8634 This filter supports the following commands:
8640 Set width/height of the output video and the horizontal/vertical position
8642 The command accepts the same syntax of the corresponding option.
8644 If the specified expression is not valid, it is kept at its current
8650 Auto-detect the crop size.
8652 It calculates the necessary cropping parameters and prints the
8653 recommended parameters via the logging system. The detected dimensions
8654 correspond to the non-black area of the input video.
8656 It accepts the following parameters:
8661 Set higher black value threshold, which can be optionally specified
8662 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8663 value greater to the set value is considered non-black. It defaults to 24.
8664 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8665 on the bitdepth of the pixel format.
8668 The value which the width/height should be divisible by. It defaults to
8669 16. The offset is automatically adjusted to center the video. Use 2 to
8670 get only even dimensions (needed for 4:2:2 video). 16 is best when
8671 encoding to most video codecs.
8673 @item reset_count, reset
8674 Set the counter that determines after how many frames cropdetect will
8675 reset the previously detected largest video area and start over to
8676 detect the current optimal crop area. Default value is 0.
8678 This can be useful when channel logos distort the video area. 0
8679 indicates 'never reset', and returns the largest area encountered during
8686 Delay video filtering until a given wallclock timestamp. The filter first
8687 passes on @option{preroll} amount of frames, then it buffers at most
8688 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8689 it forwards the buffered frames and also any subsequent frames coming in its
8692 The filter can be used synchronize the output of multiple ffmpeg processes for
8693 realtime output devices like decklink. By putting the delay in the filtering
8694 chain and pre-buffering frames the process can pass on data to output almost
8695 immediately after the target wallclock timestamp is reached.
8697 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8703 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8706 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8709 The maximum duration of content to buffer before waiting for the cue expressed
8710 in seconds. Default is 0.
8717 Apply color adjustments using curves.
8719 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8720 component (red, green and blue) has its values defined by @var{N} key points
8721 tied from each other using a smooth curve. The x-axis represents the pixel
8722 values from the input frame, and the y-axis the new pixel values to be set for
8725 By default, a component curve is defined by the two points @var{(0;0)} and
8726 @var{(1;1)}. This creates a straight line where each original pixel value is
8727 "adjusted" to its own value, which means no change to the image.
8729 The filter allows you to redefine these two points and add some more. A new
8730 curve (using a natural cubic spline interpolation) will be define to pass
8731 smoothly through all these new coordinates. The new defined points needs to be
8732 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8733 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8734 the vector spaces, the values will be clipped accordingly.
8736 The filter accepts the following options:
8740 Select one of the available color presets. This option can be used in addition
8741 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8742 options takes priority on the preset values.
8743 Available presets are:
8746 @item color_negative
8749 @item increase_contrast
8751 @item linear_contrast
8752 @item medium_contrast
8754 @item strong_contrast
8757 Default is @code{none}.
8759 Set the master key points. These points will define a second pass mapping. It
8760 is sometimes called a "luminance" or "value" mapping. It can be used with
8761 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8762 post-processing LUT.
8764 Set the key points for the red component.
8766 Set the key points for the green component.
8768 Set the key points for the blue component.
8770 Set the key points for all components (not including master).
8771 Can be used in addition to the other key points component
8772 options. In this case, the unset component(s) will fallback on this
8773 @option{all} setting.
8775 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8777 Save Gnuplot script of the curves in specified file.
8780 To avoid some filtergraph syntax conflicts, each key points list need to be
8781 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8783 @subsection Examples
8787 Increase slightly the middle level of blue:
8789 curves=blue='0/0 0.5/0.58 1/1'
8795 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'
8797 Here we obtain the following coordinates for each components:
8800 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8802 @code{(0;0) (0.50;0.48) (1;1)}
8804 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8808 The previous example can also be achieved with the associated built-in preset:
8810 curves=preset=vintage
8820 Use a Photoshop preset and redefine the points of the green component:
8822 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8826 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8827 and @command{gnuplot}:
8829 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8830 gnuplot -p /tmp/curves.plt
8836 Video data analysis filter.
8838 This filter shows hexadecimal pixel values of part of video.
8840 The filter accepts the following options:
8844 Set output video size.
8847 Set x offset from where to pick pixels.
8850 Set y offset from where to pick pixels.
8853 Set scope mode, can be one of the following:
8856 Draw hexadecimal pixel values with white color on black background.
8859 Draw hexadecimal pixel values with input video pixel color on black
8863 Draw hexadecimal pixel values on color background picked from input video,
8864 the text color is picked in such way so its always visible.
8868 Draw rows and columns numbers on left and top of video.
8871 Set background opacity.
8874 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
8878 Apply Directional blur filter.
8880 The filter accepts the following options:
8884 Set angle of directional blur. Default is @code{45}.
8887 Set radius of directional blur. Default is @code{5}.
8890 Set which planes to filter. By default all planes are filtered.
8893 @subsection Commands
8894 This filter supports same @ref{commands} as options.
8895 The command accepts the same syntax of the corresponding option.
8897 If the specified expression is not valid, it is kept at its current
8902 Denoise frames using 2D DCT (frequency domain filtering).
8904 This filter is not designed for real time.
8906 The filter accepts the following options:
8910 Set the noise sigma constant.
8912 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8913 coefficient (absolute value) below this threshold with be dropped.
8915 If you need a more advanced filtering, see @option{expr}.
8917 Default is @code{0}.
8920 Set number overlapping pixels for each block. Since the filter can be slow, you
8921 may want to reduce this value, at the cost of a less effective filter and the
8922 risk of various artefacts.
8924 If the overlapping value doesn't permit processing the whole input width or
8925 height, a warning will be displayed and according borders won't be denoised.
8927 Default value is @var{blocksize}-1, which is the best possible setting.
8930 Set the coefficient factor expression.
8932 For each coefficient of a DCT block, this expression will be evaluated as a
8933 multiplier value for the coefficient.
8935 If this is option is set, the @option{sigma} option will be ignored.
8937 The absolute value of the coefficient can be accessed through the @var{c}
8941 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8942 @var{blocksize}, which is the width and height of the processed blocks.
8944 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8945 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8946 on the speed processing. Also, a larger block size does not necessarily means a
8950 @subsection Examples
8952 Apply a denoise with a @option{sigma} of @code{4.5}:
8957 The same operation can be achieved using the expression system:
8959 dctdnoiz=e='gte(c, 4.5*3)'
8962 Violent denoise using a block size of @code{16x16}:
8969 Remove banding artifacts from input video.
8970 It works by replacing banded pixels with average value of referenced pixels.
8972 The filter accepts the following options:
8979 Set banding detection threshold for each plane. Default is 0.02.
8980 Valid range is 0.00003 to 0.5.
8981 If difference between current pixel and reference pixel is less than threshold,
8982 it will be considered as banded.
8985 Banding detection range in pixels. Default is 16. If positive, random number
8986 in range 0 to set value will be used. If negative, exact absolute value
8988 The range defines square of four pixels around current pixel.
8991 Set direction in radians from which four pixel will be compared. If positive,
8992 random direction from 0 to set direction will be picked. If negative, exact of
8993 absolute value will be picked. For example direction 0, -PI or -2*PI radians
8994 will pick only pixels on same row and -PI/2 will pick only pixels on same
8998 If enabled, current pixel is compared with average value of all four
8999 surrounding pixels. The default is enabled. If disabled current pixel is
9000 compared with all four surrounding pixels. The pixel is considered banded
9001 if only all four differences with surrounding pixels are less than threshold.
9004 If enabled, current pixel is changed if and only if all pixel components are banded,
9005 e.g. banding detection threshold is triggered for all color components.
9006 The default is disabled.
9011 Remove blocking artifacts from input video.
9013 The filter accepts the following options:
9017 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9018 This controls what kind of deblocking is applied.
9021 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9027 Set blocking detection thresholds. Allowed range is 0 to 1.
9028 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9029 Using higher threshold gives more deblocking strength.
9030 Setting @var{alpha} controls threshold detection at exact edge of block.
9031 Remaining options controls threshold detection near the edge. Each one for
9032 below/above or left/right. Setting any of those to @var{0} disables
9036 Set planes to filter. Default is to filter all available planes.
9039 @subsection Examples
9043 Deblock using weak filter and block size of 4 pixels.
9045 deblock=filter=weak:block=4
9049 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9050 deblocking more edges.
9052 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9056 Similar as above, but filter only first plane.
9058 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9062 Similar as above, but filter only second and third plane.
9064 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9071 Drop duplicated frames at regular intervals.
9073 The filter accepts the following options:
9077 Set the number of frames from which one will be dropped. Setting this to
9078 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9079 Default is @code{5}.
9082 Set the threshold for duplicate detection. If the difference metric for a frame
9083 is less than or equal to this value, then it is declared as duplicate. Default
9087 Set scene change threshold. Default is @code{15}.
9091 Set the size of the x and y-axis blocks used during metric calculations.
9092 Larger blocks give better noise suppression, but also give worse detection of
9093 small movements. Must be a power of two. Default is @code{32}.
9096 Mark main input as a pre-processed input and activate clean source input
9097 stream. This allows the input to be pre-processed with various filters to help
9098 the metrics calculation while keeping the frame selection lossless. When set to
9099 @code{1}, the first stream is for the pre-processed input, and the second
9100 stream is the clean source from where the kept frames are chosen. Default is
9104 Set whether or not chroma is considered in the metric calculations. Default is
9110 Apply 2D deconvolution of video stream in frequency domain using second stream
9113 The filter accepts the following options:
9117 Set which planes to process.
9120 Set which impulse video frames will be processed, can be @var{first}
9121 or @var{all}. Default is @var{all}.
9124 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9125 and height are not same and not power of 2 or if stream prior to convolving
9129 The @code{deconvolve} filter also supports the @ref{framesync} options.
9133 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9135 It accepts the following options:
9139 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9140 @var{rainbows} for cross-color reduction.
9143 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9146 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9149 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9152 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9157 Apply deflate effect to the video.
9159 This filter replaces the pixel by the local(3x3) average by taking into account
9160 only values lower than the pixel.
9162 It accepts the following options:
9169 Limit the maximum change for each plane, default is 65535.
9170 If 0, plane will remain unchanged.
9173 @subsection Commands
9175 This filter supports the all above options as @ref{commands}.
9179 Remove temporal frame luminance variations.
9181 It accepts the following options:
9185 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9188 Set averaging mode to smooth temporal luminance variations.
9190 Available values are:
9215 Do not actually modify frame. Useful when one only wants metadata.
9220 Remove judder produced by partially interlaced telecined content.
9222 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9223 source was partially telecined content then the output of @code{pullup,dejudder}
9224 will have a variable frame rate. May change the recorded frame rate of the
9225 container. Aside from that change, this filter will not affect constant frame
9228 The option available in this filter is:
9232 Specify the length of the window over which the judder repeats.
9234 Accepts any integer greater than 1. Useful values are:
9238 If the original was telecined from 24 to 30 fps (Film to NTSC).
9241 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9244 If a mixture of the two.
9247 The default is @samp{4}.
9252 Suppress a TV station logo by a simple interpolation of the surrounding
9253 pixels. Just set a rectangle covering the logo and watch it disappear
9254 (and sometimes something even uglier appear - your mileage may vary).
9256 It accepts the following parameters:
9261 Specify the top left corner coordinates of the logo. They must be
9266 Specify the width and height of the logo to clear. They must be
9270 Specify the thickness of the fuzzy edge of the rectangle (added to
9271 @var{w} and @var{h}). The default value is 1. This option is
9272 deprecated, setting higher values should no longer be necessary and
9276 When set to 1, a green rectangle is drawn on the screen to simplify
9277 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9278 The default value is 0.
9280 The rectangle is drawn on the outermost pixels which will be (partly)
9281 replaced with interpolated values. The values of the next pixels
9282 immediately outside this rectangle in each direction will be used to
9283 compute the interpolated pixel values inside the rectangle.
9287 @subsection Examples
9291 Set a rectangle covering the area with top left corner coordinates 0,0
9292 and size 100x77, and a band of size 10:
9294 delogo=x=0:y=0:w=100:h=77:band=10
9302 Remove the rain in the input image/video by applying the derain methods based on
9303 convolutional neural networks. Supported models:
9307 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9308 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9311 Training as well as model generation scripts are provided in
9312 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9314 Native model files (.model) can be generated from TensorFlow model
9315 files (.pb) by using tools/python/convert.py
9317 The filter accepts the following options:
9321 Specify which filter to use. This option accepts the following values:
9325 Derain filter. To conduct derain filter, you need to use a derain model.
9328 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9330 Default value is @samp{derain}.
9333 Specify which DNN backend to use for model loading and execution. This option accepts
9334 the following values:
9338 Native implementation of DNN loading and execution.
9341 TensorFlow backend. To enable this backend you
9342 need to install the TensorFlow for C library (see
9343 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9344 @code{--enable-libtensorflow}
9346 Default value is @samp{native}.
9349 Set path to model file specifying network architecture and its parameters.
9350 Note that different backends use different file formats. TensorFlow and native
9351 backend can load files for only its format.
9354 It can also be finished with @ref{dnn_processing} filter.
9358 Attempt to fix small changes in horizontal and/or vertical shift. This
9359 filter helps remove camera shake from hand-holding a camera, bumping a
9360 tripod, moving on a vehicle, etc.
9362 The filter accepts the following options:
9370 Specify a rectangular area where to limit the search for motion
9372 If desired the search for motion vectors can be limited to a
9373 rectangular area of the frame defined by its top left corner, width
9374 and height. These parameters have the same meaning as the drawbox
9375 filter which can be used to visualise the position of the bounding
9378 This is useful when simultaneous movement of subjects within the frame
9379 might be confused for camera motion by the motion vector search.
9381 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9382 then the full frame is used. This allows later options to be set
9383 without specifying the bounding box for the motion vector search.
9385 Default - search the whole frame.
9389 Specify the maximum extent of movement in x and y directions in the
9390 range 0-64 pixels. Default 16.
9393 Specify how to generate pixels to fill blanks at the edge of the
9394 frame. Available values are:
9397 Fill zeroes at blank locations
9399 Original image at blank locations
9401 Extruded edge value at blank locations
9403 Mirrored edge at blank locations
9405 Default value is @samp{mirror}.
9408 Specify the blocksize to use for motion search. Range 4-128 pixels,
9412 Specify the contrast threshold for blocks. Only blocks with more than
9413 the specified contrast (difference between darkest and lightest
9414 pixels) will be considered. Range 1-255, default 125.
9417 Specify the search strategy. Available values are:
9420 Set exhaustive search
9422 Set less exhaustive search.
9424 Default value is @samp{exhaustive}.
9427 If set then a detailed log of the motion search is written to the
9434 Remove unwanted contamination of foreground colors, caused by reflected color of
9435 greenscreen or bluescreen.
9437 This filter accepts the following options:
9441 Set what type of despill to use.
9444 Set how spillmap will be generated.
9447 Set how much to get rid of still remaining spill.
9450 Controls amount of red in spill area.
9453 Controls amount of green in spill area.
9454 Should be -1 for greenscreen.
9457 Controls amount of blue in spill area.
9458 Should be -1 for bluescreen.
9461 Controls brightness of spill area, preserving colors.
9464 Modify alpha from generated spillmap.
9467 @subsection Commands
9469 This filter supports the all above options as @ref{commands}.
9473 Apply an exact inverse of the telecine operation. It requires a predefined
9474 pattern specified using the pattern option which must be the same as that passed
9475 to the telecine filter.
9477 This filter accepts the following options:
9486 The default value is @code{top}.
9490 A string of numbers representing the pulldown pattern you wish to apply.
9491 The default value is @code{23}.
9494 A number representing position of the first frame with respect to the telecine
9495 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9500 Apply dilation effect to the video.
9502 This filter replaces the pixel by the local(3x3) maximum.
9504 It accepts the following options:
9511 Limit the maximum change for each plane, default is 65535.
9512 If 0, plane will remain unchanged.
9515 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9518 Flags to local 3x3 coordinates maps like this:
9525 @subsection Commands
9527 This filter supports the all above options as @ref{commands}.
9531 Displace pixels as indicated by second and third input stream.
9533 It takes three input streams and outputs one stream, the first input is the
9534 source, and second and third input are displacement maps.
9536 The second input specifies how much to displace pixels along the
9537 x-axis, while the third input specifies how much to displace pixels
9539 If one of displacement map streams terminates, last frame from that
9540 displacement map will be used.
9542 Note that once generated, displacements maps can be reused over and over again.
9544 A description of the accepted options follows.
9548 Set displace behavior for pixels that are out of range.
9550 Available values are:
9553 Missing pixels are replaced by black pixels.
9556 Adjacent pixels will spread out to replace missing pixels.
9559 Out of range pixels are wrapped so they point to pixels of other side.
9562 Out of range pixels will be replaced with mirrored pixels.
9564 Default is @samp{smear}.
9568 @subsection Examples
9572 Add ripple effect to rgb input of video size hd720:
9574 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
9578 Add wave effect to rgb input of video size hd720:
9580 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
9584 @anchor{dnn_processing}
9585 @section dnn_processing
9587 Do image processing with deep neural networks. It works together with another filter
9588 which converts the pixel format of the Frame to what the dnn network requires.
9590 The filter accepts the following options:
9594 Specify which DNN backend to use for model loading and execution. This option accepts
9595 the following values:
9599 Native implementation of DNN loading and execution.
9602 TensorFlow backend. To enable this backend you
9603 need to install the TensorFlow for C library (see
9604 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9605 @code{--enable-libtensorflow}
9608 OpenVINO backend. To enable this backend you
9609 need to build and install the OpenVINO for C library (see
9610 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9611 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9612 be needed if the header files and libraries are not installed into system path)
9616 Default value is @samp{native}.
9619 Set path to model file specifying network architecture and its parameters.
9620 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9621 backend can load files for only its format.
9623 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9626 Set the input name of the dnn network.
9629 Set the output name of the dnn network.
9633 @subsection Examples
9637 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9639 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9643 Halve the pixel value of the frame with format gray32f:
9645 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
9649 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9651 ./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
9655 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9657 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9664 Draw a colored box on the input image.
9666 It accepts the following parameters:
9671 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9675 The expressions which specify the width and height of the box; if 0 they are interpreted as
9676 the input width and height. It defaults to 0.
9679 Specify the color of the box to write. For the general syntax of this option,
9680 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9681 value @code{invert} is used, the box edge color is the same as the
9682 video with inverted luma.
9685 The expression which sets the thickness of the box edge.
9686 A value of @code{fill} will create a filled box. Default value is @code{3}.
9688 See below for the list of accepted constants.
9691 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9692 will overwrite the video's color and alpha pixels.
9693 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9696 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9697 following constants:
9701 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9705 horizontal and vertical chroma subsample values. For example for the
9706 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9710 The input width and height.
9713 The input sample aspect ratio.
9717 The x and y offset coordinates where the box is drawn.
9721 The width and height of the drawn box.
9724 The thickness of the drawn box.
9726 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9727 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9731 @subsection Examples
9735 Draw a black box around the edge of the input image:
9741 Draw a box with color red and an opacity of 50%:
9743 drawbox=10:20:200:60:red@@0.5
9746 The previous example can be specified as:
9748 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9752 Fill the box with pink color:
9754 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9758 Draw a 2-pixel red 2.40:1 mask:
9760 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
9764 @subsection Commands
9765 This filter supports same commands as options.
9766 The command accepts the same syntax of the corresponding option.
9768 If the specified expression is not valid, it is kept at its current
9773 Draw a graph using input video metadata.
9775 It accepts the following parameters:
9779 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9782 Set 1st foreground color expression.
9785 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9788 Set 2nd foreground color expression.
9791 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9794 Set 3rd foreground color expression.
9797 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9800 Set 4th foreground color expression.
9803 Set minimal value of metadata value.
9806 Set maximal value of metadata value.
9809 Set graph background color. Default is white.
9814 Available values for mode is:
9821 Default is @code{line}.
9826 Available values for slide is:
9829 Draw new frame when right border is reached.
9832 Replace old columns with new ones.
9835 Scroll from right to left.
9838 Scroll from left to right.
9841 Draw single picture.
9844 Default is @code{frame}.
9847 Set size of graph video. For the syntax of this option, check the
9848 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9849 The default value is @code{900x256}.
9852 Set the output frame rate. Default value is @code{25}.
9854 The foreground color expressions can use the following variables:
9857 Minimal value of metadata value.
9860 Maximal value of metadata value.
9863 Current metadata key value.
9866 The color is defined as 0xAABBGGRR.
9869 Example using metadata from @ref{signalstats} filter:
9871 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9874 Example using metadata from @ref{ebur128} filter:
9876 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9881 Draw a grid on the input image.
9883 It accepts the following parameters:
9888 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9892 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9893 input width and height, respectively, minus @code{thickness}, so image gets
9894 framed. Default to 0.
9897 Specify the color of the grid. For the general syntax of this option,
9898 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9899 value @code{invert} is used, the grid color is the same as the
9900 video with inverted luma.
9903 The expression which sets the thickness of the grid line. Default value is @code{1}.
9905 See below for the list of accepted constants.
9908 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9909 will overwrite the video's color and alpha pixels.
9910 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9913 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9914 following constants:
9918 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9922 horizontal and vertical chroma subsample values. For example for the
9923 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9927 The input grid cell width and height.
9930 The input sample aspect ratio.
9934 The x and y coordinates of some point of grid intersection (meant to configure offset).
9938 The width and height of the drawn cell.
9941 The thickness of the drawn cell.
9943 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9944 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9948 @subsection Examples
9952 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9954 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9958 Draw a white 3x3 grid with an opacity of 50%:
9960 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
9964 @subsection Commands
9965 This filter supports same commands as options.
9966 The command accepts the same syntax of the corresponding option.
9968 If the specified expression is not valid, it is kept at its current
9974 Draw a text string or text from a specified file on top of a video, using the
9975 libfreetype library.
9977 To enable compilation of this filter, you need to configure FFmpeg with
9978 @code{--enable-libfreetype}.
9979 To enable default font fallback and the @var{font} option you need to
9980 configure FFmpeg with @code{--enable-libfontconfig}.
9981 To enable the @var{text_shaping} option, you need to configure FFmpeg with
9982 @code{--enable-libfribidi}.
9986 It accepts the following parameters:
9991 Used to draw a box around text using the background color.
9992 The value must be either 1 (enable) or 0 (disable).
9993 The default value of @var{box} is 0.
9996 Set the width of the border to be drawn around the box using @var{boxcolor}.
9997 The default value of @var{boxborderw} is 0.
10000 The color to be used for drawing box around text. For the syntax of this
10001 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10003 The default value of @var{boxcolor} is "white".
10006 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10007 The default value of @var{line_spacing} is 0.
10010 Set the width of the border to be drawn around the text using @var{bordercolor}.
10011 The default value of @var{borderw} is 0.
10014 Set the color to be used for drawing border around text. For the syntax of this
10015 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10017 The default value of @var{bordercolor} is "black".
10020 Select how the @var{text} is expanded. Can be either @code{none},
10021 @code{strftime} (deprecated) or
10022 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10026 Set a start time for the count. Value is in microseconds. Only applied
10027 in the deprecated strftime expansion mode. To emulate in normal expansion
10028 mode use the @code{pts} function, supplying the start time (in seconds)
10029 as the second argument.
10032 If true, check and fix text coords to avoid clipping.
10035 The color to be used for drawing fonts. For the syntax of this option, check
10036 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10038 The default value of @var{fontcolor} is "black".
10040 @item fontcolor_expr
10041 String which is expanded the same way as @var{text} to obtain dynamic
10042 @var{fontcolor} value. By default this option has empty value and is not
10043 processed. When this option is set, it overrides @var{fontcolor} option.
10046 The font family to be used for drawing text. By default Sans.
10049 The font file to be used for drawing text. The path must be included.
10050 This parameter is mandatory if the fontconfig support is disabled.
10053 Draw the text applying alpha blending. The value can
10054 be a number between 0.0 and 1.0.
10055 The expression accepts the same variables @var{x, y} as well.
10056 The default value is 1.
10057 Please see @var{fontcolor_expr}.
10060 The font size to be used for drawing text.
10061 The default value of @var{fontsize} is 16.
10064 If set to 1, attempt to shape the text (for example, reverse the order of
10065 right-to-left text and join Arabic characters) before drawing it.
10066 Otherwise, just draw the text exactly as given.
10067 By default 1 (if supported).
10069 @item ft_load_flags
10070 The flags to be used for loading the fonts.
10072 The flags map the corresponding flags supported by libfreetype, and are
10073 a combination of the following values:
10080 @item vertical_layout
10081 @item force_autohint
10084 @item ignore_global_advance_width
10086 @item ignore_transform
10088 @item linear_design
10092 Default value is "default".
10094 For more information consult the documentation for the FT_LOAD_*
10098 The color to be used for drawing a shadow behind the drawn text. For the
10099 syntax of this option, check the @ref{color syntax,,"Color" section in the
10100 ffmpeg-utils manual,ffmpeg-utils}.
10102 The default value of @var{shadowcolor} is "black".
10106 The x and y offsets for the text shadow position with respect to the
10107 position of the text. They can be either positive or negative
10108 values. The default value for both is "0".
10111 The starting frame number for the n/frame_num variable. The default value
10115 The size in number of spaces to use for rendering the tab.
10116 Default value is 4.
10119 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10120 format. It can be used with or without text parameter. @var{timecode_rate}
10121 option must be specified.
10123 @item timecode_rate, rate, r
10124 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10125 integer. Minimum value is "1".
10126 Drop-frame timecode is supported for frame rates 30 & 60.
10129 If set to 1, the output of the timecode option will wrap around at 24 hours.
10130 Default is 0 (disabled).
10133 The text string to be drawn. The text must be a sequence of UTF-8
10134 encoded characters.
10135 This parameter is mandatory if no file is specified with the parameter
10139 A text file containing text to be drawn. The text must be a sequence
10140 of UTF-8 encoded characters.
10142 This parameter is mandatory if no text string is specified with the
10143 parameter @var{text}.
10145 If both @var{text} and @var{textfile} are specified, an error is thrown.
10148 If set to 1, the @var{textfile} will be reloaded before each frame.
10149 Be sure to update it atomically, or it may be read partially, or even fail.
10153 The expressions which specify the offsets where text will be drawn
10154 within the video frame. They are relative to the top/left border of the
10157 The default value of @var{x} and @var{y} is "0".
10159 See below for the list of accepted constants and functions.
10162 The parameters for @var{x} and @var{y} are expressions containing the
10163 following constants and functions:
10167 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10171 horizontal and vertical chroma subsample values. For example for the
10172 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10175 the height of each text line
10183 @item max_glyph_a, ascent
10184 the maximum distance from the baseline to the highest/upper grid
10185 coordinate used to place a glyph outline point, for all the rendered
10187 It is a positive value, due to the grid's orientation with the Y axis
10190 @item max_glyph_d, descent
10191 the maximum distance from the baseline to the lowest grid coordinate
10192 used to place a glyph outline point, for all the rendered glyphs.
10193 This is a negative value, due to the grid's orientation, with the Y axis
10197 maximum glyph height, that is the maximum height for all the glyphs
10198 contained in the rendered text, it is equivalent to @var{ascent} -
10202 maximum glyph width, that is the maximum width for all the glyphs
10203 contained in the rendered text
10206 the number of input frame, starting from 0
10208 @item rand(min, max)
10209 return a random number included between @var{min} and @var{max}
10212 The input sample aspect ratio.
10215 timestamp expressed in seconds, NAN if the input timestamp is unknown
10218 the height of the rendered text
10221 the width of the rendered text
10225 the x and y offset coordinates where the text is drawn.
10227 These parameters allow the @var{x} and @var{y} expressions to refer
10228 to each other, so you can for example specify @code{y=x/dar}.
10231 A one character description of the current frame's picture type.
10234 The current packet's position in the input file or stream
10235 (in bytes, from the start of the input). A value of -1 indicates
10236 this info is not available.
10239 The current packet's duration, in seconds.
10242 The current packet's size (in bytes).
10245 @anchor{drawtext_expansion}
10246 @subsection Text expansion
10248 If @option{expansion} is set to @code{strftime},
10249 the filter recognizes strftime() sequences in the provided text and
10250 expands them accordingly. Check the documentation of strftime(). This
10251 feature is deprecated.
10253 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10255 If @option{expansion} is set to @code{normal} (which is the default),
10256 the following expansion mechanism is used.
10258 The backslash character @samp{\}, followed by any character, always expands to
10259 the second character.
10261 Sequences of the form @code{%@{...@}} are expanded. The text between the
10262 braces is a function name, possibly followed by arguments separated by ':'.
10263 If the arguments contain special characters or delimiters (':' or '@}'),
10264 they should be escaped.
10266 Note that they probably must also be escaped as the value for the
10267 @option{text} option in the filter argument string and as the filter
10268 argument in the filtergraph description, and possibly also for the shell,
10269 that makes up to four levels of escaping; using a text file avoids these
10272 The following functions are available:
10277 The expression evaluation result.
10279 It must take one argument specifying the expression to be evaluated,
10280 which accepts the same constants and functions as the @var{x} and
10281 @var{y} values. Note that not all constants should be used, for
10282 example the text size is not known when evaluating the expression, so
10283 the constants @var{text_w} and @var{text_h} will have an undefined
10286 @item expr_int_format, eif
10287 Evaluate the expression's value and output as formatted integer.
10289 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10290 The second argument specifies the output format. Allowed values are @samp{x},
10291 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10292 @code{printf} function.
10293 The third parameter is optional and sets the number of positions taken by the output.
10294 It can be used to add padding with zeros from the left.
10297 The time at which the filter is running, expressed in UTC.
10298 It can accept an argument: a strftime() format string.
10301 The time at which the filter is running, expressed in the local time zone.
10302 It can accept an argument: a strftime() format string.
10305 Frame metadata. Takes one or two arguments.
10307 The first argument is mandatory and specifies the metadata key.
10309 The second argument is optional and specifies a default value, used when the
10310 metadata key is not found or empty.
10312 Available metadata can be identified by inspecting entries
10313 starting with TAG included within each frame section
10314 printed by running @code{ffprobe -show_frames}.
10316 String metadata generated in filters leading to
10317 the drawtext filter are also available.
10320 The frame number, starting from 0.
10323 A one character description of the current picture type.
10326 The timestamp of the current frame.
10327 It can take up to three arguments.
10329 The first argument is the format of the timestamp; it defaults to @code{flt}
10330 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10331 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10332 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10333 @code{localtime} stands for the timestamp of the frame formatted as
10334 local time zone time.
10336 The second argument is an offset added to the timestamp.
10338 If the format is set to @code{hms}, a third argument @code{24HH} may be
10339 supplied to present the hour part of the formatted timestamp in 24h format
10342 If the format is set to @code{localtime} or @code{gmtime},
10343 a third argument may be supplied: a strftime() format string.
10344 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10347 @subsection Commands
10349 This filter supports altering parameters via commands:
10352 Alter existing filter parameters.
10354 Syntax for the argument is the same as for filter invocation, e.g.
10357 fontsize=56:fontcolor=green:text='Hello World'
10360 Full filter invocation with sendcmd would look like this:
10363 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10367 If the entire argument can't be parsed or applied as valid values then the filter will
10368 continue with its existing parameters.
10370 @subsection Examples
10374 Draw "Test Text" with font FreeSerif, using the default values for the
10375 optional parameters.
10378 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10382 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10383 and y=50 (counting from the top-left corner of the screen), text is
10384 yellow with a red box around it. Both the text and the box have an
10388 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10389 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10392 Note that the double quotes are not necessary if spaces are not used
10393 within the parameter list.
10396 Show the text at the center of the video frame:
10398 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10402 Show the text at a random position, switching to a new position every 30 seconds:
10404 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)"
10408 Show a text line sliding from right to left in the last row of the video
10409 frame. The file @file{LONG_LINE} is assumed to contain a single line
10412 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10416 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10418 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10422 Draw a single green letter "g", at the center of the input video.
10423 The glyph baseline is placed at half screen height.
10425 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10429 Show text for 1 second every 3 seconds:
10431 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10435 Use fontconfig to set the font. Note that the colons need to be escaped.
10437 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10441 Draw "Test Text" with font size dependent on height of the video.
10443 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10447 Print the date of a real-time encoding (see strftime(3)):
10449 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10453 Show text fading in and out (appearing/disappearing):
10456 DS=1.0 # display start
10457 DE=10.0 # display end
10458 FID=1.5 # fade in duration
10459 FOD=5 # fade out duration
10460 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 @}"
10464 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10465 and the @option{fontsize} value are included in the @option{y} offset.
10467 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10468 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10472 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10473 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10474 must have option @option{-export_path_metadata 1} for the special metadata fields
10475 to be available for filters.
10477 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10482 For more information about libfreetype, check:
10483 @url{http://www.freetype.org/}.
10485 For more information about fontconfig, check:
10486 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10488 For more information about libfribidi, check:
10489 @url{http://fribidi.org/}.
10491 @section edgedetect
10493 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10495 The filter accepts the following options:
10500 Set low and high threshold values used by the Canny thresholding
10503 The high threshold selects the "strong" edge pixels, which are then
10504 connected through 8-connectivity with the "weak" edge pixels selected
10505 by the low threshold.
10507 @var{low} and @var{high} threshold values must be chosen in the range
10508 [0,1], and @var{low} should be lesser or equal to @var{high}.
10510 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10514 Define the drawing mode.
10518 Draw white/gray wires on black background.
10521 Mix the colors to create a paint/cartoon effect.
10524 Apply Canny edge detector on all selected planes.
10526 Default value is @var{wires}.
10529 Select planes for filtering. By default all available planes are filtered.
10532 @subsection Examples
10536 Standard edge detection with custom values for the hysteresis thresholding:
10538 edgedetect=low=0.1:high=0.4
10542 Painting effect without thresholding:
10544 edgedetect=mode=colormix:high=0
10550 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10552 For each input image, the filter will compute the optimal mapping from
10553 the input to the output given the codebook length, that is the number
10554 of distinct output colors.
10556 This filter accepts the following options.
10559 @item codebook_length, l
10560 Set codebook length. The value must be a positive integer, and
10561 represents the number of distinct output colors. Default value is 256.
10564 Set the maximum number of iterations to apply for computing the optimal
10565 mapping. The higher the value the better the result and the higher the
10566 computation time. Default value is 1.
10569 Set a random seed, must be an integer included between 0 and
10570 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10571 will try to use a good random seed on a best effort basis.
10574 Set pal8 output pixel format. This option does not work with codebook
10575 length greater than 256.
10580 Measure graylevel entropy in histogram of color channels of video frames.
10582 It accepts the following parameters:
10586 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10588 @var{diff} mode measures entropy of histogram delta values, absolute differences
10589 between neighbour histogram values.
10593 Set brightness, contrast, saturation and approximate gamma adjustment.
10595 The filter accepts the following options:
10599 Set the contrast expression. The value must be a float value in range
10600 @code{-1000.0} to @code{1000.0}. The default value is "1".
10603 Set the brightness expression. The value must be a float value in
10604 range @code{-1.0} to @code{1.0}. The default value is "0".
10607 Set the saturation expression. The value must be a float in
10608 range @code{0.0} to @code{3.0}. The default value is "1".
10611 Set the gamma expression. The value must be a float in range
10612 @code{0.1} to @code{10.0}. The default value is "1".
10615 Set the gamma expression for red. The value must be a float in
10616 range @code{0.1} to @code{10.0}. The default value is "1".
10619 Set the gamma expression for green. The value must be a float in range
10620 @code{0.1} to @code{10.0}. The default value is "1".
10623 Set the gamma expression for blue. The value must be a float in range
10624 @code{0.1} to @code{10.0}. The default value is "1".
10627 Set the gamma weight expression. It can be used to reduce the effect
10628 of a high gamma value on bright image areas, e.g. keep them from
10629 getting overamplified and just plain white. The value must be a float
10630 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10631 gamma correction all the way down while @code{1.0} leaves it at its
10632 full strength. Default is "1".
10635 Set when the expressions for brightness, contrast, saturation and
10636 gamma expressions are evaluated.
10638 It accepts the following values:
10641 only evaluate expressions once during the filter initialization or
10642 when a command is processed
10645 evaluate expressions for each incoming frame
10648 Default value is @samp{init}.
10651 The expressions accept the following parameters:
10654 frame count of the input frame starting from 0
10657 byte position of the corresponding packet in the input file, NAN if
10661 frame rate of the input video, NAN if the input frame rate is unknown
10664 timestamp expressed in seconds, NAN if the input timestamp is unknown
10667 @subsection Commands
10668 The filter supports the following commands:
10672 Set the contrast expression.
10675 Set the brightness expression.
10678 Set the saturation expression.
10681 Set the gamma expression.
10684 Set the gamma_r expression.
10687 Set gamma_g expression.
10690 Set gamma_b expression.
10693 Set gamma_weight expression.
10695 The command accepts the same syntax of the corresponding option.
10697 If the specified expression is not valid, it is kept at its current
10704 Apply erosion effect to the video.
10706 This filter replaces the pixel by the local(3x3) minimum.
10708 It accepts the following options:
10715 Limit the maximum change for each plane, default is 65535.
10716 If 0, plane will remain unchanged.
10719 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10722 Flags to local 3x3 coordinates maps like this:
10729 @subsection Commands
10731 This filter supports the all above options as @ref{commands}.
10733 @section extractplanes
10735 Extract color channel components from input video stream into
10736 separate grayscale video streams.
10738 The filter accepts the following option:
10742 Set plane(s) to extract.
10744 Available values for planes are:
10755 Choosing planes not available in the input will result in an error.
10756 That means you cannot select @code{r}, @code{g}, @code{b} planes
10757 with @code{y}, @code{u}, @code{v} planes at same time.
10760 @subsection Examples
10764 Extract luma, u and v color channel component from input video frame
10765 into 3 grayscale outputs:
10767 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
10773 Apply a fade-in/out effect to the input video.
10775 It accepts the following parameters:
10779 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10781 Default is @code{in}.
10783 @item start_frame, s
10784 Specify the number of the frame to start applying the fade
10785 effect at. Default is 0.
10788 The number of frames that the fade effect lasts. At the end of the
10789 fade-in effect, the output video will have the same intensity as the input video.
10790 At the end of the fade-out transition, the output video will be filled with the
10791 selected @option{color}.
10795 If set to 1, fade only alpha channel, if one exists on the input.
10796 Default value is 0.
10798 @item start_time, st
10799 Specify the timestamp (in seconds) of the frame to start to apply the fade
10800 effect. If both start_frame and start_time are specified, the fade will start at
10801 whichever comes last. Default is 0.
10804 The number of seconds for which the fade effect has to last. At the end of the
10805 fade-in effect the output video will have the same intensity as the input video,
10806 at the end of the fade-out transition the output video will be filled with the
10807 selected @option{color}.
10808 If both duration and nb_frames are specified, duration is used. Default is 0
10809 (nb_frames is used by default).
10812 Specify the color of the fade. Default is "black".
10815 @subsection Examples
10819 Fade in the first 30 frames of video:
10824 The command above is equivalent to:
10830 Fade out the last 45 frames of a 200-frame video:
10833 fade=type=out:start_frame=155:nb_frames=45
10837 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10839 fade=in:0:25, fade=out:975:25
10843 Make the first 5 frames yellow, then fade in from frame 5-24:
10845 fade=in:5:20:color=yellow
10849 Fade in alpha over first 25 frames of video:
10851 fade=in:0:25:alpha=1
10855 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10857 fade=t=in:st=5.5:d=0.5
10863 Denoise frames using 3D FFT (frequency domain filtering).
10865 The filter accepts the following options:
10869 Set the noise sigma constant. This sets denoising strength.
10870 Default value is 1. Allowed range is from 0 to 30.
10871 Using very high sigma with low overlap may give blocking artifacts.
10874 Set amount of denoising. By default all detected noise is reduced.
10875 Default value is 1. Allowed range is from 0 to 1.
10878 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10879 Actual size of block in pixels is 2 to power of @var{block}, so by default
10880 block size in pixels is 2^4 which is 16.
10883 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10886 Set number of previous frames to use for denoising. By default is set to 0.
10889 Set number of next frames to to use for denoising. By default is set to 0.
10892 Set planes which will be filtered, by default are all available filtered
10897 Apply arbitrary expressions to samples in frequency domain
10901 Adjust the dc value (gain) of the luma plane of the image. The filter
10902 accepts an integer value in range @code{0} to @code{1000}. The default
10903 value is set to @code{0}.
10906 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10907 filter accepts an integer value in range @code{0} to @code{1000}. The
10908 default value is set to @code{0}.
10911 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10912 filter accepts an integer value in range @code{0} to @code{1000}. The
10913 default value is set to @code{0}.
10916 Set the frequency domain weight expression for the luma plane.
10919 Set the frequency domain weight expression for the 1st chroma plane.
10922 Set the frequency domain weight expression for the 2nd chroma plane.
10925 Set when the expressions are evaluated.
10927 It accepts the following values:
10930 Only evaluate expressions once during the filter initialization.
10933 Evaluate expressions for each incoming frame.
10936 Default value is @samp{init}.
10938 The filter accepts the following variables:
10941 The coordinates of the current sample.
10945 The width and height of the image.
10948 The number of input frame, starting from 0.
10951 @subsection Examples
10957 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
10963 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
10969 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
10975 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
10982 Extract a single field from an interlaced image using stride
10983 arithmetic to avoid wasting CPU time. The output frames are marked as
10986 The filter accepts the following options:
10990 Specify whether to extract the top (if the value is @code{0} or
10991 @code{top}) or the bottom field (if the value is @code{1} or
10997 Create new frames by copying the top and bottom fields from surrounding frames
10998 supplied as numbers by the hint file.
11002 Set file containing hints: absolute/relative frame numbers.
11004 There must be one line for each frame in a clip. Each line must contain two
11005 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11006 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11007 is current frame number for @code{absolute} mode or out of [-1, 1] range
11008 for @code{relative} mode. First number tells from which frame to pick up top
11009 field and second number tells from which frame to pick up bottom field.
11011 If optionally followed by @code{+} output frame will be marked as interlaced,
11012 else if followed by @code{-} output frame will be marked as progressive, else
11013 it will be marked same as input frame.
11014 If optionally followed by @code{t} output frame will use only top field, or in
11015 case of @code{b} it will use only bottom field.
11016 If line starts with @code{#} or @code{;} that line is skipped.
11019 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11022 Example of first several lines of @code{hint} file for @code{relative} mode:
11024 0,0 - # first frame
11025 1,0 - # second frame, use third's frame top field and second's frame bottom field
11026 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11041 @section fieldmatch
11043 Field matching filter for inverse telecine. It is meant to reconstruct the
11044 progressive frames from a telecined stream. The filter does not drop duplicated
11045 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11046 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11048 The separation of the field matching and the decimation is notably motivated by
11049 the possibility of inserting a de-interlacing filter fallback between the two.
11050 If the source has mixed telecined and real interlaced content,
11051 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11052 But these remaining combed frames will be marked as interlaced, and thus can be
11053 de-interlaced by a later filter such as @ref{yadif} before decimation.
11055 In addition to the various configuration options, @code{fieldmatch} can take an
11056 optional second stream, activated through the @option{ppsrc} option. If
11057 enabled, the frames reconstruction will be based on the fields and frames from
11058 this second stream. This allows the first input to be pre-processed in order to
11059 help the various algorithms of the filter, while keeping the output lossless
11060 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11061 or brightness/contrast adjustments can help.
11063 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11064 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11065 which @code{fieldmatch} is based on. While the semantic and usage are very
11066 close, some behaviour and options names can differ.
11068 The @ref{decimate} filter currently only works for constant frame rate input.
11069 If your input has mixed telecined (30fps) and progressive content with a lower
11070 framerate like 24fps use the following filterchain to produce the necessary cfr
11071 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11073 The filter accepts the following options:
11077 Specify the assumed field order of the input stream. Available values are:
11081 Auto detect parity (use FFmpeg's internal parity value).
11083 Assume bottom field first.
11085 Assume top field first.
11088 Note that it is sometimes recommended not to trust the parity announced by the
11091 Default value is @var{auto}.
11094 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11095 sense that it won't risk creating jerkiness due to duplicate frames when
11096 possible, but if there are bad edits or blended fields it will end up
11097 outputting combed frames when a good match might actually exist. On the other
11098 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11099 but will almost always find a good frame if there is one. The other values are
11100 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11101 jerkiness and creating duplicate frames versus finding good matches in sections
11102 with bad edits, orphaned fields, blended fields, etc.
11104 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11106 Available values are:
11110 2-way matching (p/c)
11112 2-way matching, and trying 3rd match if still combed (p/c + n)
11114 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11116 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11117 still combed (p/c + n + u/b)
11119 3-way matching (p/c/n)
11121 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11122 detected as combed (p/c/n + u/b)
11125 The parenthesis at the end indicate the matches that would be used for that
11126 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11129 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11132 Default value is @var{pc_n}.
11135 Mark the main input stream as a pre-processed input, and enable the secondary
11136 input stream as the clean source to pick the fields from. See the filter
11137 introduction for more details. It is similar to the @option{clip2} feature from
11140 Default value is @code{0} (disabled).
11143 Set the field to match from. It is recommended to set this to the same value as
11144 @option{order} unless you experience matching failures with that setting. In
11145 certain circumstances changing the field that is used to match from can have a
11146 large impact on matching performance. Available values are:
11150 Automatic (same value as @option{order}).
11152 Match from the bottom field.
11154 Match from the top field.
11157 Default value is @var{auto}.
11160 Set whether or not chroma is included during the match comparisons. In most
11161 cases it is recommended to leave this enabled. You should set this to @code{0}
11162 only if your clip has bad chroma problems such as heavy rainbowing or other
11163 artifacts. Setting this to @code{0} could also be used to speed things up at
11164 the cost of some accuracy.
11166 Default value is @code{1}.
11170 These define an exclusion band which excludes the lines between @option{y0} and
11171 @option{y1} from being included in the field matching decision. An exclusion
11172 band can be used to ignore subtitles, a logo, or other things that may
11173 interfere with the matching. @option{y0} sets the starting scan line and
11174 @option{y1} sets the ending line; all lines in between @option{y0} and
11175 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11176 @option{y0} and @option{y1} to the same value will disable the feature.
11177 @option{y0} and @option{y1} defaults to @code{0}.
11180 Set the scene change detection threshold as a percentage of maximum change on
11181 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11182 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11183 @option{scthresh} is @code{[0.0, 100.0]}.
11185 Default value is @code{12.0}.
11188 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11189 account the combed scores of matches when deciding what match to use as the
11190 final match. Available values are:
11194 No final matching based on combed scores.
11196 Combed scores are only used when a scene change is detected.
11198 Use combed scores all the time.
11201 Default is @var{sc}.
11204 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11205 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11206 Available values are:
11210 No forced calculation.
11212 Force p/c/n calculations.
11214 Force p/c/n/u/b calculations.
11217 Default value is @var{none}.
11220 This is the area combing threshold used for combed frame detection. This
11221 essentially controls how "strong" or "visible" combing must be to be detected.
11222 Larger values mean combing must be more visible and smaller values mean combing
11223 can be less visible or strong and still be detected. Valid settings are from
11224 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11225 be detected as combed). This is basically a pixel difference value. A good
11226 range is @code{[8, 12]}.
11228 Default value is @code{9}.
11231 Sets whether or not chroma is considered in the combed frame decision. Only
11232 disable this if your source has chroma problems (rainbowing, etc.) that are
11233 causing problems for the combed frame detection with chroma enabled. Actually,
11234 using @option{chroma}=@var{0} is usually more reliable, except for the case
11235 where there is chroma only combing in the source.
11237 Default value is @code{0}.
11241 Respectively set the x-axis and y-axis size of the window used during combed
11242 frame detection. This has to do with the size of the area in which
11243 @option{combpel} pixels are required to be detected as combed for a frame to be
11244 declared combed. See the @option{combpel} parameter description for more info.
11245 Possible values are any number that is a power of 2 starting at 4 and going up
11248 Default value is @code{16}.
11251 The number of combed pixels inside any of the @option{blocky} by
11252 @option{blockx} size blocks on the frame for the frame to be detected as
11253 combed. While @option{cthresh} controls how "visible" the combing must be, this
11254 setting controls "how much" combing there must be in any localized area (a
11255 window defined by the @option{blockx} and @option{blocky} settings) on the
11256 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11257 which point no frames will ever be detected as combed). This setting is known
11258 as @option{MI} in TFM/VFM vocabulary.
11260 Default value is @code{80}.
11263 @anchor{p/c/n/u/b meaning}
11264 @subsection p/c/n/u/b meaning
11266 @subsubsection p/c/n
11268 We assume the following telecined stream:
11271 Top fields: 1 2 2 3 4
11272 Bottom fields: 1 2 3 4 4
11275 The numbers correspond to the progressive frame the fields relate to. Here, the
11276 first two frames are progressive, the 3rd and 4th are combed, and so on.
11278 When @code{fieldmatch} is configured to run a matching from bottom
11279 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11284 B 1 2 3 4 4 <-- matching reference
11293 As a result of the field matching, we can see that some frames get duplicated.
11294 To perform a complete inverse telecine, you need to rely on a decimation filter
11295 after this operation. See for instance the @ref{decimate} filter.
11297 The same operation now matching from top fields (@option{field}=@var{top})
11302 T 1 2 2 3 4 <-- matching reference
11312 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11313 basically, they refer to the frame and field of the opposite parity:
11316 @item @var{p} matches the field of the opposite parity in the previous frame
11317 @item @var{c} matches the field of the opposite parity in the current frame
11318 @item @var{n} matches the field of the opposite parity in the next frame
11323 The @var{u} and @var{b} matching are a bit special in the sense that they match
11324 from the opposite parity flag. In the following examples, we assume that we are
11325 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11326 'x' is placed above and below each matched fields.
11328 With bottom matching (@option{field}=@var{bottom}):
11333 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11334 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11342 With top matching (@option{field}=@var{top}):
11347 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11348 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11356 @subsection Examples
11358 Simple IVTC of a top field first telecined stream:
11360 fieldmatch=order=tff:combmatch=none, decimate
11363 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11365 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11368 @section fieldorder
11370 Transform the field order of the input video.
11372 It accepts the following parameters:
11377 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11378 for bottom field first.
11381 The default value is @samp{tff}.
11383 The transformation is done by shifting the picture content up or down
11384 by one line, and filling the remaining line with appropriate picture content.
11385 This method is consistent with most broadcast field order converters.
11387 If the input video is not flagged as being interlaced, or it is already
11388 flagged as being of the required output field order, then this filter does
11389 not alter the incoming video.
11391 It is very useful when converting to or from PAL DV material,
11392 which is bottom field first.
11396 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11399 @section fifo, afifo
11401 Buffer input images and send them when they are requested.
11403 It is mainly useful when auto-inserted by the libavfilter
11406 It does not take parameters.
11408 @section fillborders
11410 Fill borders of the input video, without changing video stream dimensions.
11411 Sometimes video can have garbage at the four edges and you may not want to
11412 crop video input to keep size multiple of some number.
11414 This filter accepts the following options:
11418 Number of pixels to fill from left border.
11421 Number of pixels to fill from right border.
11424 Number of pixels to fill from top border.
11427 Number of pixels to fill from bottom border.
11432 It accepts the following values:
11435 fill pixels using outermost pixels
11438 fill pixels using mirroring
11441 fill pixels with constant value
11444 Default is @var{smear}.
11447 Set color for pixels in fixed mode. Default is @var{black}.
11450 @subsection Commands
11451 This filter supports same @ref{commands} as options.
11452 The command accepts the same syntax of the corresponding option.
11454 If the specified expression is not valid, it is kept at its current
11459 Find a rectangular object
11461 It accepts the following options:
11465 Filepath of the object image, needs to be in gray8.
11468 Detection threshold, default is 0.5.
11471 Number of mipmaps, default is 3.
11473 @item xmin, ymin, xmax, ymax
11474 Specifies the rectangle in which to search.
11477 @subsection Examples
11481 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11483 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11489 Flood area with values of same pixel components with another values.
11491 It accepts the following options:
11494 Set pixel x coordinate.
11497 Set pixel y coordinate.
11500 Set source #0 component value.
11503 Set source #1 component value.
11506 Set source #2 component value.
11509 Set source #3 component value.
11512 Set destination #0 component value.
11515 Set destination #1 component value.
11518 Set destination #2 component value.
11521 Set destination #3 component value.
11527 Convert the input video to one of the specified pixel formats.
11528 Libavfilter will try to pick one that is suitable as input to
11531 It accepts the following parameters:
11535 A '|'-separated list of pixel format names, such as
11536 "pix_fmts=yuv420p|monow|rgb24".
11540 @subsection Examples
11544 Convert the input video to the @var{yuv420p} format
11546 format=pix_fmts=yuv420p
11549 Convert the input video to any of the formats in the list
11551 format=pix_fmts=yuv420p|yuv444p|yuv410p
11558 Convert the video to specified constant frame rate by duplicating or dropping
11559 frames as necessary.
11561 It accepts the following parameters:
11565 The desired output frame rate. The default is @code{25}.
11568 Assume the first PTS should be the given value, in seconds. This allows for
11569 padding/trimming at the start of stream. By default, no assumption is made
11570 about the first frame's expected PTS, so no padding or trimming is done.
11571 For example, this could be set to 0 to pad the beginning with duplicates of
11572 the first frame if a video stream starts after the audio stream or to trim any
11573 frames with a negative PTS.
11576 Timestamp (PTS) rounding method.
11578 Possible values are:
11585 round towards -infinity
11587 round towards +infinity
11591 The default is @code{near}.
11594 Action performed when reading the last frame.
11596 Possible values are:
11599 Use same timestamp rounding method as used for other frames.
11601 Pass through last frame if input duration has not been reached yet.
11603 The default is @code{round}.
11607 Alternatively, the options can be specified as a flat string:
11608 @var{fps}[:@var{start_time}[:@var{round}]].
11610 See also the @ref{setpts} filter.
11612 @subsection Examples
11616 A typical usage in order to set the fps to 25:
11622 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11624 fps=fps=film:round=near
11630 Pack two different video streams into a stereoscopic video, setting proper
11631 metadata on supported codecs. The two views should have the same size and
11632 framerate and processing will stop when the shorter video ends. Please note
11633 that you may conveniently adjust view properties with the @ref{scale} and
11636 It accepts the following parameters:
11640 The desired packing format. Supported values are:
11645 The views are next to each other (default).
11648 The views are on top of each other.
11651 The views are packed by line.
11654 The views are packed by column.
11657 The views are temporally interleaved.
11666 # Convert left and right views into a frame-sequential video
11667 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11669 # Convert views into a side-by-side video with the same output resolution as the input
11670 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
11675 Change the frame rate by interpolating new video output frames from the source
11678 This filter is not designed to function correctly with interlaced media. If
11679 you wish to change the frame rate of interlaced media then you are required
11680 to deinterlace before this filter and re-interlace after this filter.
11682 A description of the accepted options follows.
11686 Specify the output frames per second. This option can also be specified
11687 as a value alone. The default is @code{50}.
11690 Specify the start of a range where the output frame will be created as a
11691 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11692 the default is @code{15}.
11695 Specify the end of a range where the output frame will be created as a
11696 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11697 the default is @code{240}.
11700 Specify the level at which a scene change is detected as a value between
11701 0 and 100 to indicate a new scene; a low value reflects a low
11702 probability for the current frame to introduce a new scene, while a higher
11703 value means the current frame is more likely to be one.
11704 The default is @code{8.2}.
11707 Specify flags influencing the filter process.
11709 Available value for @var{flags} is:
11712 @item scene_change_detect, scd
11713 Enable scene change detection using the value of the option @var{scene}.
11714 This flag is enabled by default.
11720 Select one frame every N-th frame.
11722 This filter accepts the following option:
11725 Select frame after every @code{step} frames.
11726 Allowed values are positive integers higher than 0. Default value is @code{1}.
11729 @section freezedetect
11731 Detect frozen video.
11733 This filter logs a message and sets frame metadata when it detects that the
11734 input video has no significant change in content during a specified duration.
11735 Video freeze detection calculates the mean average absolute difference of all
11736 the components of video frames and compares it to a noise floor.
11738 The printed times and duration are expressed in seconds. The
11739 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11740 whose timestamp equals or exceeds the detection duration and it contains the
11741 timestamp of the first frame of the freeze. The
11742 @code{lavfi.freezedetect.freeze_duration} and
11743 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11746 The filter accepts the following options:
11750 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11751 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11755 Set freeze duration until notification (default is 2 seconds).
11758 @section freezeframes
11760 Freeze video frames.
11762 This filter freezes video frames using frame from 2nd input.
11764 The filter accepts the following options:
11768 Set number of first frame from which to start freeze.
11771 Set number of last frame from which to end freeze.
11774 Set number of frame from 2nd input which will be used instead of replaced frames.
11780 Apply a frei0r effect to the input video.
11782 To enable the compilation of this filter, you need to install the frei0r
11783 header and configure FFmpeg with @code{--enable-frei0r}.
11785 It accepts the following parameters:
11790 The name of the frei0r effect to load. If the environment variable
11791 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11792 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11793 Otherwise, the standard frei0r paths are searched, in this order:
11794 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11795 @file{/usr/lib/frei0r-1/}.
11797 @item filter_params
11798 A '|'-separated list of parameters to pass to the frei0r effect.
11802 A frei0r effect parameter can be a boolean (its value is either
11803 "y" or "n"), a double, a color (specified as
11804 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11805 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11806 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11807 a position (specified as @var{X}/@var{Y}, where
11808 @var{X} and @var{Y} are floating point numbers) and/or a string.
11810 The number and types of parameters depend on the loaded effect. If an
11811 effect parameter is not specified, the default value is set.
11813 @subsection Examples
11817 Apply the distort0r effect, setting the first two double parameters:
11819 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11823 Apply the colordistance effect, taking a color as the first parameter:
11825 frei0r=colordistance:0.2/0.3/0.4
11826 frei0r=colordistance:violet
11827 frei0r=colordistance:0x112233
11831 Apply the perspective effect, specifying the top left and top right image
11834 frei0r=perspective:0.2/0.2|0.8/0.2
11838 For more information, see
11839 @url{http://frei0r.dyne.org}
11841 @subsection Commands
11843 This filter supports the @option{filter_params} option as @ref{commands}.
11847 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11849 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11850 processing filter, one of them is performed once per block, not per pixel.
11851 This allows for much higher speed.
11853 The filter accepts the following options:
11857 Set quality. This option defines the number of levels for averaging. It accepts
11858 an integer in the range 4-5. Default value is @code{4}.
11861 Force a constant quantization parameter. It accepts an integer in range 0-63.
11862 If not set, the filter will use the QP from the video stream (if available).
11865 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11866 more details but also more artifacts, while higher values make the image smoother
11867 but also blurrier. Default value is @code{0} − PSNR optimal.
11869 @item use_bframe_qp
11870 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11871 option may cause flicker since the B-Frames have often larger QP. Default is
11872 @code{0} (not enabled).
11878 Apply Gaussian blur filter.
11880 The filter accepts the following options:
11884 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11887 Set number of steps for Gaussian approximation. Default is @code{1}.
11890 Set which planes to filter. By default all planes are filtered.
11893 Set vertical sigma, if negative it will be same as @code{sigma}.
11894 Default is @code{-1}.
11897 @subsection Commands
11898 This filter supports same commands as options.
11899 The command accepts the same syntax of the corresponding option.
11901 If the specified expression is not valid, it is kept at its current
11906 Apply generic equation to each pixel.
11908 The filter accepts the following options:
11911 @item lum_expr, lum
11912 Set the luminance expression.
11914 Set the chrominance blue expression.
11916 Set the chrominance red expression.
11917 @item alpha_expr, a
11918 Set the alpha expression.
11920 Set the red expression.
11921 @item green_expr, g
11922 Set the green expression.
11924 Set the blue expression.
11927 The colorspace is selected according to the specified options. If one
11928 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11929 options is specified, the filter will automatically select a YCbCr
11930 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11931 @option{blue_expr} options is specified, it will select an RGB
11934 If one of the chrominance expression is not defined, it falls back on the other
11935 one. If no alpha expression is specified it will evaluate to opaque value.
11936 If none of chrominance expressions are specified, they will evaluate
11937 to the luminance expression.
11939 The expressions can use the following variables and functions:
11943 The sequential number of the filtered frame, starting from @code{0}.
11947 The coordinates of the current sample.
11951 The width and height of the image.
11955 Width and height scale depending on the currently filtered plane. It is the
11956 ratio between the corresponding luma plane number of pixels and the current
11957 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11958 @code{0.5,0.5} for chroma planes.
11961 Time of the current frame, expressed in seconds.
11964 Return the value of the pixel at location (@var{x},@var{y}) of the current
11968 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
11972 Return the value of the pixel at location (@var{x},@var{y}) of the
11973 blue-difference chroma plane. Return 0 if there is no such plane.
11976 Return the value of the pixel at location (@var{x},@var{y}) of the
11977 red-difference chroma plane. Return 0 if there is no such plane.
11982 Return the value of the pixel at location (@var{x},@var{y}) of the
11983 red/green/blue component. Return 0 if there is no such component.
11986 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
11987 plane. Return 0 if there is no such plane.
11989 @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)
11990 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
11991 sums of samples within a rectangle. See the functions without the sum postfix.
11993 @item interpolation
11994 Set one of interpolation methods:
11999 Default is bilinear.
12002 For functions, if @var{x} and @var{y} are outside the area, the value will be
12003 automatically clipped to the closer edge.
12005 Please note that this filter can use multiple threads in which case each slice
12006 will have its own expression state. If you want to use only a single expression
12007 state because your expressions depend on previous state then you should limit
12008 the number of filter threads to 1.
12010 @subsection Examples
12014 Flip the image horizontally:
12020 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12021 wavelength of 100 pixels:
12023 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12027 Generate a fancy enigmatic moving light:
12029 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
12033 Generate a quick emboss effect:
12035 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12039 Modify RGB components depending on pixel position:
12041 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12045 Create a radial gradient that is the same size as the input (also see
12046 the @ref{vignette} filter):
12048 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12054 Fix the banding artifacts that are sometimes introduced into nearly flat
12055 regions by truncation to 8-bit color depth.
12056 Interpolate the gradients that should go where the bands are, and
12059 It is designed for playback only. Do not use it prior to
12060 lossy compression, because compression tends to lose the dither and
12061 bring back the bands.
12063 It accepts the following parameters:
12068 The maximum amount by which the filter will change any one pixel. This is also
12069 the threshold for detecting nearly flat regions. Acceptable values range from
12070 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12074 The neighborhood to fit the gradient to. A larger radius makes for smoother
12075 gradients, but also prevents the filter from modifying the pixels near detailed
12076 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12077 values will be clipped to the valid range.
12081 Alternatively, the options can be specified as a flat string:
12082 @var{strength}[:@var{radius}]
12084 @subsection Examples
12088 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12094 Specify radius, omitting the strength (which will fall-back to the default
12102 @anchor{graphmonitor}
12103 @section graphmonitor
12104 Show various filtergraph stats.
12106 With this filter one can debug complete filtergraph.
12107 Especially issues with links filling with queued frames.
12109 The filter accepts the following options:
12113 Set video output size. Default is @var{hd720}.
12116 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12119 Set output mode, can be @var{fulll} or @var{compact}.
12120 In @var{compact} mode only filters with some queued frames have displayed stats.
12123 Set flags which enable which stats are shown in video.
12125 Available values for flags are:
12128 Display number of queued frames in each link.
12130 @item frame_count_in
12131 Display number of frames taken from filter.
12133 @item frame_count_out
12134 Display number of frames given out from filter.
12137 Display current filtered frame pts.
12140 Display current filtered frame time.
12143 Display time base for filter link.
12146 Display used format for filter link.
12149 Display video size or number of audio channels in case of audio used by filter link.
12152 Display video frame rate or sample rate in case of audio used by filter link.
12155 Display link output status.
12159 Set upper limit for video rate of output stream, Default value is @var{25}.
12160 This guarantee that output video frame rate will not be higher than this value.
12164 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12165 and corrects the scene colors accordingly.
12167 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12169 The filter accepts the following options:
12173 The order of differentiation to be applied on the scene. Must be chosen in the range
12174 [0,2] and default value is 1.
12177 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12178 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12179 max value instead of calculating Minkowski distance.
12182 The standard deviation of Gaussian blur to be applied on the scene. Must be
12183 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12184 can't be equal to 0 if @var{difford} is greater than 0.
12187 @subsection Examples
12193 greyedge=difford=1:minknorm=5:sigma=2
12199 greyedge=difford=1:minknorm=0:sigma=2
12207 Apply a Hald CLUT to a video stream.
12209 First input is the video stream to process, and second one is the Hald CLUT.
12210 The Hald CLUT input can be a simple picture or a complete video stream.
12212 The filter accepts the following options:
12216 Force termination when the shortest input terminates. Default is @code{0}.
12218 Continue applying the last CLUT after the end of the stream. A value of
12219 @code{0} disable the filter after the last frame of the CLUT is reached.
12220 Default is @code{1}.
12223 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12224 filters share the same internals).
12226 This filter also supports the @ref{framesync} options.
12228 More information about the Hald CLUT can be found on Eskil Steenberg's website
12229 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12231 @subsection Workflow examples
12233 @subsubsection Hald CLUT video stream
12235 Generate an identity Hald CLUT stream altered with various effects:
12237 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
12240 Note: make sure you use a lossless codec.
12242 Then use it with @code{haldclut} to apply it on some random stream:
12244 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12247 The Hald CLUT will be applied to the 10 first seconds (duration of
12248 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12249 to the remaining frames of the @code{mandelbrot} stream.
12251 @subsubsection Hald CLUT with preview
12253 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12254 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12255 biggest possible square starting at the top left of the picture. The remaining
12256 padding pixels (bottom or right) will be ignored. This area can be used to add
12257 a preview of the Hald CLUT.
12259 Typically, the following generated Hald CLUT will be supported by the
12260 @code{haldclut} filter:
12263 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12264 pad=iw+320 [padded_clut];
12265 smptebars=s=320x256, split [a][b];
12266 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12267 [main][b] overlay=W-320" -frames:v 1 clut.png
12270 It contains the original and a preview of the effect of the CLUT: SMPTE color
12271 bars are displayed on the right-top, and below the same color bars processed by
12274 Then, the effect of this Hald CLUT can be visualized with:
12276 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12281 Flip the input video horizontally.
12283 For example, to horizontally flip the input video with @command{ffmpeg}:
12285 ffmpeg -i in.avi -vf "hflip" out.avi
12289 This filter applies a global color histogram equalization on a
12292 It can be used to correct video that has a compressed range of pixel
12293 intensities. The filter redistributes the pixel intensities to
12294 equalize their distribution across the intensity range. It may be
12295 viewed as an "automatically adjusting contrast filter". This filter is
12296 useful only for correcting degraded or poorly captured source
12299 The filter accepts the following options:
12303 Determine the amount of equalization to be applied. As the strength
12304 is reduced, the distribution of pixel intensities more-and-more
12305 approaches that of the input frame. The value must be a float number
12306 in the range [0,1] and defaults to 0.200.
12309 Set the maximum intensity that can generated and scale the output
12310 values appropriately. The strength should be set as desired and then
12311 the intensity can be limited if needed to avoid washing-out. The value
12312 must be a float number in the range [0,1] and defaults to 0.210.
12315 Set the antibanding level. If enabled the filter will randomly vary
12316 the luminance of output pixels by a small amount to avoid banding of
12317 the histogram. Possible values are @code{none}, @code{weak} or
12318 @code{strong}. It defaults to @code{none}.
12324 Compute and draw a color distribution histogram for the input video.
12326 The computed histogram is a representation of the color component
12327 distribution in an image.
12329 Standard histogram displays the color components distribution in an image.
12330 Displays color graph for each color component. Shows distribution of
12331 the Y, U, V, A or R, G, B components, depending on input format, in the
12332 current frame. Below each graph a color component scale meter is shown.
12334 The filter accepts the following options:
12338 Set height of level. Default value is @code{200}.
12339 Allowed range is [50, 2048].
12342 Set height of color scale. Default value is @code{12}.
12343 Allowed range is [0, 40].
12347 It accepts the following values:
12350 Per color component graphs are placed below each other.
12353 Per color component graphs are placed side by side.
12356 Presents information identical to that in the @code{parade}, except
12357 that the graphs representing color components are superimposed directly
12360 Default is @code{stack}.
12363 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12364 Default is @code{linear}.
12367 Set what color components to display.
12368 Default is @code{7}.
12371 Set foreground opacity. Default is @code{0.7}.
12374 Set background opacity. Default is @code{0.5}.
12377 @subsection Examples
12382 Calculate and draw histogram:
12384 ffplay -i input -vf histogram
12392 This is a high precision/quality 3d denoise filter. It aims to reduce
12393 image noise, producing smooth images and making still images really
12394 still. It should enhance compressibility.
12396 It accepts the following optional parameters:
12400 A non-negative floating point number which specifies spatial luma strength.
12401 It defaults to 4.0.
12403 @item chroma_spatial
12404 A non-negative floating point number which specifies spatial chroma strength.
12405 It defaults to 3.0*@var{luma_spatial}/4.0.
12408 A floating point number which specifies luma temporal strength. It defaults to
12409 6.0*@var{luma_spatial}/4.0.
12412 A floating point number which specifies chroma temporal strength. It defaults to
12413 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12416 @subsection Commands
12417 This filter supports same @ref{commands} as options.
12418 The command accepts the same syntax of the corresponding option.
12420 If the specified expression is not valid, it is kept at its current
12423 @anchor{hwdownload}
12424 @section hwdownload
12426 Download hardware frames to system memory.
12428 The input must be in hardware frames, and the output a non-hardware format.
12429 Not all formats will be supported on the output - it may be necessary to insert
12430 an additional @option{format} filter immediately following in the graph to get
12431 the output in a supported format.
12435 Map hardware frames to system memory or to another device.
12437 This filter has several different modes of operation; which one is used depends
12438 on the input and output formats:
12441 Hardware frame input, normal frame output
12443 Map the input frames to system memory and pass them to the output. If the
12444 original hardware frame is later required (for example, after overlaying
12445 something else on part of it), the @option{hwmap} filter can be used again
12446 in the next mode to retrieve it.
12448 Normal frame input, hardware frame output
12450 If the input is actually a software-mapped hardware frame, then unmap it -
12451 that is, return the original hardware frame.
12453 Otherwise, a device must be provided. Create new hardware surfaces on that
12454 device for the output, then map them back to the software format at the input
12455 and give those frames to the preceding filter. This will then act like the
12456 @option{hwupload} filter, but may be able to avoid an additional copy when
12457 the input is already in a compatible format.
12459 Hardware frame input and output
12461 A device must be supplied for the output, either directly or with the
12462 @option{derive_device} option. The input and output devices must be of
12463 different types and compatible - the exact meaning of this is
12464 system-dependent, but typically it means that they must refer to the same
12465 underlying hardware context (for example, refer to the same graphics card).
12467 If the input frames were originally created on the output device, then unmap
12468 to retrieve the original frames.
12470 Otherwise, map the frames to the output device - create new hardware frames
12471 on the output corresponding to the frames on the input.
12474 The following additional parameters are accepted:
12478 Set the frame mapping mode. Some combination of:
12481 The mapped frame should be readable.
12483 The mapped frame should be writeable.
12485 The mapping will always overwrite the entire frame.
12487 This may improve performance in some cases, as the original contents of the
12488 frame need not be loaded.
12490 The mapping must not involve any copying.
12492 Indirect mappings to copies of frames are created in some cases where either
12493 direct mapping is not possible or it would have unexpected properties.
12494 Setting this flag ensures that the mapping is direct and will fail if that is
12497 Defaults to @var{read+write} if not specified.
12499 @item derive_device @var{type}
12500 Rather than using the device supplied at initialisation, instead derive a new
12501 device of type @var{type} from the device the input frames exist on.
12504 In a hardware to hardware mapping, map in reverse - create frames in the sink
12505 and map them back to the source. This may be necessary in some cases where
12506 a mapping in one direction is required but only the opposite direction is
12507 supported by the devices being used.
12509 This option is dangerous - it may break the preceding filter in undefined
12510 ways if there are any additional constraints on that filter's output.
12511 Do not use it without fully understanding the implications of its use.
12517 Upload system memory frames to hardware surfaces.
12519 The device to upload to must be supplied when the filter is initialised. If
12520 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12521 option or with the @option{derive_device} option. The input and output devices
12522 must be of different types and compatible - the exact meaning of this is
12523 system-dependent, but typically it means that they must refer to the same
12524 underlying hardware context (for example, refer to the same graphics card).
12526 The following additional parameters are accepted:
12529 @item derive_device @var{type}
12530 Rather than using the device supplied at initialisation, instead derive a new
12531 device of type @var{type} from the device the input frames exist on.
12534 @anchor{hwupload_cuda}
12535 @section hwupload_cuda
12537 Upload system memory frames to a CUDA device.
12539 It accepts the following optional parameters:
12543 The number of the CUDA device to use
12548 Apply a high-quality magnification filter designed for pixel art. This filter
12549 was originally created by Maxim Stepin.
12551 It accepts the following option:
12555 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12556 @code{hq3x} and @code{4} for @code{hq4x}.
12557 Default is @code{3}.
12561 Stack input videos horizontally.
12563 All streams must be of same pixel format and of same height.
12565 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12566 to create same output.
12568 The filter accepts the following option:
12572 Set number of input streams. Default is 2.
12575 If set to 1, force the output to terminate when the shortest input
12576 terminates. Default value is 0.
12581 Modify the hue and/or the saturation of the input.
12583 It accepts the following parameters:
12587 Specify the hue angle as a number of degrees. It accepts an expression,
12588 and defaults to "0".
12591 Specify the saturation in the [-10,10] range. It accepts an expression and
12595 Specify the hue angle as a number of radians. It accepts an
12596 expression, and defaults to "0".
12599 Specify the brightness in the [-10,10] range. It accepts an expression and
12603 @option{h} and @option{H} are mutually exclusive, and can't be
12604 specified at the same time.
12606 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12607 expressions containing the following constants:
12611 frame count of the input frame starting from 0
12614 presentation timestamp of the input frame expressed in time base units
12617 frame rate of the input video, NAN if the input frame rate is unknown
12620 timestamp expressed in seconds, NAN if the input timestamp is unknown
12623 time base of the input video
12626 @subsection Examples
12630 Set the hue to 90 degrees and the saturation to 1.0:
12636 Same command but expressing the hue in radians:
12642 Rotate hue and make the saturation swing between 0
12643 and 2 over a period of 1 second:
12645 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12649 Apply a 3 seconds saturation fade-in effect starting at 0:
12651 hue="s=min(t/3\,1)"
12654 The general fade-in expression can be written as:
12656 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12660 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12662 hue="s=max(0\, min(1\, (8-t)/3))"
12665 The general fade-out expression can be written as:
12667 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12672 @subsection Commands
12674 This filter supports the following commands:
12680 Modify the hue and/or the saturation and/or brightness of the input video.
12681 The command accepts the same syntax of the corresponding option.
12683 If the specified expression is not valid, it is kept at its current
12687 @section hysteresis
12689 Grow first stream into second stream by connecting components.
12690 This makes it possible to build more robust edge masks.
12692 This filter accepts the following options:
12696 Set which planes will be processed as bitmap, unprocessed planes will be
12697 copied from first stream.
12698 By default value 0xf, all planes will be processed.
12701 Set threshold which is used in filtering. If pixel component value is higher than
12702 this value filter algorithm for connecting components is activated.
12703 By default value is 0.
12706 The @code{hysteresis} filter also supports the @ref{framesync} options.
12710 Detect video interlacing type.
12712 This filter tries to detect if the input frames are interlaced, progressive,
12713 top or bottom field first. It will also try to detect fields that are
12714 repeated between adjacent frames (a sign of telecine).
12716 Single frame detection considers only immediately adjacent frames when classifying each frame.
12717 Multiple frame detection incorporates the classification history of previous frames.
12719 The filter will log these metadata values:
12722 @item single.current_frame
12723 Detected type of current frame using single-frame detection. One of:
12724 ``tff'' (top field first), ``bff'' (bottom field first),
12725 ``progressive'', or ``undetermined''
12728 Cumulative number of frames detected as top field first using single-frame detection.
12731 Cumulative number of frames detected as top field first using multiple-frame detection.
12734 Cumulative number of frames detected as bottom field first using single-frame detection.
12736 @item multiple.current_frame
12737 Detected type of current frame using multiple-frame detection. One of:
12738 ``tff'' (top field first), ``bff'' (bottom field first),
12739 ``progressive'', or ``undetermined''
12742 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12744 @item single.progressive
12745 Cumulative number of frames detected as progressive using single-frame detection.
12747 @item multiple.progressive
12748 Cumulative number of frames detected as progressive using multiple-frame detection.
12750 @item single.undetermined
12751 Cumulative number of frames that could not be classified using single-frame detection.
12753 @item multiple.undetermined
12754 Cumulative number of frames that could not be classified using multiple-frame detection.
12756 @item repeated.current_frame
12757 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12759 @item repeated.neither
12760 Cumulative number of frames with no repeated field.
12763 Cumulative number of frames with the top field repeated from the previous frame's top field.
12765 @item repeated.bottom
12766 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12769 The filter accepts the following options:
12773 Set interlacing threshold.
12775 Set progressive threshold.
12777 Threshold for repeated field detection.
12779 Number of frames after which a given frame's contribution to the
12780 statistics is halved (i.e., it contributes only 0.5 to its
12781 classification). The default of 0 means that all frames seen are given
12782 full weight of 1.0 forever.
12783 @item analyze_interlaced_flag
12784 When this is not 0 then idet will use the specified number of frames to determine
12785 if the interlaced flag is accurate, it will not count undetermined frames.
12786 If the flag is found to be accurate it will be used without any further
12787 computations, if it is found to be inaccurate it will be cleared without any
12788 further computations. This allows inserting the idet filter as a low computational
12789 method to clean up the interlaced flag
12794 Deinterleave or interleave fields.
12796 This filter allows one to process interlaced images fields without
12797 deinterlacing them. Deinterleaving splits the input frame into 2
12798 fields (so called half pictures). Odd lines are moved to the top
12799 half of the output image, even lines to the bottom half.
12800 You can process (filter) them independently and then re-interleave them.
12802 The filter accepts the following options:
12806 @item chroma_mode, c
12807 @item alpha_mode, a
12808 Available values for @var{luma_mode}, @var{chroma_mode} and
12809 @var{alpha_mode} are:
12815 @item deinterleave, d
12816 Deinterleave fields, placing one above the other.
12818 @item interleave, i
12819 Interleave fields. Reverse the effect of deinterleaving.
12821 Default value is @code{none}.
12823 @item luma_swap, ls
12824 @item chroma_swap, cs
12825 @item alpha_swap, as
12826 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12829 @subsection Commands
12831 This filter supports the all above options as @ref{commands}.
12835 Apply inflate effect to the video.
12837 This filter replaces the pixel by the local(3x3) average by taking into account
12838 only values higher than the pixel.
12840 It accepts the following options:
12847 Limit the maximum change for each plane, default is 65535.
12848 If 0, plane will remain unchanged.
12851 @subsection Commands
12853 This filter supports the all above options as @ref{commands}.
12857 Simple interlacing filter from progressive contents. This interleaves upper (or
12858 lower) lines from odd frames with lower (or upper) lines from even frames,
12859 halving the frame rate and preserving image height.
12862 Original Original New Frame
12863 Frame 'j' Frame 'j+1' (tff)
12864 ========== =========== ==================
12865 Line 0 --------------------> Frame 'j' Line 0
12866 Line 1 Line 1 ----> Frame 'j+1' Line 1
12867 Line 2 ---------------------> Frame 'j' Line 2
12868 Line 3 Line 3 ----> Frame 'j+1' Line 3
12870 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12873 It accepts the following optional parameters:
12877 This determines whether the interlaced frame is taken from the even
12878 (tff - default) or odd (bff) lines of the progressive frame.
12881 Vertical lowpass filter to avoid twitter interlacing and
12882 reduce moire patterns.
12886 Disable vertical lowpass filter
12889 Enable linear filter (default)
12892 Enable complex filter. This will slightly less reduce twitter and moire
12893 but better retain detail and subjective sharpness impression.
12900 Deinterlace input video by applying Donald Graft's adaptive kernel
12901 deinterling. Work on interlaced parts of a video to produce
12902 progressive frames.
12904 The description of the accepted parameters follows.
12908 Set the threshold which affects the filter's tolerance when
12909 determining if a pixel line must be processed. It must be an integer
12910 in the range [0,255] and defaults to 10. A value of 0 will result in
12911 applying the process on every pixels.
12914 Paint pixels exceeding the threshold value to white if set to 1.
12918 Set the fields order. Swap fields if set to 1, leave fields alone if
12922 Enable additional sharpening if set to 1. Default is 0.
12925 Enable twoway sharpening if set to 1. Default is 0.
12928 @subsection Examples
12932 Apply default values:
12934 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12938 Enable additional sharpening:
12944 Paint processed pixels in white:
12952 Slowly update darker pixels.
12954 This filter makes short flashes of light appear longer.
12955 This filter accepts the following options:
12959 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
12962 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
12965 @section lenscorrection
12967 Correct radial lens distortion
12969 This filter can be used to correct for radial distortion as can result from the use
12970 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
12971 one can use tools available for example as part of opencv or simply trial-and-error.
12972 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
12973 and extract the k1 and k2 coefficients from the resulting matrix.
12975 Note that effectively the same filter is available in the open-source tools Krita and
12976 Digikam from the KDE project.
12978 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
12979 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
12980 brightness distribution, so you may want to use both filters together in certain
12981 cases, though you will have to take care of ordering, i.e. whether vignetting should
12982 be applied before or after lens correction.
12984 @subsection Options
12986 The filter accepts the following options:
12990 Relative x-coordinate of the focal point of the image, and thereby the center of the
12991 distortion. This value has a range [0,1] and is expressed as fractions of the image
12992 width. Default is 0.5.
12994 Relative y-coordinate of the focal point of the image, and thereby the center of the
12995 distortion. This value has a range [0,1] and is expressed as fractions of the image
12996 height. Default is 0.5.
12998 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
12999 no correction. Default is 0.
13001 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13002 0 means no correction. Default is 0.
13005 The formula that generates the correction is:
13007 @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)
13009 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13010 distances from the focal point in the source and target images, respectively.
13014 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13016 The @code{lensfun} filter requires the camera make, camera model, and lens model
13017 to apply the lens correction. The filter will load the lensfun database and
13018 query it to find the corresponding camera and lens entries in the database. As
13019 long as these entries can be found with the given options, the filter can
13020 perform corrections on frames. Note that incomplete strings will result in the
13021 filter choosing the best match with the given options, and the filter will
13022 output the chosen camera and lens models (logged with level "info"). You must
13023 provide the make, camera model, and lens model as they are required.
13025 The filter accepts the following options:
13029 The make of the camera (for example, "Canon"). This option is required.
13032 The model of the camera (for example, "Canon EOS 100D"). This option is
13036 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13037 option is required.
13040 The type of correction to apply. The following values are valid options:
13044 Enables fixing lens vignetting.
13047 Enables fixing lens geometry. This is the default.
13050 Enables fixing chromatic aberrations.
13053 Enables fixing lens vignetting and lens geometry.
13056 Enables fixing lens vignetting and chromatic aberrations.
13059 Enables fixing both lens geometry and chromatic aberrations.
13062 Enables all possible corrections.
13066 The focal length of the image/video (zoom; expected constant for video). For
13067 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13068 range should be chosen when using that lens. Default 18.
13071 The aperture of the image/video (expected constant for video). Note that
13072 aperture is only used for vignetting correction. Default 3.5.
13074 @item focus_distance
13075 The focus distance of the image/video (expected constant for video). Note that
13076 focus distance is only used for vignetting and only slightly affects the
13077 vignetting correction process. If unknown, leave it at the default value (which
13081 The scale factor which is applied after transformation. After correction the
13082 video is no longer necessarily rectangular. This parameter controls how much of
13083 the resulting image is visible. The value 0 means that a value will be chosen
13084 automatically such that there is little or no unmapped area in the output
13085 image. 1.0 means that no additional scaling is done. Lower values may result
13086 in more of the corrected image being visible, while higher values may avoid
13087 unmapped areas in the output.
13089 @item target_geometry
13090 The target geometry of the output image/video. The following values are valid
13094 @item rectilinear (default)
13097 @item equirectangular
13098 @item fisheye_orthographic
13099 @item fisheye_stereographic
13100 @item fisheye_equisolid
13101 @item fisheye_thoby
13104 Apply the reverse of image correction (instead of correcting distortion, apply
13107 @item interpolation
13108 The type of interpolation used when correcting distortion. The following values
13113 @item linear (default)
13118 @subsection Examples
13122 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13123 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13127 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
13131 Apply the same as before, but only for the first 5 seconds of video.
13134 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
13141 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13142 score between two input videos.
13144 The obtained VMAF score is printed through the logging system.
13146 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13147 After installing the library it can be enabled using:
13148 @code{./configure --enable-libvmaf}.
13149 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13151 The filter has following options:
13155 Set the model path which is to be used for SVM.
13156 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13159 Set the file path to be used to store logs.
13162 Set the format of the log file (csv, json or xml).
13164 @item enable_transform
13165 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13166 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13167 Default value: @code{false}
13170 Invokes the phone model which will generate VMAF scores higher than in the
13171 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13172 Default value: @code{false}
13175 Enables computing psnr along with vmaf.
13176 Default value: @code{false}
13179 Enables computing ssim along with vmaf.
13180 Default value: @code{false}
13183 Enables computing ms_ssim along with vmaf.
13184 Default value: @code{false}
13187 Set the pool method to be used for computing vmaf.
13188 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13191 Set number of threads to be used when computing vmaf.
13192 Default value: @code{0}, which makes use of all available logical processors.
13195 Set interval for frame subsampling used when computing vmaf.
13196 Default value: @code{1}
13198 @item enable_conf_interval
13199 Enables confidence interval.
13200 Default value: @code{false}
13203 This filter also supports the @ref{framesync} options.
13205 @subsection Examples
13208 On the below examples the input file @file{main.mpg} being processed is
13209 compared with the reference file @file{ref.mpg}.
13212 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13216 Example with options:
13218 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13222 Example with options and different containers:
13224 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 -
13230 Limits the pixel components values to the specified range [min, max].
13232 The filter accepts the following options:
13236 Lower bound. Defaults to the lowest allowed value for the input.
13239 Upper bound. Defaults to the highest allowed value for the input.
13242 Specify which planes will be processed. Defaults to all available.
13249 The filter accepts the following options:
13253 Set the number of loops. Setting this value to -1 will result in infinite loops.
13257 Set maximal size in number of frames. Default is 0.
13260 Set first frame of loop. Default is 0.
13263 @subsection Examples
13267 Loop single first frame infinitely:
13269 loop=loop=-1:size=1:start=0
13273 Loop single first frame 10 times:
13275 loop=loop=10:size=1:start=0
13279 Loop 10 first frames 5 times:
13281 loop=loop=5:size=10:start=0
13287 Apply a 1D LUT to an input video.
13289 The filter accepts the following options:
13293 Set the 1D LUT file name.
13295 Currently supported formats:
13304 Select interpolation mode.
13306 Available values are:
13310 Use values from the nearest defined point.
13312 Interpolate values using the linear interpolation.
13314 Interpolate values using the cosine interpolation.
13316 Interpolate values using the cubic interpolation.
13318 Interpolate values using the spline interpolation.
13325 Apply a 3D LUT to an input video.
13327 The filter accepts the following options:
13331 Set the 3D LUT file name.
13333 Currently supported formats:
13347 Select interpolation mode.
13349 Available values are:
13353 Use values from the nearest defined point.
13355 Interpolate values using the 8 points defining a cube.
13357 Interpolate values using a tetrahedron.
13363 Turn certain luma values into transparency.
13365 The filter accepts the following options:
13369 Set the luma which will be used as base for transparency.
13370 Default value is @code{0}.
13373 Set the range of luma values to be keyed out.
13374 Default value is @code{0.01}.
13377 Set the range of softness. Default value is @code{0}.
13378 Use this to control gradual transition from zero to full transparency.
13381 @subsection Commands
13382 This filter supports same @ref{commands} as options.
13383 The command accepts the same syntax of the corresponding option.
13385 If the specified expression is not valid, it is kept at its current
13388 @section lut, lutrgb, lutyuv
13390 Compute a look-up table for binding each pixel component input value
13391 to an output value, and apply it to the input video.
13393 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13394 to an RGB input video.
13396 These filters accept the following parameters:
13399 set first pixel component expression
13401 set second pixel component expression
13403 set third pixel component expression
13405 set fourth pixel component expression, corresponds to the alpha component
13408 set red component expression
13410 set green component expression
13412 set blue component expression
13414 alpha component expression
13417 set Y/luminance component expression
13419 set U/Cb component expression
13421 set V/Cr component expression
13424 Each of them specifies the expression to use for computing the lookup table for
13425 the corresponding pixel component values.
13427 The exact component associated to each of the @var{c*} options depends on the
13430 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13431 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13433 The expressions can contain the following constants and functions:
13438 The input width and height.
13441 The input value for the pixel component.
13444 The input value, clipped to the @var{minval}-@var{maxval} range.
13447 The maximum value for the pixel component.
13450 The minimum value for the pixel component.
13453 The negated value for the pixel component value, clipped to the
13454 @var{minval}-@var{maxval} range; it corresponds to the expression
13455 "maxval-clipval+minval".
13458 The computed value in @var{val}, clipped to the
13459 @var{minval}-@var{maxval} range.
13461 @item gammaval(gamma)
13462 The computed gamma correction value of the pixel component value,
13463 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13465 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13469 All expressions default to "val".
13471 @subsection Examples
13475 Negate input video:
13477 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13478 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13481 The above is the same as:
13483 lutrgb="r=negval:g=negval:b=negval"
13484 lutyuv="y=negval:u=negval:v=negval"
13494 Remove chroma components, turning the video into a graytone image:
13496 lutyuv="u=128:v=128"
13500 Apply a luma burning effect:
13506 Remove green and blue components:
13512 Set a constant alpha channel value on input:
13514 format=rgba,lutrgb=a="maxval-minval/2"
13518 Correct luminance gamma by a factor of 0.5:
13520 lutyuv=y=gammaval(0.5)
13524 Discard least significant bits of luma:
13526 lutyuv=y='bitand(val, 128+64+32)'
13530 Technicolor like effect:
13532 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13536 @section lut2, tlut2
13538 The @code{lut2} filter takes two input streams and outputs one
13541 The @code{tlut2} (time lut2) filter takes two consecutive frames
13542 from one single stream.
13544 This filter accepts the following parameters:
13547 set first pixel component expression
13549 set second pixel component expression
13551 set third pixel component expression
13553 set fourth pixel component expression, corresponds to the alpha component
13556 set output bit depth, only available for @code{lut2} filter. By default is 0,
13557 which means bit depth is automatically picked from first input format.
13560 The @code{lut2} filter also supports the @ref{framesync} options.
13562 Each of them specifies the expression to use for computing the lookup table for
13563 the corresponding pixel component values.
13565 The exact component associated to each of the @var{c*} options depends on the
13568 The expressions can contain the following constants:
13573 The input width and height.
13576 The first input value for the pixel component.
13579 The second input value for the pixel component.
13582 The first input video bit depth.
13585 The second input video bit depth.
13588 All expressions default to "x".
13590 @subsection Examples
13594 Highlight differences between two RGB video streams:
13596 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)'
13600 Highlight differences between two YUV video streams:
13602 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)'
13606 Show max difference between two video streams:
13608 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)))'
13612 @section maskedclamp
13614 Clamp the first input stream with the second input and third input stream.
13616 Returns the value of first stream to be between second input
13617 stream - @code{undershoot} and third input stream + @code{overshoot}.
13619 This filter accepts the following options:
13622 Default value is @code{0}.
13625 Default value is @code{0}.
13628 Set which planes will be processed as bitmap, unprocessed planes will be
13629 copied from first stream.
13630 By default value 0xf, all planes will be processed.
13635 Merge the second and third input stream into output stream using absolute differences
13636 between second input stream and first input stream and absolute difference between
13637 third input stream and first input stream. The picked value will be from second input
13638 stream if second absolute difference is greater than first one or from third input stream
13641 This filter accepts the following options:
13644 Set which planes will be processed as bitmap, unprocessed planes will be
13645 copied from first stream.
13646 By default value 0xf, all planes will be processed.
13649 @section maskedmerge
13651 Merge the first input stream with the second input stream using per pixel
13652 weights in the third input stream.
13654 A value of 0 in the third stream pixel component means that pixel component
13655 from first stream is returned unchanged, while maximum value (eg. 255 for
13656 8-bit videos) means that pixel component from second stream is returned
13657 unchanged. Intermediate values define the amount of merging between both
13658 input stream's pixel components.
13660 This filter accepts the following options:
13663 Set which planes will be processed as bitmap, unprocessed planes will be
13664 copied from first stream.
13665 By default value 0xf, all planes will be processed.
13670 Merge the second and third input stream into output stream using absolute differences
13671 between second input stream and first input stream and absolute difference between
13672 third input stream and first input stream. The picked value will be from second input
13673 stream if second absolute difference is less than first one or from third input stream
13676 This filter accepts the following options:
13679 Set which planes will be processed as bitmap, unprocessed planes will be
13680 copied from first stream.
13681 By default value 0xf, all planes will be processed.
13684 @section maskedthreshold
13685 Pick pixels comparing absolute difference of two video streams with fixed
13688 If absolute difference between pixel component of first and second video
13689 stream is equal or lower than user supplied threshold than pixel component
13690 from first video stream is picked, otherwise pixel component from second
13691 video stream is picked.
13693 This filter accepts the following options:
13696 Set threshold used when picking pixels from absolute difference from two input
13700 Set which planes will be processed as bitmap, unprocessed planes will be
13701 copied from second stream.
13702 By default value 0xf, all planes will be processed.
13706 Create mask from input video.
13708 For example it is useful to create motion masks after @code{tblend} filter.
13710 This filter accepts the following options:
13714 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13717 Set high threshold. Any pixel component higher than this value will be set to max value
13718 allowed for current pixel format.
13721 Set planes to filter, by default all available planes are filtered.
13724 Fill all frame pixels with this value.
13727 Set max average pixel value for frame. If sum of all pixel components is higher that this
13728 average, output frame will be completely filled with value set by @var{fill} option.
13729 Typically useful for scene changes when used in combination with @code{tblend} filter.
13734 Apply motion-compensation deinterlacing.
13736 It needs one field per frame as input and must thus be used together
13737 with yadif=1/3 or equivalent.
13739 This filter accepts the following options:
13742 Set the deinterlacing mode.
13744 It accepts one of the following values:
13749 use iterative motion estimation
13751 like @samp{slow}, but use multiple reference frames.
13753 Default value is @samp{fast}.
13756 Set the picture field parity assumed for the input video. It must be
13757 one of the following values:
13761 assume top field first
13763 assume bottom field first
13766 Default value is @samp{bff}.
13769 Set per-block quantization parameter (QP) used by the internal
13772 Higher values should result in a smoother motion vector field but less
13773 optimal individual vectors. Default value is 1.
13778 Pick median pixel from certain rectangle defined by radius.
13780 This filter accepts the following options:
13784 Set horizontal radius size. Default value is @code{1}.
13785 Allowed range is integer from 1 to 127.
13788 Set which planes to process. Default is @code{15}, which is all available planes.
13791 Set vertical radius size. Default value is @code{0}.
13792 Allowed range is integer from 0 to 127.
13793 If it is 0, value will be picked from horizontal @code{radius} option.
13796 Set median percentile. Default value is @code{0.5}.
13797 Default value of @code{0.5} will pick always median values, while @code{0} will pick
13798 minimum values, and @code{1} maximum values.
13801 @subsection Commands
13802 This filter supports same @ref{commands} as options.
13803 The command accepts the same syntax of the corresponding option.
13805 If the specified expression is not valid, it is kept at its current
13808 @section mergeplanes
13810 Merge color channel components from several video streams.
13812 The filter accepts up to 4 input streams, and merge selected input
13813 planes to the output video.
13815 This filter accepts the following options:
13818 Set input to output plane mapping. Default is @code{0}.
13820 The mappings is specified as a bitmap. It should be specified as a
13821 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13822 mapping for the first plane of the output stream. 'A' sets the number of
13823 the input stream to use (from 0 to 3), and 'a' the plane number of the
13824 corresponding input to use (from 0 to 3). The rest of the mappings is
13825 similar, 'Bb' describes the mapping for the output stream second
13826 plane, 'Cc' describes the mapping for the output stream third plane and
13827 'Dd' describes the mapping for the output stream fourth plane.
13830 Set output pixel format. Default is @code{yuva444p}.
13833 @subsection Examples
13837 Merge three gray video streams of same width and height into single video stream:
13839 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13843 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13845 [a0][a1]mergeplanes=0x00010210:yuva444p
13849 Swap Y and A plane in yuva444p stream:
13851 format=yuva444p,mergeplanes=0x03010200:yuva444p
13855 Swap U and V plane in yuv420p stream:
13857 format=yuv420p,mergeplanes=0x000201:yuv420p
13861 Cast a rgb24 clip to yuv444p:
13863 format=rgb24,mergeplanes=0x000102:yuv444p
13869 Estimate and export motion vectors using block matching algorithms.
13870 Motion vectors are stored in frame side data to be used by other filters.
13872 This filter accepts the following options:
13875 Specify the motion estimation method. Accepts one of the following values:
13879 Exhaustive search algorithm.
13881 Three step search algorithm.
13883 Two dimensional logarithmic search algorithm.
13885 New three step search algorithm.
13887 Four step search algorithm.
13889 Diamond search algorithm.
13891 Hexagon-based search algorithm.
13893 Enhanced predictive zonal search algorithm.
13895 Uneven multi-hexagon search algorithm.
13897 Default value is @samp{esa}.
13900 Macroblock size. Default @code{16}.
13903 Search parameter. Default @code{7}.
13906 @section midequalizer
13908 Apply Midway Image Equalization effect using two video streams.
13910 Midway Image Equalization adjusts a pair of images to have the same
13911 histogram, while maintaining their dynamics as much as possible. It's
13912 useful for e.g. matching exposures from a pair of stereo cameras.
13914 This filter has two inputs and one output, which must be of same pixel format, but
13915 may be of different sizes. The output of filter is first input adjusted with
13916 midway histogram of both inputs.
13918 This filter accepts the following option:
13922 Set which planes to process. Default is @code{15}, which is all available planes.
13925 @section minterpolate
13927 Convert the video to specified frame rate using motion interpolation.
13929 This filter accepts the following options:
13932 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}.
13935 Motion interpolation mode. Following values are accepted:
13938 Duplicate previous or next frame for interpolating new ones.
13940 Blend source frames. Interpolated frame is mean of previous and next frames.
13942 Motion compensated interpolation. Following options are effective when this mode is selected:
13946 Motion compensation mode. Following values are accepted:
13949 Overlapped block motion compensation.
13951 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13953 Default mode is @samp{obmc}.
13956 Motion estimation mode. Following values are accepted:
13959 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
13961 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
13963 Default mode is @samp{bilat}.
13966 The algorithm to be used for motion estimation. Following values are accepted:
13969 Exhaustive search algorithm.
13971 Three step search algorithm.
13973 Two dimensional logarithmic search algorithm.
13975 New three step search algorithm.
13977 Four step search algorithm.
13979 Diamond search algorithm.
13981 Hexagon-based search algorithm.
13983 Enhanced predictive zonal search algorithm.
13985 Uneven multi-hexagon search algorithm.
13987 Default algorithm is @samp{epzs}.
13990 Macroblock size. Default @code{16}.
13993 Motion estimation search parameter. Default @code{32}.
13996 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).
14001 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:
14004 Disable scene change detection.
14006 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14008 Default method is @samp{fdiff}.
14010 @item scd_threshold
14011 Scene change detection threshold. Default is @code{10.}.
14016 Mix several video input streams into one video stream.
14018 A description of the accepted options follows.
14022 The number of inputs. If unspecified, it defaults to 2.
14025 Specify weight of each input video stream as sequence.
14026 Each weight is separated by space. If number of weights
14027 is smaller than number of @var{frames} last specified
14028 weight will be used for all remaining unset weights.
14031 Specify scale, if it is set it will be multiplied with sum
14032 of each weight multiplied with pixel values to give final destination
14033 pixel value. By default @var{scale} is auto scaled to sum of weights.
14036 Specify how end of stream is determined.
14039 The duration of the longest input. (default)
14042 The duration of the shortest input.
14045 The duration of the first input.
14049 @section mpdecimate
14051 Drop frames that do not differ greatly from the previous frame in
14052 order to reduce frame rate.
14054 The main use of this filter is for very-low-bitrate encoding
14055 (e.g. streaming over dialup modem), but it could in theory be used for
14056 fixing movies that were inverse-telecined incorrectly.
14058 A description of the accepted options follows.
14062 Set the maximum number of consecutive frames which can be dropped (if
14063 positive), or the minimum interval between dropped frames (if
14064 negative). If the value is 0, the frame is dropped disregarding the
14065 number of previous sequentially dropped frames.
14067 Default value is 0.
14072 Set the dropping threshold values.
14074 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14075 represent actual pixel value differences, so a threshold of 64
14076 corresponds to 1 unit of difference for each pixel, or the same spread
14077 out differently over the block.
14079 A frame is a candidate for dropping if no 8x8 blocks differ by more
14080 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14081 meaning the whole image) differ by more than a threshold of @option{lo}.
14083 Default value for @option{hi} is 64*12, default value for @option{lo} is
14084 64*5, and default value for @option{frac} is 0.33.
14090 Negate (invert) the input video.
14092 It accepts the following option:
14097 With value 1, it negates the alpha component, if present. Default value is 0.
14103 Denoise frames using Non-Local Means algorithm.
14105 Each pixel is adjusted by looking for other pixels with similar contexts. This
14106 context similarity is defined by comparing their surrounding patches of size
14107 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14110 Note that the research area defines centers for patches, which means some
14111 patches will be made of pixels outside that research area.
14113 The filter accepts the following options.
14117 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14120 Set patch size. Default is 7. Must be odd number in range [0, 99].
14123 Same as @option{p} but for chroma planes.
14125 The default value is @var{0} and means automatic.
14128 Set research size. Default is 15. Must be odd number in range [0, 99].
14131 Same as @option{r} but for chroma planes.
14133 The default value is @var{0} and means automatic.
14138 Deinterlace video using neural network edge directed interpolation.
14140 This filter accepts the following options:
14144 Mandatory option, without binary file filter can not work.
14145 Currently file can be found here:
14146 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14149 Set which frames to deinterlace, by default it is @code{all}.
14150 Can be @code{all} or @code{interlaced}.
14153 Set mode of operation.
14155 Can be one of the following:
14159 Use frame flags, both fields.
14161 Use frame flags, single field.
14163 Use top field only.
14165 Use bottom field only.
14167 Use both fields, top first.
14169 Use both fields, bottom first.
14173 Set which planes to process, by default filter process all frames.
14176 Set size of local neighborhood around each pixel, used by the predictor neural
14179 Can be one of the following:
14192 Set the number of neurons in predictor neural network.
14193 Can be one of the following:
14204 Controls the number of different neural network predictions that are blended
14205 together to compute the final output value. Can be @code{fast}, default or
14209 Set which set of weights to use in the predictor.
14210 Can be one of the following:
14214 weights trained to minimize absolute error
14216 weights trained to minimize squared error
14220 Controls whether or not the prescreener neural network is used to decide
14221 which pixels should be processed by the predictor neural network and which
14222 can be handled by simple cubic interpolation.
14223 The prescreener is trained to know whether cubic interpolation will be
14224 sufficient for a pixel or whether it should be predicted by the predictor nn.
14225 The computational complexity of the prescreener nn is much less than that of
14226 the predictor nn. Since most pixels can be handled by cubic interpolation,
14227 using the prescreener generally results in much faster processing.
14228 The prescreener is pretty accurate, so the difference between using it and not
14229 using it is almost always unnoticeable.
14231 Can be one of the following:
14239 Default is @code{new}.
14242 Set various debugging flags.
14247 Force libavfilter not to use any of the specified pixel formats for the
14248 input to the next filter.
14250 It accepts the following parameters:
14254 A '|'-separated list of pixel format names, such as
14255 pix_fmts=yuv420p|monow|rgb24".
14259 @subsection Examples
14263 Force libavfilter to use a format different from @var{yuv420p} for the
14264 input to the vflip filter:
14266 noformat=pix_fmts=yuv420p,vflip
14270 Convert the input video to any of the formats not contained in the list:
14272 noformat=yuv420p|yuv444p|yuv410p
14278 Add noise on video input frame.
14280 The filter accepts the following options:
14288 Set noise seed for specific pixel component or all pixel components in case
14289 of @var{all_seed}. Default value is @code{123457}.
14291 @item all_strength, alls
14292 @item c0_strength, c0s
14293 @item c1_strength, c1s
14294 @item c2_strength, c2s
14295 @item c3_strength, c3s
14296 Set noise strength for specific pixel component or all pixel components in case
14297 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14299 @item all_flags, allf
14300 @item c0_flags, c0f
14301 @item c1_flags, c1f
14302 @item c2_flags, c2f
14303 @item c3_flags, c3f
14304 Set pixel component flags or set flags for all components if @var{all_flags}.
14305 Available values for component flags are:
14308 averaged temporal noise (smoother)
14310 mix random noise with a (semi)regular pattern
14312 temporal noise (noise pattern changes between frames)
14314 uniform noise (gaussian otherwise)
14318 @subsection Examples
14320 Add temporal and uniform noise to input video:
14322 noise=alls=20:allf=t+u
14327 Normalize RGB video (aka histogram stretching, contrast stretching).
14328 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14330 For each channel of each frame, the filter computes the input range and maps
14331 it linearly to the user-specified output range. The output range defaults
14332 to the full dynamic range from pure black to pure white.
14334 Temporal smoothing can be used on the input range to reduce flickering (rapid
14335 changes in brightness) caused when small dark or bright objects enter or leave
14336 the scene. This is similar to the auto-exposure (automatic gain control) on a
14337 video camera, and, like a video camera, it may cause a period of over- or
14338 under-exposure of the video.
14340 The R,G,B channels can be normalized independently, which may cause some
14341 color shifting, or linked together as a single channel, which prevents
14342 color shifting. Linked normalization preserves hue. Independent normalization
14343 does not, so it can be used to remove some color casts. Independent and linked
14344 normalization can be combined in any ratio.
14346 The normalize filter accepts the following options:
14351 Colors which define the output range. The minimum input value is mapped to
14352 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14353 The defaults are black and white respectively. Specifying white for
14354 @var{blackpt} and black for @var{whitept} will give color-inverted,
14355 normalized video. Shades of grey can be used to reduce the dynamic range
14356 (contrast). Specifying saturated colors here can create some interesting
14360 The number of previous frames to use for temporal smoothing. The input range
14361 of each channel is smoothed using a rolling average over the current frame
14362 and the @var{smoothing} previous frames. The default is 0 (no temporal
14366 Controls the ratio of independent (color shifting) channel normalization to
14367 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14368 independent. Defaults to 1.0 (fully independent).
14371 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14372 expensive no-op. Defaults to 1.0 (full strength).
14376 @subsection Commands
14377 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14378 The command accepts the same syntax of the corresponding option.
14380 If the specified expression is not valid, it is kept at its current
14383 @subsection Examples
14385 Stretch video contrast to use the full dynamic range, with no temporal
14386 smoothing; may flicker depending on the source content:
14388 normalize=blackpt=black:whitept=white:smoothing=0
14391 As above, but with 50 frames of temporal smoothing; flicker should be
14392 reduced, depending on the source content:
14394 normalize=blackpt=black:whitept=white:smoothing=50
14397 As above, but with hue-preserving linked channel normalization:
14399 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14402 As above, but with half strength:
14404 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14407 Map the darkest input color to red, the brightest input color to cyan:
14409 normalize=blackpt=red:whitept=cyan
14414 Pass the video source unchanged to the output.
14417 Optical Character Recognition
14419 This filter uses Tesseract for optical character recognition. To enable
14420 compilation of this filter, you need to configure FFmpeg with
14421 @code{--enable-libtesseract}.
14423 It accepts the following options:
14427 Set datapath to tesseract data. Default is to use whatever was
14428 set at installation.
14431 Set language, default is "eng".
14434 Set character whitelist.
14437 Set character blacklist.
14440 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14441 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14445 Apply a video transform using libopencv.
14447 To enable this filter, install the libopencv library and headers and
14448 configure FFmpeg with @code{--enable-libopencv}.
14450 It accepts the following parameters:
14455 The name of the libopencv filter to apply.
14457 @item filter_params
14458 The parameters to pass to the libopencv filter. If not specified, the default
14459 values are assumed.
14463 Refer to the official libopencv documentation for more precise
14465 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14467 Several libopencv filters are supported; see the following subsections.
14472 Dilate an image by using a specific structuring element.
14473 It corresponds to the libopencv function @code{cvDilate}.
14475 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14477 @var{struct_el} represents a structuring element, and has the syntax:
14478 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14480 @var{cols} and @var{rows} represent the number of columns and rows of
14481 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14482 point, and @var{shape} the shape for the structuring element. @var{shape}
14483 must be "rect", "cross", "ellipse", or "custom".
14485 If the value for @var{shape} is "custom", it must be followed by a
14486 string of the form "=@var{filename}". The file with name
14487 @var{filename} is assumed to represent a binary image, with each
14488 printable character corresponding to a bright pixel. When a custom
14489 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14490 or columns and rows of the read file are assumed instead.
14492 The default value for @var{struct_el} is "3x3+0x0/rect".
14494 @var{nb_iterations} specifies the number of times the transform is
14495 applied to the image, and defaults to 1.
14499 # Use the default values
14502 # Dilate using a structuring element with a 5x5 cross, iterating two times
14503 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14505 # Read the shape from the file diamond.shape, iterating two times.
14506 # The file diamond.shape may contain a pattern of characters like this
14512 # The specified columns and rows are ignored
14513 # but the anchor point coordinates are not
14514 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14519 Erode an image by using a specific structuring element.
14520 It corresponds to the libopencv function @code{cvErode}.
14522 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14523 with the same syntax and semantics as the @ref{dilate} filter.
14527 Smooth the input video.
14529 The filter takes the following parameters:
14530 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14532 @var{type} is the type of smooth filter to apply, and must be one of
14533 the following values: "blur", "blur_no_scale", "median", "gaussian",
14534 or "bilateral". The default value is "gaussian".
14536 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14537 depends on the smooth type. @var{param1} and
14538 @var{param2} accept integer positive values or 0. @var{param3} and
14539 @var{param4} accept floating point values.
14541 The default value for @var{param1} is 3. The default value for the
14542 other parameters is 0.
14544 These parameters correspond to the parameters assigned to the
14545 libopencv function @code{cvSmooth}.
14547 @section oscilloscope
14549 2D Video Oscilloscope.
14551 Useful to measure spatial impulse, step responses, chroma delays, etc.
14553 It accepts the following parameters:
14557 Set scope center x position.
14560 Set scope center y position.
14563 Set scope size, relative to frame diagonal.
14566 Set scope tilt/rotation.
14572 Set trace center x position.
14575 Set trace center y position.
14578 Set trace width, relative to width of frame.
14581 Set trace height, relative to height of frame.
14584 Set which components to trace. By default it traces first three components.
14587 Draw trace grid. By default is enabled.
14590 Draw some statistics. By default is enabled.
14593 Draw scope. By default is enabled.
14596 @subsection Commands
14597 This filter supports same @ref{commands} as options.
14598 The command accepts the same syntax of the corresponding option.
14600 If the specified expression is not valid, it is kept at its current
14603 @subsection Examples
14607 Inspect full first row of video frame.
14609 oscilloscope=x=0.5:y=0:s=1
14613 Inspect full last row of video frame.
14615 oscilloscope=x=0.5:y=1:s=1
14619 Inspect full 5th line of video frame of height 1080.
14621 oscilloscope=x=0.5:y=5/1080:s=1
14625 Inspect full last column of video frame.
14627 oscilloscope=x=1:y=0.5:s=1:t=1
14635 Overlay one video on top of another.
14637 It takes two inputs and has one output. The first input is the "main"
14638 video on which the second input is overlaid.
14640 It accepts the following parameters:
14642 A description of the accepted options follows.
14647 Set the expression for the x and y coordinates of the overlaid video
14648 on the main video. Default value is "0" for both expressions. In case
14649 the expression is invalid, it is set to a huge value (meaning that the
14650 overlay will not be displayed within the output visible area).
14653 See @ref{framesync}.
14656 Set when the expressions for @option{x}, and @option{y} are evaluated.
14658 It accepts the following values:
14661 only evaluate expressions once during the filter initialization or
14662 when a command is processed
14665 evaluate expressions for each incoming frame
14668 Default value is @samp{frame}.
14671 See @ref{framesync}.
14674 Set the format for the output video.
14676 It accepts the following values:
14679 force YUV420 output
14682 force YUV420p10 output
14685 force YUV422 output
14688 force YUV422p10 output
14691 force YUV444 output
14694 force packed RGB output
14697 force planar RGB output
14700 automatically pick format
14703 Default value is @samp{yuv420}.
14706 See @ref{framesync}.
14709 Set format of alpha of the overlaid video, it can be @var{straight} or
14710 @var{premultiplied}. Default is @var{straight}.
14713 The @option{x}, and @option{y} expressions can contain the following
14719 The main input width and height.
14723 The overlay input width and height.
14727 The computed values for @var{x} and @var{y}. They are evaluated for
14732 horizontal and vertical chroma subsample values of the output
14733 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14737 the number of input frame, starting from 0
14740 the position in the file of the input frame, NAN if unknown
14743 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14747 This filter also supports the @ref{framesync} options.
14749 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14750 when evaluation is done @emph{per frame}, and will evaluate to NAN
14751 when @option{eval} is set to @samp{init}.
14753 Be aware that frames are taken from each input video in timestamp
14754 order, hence, if their initial timestamps differ, it is a good idea
14755 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14756 have them begin in the same zero timestamp, as the example for
14757 the @var{movie} filter does.
14759 You can chain together more overlays but you should test the
14760 efficiency of such approach.
14762 @subsection Commands
14764 This filter supports the following commands:
14768 Modify the x and y of the overlay input.
14769 The command accepts the same syntax of the corresponding option.
14771 If the specified expression is not valid, it is kept at its current
14775 @subsection Examples
14779 Draw the overlay at 10 pixels from the bottom right corner of the main
14782 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14785 Using named options the example above becomes:
14787 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14791 Insert a transparent PNG logo in the bottom left corner of the input,
14792 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14794 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14798 Insert 2 different transparent PNG logos (second logo on bottom
14799 right corner) using the @command{ffmpeg} tool:
14801 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
14805 Add a transparent color layer on top of the main video; @code{WxH}
14806 must specify the size of the main input to the overlay filter:
14808 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14812 Play an original video and a filtered version (here with the deshake
14813 filter) side by side using the @command{ffplay} tool:
14815 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14818 The above command is the same as:
14820 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14824 Make a sliding overlay appearing from the left to the right top part of the
14825 screen starting since time 2:
14827 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14831 Compose output by putting two input videos side to side:
14833 ffmpeg -i left.avi -i right.avi -filter_complex "
14834 nullsrc=size=200x100 [background];
14835 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14836 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14837 [background][left] overlay=shortest=1 [background+left];
14838 [background+left][right] overlay=shortest=1:x=100 [left+right]
14843 Mask 10-20 seconds of a video by applying the delogo filter to a section
14845 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14846 -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]'
14851 Chain several overlays in cascade:
14853 nullsrc=s=200x200 [bg];
14854 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14855 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14856 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14857 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14858 [in3] null, [mid2] overlay=100:100 [out0]
14863 @anchor{overlay_cuda}
14864 @section overlay_cuda
14866 Overlay one video on top of another.
14868 This is the CUDA variant of the @ref{overlay} filter.
14869 It only accepts CUDA frames. The underlying input pixel formats have to match.
14871 It takes two inputs and has one output. The first input is the "main"
14872 video on which the second input is overlaid.
14874 It accepts the following parameters:
14879 Set the x and y coordinates of the overlaid video on the main video.
14880 Default value is "0" for both expressions.
14883 See @ref{framesync}.
14886 See @ref{framesync}.
14889 See @ref{framesync}.
14893 This filter also supports the @ref{framesync} options.
14897 Apply Overcomplete Wavelet denoiser.
14899 The filter accepts the following options:
14905 Larger depth values will denoise lower frequency components more, but
14906 slow down filtering.
14908 Must be an int in the range 8-16, default is @code{8}.
14910 @item luma_strength, ls
14913 Must be a double value in the range 0-1000, default is @code{1.0}.
14915 @item chroma_strength, cs
14916 Set chroma strength.
14918 Must be a double value in the range 0-1000, default is @code{1.0}.
14924 Add paddings to the input image, and place the original input at the
14925 provided @var{x}, @var{y} coordinates.
14927 It accepts the following parameters:
14932 Specify an expression for the size of the output image with the
14933 paddings added. If the value for @var{width} or @var{height} is 0, the
14934 corresponding input size is used for the output.
14936 The @var{width} expression can reference the value set by the
14937 @var{height} expression, and vice versa.
14939 The default value of @var{width} and @var{height} is 0.
14943 Specify the offsets to place the input image at within the padded area,
14944 with respect to the top/left border of the output image.
14946 The @var{x} expression can reference the value set by the @var{y}
14947 expression, and vice versa.
14949 The default value of @var{x} and @var{y} is 0.
14951 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14952 so the input image is centered on the padded area.
14955 Specify the color of the padded area. For the syntax of this option,
14956 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14957 manual,ffmpeg-utils}.
14959 The default value of @var{color} is "black".
14962 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
14964 It accepts the following values:
14968 Only evaluate expressions once during the filter initialization or when
14969 a command is processed.
14972 Evaluate expressions for each incoming frame.
14976 Default value is @samp{init}.
14979 Pad to aspect instead to a resolution.
14983 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
14984 options are expressions containing the following constants:
14989 The input video width and height.
14993 These are the same as @var{in_w} and @var{in_h}.
14997 The output width and height (the size of the padded area), as
14998 specified by the @var{width} and @var{height} expressions.
15002 These are the same as @var{out_w} and @var{out_h}.
15006 The x and y offsets as specified by the @var{x} and @var{y}
15007 expressions, or NAN if not yet specified.
15010 same as @var{iw} / @var{ih}
15013 input sample aspect ratio
15016 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15020 The horizontal and vertical chroma subsample values. For example for the
15021 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15024 @subsection Examples
15028 Add paddings with the color "violet" to the input video. The output video
15029 size is 640x480, and the top-left corner of the input video is placed at
15032 pad=640:480:0:40:violet
15035 The example above is equivalent to the following command:
15037 pad=width=640:height=480:x=0:y=40:color=violet
15041 Pad the input to get an output with dimensions increased by 3/2,
15042 and put the input video at the center of the padded area:
15044 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15048 Pad the input to get a squared output with size equal to the maximum
15049 value between the input width and height, and put the input video at
15050 the center of the padded area:
15052 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15056 Pad the input to get a final w/h ratio of 16:9:
15058 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15062 In case of anamorphic video, in order to set the output display aspect
15063 correctly, it is necessary to use @var{sar} in the expression,
15064 according to the relation:
15066 (ih * X / ih) * sar = output_dar
15067 X = output_dar / sar
15070 Thus the previous example needs to be modified to:
15072 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15076 Double the output size and put the input video in the bottom-right
15077 corner of the output padded area:
15079 pad="2*iw:2*ih:ow-iw:oh-ih"
15083 @anchor{palettegen}
15084 @section palettegen
15086 Generate one palette for a whole video stream.
15088 It accepts the following options:
15092 Set the maximum number of colors to quantize in the palette.
15093 Note: the palette will still contain 256 colors; the unused palette entries
15096 @item reserve_transparent
15097 Create a palette of 255 colors maximum and reserve the last one for
15098 transparency. Reserving the transparency color is useful for GIF optimization.
15099 If not set, the maximum of colors in the palette will be 256. You probably want
15100 to disable this option for a standalone image.
15103 @item transparency_color
15104 Set the color that will be used as background for transparency.
15107 Set statistics mode.
15109 It accepts the following values:
15112 Compute full frame histograms.
15114 Compute histograms only for the part that differs from previous frame. This
15115 might be relevant to give more importance to the moving part of your input if
15116 the background is static.
15118 Compute new histogram for each frame.
15121 Default value is @var{full}.
15124 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15125 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15126 color quantization of the palette. This information is also visible at
15127 @var{info} logging level.
15129 @subsection Examples
15133 Generate a representative palette of a given video using @command{ffmpeg}:
15135 ffmpeg -i input.mkv -vf palettegen palette.png
15139 @section paletteuse
15141 Use a palette to downsample an input video stream.
15143 The filter takes two inputs: one video stream and a palette. The palette must
15144 be a 256 pixels image.
15146 It accepts the following options:
15150 Select dithering mode. Available algorithms are:
15153 Ordered 8x8 bayer dithering (deterministic)
15155 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15156 Note: this dithering is sometimes considered "wrong" and is included as a
15158 @item floyd_steinberg
15159 Floyd and Steingberg dithering (error diffusion)
15161 Frankie Sierra dithering v2 (error diffusion)
15163 Frankie Sierra dithering v2 "Lite" (error diffusion)
15166 Default is @var{sierra2_4a}.
15169 When @var{bayer} dithering is selected, this option defines the scale of the
15170 pattern (how much the crosshatch pattern is visible). A low value means more
15171 visible pattern for less banding, and higher value means less visible pattern
15172 at the cost of more banding.
15174 The option must be an integer value in the range [0,5]. Default is @var{2}.
15177 If set, define the zone to process
15181 Only the changing rectangle will be reprocessed. This is similar to GIF
15182 cropping/offsetting compression mechanism. This option can be useful for speed
15183 if only a part of the image is changing, and has use cases such as limiting the
15184 scope of the error diffusal @option{dither} to the rectangle that bounds the
15185 moving scene (it leads to more deterministic output if the scene doesn't change
15186 much, and as a result less moving noise and better GIF compression).
15189 Default is @var{none}.
15192 Take new palette for each output frame.
15194 @item alpha_threshold
15195 Sets the alpha threshold for transparency. Alpha values above this threshold
15196 will be treated as completely opaque, and values below this threshold will be
15197 treated as completely transparent.
15199 The option must be an integer value in the range [0,255]. Default is @var{128}.
15202 @subsection Examples
15206 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15207 using @command{ffmpeg}:
15209 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15213 @section perspective
15215 Correct perspective of video not recorded perpendicular to the screen.
15217 A description of the accepted parameters follows.
15228 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15229 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15230 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15231 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15232 then the corners of the source will be sent to the specified coordinates.
15234 The expressions can use the following variables:
15239 the width and height of video frame.
15243 Output frame count.
15246 @item interpolation
15247 Set interpolation for perspective correction.
15249 It accepts the following values:
15255 Default value is @samp{linear}.
15258 Set interpretation of coordinate options.
15260 It accepts the following values:
15264 Send point in the source specified by the given coordinates to
15265 the corners of the destination.
15267 @item 1, destination
15269 Send the corners of the source to the point in the destination specified
15270 by the given coordinates.
15272 Default value is @samp{source}.
15276 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15278 It accepts the following values:
15281 only evaluate expressions once during the filter initialization or
15282 when a command is processed
15285 evaluate expressions for each incoming frame
15288 Default value is @samp{init}.
15293 Delay interlaced video by one field time so that the field order changes.
15295 The intended use is to fix PAL movies that have been captured with the
15296 opposite field order to the film-to-video transfer.
15298 A description of the accepted parameters follows.
15304 It accepts the following values:
15307 Capture field order top-first, transfer bottom-first.
15308 Filter will delay the bottom field.
15311 Capture field order bottom-first, transfer top-first.
15312 Filter will delay the top field.
15315 Capture and transfer with the same field order. This mode only exists
15316 for the documentation of the other options to refer to, but if you
15317 actually select it, the filter will faithfully do nothing.
15320 Capture field order determined automatically by field flags, transfer
15322 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15323 basis using field flags. If no field information is available,
15324 then this works just like @samp{u}.
15327 Capture unknown or varying, transfer opposite.
15328 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15329 analyzing the images and selecting the alternative that produces best
15330 match between the fields.
15333 Capture top-first, transfer unknown or varying.
15334 Filter selects among @samp{t} and @samp{p} using image analysis.
15337 Capture bottom-first, transfer unknown or varying.
15338 Filter selects among @samp{b} and @samp{p} using image analysis.
15341 Capture determined by field flags, transfer unknown or varying.
15342 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15343 image analysis. If no field information is available, then this works just
15344 like @samp{U}. This is the default mode.
15347 Both capture and transfer unknown or varying.
15348 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15352 @section photosensitivity
15353 Reduce various flashes in video, so to help users with epilepsy.
15355 It accepts the following options:
15358 Set how many frames to use when filtering. Default is 30.
15361 Set detection threshold factor. Default is 1.
15365 Set how many pixels to skip when sampling frames. Default is 1.
15366 Allowed range is from 1 to 1024.
15369 Leave frames unchanged. Default is disabled.
15372 @section pixdesctest
15374 Pixel format descriptor test filter, mainly useful for internal
15375 testing. The output video should be equal to the input video.
15379 format=monow, pixdesctest
15382 can be used to test the monowhite pixel format descriptor definition.
15386 Display sample values of color channels. Mainly useful for checking color
15387 and levels. Minimum supported resolution is 640x480.
15389 The filters accept the following options:
15393 Set scope X position, relative offset on X axis.
15396 Set scope Y position, relative offset on Y axis.
15405 Set window opacity. This window also holds statistics about pixel area.
15408 Set window X position, relative offset on X axis.
15411 Set window Y position, relative offset on Y axis.
15416 Enable the specified chain of postprocessing subfilters using libpostproc. This
15417 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15418 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15419 Each subfilter and some options have a short and a long name that can be used
15420 interchangeably, i.e. dr/dering are the same.
15422 The filters accept the following options:
15426 Set postprocessing subfilters string.
15429 All subfilters share common options to determine their scope:
15433 Honor the quality commands for this subfilter.
15436 Do chrominance filtering, too (default).
15439 Do luminance filtering only (no chrominance).
15442 Do chrominance filtering only (no luminance).
15445 These options can be appended after the subfilter name, separated by a '|'.
15447 Available subfilters are:
15450 @item hb/hdeblock[|difference[|flatness]]
15451 Horizontal deblocking filter
15454 Difference factor where higher values mean more deblocking (default: @code{32}).
15456 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15459 @item vb/vdeblock[|difference[|flatness]]
15460 Vertical deblocking filter
15463 Difference factor where higher values mean more deblocking (default: @code{32}).
15465 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15468 @item ha/hadeblock[|difference[|flatness]]
15469 Accurate horizontal deblocking filter
15472 Difference factor where higher values mean more deblocking (default: @code{32}).
15474 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15477 @item va/vadeblock[|difference[|flatness]]
15478 Accurate vertical deblocking filter
15481 Difference factor where higher values mean more deblocking (default: @code{32}).
15483 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15487 The horizontal and vertical deblocking filters share the difference and
15488 flatness values so you cannot set different horizontal and vertical
15492 @item h1/x1hdeblock
15493 Experimental horizontal deblocking filter
15495 @item v1/x1vdeblock
15496 Experimental vertical deblocking filter
15501 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15504 larger -> stronger filtering
15506 larger -> stronger filtering
15508 larger -> stronger filtering
15511 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15514 Stretch luminance to @code{0-255}.
15517 @item lb/linblenddeint
15518 Linear blend deinterlacing filter that deinterlaces the given block by
15519 filtering all lines with a @code{(1 2 1)} filter.
15521 @item li/linipoldeint
15522 Linear interpolating deinterlacing filter that deinterlaces the given block by
15523 linearly interpolating every second line.
15525 @item ci/cubicipoldeint
15526 Cubic interpolating deinterlacing filter deinterlaces the given block by
15527 cubically interpolating every second line.
15529 @item md/mediandeint
15530 Median deinterlacing filter that deinterlaces the given block by applying a
15531 median filter to every second line.
15533 @item fd/ffmpegdeint
15534 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15535 second line with a @code{(-1 4 2 4 -1)} filter.
15538 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15539 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15541 @item fq/forceQuant[|quantizer]
15542 Overrides the quantizer table from the input with the constant quantizer you
15550 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15553 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15556 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15559 @subsection Examples
15563 Apply horizontal and vertical deblocking, deringing and automatic
15564 brightness/contrast:
15570 Apply default filters without brightness/contrast correction:
15576 Apply default filters and temporal denoiser:
15578 pp=default/tmpnoise|1|2|3
15582 Apply deblocking on luminance only, and switch vertical deblocking on or off
15583 automatically depending on available CPU time:
15590 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15591 similar to spp = 6 with 7 point DCT, where only the center sample is
15594 The filter accepts the following options:
15598 Force a constant quantization parameter. It accepts an integer in range
15599 0 to 63. If not set, the filter will use the QP from the video stream
15603 Set thresholding mode. Available modes are:
15607 Set hard thresholding.
15609 Set soft thresholding (better de-ringing effect, but likely blurrier).
15611 Set medium thresholding (good results, default).
15615 @section premultiply
15616 Apply alpha premultiply effect to input video stream using first plane
15617 of second stream as alpha.
15619 Both streams must have same dimensions and same pixel format.
15621 The filter accepts the following option:
15625 Set which planes will be processed, unprocessed planes will be copied.
15626 By default value 0xf, all planes will be processed.
15629 Do not require 2nd input for processing, instead use alpha plane from input stream.
15633 Apply prewitt operator to input video stream.
15635 The filter accepts the following option:
15639 Set which planes will be processed, unprocessed planes will be copied.
15640 By default value 0xf, all planes will be processed.
15643 Set value which will be multiplied with filtered result.
15646 Set value which will be added to filtered result.
15649 @section pseudocolor
15651 Alter frame colors in video with pseudocolors.
15653 This filter accepts the following options:
15657 set pixel first component expression
15660 set pixel second component expression
15663 set pixel third component expression
15666 set pixel fourth component expression, corresponds to the alpha component
15669 set component to use as base for altering colors
15672 Each of them specifies the expression to use for computing the lookup table for
15673 the corresponding pixel component values.
15675 The expressions can contain the following constants and functions:
15680 The input width and height.
15683 The input value for the pixel component.
15685 @item ymin, umin, vmin, amin
15686 The minimum allowed component value.
15688 @item ymax, umax, vmax, amax
15689 The maximum allowed component value.
15692 All expressions default to "val".
15694 @subsection Examples
15698 Change too high luma values to gradient:
15700 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'"
15706 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15707 Ratio) between two input videos.
15709 This filter takes in input two input videos, the first input is
15710 considered the "main" source and is passed unchanged to the
15711 output. The second input is used as a "reference" video for computing
15714 Both video inputs must have the same resolution and pixel format for
15715 this filter to work correctly. Also it assumes that both inputs
15716 have the same number of frames, which are compared one by one.
15718 The obtained average PSNR is printed through the logging system.
15720 The filter stores the accumulated MSE (mean squared error) of each
15721 frame, and at the end of the processing it is averaged across all frames
15722 equally, and the following formula is applied to obtain the PSNR:
15725 PSNR = 10*log10(MAX^2/MSE)
15728 Where MAX is the average of the maximum values of each component of the
15731 The description of the accepted parameters follows.
15734 @item stats_file, f
15735 If specified the filter will use the named file to save the PSNR of
15736 each individual frame. When filename equals "-" the data is sent to
15739 @item stats_version
15740 Specifies which version of the stats file format to use. Details of
15741 each format are written below.
15742 Default value is 1.
15744 @item stats_add_max
15745 Determines whether the max value is output to the stats log.
15746 Default value is 0.
15747 Requires stats_version >= 2. If this is set and stats_version < 2,
15748 the filter will return an error.
15751 This filter also supports the @ref{framesync} options.
15753 The file printed if @var{stats_file} is selected, contains a sequence of
15754 key/value pairs of the form @var{key}:@var{value} for each compared
15757 If a @var{stats_version} greater than 1 is specified, a header line precedes
15758 the list of per-frame-pair stats, with key value pairs following the frame
15759 format with the following parameters:
15762 @item psnr_log_version
15763 The version of the log file format. Will match @var{stats_version}.
15766 A comma separated list of the per-frame-pair parameters included in
15770 A description of each shown per-frame-pair parameter follows:
15774 sequential number of the input frame, starting from 1
15777 Mean Square Error pixel-by-pixel average difference of the compared
15778 frames, averaged over all the image components.
15780 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15781 Mean Square Error pixel-by-pixel average difference of the compared
15782 frames for the component specified by the suffix.
15784 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15785 Peak Signal to Noise ratio of the compared frames for the component
15786 specified by the suffix.
15788 @item max_avg, max_y, max_u, max_v
15789 Maximum allowed value for each channel, and average over all
15793 @subsection Examples
15798 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15799 [main][ref] psnr="stats_file=stats.log" [out]
15802 On this example the input file being processed is compared with the
15803 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15804 is stored in @file{stats.log}.
15807 Another example with different containers:
15809 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 -
15816 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15817 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15820 The pullup filter is designed to take advantage of future context in making
15821 its decisions. This filter is stateless in the sense that it does not lock
15822 onto a pattern to follow, but it instead looks forward to the following
15823 fields in order to identify matches and rebuild progressive frames.
15825 To produce content with an even framerate, insert the fps filter after
15826 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15827 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15829 The filter accepts the following options:
15836 These options set the amount of "junk" to ignore at the left, right, top, and
15837 bottom of the image, respectively. Left and right are in units of 8 pixels,
15838 while top and bottom are in units of 2 lines.
15839 The default is 8 pixels on each side.
15842 Set the strict breaks. Setting this option to 1 will reduce the chances of
15843 filter generating an occasional mismatched frame, but it may also cause an
15844 excessive number of frames to be dropped during high motion sequences.
15845 Conversely, setting it to -1 will make filter match fields more easily.
15846 This may help processing of video where there is slight blurring between
15847 the fields, but may also cause there to be interlaced frames in the output.
15848 Default value is @code{0}.
15851 Set the metric plane to use. It accepts the following values:
15857 Use chroma blue plane.
15860 Use chroma red plane.
15863 This option may be set to use chroma plane instead of the default luma plane
15864 for doing filter's computations. This may improve accuracy on very clean
15865 source material, but more likely will decrease accuracy, especially if there
15866 is chroma noise (rainbow effect) or any grayscale video.
15867 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15868 load and make pullup usable in realtime on slow machines.
15871 For best results (without duplicated frames in the output file) it is
15872 necessary to change the output frame rate. For example, to inverse
15873 telecine NTSC input:
15875 ffmpeg -i input -vf pullup -r 24000/1001 ...
15880 Change video quantization parameters (QP).
15882 The filter accepts the following option:
15886 Set expression for quantization parameter.
15889 The expression is evaluated through the eval API and can contain, among others,
15890 the following constants:
15894 1 if index is not 129, 0 otherwise.
15897 Sequential index starting from -129 to 128.
15900 @subsection Examples
15904 Some equation like:
15912 Flush video frames from internal cache of frames into a random order.
15913 No frame is discarded.
15914 Inspired by @ref{frei0r} nervous filter.
15918 Set size in number of frames of internal cache, in range from @code{2} to
15919 @code{512}. Default is @code{30}.
15922 Set seed for random number generator, must be an integer included between
15923 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15924 less than @code{0}, the filter will try to use a good random seed on a
15928 @section readeia608
15930 Read closed captioning (EIA-608) information from the top lines of a video frame.
15932 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15933 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15934 with EIA-608 data (starting from 0). A description of each metadata value follows:
15937 @item lavfi.readeia608.X.cc
15938 The two bytes stored as EIA-608 data (printed in hexadecimal).
15940 @item lavfi.readeia608.X.line
15941 The number of the line on which the EIA-608 data was identified and read.
15944 This filter accepts the following options:
15948 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15951 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15954 Set the ratio of width reserved for sync code detection.
15955 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
15958 Enable checking the parity bit. In the event of a parity error, the filter will output
15959 @code{0x00} for that character. Default is false.
15962 Lowpass lines prior to further processing. Default is enabled.
15965 @subsection Commands
15967 This filter supports the all above options as @ref{commands}.
15969 @subsection Examples
15973 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
15975 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
15981 Read vertical interval timecode (VITC) information from the top lines of a
15984 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
15985 timecode value, if a valid timecode has been detected. Further metadata key
15986 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
15987 timecode data has been found or not.
15989 This filter accepts the following options:
15993 Set the maximum number of lines to scan for VITC data. If the value is set to
15994 @code{-1} the full video frame is scanned. Default is @code{45}.
15997 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
15998 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16001 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16002 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16005 @subsection Examples
16009 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16010 draw @code{--:--:--:--} as a placeholder:
16012 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16018 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16020 Destination pixel at position (X, Y) will be picked from source (x, y) position
16021 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16022 value for pixel will be used for destination pixel.
16024 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16025 will have Xmap/Ymap video stream dimensions.
16026 Xmap and Ymap input video streams are 16bit depth, single channel.
16030 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16031 Default is @code{color}.
16034 Specify the color of the unmapped pixels. For the syntax of this option,
16035 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16036 manual,ffmpeg-utils}. Default color is @code{black}.
16039 @section removegrain
16041 The removegrain filter is a spatial denoiser for progressive video.
16045 Set mode for the first plane.
16048 Set mode for the second plane.
16051 Set mode for the third plane.
16054 Set mode for the fourth plane.
16057 Range of mode is from 0 to 24. Description of each mode follows:
16061 Leave input plane unchanged. Default.
16064 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16067 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16070 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16073 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16074 This is equivalent to a median filter.
16077 Line-sensitive clipping giving the minimal change.
16080 Line-sensitive clipping, intermediate.
16083 Line-sensitive clipping, intermediate.
16086 Line-sensitive clipping, intermediate.
16089 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16092 Replaces the target pixel with the closest neighbour.
16095 [1 2 1] horizontal and vertical kernel blur.
16101 Bob mode, interpolates top field from the line where the neighbours
16102 pixels are the closest.
16105 Bob mode, interpolates bottom field from the line where the neighbours
16106 pixels are the closest.
16109 Bob mode, interpolates top field. Same as 13 but with a more complicated
16110 interpolation formula.
16113 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16114 interpolation formula.
16117 Clips the pixel with the minimum and maximum of respectively the maximum and
16118 minimum of each pair of opposite neighbour pixels.
16121 Line-sensitive clipping using opposite neighbours whose greatest distance from
16122 the current pixel is minimal.
16125 Replaces the pixel with the average of its 8 neighbours.
16128 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16131 Clips pixels using the averages of opposite neighbour.
16134 Same as mode 21 but simpler and faster.
16137 Small edge and halo removal, but reputed useless.
16143 @section removelogo
16145 Suppress a TV station logo, using an image file to determine which
16146 pixels comprise the logo. It works by filling in the pixels that
16147 comprise the logo with neighboring pixels.
16149 The filter accepts the following options:
16153 Set the filter bitmap file, which can be any image format supported by
16154 libavformat. The width and height of the image file must match those of the
16155 video stream being processed.
16158 Pixels in the provided bitmap image with a value of zero are not
16159 considered part of the logo, non-zero pixels are considered part of
16160 the logo. If you use white (255) for the logo and black (0) for the
16161 rest, you will be safe. For making the filter bitmap, it is
16162 recommended to take a screen capture of a black frame with the logo
16163 visible, and then using a threshold filter followed by the erode
16164 filter once or twice.
16166 If needed, little splotches can be fixed manually. Remember that if
16167 logo pixels are not covered, the filter quality will be much
16168 reduced. Marking too many pixels as part of the logo does not hurt as
16169 much, but it will increase the amount of blurring needed to cover over
16170 the image and will destroy more information than necessary, and extra
16171 pixels will slow things down on a large logo.
16173 @section repeatfields
16175 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16176 fields based on its value.
16180 Reverse a video clip.
16182 Warning: This filter requires memory to buffer the entire clip, so trimming
16185 @subsection Examples
16189 Take the first 5 seconds of a clip, and reverse it.
16196 Shift R/G/B/A pixels horizontally and/or vertically.
16198 The filter accepts the following options:
16201 Set amount to shift red horizontally.
16203 Set amount to shift red vertically.
16205 Set amount to shift green horizontally.
16207 Set amount to shift green vertically.
16209 Set amount to shift blue horizontally.
16211 Set amount to shift blue vertically.
16213 Set amount to shift alpha horizontally.
16215 Set amount to shift alpha vertically.
16217 Set edge mode, can be @var{smear}, default, or @var{warp}.
16220 @subsection Commands
16222 This filter supports the all above options as @ref{commands}.
16225 Apply roberts cross operator to input video stream.
16227 The filter accepts the following option:
16231 Set which planes will be processed, unprocessed planes will be copied.
16232 By default value 0xf, all planes will be processed.
16235 Set value which will be multiplied with filtered result.
16238 Set value which will be added to filtered result.
16243 Rotate video by an arbitrary angle expressed in radians.
16245 The filter accepts the following options:
16247 A description of the optional parameters follows.
16250 Set an expression for the angle by which to rotate the input video
16251 clockwise, expressed as a number of radians. A negative value will
16252 result in a counter-clockwise rotation. By default it is set to "0".
16254 This expression is evaluated for each frame.
16257 Set the output width expression, default value is "iw".
16258 This expression is evaluated just once during configuration.
16261 Set the output height expression, default value is "ih".
16262 This expression is evaluated just once during configuration.
16265 Enable bilinear interpolation if set to 1, a value of 0 disables
16266 it. Default value is 1.
16269 Set the color used to fill the output area not covered by the rotated
16270 image. For the general syntax of this option, check the
16271 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16272 If the special value "none" is selected then no
16273 background is printed (useful for example if the background is never shown).
16275 Default value is "black".
16278 The expressions for the angle and the output size can contain the
16279 following constants and functions:
16283 sequential number of the input frame, starting from 0. It is always NAN
16284 before the first frame is filtered.
16287 time in seconds of the input frame, it is set to 0 when the filter is
16288 configured. It is always NAN before the first frame is filtered.
16292 horizontal and vertical chroma subsample values. For example for the
16293 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16297 the input video width and height
16301 the output width and height, that is the size of the padded area as
16302 specified by the @var{width} and @var{height} expressions
16306 the minimal width/height required for completely containing the input
16307 video rotated by @var{a} radians.
16309 These are only available when computing the @option{out_w} and
16310 @option{out_h} expressions.
16313 @subsection Examples
16317 Rotate the input by PI/6 radians clockwise:
16323 Rotate the input by PI/6 radians counter-clockwise:
16329 Rotate the input by 45 degrees clockwise:
16335 Apply a constant rotation with period T, starting from an angle of PI/3:
16337 rotate=PI/3+2*PI*t/T
16341 Make the input video rotation oscillating with a period of T
16342 seconds and an amplitude of A radians:
16344 rotate=A*sin(2*PI/T*t)
16348 Rotate the video, output size is chosen so that the whole rotating
16349 input video is always completely contained in the output:
16351 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16355 Rotate the video, reduce the output size so that no background is ever
16358 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16362 @subsection Commands
16364 The filter supports the following commands:
16368 Set the angle expression.
16369 The command accepts the same syntax of the corresponding option.
16371 If the specified expression is not valid, it is kept at its current
16377 Apply Shape Adaptive Blur.
16379 The filter accepts the following options:
16382 @item luma_radius, lr
16383 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16384 value is 1.0. A greater value will result in a more blurred image, and
16385 in slower processing.
16387 @item luma_pre_filter_radius, lpfr
16388 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16391 @item luma_strength, ls
16392 Set luma maximum difference between pixels to still be considered, must
16393 be a value in the 0.1-100.0 range, default value is 1.0.
16395 @item chroma_radius, cr
16396 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16397 greater value will result in a more blurred image, and in slower
16400 @item chroma_pre_filter_radius, cpfr
16401 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16403 @item chroma_strength, cs
16404 Set chroma maximum difference between pixels to still be considered,
16405 must be a value in the -0.9-100.0 range.
16408 Each chroma option value, if not explicitly specified, is set to the
16409 corresponding luma option value.
16414 Scale (resize) the input video, using the libswscale library.
16416 The scale filter forces the output display aspect ratio to be the same
16417 of the input, by changing the output sample aspect ratio.
16419 If the input image format is different from the format requested by
16420 the next filter, the scale filter will convert the input to the
16423 @subsection Options
16424 The filter accepts the following options, or any of the options
16425 supported by the libswscale scaler.
16427 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16428 the complete list of scaler options.
16433 Set the output video dimension expression. Default value is the input
16436 If the @var{width} or @var{w} value is 0, the input width is used for
16437 the output. If the @var{height} or @var{h} value is 0, the input height
16438 is used for the output.
16440 If one and only one of the values is -n with n >= 1, the scale filter
16441 will use a value that maintains the aspect ratio of the input image,
16442 calculated from the other specified dimension. After that it will,
16443 however, make sure that the calculated dimension is divisible by n and
16444 adjust the value if necessary.
16446 If both values are -n with n >= 1, the behavior will be identical to
16447 both values being set to 0 as previously detailed.
16449 See below for the list of accepted constants for use in the dimension
16453 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16457 Only evaluate expressions once during the filter initialization or when a command is processed.
16460 Evaluate expressions for each incoming frame.
16464 Default value is @samp{init}.
16468 Set the interlacing mode. It accepts the following values:
16472 Force interlaced aware scaling.
16475 Do not apply interlaced scaling.
16478 Select interlaced aware scaling depending on whether the source frames
16479 are flagged as interlaced or not.
16482 Default value is @samp{0}.
16485 Set libswscale scaling flags. See
16486 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16487 complete list of values. If not explicitly specified the filter applies
16491 @item param0, param1
16492 Set libswscale input parameters for scaling algorithms that need them. See
16493 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16494 complete documentation. If not explicitly specified the filter applies
16500 Set the video size. For the syntax of this option, check the
16501 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16503 @item in_color_matrix
16504 @item out_color_matrix
16505 Set in/output YCbCr color space type.
16507 This allows the autodetected value to be overridden as well as allows forcing
16508 a specific value used for the output and encoder.
16510 If not specified, the color space type depends on the pixel format.
16516 Choose automatically.
16519 Format conforming to International Telecommunication Union (ITU)
16520 Recommendation BT.709.
16523 Set color space conforming to the United States Federal Communications
16524 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16529 Set color space conforming to:
16533 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16536 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16539 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16544 Set color space conforming to SMPTE ST 240:1999.
16547 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16552 Set in/output YCbCr sample range.
16554 This allows the autodetected value to be overridden as well as allows forcing
16555 a specific value used for the output and encoder. If not specified, the
16556 range depends on the pixel format. Possible values:
16560 Choose automatically.
16563 Set full range (0-255 in case of 8-bit luma).
16565 @item mpeg/limited/tv
16566 Set "MPEG" range (16-235 in case of 8-bit luma).
16569 @item force_original_aspect_ratio
16570 Enable decreasing or increasing output video width or height if necessary to
16571 keep the original aspect ratio. Possible values:
16575 Scale the video as specified and disable this feature.
16578 The output video dimensions will automatically be decreased if needed.
16581 The output video dimensions will automatically be increased if needed.
16585 One useful instance of this option is that when you know a specific device's
16586 maximum allowed resolution, you can use this to limit the output video to
16587 that, while retaining the aspect ratio. For example, device A allows
16588 1280x720 playback, and your video is 1920x800. Using this option (set it to
16589 decrease) and specifying 1280x720 to the command line makes the output
16592 Please note that this is a different thing than specifying -1 for @option{w}
16593 or @option{h}, you still need to specify the output resolution for this option
16596 @item force_divisible_by
16597 Ensures that both the output dimensions, width and height, are divisible by the
16598 given integer when used together with @option{force_original_aspect_ratio}. This
16599 works similar to using @code{-n} in the @option{w} and @option{h} options.
16601 This option respects the value set for @option{force_original_aspect_ratio},
16602 increasing or decreasing the resolution accordingly. The video's aspect ratio
16603 may be slightly modified.
16605 This option can be handy if you need to have a video fit within or exceed
16606 a defined resolution using @option{force_original_aspect_ratio} but also have
16607 encoder restrictions on width or height divisibility.
16611 The values of the @option{w} and @option{h} options are expressions
16612 containing the following constants:
16617 The input width and height
16621 These are the same as @var{in_w} and @var{in_h}.
16625 The output (scaled) width and height
16629 These are the same as @var{out_w} and @var{out_h}
16632 The same as @var{iw} / @var{ih}
16635 input sample aspect ratio
16638 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16642 horizontal and vertical input chroma subsample values. For example for the
16643 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16647 horizontal and vertical output chroma subsample values. For example for the
16648 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16651 The (sequential) number of the input frame, starting from 0.
16652 Only available with @code{eval=frame}.
16655 The presentation timestamp of the input frame, expressed as a number of
16656 seconds. Only available with @code{eval=frame}.
16659 The position (byte offset) of the frame in the input stream, or NaN if
16660 this information is unavailable and/or meaningless (for example in case of synthetic video).
16661 Only available with @code{eval=frame}.
16664 @subsection Examples
16668 Scale the input video to a size of 200x100
16673 This is equivalent to:
16684 Specify a size abbreviation for the output size:
16689 which can also be written as:
16695 Scale the input to 2x:
16697 scale=w=2*iw:h=2*ih
16701 The above is the same as:
16703 scale=2*in_w:2*in_h
16707 Scale the input to 2x with forced interlaced scaling:
16709 scale=2*iw:2*ih:interl=1
16713 Scale the input to half size:
16715 scale=w=iw/2:h=ih/2
16719 Increase the width, and set the height to the same size:
16725 Seek Greek harmony:
16732 Increase the height, and set the width to 3/2 of the height:
16734 scale=w=3/2*oh:h=3/5*ih
16738 Increase the size, making the size a multiple of the chroma
16741 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16745 Increase the width to a maximum of 500 pixels,
16746 keeping the same aspect ratio as the input:
16748 scale=w='min(500\, iw*3/2):h=-1'
16752 Make pixels square by combining scale and setsar:
16754 scale='trunc(ih*dar):ih',setsar=1/1
16758 Make pixels square by combining scale and setsar,
16759 making sure the resulting resolution is even (required by some codecs):
16761 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16765 @subsection Commands
16767 This filter supports the following commands:
16771 Set the output video dimension expression.
16772 The command accepts the same syntax of the corresponding option.
16774 If the specified expression is not valid, it is kept at its current
16780 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16781 format conversion on CUDA video frames. Setting the output width and height
16782 works in the same way as for the @var{scale} filter.
16784 The following additional options are accepted:
16787 The pixel format of the output CUDA frames. If set to the string "same" (the
16788 default), the input format will be kept. Note that automatic format negotiation
16789 and conversion is not yet supported for hardware frames
16792 The interpolation algorithm used for resizing. One of the following:
16799 @item cubic2p_bspline
16800 2-parameter cubic (B=1, C=0)
16802 @item cubic2p_catmullrom
16803 2-parameter cubic (B=0, C=1/2)
16805 @item cubic2p_b05c03
16806 2-parameter cubic (B=1/2, C=3/10)
16814 @item force_original_aspect_ratio
16815 Enable decreasing or increasing output video width or height if necessary to
16816 keep the original aspect ratio. Possible values:
16820 Scale the video as specified and disable this feature.
16823 The output video dimensions will automatically be decreased if needed.
16826 The output video dimensions will automatically be increased if needed.
16830 One useful instance of this option is that when you know a specific device's
16831 maximum allowed resolution, you can use this to limit the output video to
16832 that, while retaining the aspect ratio. For example, device A allows
16833 1280x720 playback, and your video is 1920x800. Using this option (set it to
16834 decrease) and specifying 1280x720 to the command line makes the output
16837 Please note that this is a different thing than specifying -1 for @option{w}
16838 or @option{h}, you still need to specify the output resolution for this option
16841 @item force_divisible_by
16842 Ensures that both the output dimensions, width and height, are divisible by the
16843 given integer when used together with @option{force_original_aspect_ratio}. This
16844 works similar to using @code{-n} in the @option{w} and @option{h} options.
16846 This option respects the value set for @option{force_original_aspect_ratio},
16847 increasing or decreasing the resolution accordingly. The video's aspect ratio
16848 may be slightly modified.
16850 This option can be handy if you need to have a video fit within or exceed
16851 a defined resolution using @option{force_original_aspect_ratio} but also have
16852 encoder restrictions on width or height divisibility.
16858 Scale (resize) the input video, based on a reference video.
16860 See the scale filter for available options, scale2ref supports the same but
16861 uses the reference video instead of the main input as basis. scale2ref also
16862 supports the following additional constants for the @option{w} and
16863 @option{h} options:
16868 The main input video's width and height
16871 The same as @var{main_w} / @var{main_h}
16874 The main input video's sample aspect ratio
16876 @item main_dar, mdar
16877 The main input video's display aspect ratio. Calculated from
16878 @code{(main_w / main_h) * main_sar}.
16882 The main input video's horizontal and vertical chroma subsample values.
16883 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16887 The (sequential) number of the main input frame, starting from 0.
16888 Only available with @code{eval=frame}.
16891 The presentation timestamp of the main input frame, expressed as a number of
16892 seconds. Only available with @code{eval=frame}.
16895 The position (byte offset) of the frame in the main input stream, or NaN if
16896 this information is unavailable and/or meaningless (for example in case of synthetic video).
16897 Only available with @code{eval=frame}.
16900 @subsection Examples
16904 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16906 'scale2ref[b][a];[a][b]overlay'
16910 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16912 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16916 @subsection Commands
16918 This filter supports the following commands:
16922 Set the output video dimension expression.
16923 The command accepts the same syntax of the corresponding option.
16925 If the specified expression is not valid, it is kept at its current
16930 Scroll input video horizontally and/or vertically by constant speed.
16932 The filter accepts the following options:
16934 @item horizontal, h
16935 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16936 Negative values changes scrolling direction.
16939 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16940 Negative values changes scrolling direction.
16943 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16946 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16949 @subsection Commands
16951 This filter supports the following @ref{commands}:
16953 @item horizontal, h
16954 Set the horizontal scrolling speed.
16956 Set the vertical scrolling speed.
16962 Detect video scene change.
16964 This filter sets frame metadata with mafd between frame, the scene score, and
16965 forward the frame to the next filter, so they can use these metadata to detect
16966 scene change or others.
16968 In addition, this filter logs a message and sets frame metadata when it detects
16969 a scene change by @option{threshold}.
16971 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
16973 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
16974 to detect scene change.
16976 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
16977 detect scene change with @option{threshold}.
16979 The filter accepts the following options:
16983 Set the scene change detection threshold as a percentage of maximum change. Good
16984 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
16987 Default value is @code{10.}.
16990 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
16991 You can enable it if you want to get snapshot of scene change frames only.
16994 @anchor{selectivecolor}
16995 @section selectivecolor
16997 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
16998 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
16999 by the "purity" of the color (that is, how saturated it already is).
17001 This filter is similar to the Adobe Photoshop Selective Color tool.
17003 The filter accepts the following options:
17006 @item correction_method
17007 Select color correction method.
17009 Available values are:
17012 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17015 Specified adjustments are relative to the original component value.
17017 Default is @code{absolute}.
17019 Adjustments for red pixels (pixels where the red component is the maximum)
17021 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17023 Adjustments for green pixels (pixels where the green component is the maximum)
17025 Adjustments for cyan pixels (pixels where the red component is the minimum)
17027 Adjustments for blue pixels (pixels where the blue component is the maximum)
17029 Adjustments for magenta pixels (pixels where the green component is the minimum)
17031 Adjustments for white pixels (pixels where all components are greater than 128)
17033 Adjustments for all pixels except pure black and pure white
17035 Adjustments for black pixels (pixels where all components are lesser than 128)
17037 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17040 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17041 4 space separated floating point adjustment values in the [-1,1] range,
17042 respectively to adjust the amount of cyan, magenta, yellow and black for the
17043 pixels of its range.
17045 @subsection Examples
17049 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17050 increase magenta by 27% in blue areas:
17052 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17056 Use a Photoshop selective color preset:
17058 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17062 @anchor{separatefields}
17063 @section separatefields
17065 The @code{separatefields} takes a frame-based video input and splits
17066 each frame into its components fields, producing a new half height clip
17067 with twice the frame rate and twice the frame count.
17069 This filter use field-dominance information in frame to decide which
17070 of each pair of fields to place first in the output.
17071 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17073 @section setdar, setsar
17075 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17078 This is done by changing the specified Sample (aka Pixel) Aspect
17079 Ratio, according to the following equation:
17081 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17084 Keep in mind that the @code{setdar} filter does not modify the pixel
17085 dimensions of the video frame. Also, the display aspect ratio set by
17086 this filter may be changed by later filters in the filterchain,
17087 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17090 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17091 the filter output video.
17093 Note that as a consequence of the application of this filter, the
17094 output display aspect ratio will change according to the equation
17097 Keep in mind that the sample aspect ratio set by the @code{setsar}
17098 filter may be changed by later filters in the filterchain, e.g. if
17099 another "setsar" or a "setdar" filter is applied.
17101 It accepts the following parameters:
17104 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17105 Set the aspect ratio used by the filter.
17107 The parameter can be a floating point number string, an expression, or
17108 a string of the form @var{num}:@var{den}, where @var{num} and
17109 @var{den} are the numerator and denominator of the aspect ratio. If
17110 the parameter is not specified, it is assumed the value "0".
17111 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17115 Set the maximum integer value to use for expressing numerator and
17116 denominator when reducing the expressed aspect ratio to a rational.
17117 Default value is @code{100}.
17121 The parameter @var{sar} is an expression containing
17122 the following constants:
17126 These are approximated values for the mathematical constants e
17127 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17130 The input width and height.
17133 These are the same as @var{w} / @var{h}.
17136 The input sample aspect ratio.
17139 The input display aspect ratio. It is the same as
17140 (@var{w} / @var{h}) * @var{sar}.
17143 Horizontal and vertical chroma subsample values. For example, for the
17144 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17147 @subsection Examples
17152 To change the display aspect ratio to 16:9, specify one of the following:
17159 To change the sample aspect ratio to 10:11, specify:
17165 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17166 1000 in the aspect ratio reduction, use the command:
17168 setdar=ratio=16/9:max=1000
17176 Force field for the output video frame.
17178 The @code{setfield} filter marks the interlace type field for the
17179 output frames. It does not change the input frame, but only sets the
17180 corresponding property, which affects how the frame is treated by
17181 following filters (e.g. @code{fieldorder} or @code{yadif}).
17183 The filter accepts the following options:
17188 Available values are:
17192 Keep the same field property.
17195 Mark the frame as bottom-field-first.
17198 Mark the frame as top-field-first.
17201 Mark the frame as progressive.
17208 Force frame parameter for the output video frame.
17210 The @code{setparams} filter marks interlace and color range for the
17211 output frames. It does not change the input frame, but only sets the
17212 corresponding property, which affects how the frame is treated by
17217 Available values are:
17221 Keep the same field property (default).
17224 Mark the frame as bottom-field-first.
17227 Mark the frame as top-field-first.
17230 Mark the frame as progressive.
17234 Available values are:
17238 Keep the same color range property (default).
17240 @item unspecified, unknown
17241 Mark the frame as unspecified color range.
17243 @item limited, tv, mpeg
17244 Mark the frame as limited range.
17246 @item full, pc, jpeg
17247 Mark the frame as full range.
17250 @item color_primaries
17251 Set the color primaries.
17252 Available values are:
17256 Keep the same color primaries property (default).
17273 Set the color transfer.
17274 Available values are:
17278 Keep the same color trc property (default).
17300 Set the colorspace.
17301 Available values are:
17305 Keep the same colorspace property (default).
17318 @item chroma-derived-nc
17319 @item chroma-derived-c
17326 Show a line containing various information for each input video frame.
17327 The input video is not modified.
17329 This filter supports the following options:
17333 Calculate checksums of each plane. By default enabled.
17336 The shown line contains a sequence of key/value pairs of the form
17337 @var{key}:@var{value}.
17339 The following values are shown in the output:
17343 The (sequential) number of the input frame, starting from 0.
17346 The Presentation TimeStamp of the input frame, expressed as a number of
17347 time base units. The time base unit depends on the filter input pad.
17350 The Presentation TimeStamp of the input frame, expressed as a number of
17354 The position of the frame in the input stream, or -1 if this information is
17355 unavailable and/or meaningless (for example in case of synthetic video).
17358 The pixel format name.
17361 The sample aspect ratio of the input frame, expressed in the form
17362 @var{num}/@var{den}.
17365 The size of the input frame. For the syntax of this option, check the
17366 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17369 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17370 for bottom field first).
17373 This is 1 if the frame is a key frame, 0 otherwise.
17376 The picture type of the input frame ("I" for an I-frame, "P" for a
17377 P-frame, "B" for a B-frame, or "?" for an unknown type).
17378 Also refer to the documentation of the @code{AVPictureType} enum and of
17379 the @code{av_get_picture_type_char} function defined in
17380 @file{libavutil/avutil.h}.
17383 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17385 @item plane_checksum
17386 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17387 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17390 The mean value of pixels in each plane of the input frame, expressed in the form
17391 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17394 The standard deviation of pixel values in each plane of the input frame, expressed
17395 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17399 @section showpalette
17401 Displays the 256 colors palette of each frame. This filter is only relevant for
17402 @var{pal8} pixel format frames.
17404 It accepts the following option:
17408 Set the size of the box used to represent one palette color entry. Default is
17409 @code{30} (for a @code{30x30} pixel box).
17412 @section shuffleframes
17414 Reorder and/or duplicate and/or drop video frames.
17416 It accepts the following parameters:
17420 Set the destination indexes of input frames.
17421 This is space or '|' separated list of indexes that maps input frames to output
17422 frames. Number of indexes also sets maximal value that each index may have.
17423 '-1' index have special meaning and that is to drop frame.
17426 The first frame has the index 0. The default is to keep the input unchanged.
17428 @subsection Examples
17432 Swap second and third frame of every three frames of the input:
17434 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17438 Swap 10th and 1st frame of every ten frames of the input:
17440 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17444 @section shuffleplanes
17446 Reorder and/or duplicate video planes.
17448 It accepts the following parameters:
17453 The index of the input plane to be used as the first output plane.
17456 The index of the input plane to be used as the second output plane.
17459 The index of the input plane to be used as the third output plane.
17462 The index of the input plane to be used as the fourth output plane.
17466 The first plane has the index 0. The default is to keep the input unchanged.
17468 @subsection Examples
17472 Swap the second and third planes of the input:
17474 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17478 @anchor{signalstats}
17479 @section signalstats
17480 Evaluate various visual metrics that assist in determining issues associated
17481 with the digitization of analog video media.
17483 By default the filter will log these metadata values:
17487 Display the minimal Y value contained within the input frame. Expressed in
17491 Display the Y value at the 10% percentile within the input frame. Expressed in
17495 Display the average Y value within the input frame. Expressed in range of
17499 Display the Y value at the 90% percentile within the input frame. Expressed in
17503 Display the maximum Y value contained within the input frame. Expressed in
17507 Display the minimal U value contained within the input frame. Expressed in
17511 Display the U value at the 10% percentile within the input frame. Expressed in
17515 Display the average U value within the input frame. Expressed in range of
17519 Display the U value at the 90% percentile within the input frame. Expressed in
17523 Display the maximum U value contained within the input frame. Expressed in
17527 Display the minimal V value contained within the input frame. Expressed in
17531 Display the V value at the 10% percentile within the input frame. Expressed in
17535 Display the average V value within the input frame. Expressed in range of
17539 Display the V value at the 90% percentile within the input frame. Expressed in
17543 Display the maximum V value contained within the input frame. Expressed in
17547 Display the minimal saturation value contained within the input frame.
17548 Expressed in range of [0-~181.02].
17551 Display the saturation value at the 10% percentile within the input frame.
17552 Expressed in range of [0-~181.02].
17555 Display the average saturation value within the input frame. Expressed in range
17559 Display the saturation value at the 90% percentile within the input frame.
17560 Expressed in range of [0-~181.02].
17563 Display the maximum saturation value contained within the input frame.
17564 Expressed in range of [0-~181.02].
17567 Display the median value for hue within the input frame. Expressed in range of
17571 Display the average value for hue within the input frame. Expressed in range of
17575 Display the average of sample value difference between all values of the Y
17576 plane in the current frame and corresponding values of the previous input frame.
17577 Expressed in range of [0-255].
17580 Display the average of sample value difference between all values of the U
17581 plane in the current frame and corresponding values of the previous input frame.
17582 Expressed in range of [0-255].
17585 Display the average of sample value difference between all values of the V
17586 plane in the current frame and corresponding values of the previous input frame.
17587 Expressed in range of [0-255].
17590 Display bit depth of Y plane in current frame.
17591 Expressed in range of [0-16].
17594 Display bit depth of U plane in current frame.
17595 Expressed in range of [0-16].
17598 Display bit depth of V plane in current frame.
17599 Expressed in range of [0-16].
17602 The filter accepts the following options:
17608 @option{stat} specify an additional form of image analysis.
17609 @option{out} output video with the specified type of pixel highlighted.
17611 Both options accept the following values:
17615 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17616 unlike the neighboring pixels of the same field. Examples of temporal outliers
17617 include the results of video dropouts, head clogs, or tape tracking issues.
17620 Identify @var{vertical line repetition}. Vertical line repetition includes
17621 similar rows of pixels within a frame. In born-digital video vertical line
17622 repetition is common, but this pattern is uncommon in video digitized from an
17623 analog source. When it occurs in video that results from the digitization of an
17624 analog source it can indicate concealment from a dropout compensator.
17627 Identify pixels that fall outside of legal broadcast range.
17631 Set the highlight color for the @option{out} option. The default color is
17635 @subsection Examples
17639 Output data of various video metrics:
17641 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17645 Output specific data about the minimum and maximum values of the Y plane per frame:
17647 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17651 Playback video while highlighting pixels that are outside of broadcast range in red.
17653 ffplay example.mov -vf signalstats="out=brng:color=red"
17657 Playback video with signalstats metadata drawn over the frame.
17659 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17662 The contents of signalstat_drawtext.txt used in the command are:
17665 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
17666 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
17667 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
17668 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
17676 Calculates the MPEG-7 Video Signature. The filter can handle more than one
17677 input. In this case the matching between the inputs can be calculated additionally.
17678 The filter always passes through the first input. The signature of each stream can
17679 be written into a file.
17681 It accepts the following options:
17685 Enable or disable the matching process.
17687 Available values are:
17691 Disable the calculation of a matching (default).
17693 Calculate the matching for the whole video and output whether the whole video
17694 matches or only parts.
17696 Calculate only until a matching is found or the video ends. Should be faster in
17701 Set the number of inputs. The option value must be a non negative integer.
17702 Default value is 1.
17705 Set the path to which the output is written. If there is more than one input,
17706 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
17707 integer), that will be replaced with the input number. If no filename is
17708 specified, no output will be written. This is the default.
17711 Choose the output format.
17713 Available values are:
17717 Use the specified binary representation (default).
17719 Use the specified xml representation.
17723 Set threshold to detect one word as similar. The option value must be an integer
17724 greater than zero. The default value is 9000.
17727 Set threshold to detect all words as similar. The option value must be an integer
17728 greater than zero. The default value is 60000.
17731 Set threshold to detect frames as similar. The option value must be an integer
17732 greater than zero. The default value is 116.
17735 Set the minimum length of a sequence in frames to recognize it as matching
17736 sequence. The option value must be a non negative integer value.
17737 The default value is 0.
17740 Set the minimum relation, that matching frames to all frames must have.
17741 The option value must be a double value between 0 and 1. The default value is 0.5.
17744 @subsection Examples
17748 To calculate the signature of an input video and store it in signature.bin:
17750 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17754 To detect whether two videos match and store the signatures in XML format in
17755 signature0.xml and signature1.xml:
17757 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 -
17765 Blur the input video without impacting the outlines.
17767 It accepts the following options:
17770 @item luma_radius, lr
17771 Set the luma radius. The option value must be a float number in
17772 the range [0.1,5.0] that specifies the variance of the gaussian filter
17773 used to blur the image (slower if larger). Default value is 1.0.
17775 @item luma_strength, ls
17776 Set the luma strength. The option value must be a float number
17777 in the range [-1.0,1.0] that configures the blurring. A value included
17778 in [0.0,1.0] will blur the image whereas a value included in
17779 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17781 @item luma_threshold, lt
17782 Set the luma threshold used as a coefficient to determine
17783 whether a pixel should be blurred or not. The option value must be an
17784 integer in the range [-30,30]. A value of 0 will filter all the image,
17785 a value included in [0,30] will filter flat areas and a value included
17786 in [-30,0] will filter edges. Default value is 0.
17788 @item chroma_radius, cr
17789 Set the chroma radius. The option value must be a float number in
17790 the range [0.1,5.0] that specifies the variance of the gaussian filter
17791 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17793 @item chroma_strength, cs
17794 Set the chroma strength. The option value must be a float number
17795 in the range [-1.0,1.0] that configures the blurring. A value included
17796 in [0.0,1.0] will blur the image whereas a value included in
17797 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17799 @item chroma_threshold, ct
17800 Set the chroma threshold used as a coefficient to determine
17801 whether a pixel should be blurred or not. The option value must be an
17802 integer in the range [-30,30]. A value of 0 will filter all the image,
17803 a value included in [0,30] will filter flat areas and a value included
17804 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17807 If a chroma option is not explicitly set, the corresponding luma value
17811 Apply sobel operator to input video stream.
17813 The filter accepts the following option:
17817 Set which planes will be processed, unprocessed planes will be copied.
17818 By default value 0xf, all planes will be processed.
17821 Set value which will be multiplied with filtered result.
17824 Set value which will be added to filtered result.
17830 Apply a simple postprocessing filter that compresses and decompresses the image
17831 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17832 and average the results.
17834 The filter accepts the following options:
17838 Set quality. This option defines the number of levels for averaging. It accepts
17839 an integer in the range 0-6. If set to @code{0}, the filter will have no
17840 effect. A value of @code{6} means the higher quality. For each increment of
17841 that value the speed drops by a factor of approximately 2. Default value is
17845 Force a constant quantization parameter. If not set, the filter will use the QP
17846 from the video stream (if available).
17849 Set thresholding mode. Available modes are:
17853 Set hard thresholding (default).
17855 Set soft thresholding (better de-ringing effect, but likely blurrier).
17858 @item use_bframe_qp
17859 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17860 option may cause flicker since the B-Frames have often larger QP. Default is
17861 @code{0} (not enabled).
17864 @subsection Commands
17866 This filter supports the following commands:
17868 @item quality, level
17869 Set quality level. The value @code{max} can be used to set the maximum level,
17870 currently @code{6}.
17876 Scale the input by applying one of the super-resolution methods based on
17877 convolutional neural networks. Supported models:
17881 Super-Resolution Convolutional Neural Network model (SRCNN).
17882 See @url{https://arxiv.org/abs/1501.00092}.
17885 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17886 See @url{https://arxiv.org/abs/1609.05158}.
17889 Training scripts as well as scripts for model file (.pb) saving can be found at
17890 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17891 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17893 Native model files (.model) can be generated from TensorFlow model
17894 files (.pb) by using tools/python/convert.py
17896 The filter accepts the following options:
17900 Specify which DNN backend to use for model loading and execution. This option accepts
17901 the following values:
17905 Native implementation of DNN loading and execution.
17908 TensorFlow backend. To enable this backend you
17909 need to install the TensorFlow for C library (see
17910 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17911 @code{--enable-libtensorflow}
17914 Default value is @samp{native}.
17917 Set path to model file specifying network architecture and its parameters.
17918 Note that different backends use different file formats. TensorFlow backend
17919 can load files for both formats, while native backend can load files for only
17923 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17924 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17925 input upscaled using bicubic upscaling with proper scale factor.
17928 This feature can also be finished with @ref{dnn_processing} filter.
17932 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17934 This filter takes in input two input videos, the first input is
17935 considered the "main" source and is passed unchanged to the
17936 output. The second input is used as a "reference" video for computing
17939 Both video inputs must have the same resolution and pixel format for
17940 this filter to work correctly. Also it assumes that both inputs
17941 have the same number of frames, which are compared one by one.
17943 The filter stores the calculated SSIM of each frame.
17945 The description of the accepted parameters follows.
17948 @item stats_file, f
17949 If specified the filter will use the named file to save the SSIM of
17950 each individual frame. When filename equals "-" the data is sent to
17954 The file printed if @var{stats_file} is selected, contains a sequence of
17955 key/value pairs of the form @var{key}:@var{value} for each compared
17958 A description of each shown parameter follows:
17962 sequential number of the input frame, starting from 1
17964 @item Y, U, V, R, G, B
17965 SSIM of the compared frames for the component specified by the suffix.
17968 SSIM of the compared frames for the whole frame.
17971 Same as above but in dB representation.
17974 This filter also supports the @ref{framesync} options.
17976 @subsection Examples
17981 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
17982 [main][ref] ssim="stats_file=stats.log" [out]
17985 On this example the input file being processed is compared with the
17986 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
17987 is stored in @file{stats.log}.
17990 Another example with both psnr and ssim at same time:
17992 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
17996 Another example with different containers:
17998 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 -
18004 Convert between different stereoscopic image formats.
18006 The filters accept the following options:
18010 Set stereoscopic image format of input.
18012 Available values for input image formats are:
18015 side by side parallel (left eye left, right eye right)
18018 side by side crosseye (right eye left, left eye right)
18021 side by side parallel with half width resolution
18022 (left eye left, right eye right)
18025 side by side crosseye with half width resolution
18026 (right eye left, left eye right)
18030 above-below (left eye above, right eye below)
18034 above-below (right eye above, left eye below)
18038 above-below with half height resolution
18039 (left eye above, right eye below)
18043 above-below with half height resolution
18044 (right eye above, left eye below)
18047 alternating frames (left eye first, right eye second)
18050 alternating frames (right eye first, left eye second)
18053 interleaved rows (left eye has top row, right eye starts on next row)
18056 interleaved rows (right eye has top row, left eye starts on next row)
18059 interleaved columns, left eye first
18062 interleaved columns, right eye first
18064 Default value is @samp{sbsl}.
18068 Set stereoscopic image format of output.
18072 side by side parallel (left eye left, right eye right)
18075 side by side crosseye (right eye left, left eye right)
18078 side by side parallel with half width resolution
18079 (left eye left, right eye right)
18082 side by side crosseye with half width resolution
18083 (right eye left, left eye right)
18087 above-below (left eye above, right eye below)
18091 above-below (right eye above, left eye below)
18095 above-below with half height resolution
18096 (left eye above, right eye below)
18100 above-below with half height resolution
18101 (right eye above, left eye below)
18104 alternating frames (left eye first, right eye second)
18107 alternating frames (right eye first, left eye second)
18110 interleaved rows (left eye has top row, right eye starts on next row)
18113 interleaved rows (right eye has top row, left eye starts on next row)
18116 anaglyph red/blue gray
18117 (red filter on left eye, blue filter on right eye)
18120 anaglyph red/green gray
18121 (red filter on left eye, green filter on right eye)
18124 anaglyph red/cyan gray
18125 (red filter on left eye, cyan filter on right eye)
18128 anaglyph red/cyan half colored
18129 (red filter on left eye, cyan filter on right eye)
18132 anaglyph red/cyan color
18133 (red filter on left eye, cyan filter on right eye)
18136 anaglyph red/cyan color optimized with the least squares projection of dubois
18137 (red filter on left eye, cyan filter on right eye)
18140 anaglyph green/magenta gray
18141 (green filter on left eye, magenta filter on right eye)
18144 anaglyph green/magenta half colored
18145 (green filter on left eye, magenta filter on right eye)
18148 anaglyph green/magenta colored
18149 (green filter on left eye, magenta filter on right eye)
18152 anaglyph green/magenta color optimized with the least squares projection of dubois
18153 (green filter on left eye, magenta filter on right eye)
18156 anaglyph yellow/blue gray
18157 (yellow filter on left eye, blue filter on right eye)
18160 anaglyph yellow/blue half colored
18161 (yellow filter on left eye, blue filter on right eye)
18164 anaglyph yellow/blue colored
18165 (yellow filter on left eye, blue filter on right eye)
18168 anaglyph yellow/blue color optimized with the least squares projection of dubois
18169 (yellow filter on left eye, blue filter on right eye)
18172 mono output (left eye only)
18175 mono output (right eye only)
18178 checkerboard, left eye first
18181 checkerboard, right eye first
18184 interleaved columns, left eye first
18187 interleaved columns, right eye first
18193 Default value is @samp{arcd}.
18196 @subsection Examples
18200 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18206 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18212 @section streamselect, astreamselect
18213 Select video or audio streams.
18215 The filter accepts the following options:
18219 Set number of inputs. Default is 2.
18222 Set input indexes to remap to outputs.
18225 @subsection Commands
18227 The @code{streamselect} and @code{astreamselect} filter supports the following
18232 Set input indexes to remap to outputs.
18235 @subsection Examples
18239 Select first 5 seconds 1st stream and rest of time 2nd stream:
18241 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18245 Same as above, but for audio:
18247 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18254 Draw subtitles on top of input video using the libass library.
18256 To enable compilation of this filter you need to configure FFmpeg with
18257 @code{--enable-libass}. This filter also requires a build with libavcodec and
18258 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18259 Alpha) subtitles format.
18261 The filter accepts the following options:
18265 Set the filename of the subtitle file to read. It must be specified.
18267 @item original_size
18268 Specify the size of the original video, the video for which the ASS file
18269 was composed. For the syntax of this option, check the
18270 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18271 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18272 correctly scale the fonts if the aspect ratio has been changed.
18275 Set a directory path containing fonts that can be used by the filter.
18276 These fonts will be used in addition to whatever the font provider uses.
18279 Process alpha channel, by default alpha channel is untouched.
18282 Set subtitles input character encoding. @code{subtitles} filter only. Only
18283 useful if not UTF-8.
18285 @item stream_index, si
18286 Set subtitles stream index. @code{subtitles} filter only.
18289 Override default style or script info parameters of the subtitles. It accepts a
18290 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18293 If the first key is not specified, it is assumed that the first value
18294 specifies the @option{filename}.
18296 For example, to render the file @file{sub.srt} on top of the input
18297 video, use the command:
18302 which is equivalent to:
18304 subtitles=filename=sub.srt
18307 To render the default subtitles stream from file @file{video.mkv}, use:
18309 subtitles=video.mkv
18312 To render the second subtitles stream from that file, use:
18314 subtitles=video.mkv:si=1
18317 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18318 @code{DejaVu Serif}, use:
18320 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18323 @section super2xsai
18325 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18326 Interpolate) pixel art scaling algorithm.
18328 Useful for enlarging pixel art images without reducing sharpness.
18332 Swap two rectangular objects in video.
18334 This filter accepts the following options:
18344 Set 1st rect x coordinate.
18347 Set 1st rect y coordinate.
18350 Set 2nd rect x coordinate.
18353 Set 2nd rect y coordinate.
18355 All expressions are evaluated once for each frame.
18358 The all options are expressions containing the following constants:
18363 The input width and height.
18366 same as @var{w} / @var{h}
18369 input sample aspect ratio
18372 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18375 The number of the input frame, starting from 0.
18378 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18381 the position in the file of the input frame, NAN if unknown
18388 Blend successive video frames.
18394 Apply telecine process to the video.
18396 This filter accepts the following options:
18405 The default value is @code{top}.
18409 A string of numbers representing the pulldown pattern you wish to apply.
18410 The default value is @code{23}.
18414 Some typical patterns:
18419 24p: 2332 (preferred)
18426 24p: 222222222223 ("Euro pulldown")
18431 @section thistogram
18433 Compute and draw a color distribution histogram for the input video across time.
18435 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18436 at certain time, this filter shows also past histograms of number of frames defined
18437 by @code{width} option.
18439 The computed histogram is a representation of the color component
18440 distribution in an image.
18442 The filter accepts the following options:
18446 Set width of single color component output. Default value is @code{0}.
18447 Value of @code{0} means width will be picked from input video.
18448 This also set number of passed histograms to keep.
18449 Allowed range is [0, 8192].
18451 @item display_mode, d
18453 It accepts the following values:
18456 Per color component graphs are placed below each other.
18459 Per color component graphs are placed side by side.
18462 Presents information identical to that in the @code{parade}, except
18463 that the graphs representing color components are superimposed directly
18466 Default is @code{stack}.
18468 @item levels_mode, m
18469 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18470 Default is @code{linear}.
18472 @item components, c
18473 Set what color components to display.
18474 Default is @code{7}.
18477 Set background opacity. Default is @code{0.9}.
18480 Show envelope. Default is disabled.
18483 Set envelope color. Default is @code{gold}.
18488 Available values for slide is:
18491 Draw new frame when right border is reached.
18494 Replace old columns with new ones.
18497 Scroll from right to left.
18500 Scroll from left to right.
18503 Draw single picture.
18506 Default is @code{replace}.
18511 Apply threshold effect to video stream.
18513 This filter needs four video streams to perform thresholding.
18514 First stream is stream we are filtering.
18515 Second stream is holding threshold values, third stream is holding min values,
18516 and last, fourth stream is holding max values.
18518 The filter accepts the following option:
18522 Set which planes will be processed, unprocessed planes will be copied.
18523 By default value 0xf, all planes will be processed.
18526 For example if first stream pixel's component value is less then threshold value
18527 of pixel component from 2nd threshold stream, third stream value will picked,
18528 otherwise fourth stream pixel component value will be picked.
18530 Using color source filter one can perform various types of thresholding:
18532 @subsection Examples
18536 Binary threshold, using gray color as threshold:
18538 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18542 Inverted binary threshold, using gray color as threshold:
18544 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18548 Truncate binary threshold, using gray color as threshold:
18550 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18554 Threshold to zero, using gray color as threshold:
18556 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18560 Inverted threshold to zero, using gray color as threshold:
18562 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18567 Select the most representative frame in a given sequence of consecutive frames.
18569 The filter accepts the following options:
18573 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18574 will pick one of them, and then handle the next batch of @var{n} frames until
18575 the end. Default is @code{100}.
18578 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18579 value will result in a higher memory usage, so a high value is not recommended.
18581 @subsection Examples
18585 Extract one picture each 50 frames:
18591 Complete example of a thumbnail creation with @command{ffmpeg}:
18593 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18600 Tile several successive frames together.
18602 The @ref{untile} filter can do the reverse.
18604 The filter accepts the following options:
18609 Set the grid size (i.e. the number of lines and columns). For the syntax of
18610 this option, check the
18611 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18614 Set the maximum number of frames to render in the given area. It must be less
18615 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18616 the area will be used.
18619 Set the outer border margin in pixels.
18622 Set the inner border thickness (i.e. the number of pixels between frames). For
18623 more advanced padding options (such as having different values for the edges),
18624 refer to the pad video filter.
18627 Specify the color of the unused area. For the syntax of this option, check the
18628 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18629 The default value of @var{color} is "black".
18632 Set the number of frames to overlap when tiling several successive frames together.
18633 The value must be between @code{0} and @var{nb_frames - 1}.
18636 Set the number of frames to initially be empty before displaying first output frame.
18637 This controls how soon will one get first output frame.
18638 The value must be between @code{0} and @var{nb_frames - 1}.
18641 @subsection Examples
18645 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18647 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18649 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18650 duplicating each output frame to accommodate the originally detected frame
18654 Display @code{5} pictures in an area of @code{3x2} frames,
18655 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18656 mixed flat and named options:
18658 tile=3x2:nb_frames=5:padding=7:margin=2
18662 @section tinterlace
18664 Perform various types of temporal field interlacing.
18666 Frames are counted starting from 1, so the first input frame is
18669 The filter accepts the following options:
18674 Specify the mode of the interlacing. This option can also be specified
18675 as a value alone. See below for a list of values for this option.
18677 Available values are:
18681 Move odd frames into the upper field, even into the lower field,
18682 generating a double height frame at half frame rate.
18686 Frame 1 Frame 2 Frame 3 Frame 4
18688 11111 22222 33333 44444
18689 11111 22222 33333 44444
18690 11111 22222 33333 44444
18691 11111 22222 33333 44444
18705 Only output odd frames, even frames are dropped, generating a frame with
18706 unchanged height at half frame rate.
18711 Frame 1 Frame 2 Frame 3 Frame 4
18713 11111 22222 33333 44444
18714 11111 22222 33333 44444
18715 11111 22222 33333 44444
18716 11111 22222 33333 44444
18726 Only output even frames, odd frames are dropped, generating a frame with
18727 unchanged height at half frame rate.
18732 Frame 1 Frame 2 Frame 3 Frame 4
18734 11111 22222 33333 44444
18735 11111 22222 33333 44444
18736 11111 22222 33333 44444
18737 11111 22222 33333 44444
18747 Expand each frame to full height, but pad alternate lines with black,
18748 generating a frame with double height at the same input frame rate.
18753 Frame 1 Frame 2 Frame 3 Frame 4
18755 11111 22222 33333 44444
18756 11111 22222 33333 44444
18757 11111 22222 33333 44444
18758 11111 22222 33333 44444
18761 11111 ..... 33333 .....
18762 ..... 22222 ..... 44444
18763 11111 ..... 33333 .....
18764 ..... 22222 ..... 44444
18765 11111 ..... 33333 .....
18766 ..... 22222 ..... 44444
18767 11111 ..... 33333 .....
18768 ..... 22222 ..... 44444
18772 @item interleave_top, 4
18773 Interleave the upper field from odd frames with the lower field from
18774 even frames, generating a frame with unchanged height at half frame rate.
18779 Frame 1 Frame 2 Frame 3 Frame 4
18781 11111<- 22222 33333<- 44444
18782 11111 22222<- 33333 44444<-
18783 11111<- 22222 33333<- 44444
18784 11111 22222<- 33333 44444<-
18794 @item interleave_bottom, 5
18795 Interleave the lower field from odd frames with the upper field from
18796 even frames, generating a frame with unchanged height at half frame rate.
18801 Frame 1 Frame 2 Frame 3 Frame 4
18803 11111 22222<- 33333 44444<-
18804 11111<- 22222 33333<- 44444
18805 11111 22222<- 33333 44444<-
18806 11111<- 22222 33333<- 44444
18816 @item interlacex2, 6
18817 Double frame rate with unchanged height. Frames are inserted each
18818 containing the second temporal field from the previous input frame and
18819 the first temporal field from the next input frame. This mode relies on
18820 the top_field_first flag. Useful for interlaced video displays with no
18821 field synchronisation.
18826 Frame 1 Frame 2 Frame 3 Frame 4
18828 11111 22222 33333 44444
18829 11111 22222 33333 44444
18830 11111 22222 33333 44444
18831 11111 22222 33333 44444
18834 11111 22222 22222 33333 33333 44444 44444
18835 11111 11111 22222 22222 33333 33333 44444
18836 11111 22222 22222 33333 33333 44444 44444
18837 11111 11111 22222 22222 33333 33333 44444
18842 Move odd frames into the upper field, even into the lower field,
18843 generating a double height frame at same frame rate.
18848 Frame 1 Frame 2 Frame 3 Frame 4
18850 11111 22222 33333 44444
18851 11111 22222 33333 44444
18852 11111 22222 33333 44444
18853 11111 22222 33333 44444
18856 11111 33333 33333 55555
18857 22222 22222 44444 44444
18858 11111 33333 33333 55555
18859 22222 22222 44444 44444
18860 11111 33333 33333 55555
18861 22222 22222 44444 44444
18862 11111 33333 33333 55555
18863 22222 22222 44444 44444
18868 Numeric values are deprecated but are accepted for backward
18869 compatibility reasons.
18871 Default mode is @code{merge}.
18874 Specify flags influencing the filter process.
18876 Available value for @var{flags} is:
18879 @item low_pass_filter, vlpf
18880 Enable linear vertical low-pass filtering in the filter.
18881 Vertical low-pass filtering is required when creating an interlaced
18882 destination from a progressive source which contains high-frequency
18883 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18886 @item complex_filter, cvlpf
18887 Enable complex vertical low-pass filtering.
18888 This will slightly less reduce interlace 'twitter' and Moire
18889 patterning but better retain detail and subjective sharpness impression.
18892 Bypass already interlaced frames, only adjust the frame rate.
18895 Vertical low-pass filtering and bypassing already interlaced frames can only be
18896 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
18901 Pick median pixels from several successive input video frames.
18903 The filter accepts the following options:
18907 Set radius of median filter.
18908 Default is 1. Allowed range is from 1 to 127.
18911 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
18914 Set median percentile. Default value is @code{0.5}.
18915 Default value of @code{0.5} will pick always median values, while @code{0} will pick
18916 minimum values, and @code{1} maximum values.
18921 Mix successive video frames.
18923 A description of the accepted options follows.
18927 The number of successive frames to mix. If unspecified, it defaults to 3.
18930 Specify weight of each input video frame.
18931 Each weight is separated by space. If number of weights is smaller than
18932 number of @var{frames} last specified weight will be used for all remaining
18936 Specify scale, if it is set it will be multiplied with sum
18937 of each weight multiplied with pixel values to give final destination
18938 pixel value. By default @var{scale} is auto scaled to sum of weights.
18941 @subsection Examples
18945 Average 7 successive frames:
18947 tmix=frames=7:weights="1 1 1 1 1 1 1"
18951 Apply simple temporal convolution:
18953 tmix=frames=3:weights="-1 3 -1"
18957 Similar as above but only showing temporal differences:
18959 tmix=frames=3:weights="-1 2 -1":scale=1
18965 Tone map colors from different dynamic ranges.
18967 This filter expects data in single precision floating point, as it needs to
18968 operate on (and can output) out-of-range values. Another filter, such as
18969 @ref{zscale}, is needed to convert the resulting frame to a usable format.
18971 The tonemapping algorithms implemented only work on linear light, so input
18972 data should be linearized beforehand (and possibly correctly tagged).
18975 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
18978 @subsection Options
18979 The filter accepts the following options.
18983 Set the tone map algorithm to use.
18985 Possible values are:
18988 Do not apply any tone map, only desaturate overbright pixels.
18991 Hard-clip any out-of-range values. Use it for perfect color accuracy for
18992 in-range values, while distorting out-of-range values.
18995 Stretch the entire reference gamut to a linear multiple of the display.
18998 Fit a logarithmic transfer between the tone curves.
19001 Preserve overall image brightness with a simple curve, using nonlinear
19002 contrast, which results in flattening details and degrading color accuracy.
19005 Preserve both dark and bright details better than @var{reinhard}, at the cost
19006 of slightly darkening everything. Use it when detail preservation is more
19007 important than color and brightness accuracy.
19010 Smoothly map out-of-range values, while retaining contrast and colors for
19011 in-range material as much as possible. Use it when color accuracy is more
19012 important than detail preservation.
19018 Tune the tone mapping algorithm.
19020 This affects the following algorithms:
19026 Specifies the scale factor to use while stretching.
19030 Specifies the exponent of the function.
19034 Specify an extra linear coefficient to multiply into the signal before clipping.
19038 Specify the local contrast coefficient at the display peak.
19039 Default to 0.5, which means that in-gamut values will be about half as bright
19046 Specify the transition point from linear to mobius transform. Every value
19047 below this point is guaranteed to be mapped 1:1. The higher the value, the
19048 more accurate the result will be, at the cost of losing bright details.
19049 Default to 0.3, which due to the steep initial slope still preserves in-range
19050 colors fairly accurately.
19054 Apply desaturation for highlights that exceed this level of brightness. The
19055 higher the parameter, the more color information will be preserved. This
19056 setting helps prevent unnaturally blown-out colors for super-highlights, by
19057 (smoothly) turning into white instead. This makes images feel more natural,
19058 at the cost of reducing information about out-of-range colors.
19060 The default of 2.0 is somewhat conservative and will mostly just apply to
19061 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19063 This option works only if the input frame has a supported color tag.
19066 Override signal/nominal/reference peak with this value. Useful when the
19067 embedded peak information in display metadata is not reliable or when tone
19068 mapping from a lower range to a higher range.
19073 Temporarily pad video frames.
19075 The filter accepts the following options:
19079 Specify number of delay frames before input video stream. Default is 0.
19082 Specify number of padding frames after input video stream.
19083 Set to -1 to pad indefinitely. Default is 0.
19086 Set kind of frames added to beginning of stream.
19087 Can be either @var{add} or @var{clone}.
19088 With @var{add} frames of solid-color are added.
19089 With @var{clone} frames are clones of first frame.
19090 Default is @var{add}.
19093 Set kind of frames added to end of stream.
19094 Can be either @var{add} or @var{clone}.
19095 With @var{add} frames of solid-color are added.
19096 With @var{clone} frames are clones of last frame.
19097 Default is @var{add}.
19099 @item start_duration, stop_duration
19100 Specify the duration of the start/stop delay. See
19101 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19102 for the accepted syntax.
19103 These options override @var{start} and @var{stop}. Default is 0.
19106 Specify the color of the padded area. For the syntax of this option,
19107 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19108 manual,ffmpeg-utils}.
19110 The default value of @var{color} is "black".
19116 Transpose rows with columns in the input video and optionally flip it.
19118 It accepts the following parameters:
19123 Specify the transposition direction.
19125 Can assume the following values:
19127 @item 0, 4, cclock_flip
19128 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19136 Rotate by 90 degrees clockwise, that is:
19144 Rotate by 90 degrees counterclockwise, that is:
19151 @item 3, 7, clock_flip
19152 Rotate by 90 degrees clockwise and vertically flip, that is:
19160 For values between 4-7, the transposition is only done if the input
19161 video geometry is portrait and not landscape. These values are
19162 deprecated, the @code{passthrough} option should be used instead.
19164 Numerical values are deprecated, and should be dropped in favor of
19165 symbolic constants.
19168 Do not apply the transposition if the input geometry matches the one
19169 specified by the specified value. It accepts the following values:
19172 Always apply transposition.
19174 Preserve portrait geometry (when @var{height} >= @var{width}).
19176 Preserve landscape geometry (when @var{width} >= @var{height}).
19179 Default value is @code{none}.
19182 For example to rotate by 90 degrees clockwise and preserve portrait
19185 transpose=dir=1:passthrough=portrait
19188 The command above can also be specified as:
19190 transpose=1:portrait
19193 @section transpose_npp
19195 Transpose rows with columns in the input video and optionally flip it.
19196 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19198 It accepts the following parameters:
19203 Specify the transposition direction.
19205 Can assume the following values:
19208 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19211 Rotate by 90 degrees clockwise.
19214 Rotate by 90 degrees counterclockwise.
19217 Rotate by 90 degrees clockwise and vertically flip.
19221 Do not apply the transposition if the input geometry matches the one
19222 specified by the specified value. It accepts the following values:
19225 Always apply transposition. (default)
19227 Preserve portrait geometry (when @var{height} >= @var{width}).
19229 Preserve landscape geometry (when @var{width} >= @var{height}).
19235 Trim the input so that the output contains one continuous subpart of the input.
19237 It accepts the following parameters:
19240 Specify the time of the start of the kept section, i.e. the frame with the
19241 timestamp @var{start} will be the first frame in the output.
19244 Specify the time of the first frame that will be dropped, i.e. the frame
19245 immediately preceding the one with the timestamp @var{end} will be the last
19246 frame in the output.
19249 This is the same as @var{start}, except this option sets the start timestamp
19250 in timebase units instead of seconds.
19253 This is the same as @var{end}, except this option sets the end timestamp
19254 in timebase units instead of seconds.
19257 The maximum duration of the output in seconds.
19260 The number of the first frame that should be passed to the output.
19263 The number of the first frame that should be dropped.
19266 @option{start}, @option{end}, and @option{duration} are expressed as time
19267 duration specifications; see
19268 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19269 for the accepted syntax.
19271 Note that the first two sets of the start/end options and the @option{duration}
19272 option look at the frame timestamp, while the _frame variants simply count the
19273 frames that pass through the filter. Also note that this filter does not modify
19274 the timestamps. If you wish for the output timestamps to start at zero, insert a
19275 setpts filter after the trim filter.
19277 If multiple start or end options are set, this filter tries to be greedy and
19278 keep all the frames that match at least one of the specified constraints. To keep
19279 only the part that matches all the constraints at once, chain multiple trim
19282 The defaults are such that all the input is kept. So it is possible to set e.g.
19283 just the end values to keep everything before the specified time.
19288 Drop everything except the second minute of input:
19290 ffmpeg -i INPUT -vf trim=60:120
19294 Keep only the first second:
19296 ffmpeg -i INPUT -vf trim=duration=1
19301 @section unpremultiply
19302 Apply alpha unpremultiply effect to input video stream using first plane
19303 of second stream as alpha.
19305 Both streams must have same dimensions and same pixel format.
19307 The filter accepts the following option:
19311 Set which planes will be processed, unprocessed planes will be copied.
19312 By default value 0xf, all planes will be processed.
19314 If the format has 1 or 2 components, then luma is bit 0.
19315 If the format has 3 or 4 components:
19316 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19317 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19318 If present, the alpha channel is always the last bit.
19321 Do not require 2nd input for processing, instead use alpha plane from input stream.
19327 Sharpen or blur the input video.
19329 It accepts the following parameters:
19332 @item luma_msize_x, lx
19333 Set the luma matrix horizontal size. It must be an odd integer between
19334 3 and 23. The default value is 5.
19336 @item luma_msize_y, ly
19337 Set the luma matrix vertical size. It must be an odd integer between 3
19338 and 23. The default value is 5.
19340 @item luma_amount, la
19341 Set the luma effect strength. It must be a floating point number, reasonable
19342 values lay between -1.5 and 1.5.
19344 Negative values will blur the input video, while positive values will
19345 sharpen it, a value of zero will disable the effect.
19347 Default value is 1.0.
19349 @item chroma_msize_x, cx
19350 Set the chroma matrix horizontal size. It must be an odd integer
19351 between 3 and 23. The default value is 5.
19353 @item chroma_msize_y, cy
19354 Set the chroma matrix vertical size. It must be an odd integer
19355 between 3 and 23. The default value is 5.
19357 @item chroma_amount, ca
19358 Set the chroma effect strength. It must be a floating point number, reasonable
19359 values lay between -1.5 and 1.5.
19361 Negative values will blur the input video, while positive values will
19362 sharpen it, a value of zero will disable the effect.
19364 Default value is 0.0.
19368 All parameters are optional and default to the equivalent of the
19369 string '5:5:1.0:5:5:0.0'.
19371 @subsection Examples
19375 Apply strong luma sharpen effect:
19377 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19381 Apply a strong blur of both luma and chroma parameters:
19383 unsharp=7:7:-2:7:7:-2
19390 Decompose a video made of tiled images into the individual images.
19392 The frame rate of the output video is the frame rate of the input video
19393 multiplied by the number of tiles.
19395 This filter does the reverse of @ref{tile}.
19397 The filter accepts the following options:
19402 Set the grid size (i.e. the number of lines and columns). For the syntax of
19403 this option, check the
19404 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19407 @subsection Examples
19411 Produce a 1-second video from a still image file made of 25 frames stacked
19412 vertically, like an analogic film reel:
19414 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19420 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19421 the image at several (or - in the case of @option{quality} level @code{8} - all)
19422 shifts and average the results.
19424 The way this differs from the behavior of spp is that uspp actually encodes &
19425 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19426 DCT similar to MJPEG.
19428 The filter accepts the following options:
19432 Set quality. This option defines the number of levels for averaging. It accepts
19433 an integer in the range 0-8. If set to @code{0}, the filter will have no
19434 effect. A value of @code{8} means the higher quality. For each increment of
19435 that value the speed drops by a factor of approximately 2. Default value is
19439 Force a constant quantization parameter. If not set, the filter will use the QP
19440 from the video stream (if available).
19445 Convert 360 videos between various formats.
19447 The filter accepts the following options:
19453 Set format of the input/output video.
19461 Equirectangular projection.
19466 Cubemap with 3x2/6x1/1x6 layout.
19468 Format specific options:
19473 Set padding proportion for the input/output cubemap. Values in decimals.
19480 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)
19483 Default value is @b{@samp{0}}.
19484 Maximum value is @b{@samp{0.1}}.
19488 Set fixed padding for the input/output cubemap. Values in pixels.
19490 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19494 Set order of faces for the input/output cubemap. Choose one direction for each position.
19496 Designation of directions:
19512 Default value is @b{@samp{rludfb}}.
19516 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19518 Designation of angles:
19521 0 degrees clockwise
19523 90 degrees clockwise
19525 180 degrees clockwise
19527 270 degrees clockwise
19530 Default value is @b{@samp{000000}}.
19534 Equi-Angular Cubemap.
19541 Format specific options:
19546 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19548 If diagonal field of view is set it overrides horizontal and vertical field of view.
19553 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19555 If diagonal field of view is set it overrides horizontal and vertical field of view.
19561 Format specific options:
19566 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19568 If diagonal field of view is set it overrides horizontal and vertical field of view.
19573 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19575 If diagonal field of view is set it overrides horizontal and vertical field of view.
19581 Facebook's 360 formats.
19584 Stereographic format.
19586 Format specific options:
19591 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19593 If diagonal field of view is set it overrides horizontal and vertical field of view.
19598 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19600 If diagonal field of view is set it overrides horizontal and vertical field of view.
19607 Ball format, gives significant distortion toward the back.
19610 Hammer-Aitoff map projection format.
19613 Sinusoidal map projection format.
19616 Fisheye projection.
19618 Format specific options:
19623 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19625 If diagonal field of view is set it overrides horizontal and vertical field of view.
19630 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19632 If diagonal field of view is set it overrides horizontal and vertical field of view.
19636 Pannini projection.
19638 Format specific options:
19641 Set output pannini parameter.
19644 Set input pannini parameter.
19648 Cylindrical projection.
19650 Format specific options:
19655 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19657 If diagonal field of view is set it overrides horizontal and vertical field of view.
19662 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19664 If diagonal field of view is set it overrides horizontal and vertical field of view.
19668 Perspective projection. @i{(output only)}
19670 Format specific options:
19673 Set perspective parameter.
19677 Tetrahedron projection.
19680 Truncated square pyramid projection.
19684 Half equirectangular projection.
19689 Format specific options:
19694 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19696 If diagonal field of view is set it overrides horizontal and vertical field of view.
19701 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19703 If diagonal field of view is set it overrides horizontal and vertical field of view.
19707 Orthographic format.
19709 Format specific options:
19714 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19716 If diagonal field of view is set it overrides horizontal and vertical field of view.
19721 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19723 If diagonal field of view is set it overrides horizontal and vertical field of view.
19727 Octahedron projection.
19731 Set interpolation method.@*
19732 @i{Note: more complex interpolation methods require much more memory to run.}
19742 Bilinear interpolation.
19744 Lagrange9 interpolation.
19747 Bicubic interpolation.
19750 Lanczos interpolation.
19753 Spline16 interpolation.
19756 Gaussian interpolation.
19758 Mitchell interpolation.
19761 Default value is @b{@samp{line}}.
19765 Set the output video resolution.
19767 Default resolution depends on formats.
19771 Set the input/output stereo format.
19782 Default value is @b{@samp{2d}} for input and output format.
19787 Set rotation for the output video. Values in degrees.
19790 Set rotation order for the output video. Choose one item for each position.
19801 Default value is @b{@samp{ypr}}.
19806 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
19810 Set if input video is flipped horizontally/vertically. Boolean values.
19813 Set if input video is transposed. Boolean value, by default disabled.
19816 Set if output video needs to be transposed. Boolean value, by default disabled.
19819 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
19822 @subsection Examples
19826 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
19828 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19831 Extract back view of Equi-Angular Cubemap:
19833 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19836 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
19838 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19842 @subsection Commands
19844 This filter supports subset of above options as @ref{commands}.
19846 @section vaguedenoiser
19848 Apply a wavelet based denoiser.
19850 It transforms each frame from the video input into the wavelet domain,
19851 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19852 the obtained coefficients. It does an inverse wavelet transform after.
19853 Due to wavelet properties, it should give a nice smoothed result, and
19854 reduced noise, without blurring picture features.
19856 This filter accepts the following options:
19860 The filtering strength. The higher, the more filtered the video will be.
19861 Hard thresholding can use a higher threshold than soft thresholding
19862 before the video looks overfiltered. Default value is 2.
19865 The filtering method the filter will use.
19867 It accepts the following values:
19870 All values under the threshold will be zeroed.
19873 All values under the threshold will be zeroed. All values above will be
19874 reduced by the threshold.
19877 Scales or nullifies coefficients - intermediary between (more) soft and
19878 (less) hard thresholding.
19881 Default is garrote.
19884 Number of times, the wavelet will decompose the picture. Picture can't
19885 be decomposed beyond a particular point (typically, 8 for a 640x480
19886 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
19889 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
19892 A list of the planes to process. By default all planes are processed.
19895 The threshold type the filter will use.
19897 It accepts the following values:
19900 Threshold used is same for all decompositions.
19903 Threshold used depends also on each decomposition coefficients.
19906 Default is universal.
19909 @section vectorscope
19911 Display 2 color component values in the two dimensional graph (which is called
19914 This filter accepts the following options:
19918 Set vectorscope mode.
19920 It accepts the following values:
19924 Gray values are displayed on graph, higher brightness means more pixels have
19925 same component color value on location in graph. This is the default mode.
19928 Gray values are displayed on graph. Surrounding pixels values which are not
19929 present in video frame are drawn in gradient of 2 color components which are
19930 set by option @code{x} and @code{y}. The 3rd color component is static.
19933 Actual color components values present in video frame are displayed on graph.
19936 Similar as color2 but higher frequency of same values @code{x} and @code{y}
19937 on graph increases value of another color component, which is luminance by
19938 default values of @code{x} and @code{y}.
19941 Actual colors present in video frame are displayed on graph. If two different
19942 colors map to same position on graph then color with higher value of component
19943 not present in graph is picked.
19946 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
19947 component picked from radial gradient.
19951 Set which color component will be represented on X-axis. Default is @code{1}.
19954 Set which color component will be represented on Y-axis. Default is @code{2}.
19957 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
19958 of color component which represents frequency of (X, Y) location in graph.
19963 No envelope, this is default.
19966 Instant envelope, even darkest single pixel will be clearly highlighted.
19969 Hold maximum and minimum values presented in graph over time. This way you
19970 can still spot out of range values without constantly looking at vectorscope.
19973 Peak and instant envelope combined together.
19977 Set what kind of graticule to draw.
19986 Set graticule opacity.
19989 Set graticule flags.
19993 Draw graticule for white point.
19996 Draw graticule for black point.
19999 Draw color points short names.
20003 Set background opacity.
20005 @item lthreshold, l
20006 Set low threshold for color component not represented on X or Y axis.
20007 Values lower than this value will be ignored. Default is 0.
20008 Note this value is multiplied with actual max possible value one pixel component
20009 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20012 @item hthreshold, h
20013 Set high threshold for color component not represented on X or Y axis.
20014 Values higher than this value will be ignored. Default is 1.
20015 Note this value is multiplied with actual max possible value one pixel component
20016 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20017 is 0.9 * 255 = 230.
20019 @item colorspace, c
20020 Set what kind of colorspace to use when drawing graticule.
20030 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20031 This means no tint, and output will remain gray.
20034 @anchor{vidstabdetect}
20035 @section vidstabdetect
20037 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20038 @ref{vidstabtransform} for pass 2.
20040 This filter generates a file with relative translation and rotation
20041 transform information about subsequent frames, which is then used by
20042 the @ref{vidstabtransform} filter.
20044 To enable compilation of this filter you need to configure FFmpeg with
20045 @code{--enable-libvidstab}.
20047 This filter accepts the following options:
20051 Set the path to the file used to write the transforms information.
20052 Default value is @file{transforms.trf}.
20055 Set how shaky the video is and how quick the camera is. It accepts an
20056 integer in the range 1-10, a value of 1 means little shakiness, a
20057 value of 10 means strong shakiness. Default value is 5.
20060 Set the accuracy of the detection process. It must be a value in the
20061 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20062 accuracy. Default value is 15.
20065 Set stepsize of the search process. The region around minimum is
20066 scanned with 1 pixel resolution. Default value is 6.
20069 Set minimum contrast. Below this value a local measurement field is
20070 discarded. Must be a floating point value in the range 0-1. Default
20074 Set reference frame number for tripod mode.
20076 If enabled, the motion of the frames is compared to a reference frame
20077 in the filtered stream, identified by the specified number. The idea
20078 is to compensate all movements in a more-or-less static scene and keep
20079 the camera view absolutely still.
20081 If set to 0, it is disabled. The frames are counted starting from 1.
20084 Show fields and transforms in the resulting frames. It accepts an
20085 integer in the range 0-2. Default value is 0, which disables any
20089 @subsection Examples
20093 Use default values:
20099 Analyze strongly shaky movie and put the results in file
20100 @file{mytransforms.trf}:
20102 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20106 Visualize the result of internal transformations in the resulting
20109 vidstabdetect=show=1
20113 Analyze a video with medium shakiness using @command{ffmpeg}:
20115 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20119 @anchor{vidstabtransform}
20120 @section vidstabtransform
20122 Video stabilization/deshaking: pass 2 of 2,
20123 see @ref{vidstabdetect} for pass 1.
20125 Read a file with transform information for each frame and
20126 apply/compensate them. Together with the @ref{vidstabdetect}
20127 filter this can be used to deshake videos. See also
20128 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20129 the @ref{unsharp} filter, see below.
20131 To enable compilation of this filter you need to configure FFmpeg with
20132 @code{--enable-libvidstab}.
20134 @subsection Options
20138 Set path to the file used to read the transforms. Default value is
20139 @file{transforms.trf}.
20142 Set the number of frames (value*2 + 1) used for lowpass filtering the
20143 camera movements. Default value is 10.
20145 For example a number of 10 means that 21 frames are used (10 in the
20146 past and 10 in the future) to smoothen the motion in the video. A
20147 larger value leads to a smoother video, but limits the acceleration of
20148 the camera (pan/tilt movements). 0 is a special case where a static
20149 camera is simulated.
20152 Set the camera path optimization algorithm.
20154 Accepted values are:
20157 gaussian kernel low-pass filter on camera motion (default)
20159 averaging on transformations
20163 Set maximal number of pixels to translate frames. Default value is -1,
20167 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20168 value is -1, meaning no limit.
20171 Specify how to deal with borders that may be visible due to movement
20174 Available values are:
20177 keep image information from previous frame (default)
20179 fill the border black
20183 Invert transforms if set to 1. Default value is 0.
20186 Consider transforms as relative to previous frame if set to 1,
20187 absolute if set to 0. Default value is 0.
20190 Set percentage to zoom. A positive value will result in a zoom-in
20191 effect, a negative value in a zoom-out effect. Default value is 0 (no
20195 Set optimal zooming to avoid borders.
20197 Accepted values are:
20202 optimal static zoom value is determined (only very strong movements
20203 will lead to visible borders) (default)
20205 optimal adaptive zoom value is determined (no borders will be
20206 visible), see @option{zoomspeed}
20209 Note that the value given at zoom is added to the one calculated here.
20212 Set percent to zoom maximally each frame (enabled when
20213 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20217 Specify type of interpolation.
20219 Available values are:
20224 linear only horizontal
20226 linear in both directions (default)
20228 cubic in both directions (slow)
20232 Enable virtual tripod mode if set to 1, which is equivalent to
20233 @code{relative=0:smoothing=0}. Default value is 0.
20235 Use also @code{tripod} option of @ref{vidstabdetect}.
20238 Increase log verbosity if set to 1. Also the detected global motions
20239 are written to the temporary file @file{global_motions.trf}. Default
20243 @subsection Examples
20247 Use @command{ffmpeg} for a typical stabilization with default values:
20249 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20252 Note the use of the @ref{unsharp} filter which is always recommended.
20255 Zoom in a bit more and load transform data from a given file:
20257 vidstabtransform=zoom=5:input="mytransforms.trf"
20261 Smoothen the video even more:
20263 vidstabtransform=smoothing=30
20269 Flip the input video vertically.
20271 For example, to vertically flip a video with @command{ffmpeg}:
20273 ffmpeg -i in.avi -vf "vflip" out.avi
20278 Detect variable frame rate video.
20280 This filter tries to detect if the input is variable or constant frame rate.
20282 At end it will output number of frames detected as having variable delta pts,
20283 and ones with constant delta pts.
20284 If there was frames with variable delta, than it will also show min, max and
20285 average delta encountered.
20289 Boost or alter saturation.
20291 The filter accepts the following options:
20294 Set strength of boost if positive value or strength of alter if negative value.
20295 Default is 0. Allowed range is from -2 to 2.
20298 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20301 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20304 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20307 Set the red luma coefficient.
20310 Set the green luma coefficient.
20313 Set the blue luma coefficient.
20316 If @code{intensity} is negative and this is set to 1, colors will change,
20317 otherwise colors will be less saturated, more towards gray.
20320 @subsection Commands
20322 This filter supports the all above options as @ref{commands}.
20327 Make or reverse a natural vignetting effect.
20329 The filter accepts the following options:
20333 Set lens angle expression as a number of radians.
20335 The value is clipped in the @code{[0,PI/2]} range.
20337 Default value: @code{"PI/5"}
20341 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20345 Set forward/backward mode.
20347 Available modes are:
20350 The larger the distance from the central point, the darker the image becomes.
20353 The larger the distance from the central point, the brighter the image becomes.
20354 This can be used to reverse a vignette effect, though there is no automatic
20355 detection to extract the lens @option{angle} and other settings (yet). It can
20356 also be used to create a burning effect.
20359 Default value is @samp{forward}.
20362 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20364 It accepts the following values:
20367 Evaluate expressions only once during the filter initialization.
20370 Evaluate expressions for each incoming frame. This is way slower than the
20371 @samp{init} mode since it requires all the scalers to be re-computed, but it
20372 allows advanced dynamic expressions.
20375 Default value is @samp{init}.
20378 Set dithering to reduce the circular banding effects. Default is @code{1}
20382 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20383 Setting this value to the SAR of the input will make a rectangular vignetting
20384 following the dimensions of the video.
20386 Default is @code{1/1}.
20389 @subsection Expressions
20391 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20392 following parameters.
20397 input width and height
20400 the number of input frame, starting from 0
20403 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20404 @var{TB} units, NAN if undefined
20407 frame rate of the input video, NAN if the input frame rate is unknown
20410 the PTS (Presentation TimeStamp) of the filtered video frame,
20411 expressed in seconds, NAN if undefined
20414 time base of the input video
20418 @subsection Examples
20422 Apply simple strong vignetting effect:
20428 Make a flickering vignetting:
20430 vignette='PI/4+random(1)*PI/50':eval=frame
20435 @section vmafmotion
20437 Obtain the average VMAF motion score of a video.
20438 It is one of the component metrics of VMAF.
20440 The obtained average motion score is printed through the logging system.
20442 The filter accepts the following options:
20446 If specified, the filter will use the named file to save the motion score of
20447 each frame with respect to the previous frame.
20448 When filename equals "-" the data is sent to standard output.
20453 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20457 Stack input videos vertically.
20459 All streams must be of same pixel format and of same width.
20461 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20462 to create same output.
20464 The filter accepts the following options:
20468 Set number of input streams. Default is 2.
20471 If set to 1, force the output to terminate when the shortest input
20472 terminates. Default value is 0.
20477 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20478 Deinterlacing Filter").
20480 Based on the process described by Martin Weston for BBC R&D, and
20481 implemented based on the de-interlace algorithm written by Jim
20482 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20483 uses filter coefficients calculated by BBC R&D.
20485 This filter uses field-dominance information in frame to decide which
20486 of each pair of fields to place first in the output.
20487 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20489 There are two sets of filter coefficients, so called "simple"
20490 and "complex". Which set of filter coefficients is used can
20491 be set by passing an optional parameter:
20495 Set the interlacing filter coefficients. Accepts one of the following values:
20499 Simple filter coefficient set.
20501 More-complex filter coefficient set.
20503 Default value is @samp{complex}.
20506 Specify which frames to deinterlace. Accepts one of the following values:
20510 Deinterlace all frames,
20512 Only deinterlace frames marked as interlaced.
20515 Default value is @samp{all}.
20519 Video waveform monitor.
20521 The waveform monitor plots color component intensity. By default luminance
20522 only. Each column of the waveform corresponds to a column of pixels in the
20525 It accepts the following options:
20529 Can be either @code{row}, or @code{column}. Default is @code{column}.
20530 In row mode, the graph on the left side represents color component value 0 and
20531 the right side represents value = 255. In column mode, the top side represents
20532 color component value = 0 and bottom side represents value = 255.
20535 Set intensity. Smaller values are useful to find out how many values of the same
20536 luminance are distributed across input rows/columns.
20537 Default value is @code{0.04}. Allowed range is [0, 1].
20540 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20541 In mirrored mode, higher values will be represented on the left
20542 side for @code{row} mode and at the top for @code{column} mode. Default is
20543 @code{1} (mirrored).
20547 It accepts the following values:
20550 Presents information identical to that in the @code{parade}, except
20551 that the graphs representing color components are superimposed directly
20554 This display mode makes it easier to spot relative differences or similarities
20555 in overlapping areas of the color components that are supposed to be identical,
20556 such as neutral whites, grays, or blacks.
20559 Display separate graph for the color components side by side in
20560 @code{row} mode or one below the other in @code{column} mode.
20563 Display separate graph for the color components side by side in
20564 @code{column} mode or one below the other in @code{row} mode.
20566 Using this display mode makes it easy to spot color casts in the highlights
20567 and shadows of an image, by comparing the contours of the top and the bottom
20568 graphs of each waveform. Since whites, grays, and blacks are characterized
20569 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20570 should display three waveforms of roughly equal width/height. If not, the
20571 correction is easy to perform by making level adjustments the three waveforms.
20573 Default is @code{stack}.
20575 @item components, c
20576 Set which color components to display. Default is 1, which means only luminance
20577 or red color component if input is in RGB colorspace. If is set for example to
20578 7 it will display all 3 (if) available color components.
20583 No envelope, this is default.
20586 Instant envelope, minimum and maximum values presented in graph will be easily
20587 visible even with small @code{step} value.
20590 Hold minimum and maximum values presented in graph across time. This way you
20591 can still spot out of range values without constantly looking at waveforms.
20594 Peak and instant envelope combined together.
20600 No filtering, this is default.
20603 Luma and chroma combined together.
20606 Similar as above, but shows difference between blue and red chroma.
20609 Similar as above, but use different colors.
20612 Similar as above, but again with different colors.
20615 Displays only chroma.
20618 Displays actual color value on waveform.
20621 Similar as above, but with luma showing frequency of chroma values.
20625 Set which graticule to display.
20629 Do not display graticule.
20632 Display green graticule showing legal broadcast ranges.
20635 Display orange graticule showing legal broadcast ranges.
20638 Display invert graticule showing legal broadcast ranges.
20642 Set graticule opacity.
20645 Set graticule flags.
20649 Draw numbers above lines. By default enabled.
20652 Draw dots instead of lines.
20656 Set scale used for displaying graticule.
20663 Default is digital.
20666 Set background opacity.
20670 Set tint for output.
20671 Only used with lowpass filter and when display is not overlay and input
20672 pixel formats are not RGB.
20675 @section weave, doubleweave
20677 The @code{weave} takes a field-based video input and join
20678 each two sequential fields into single frame, producing a new double
20679 height clip with half the frame rate and half the frame count.
20681 The @code{doubleweave} works same as @code{weave} but without
20682 halving frame rate and frame count.
20684 It accepts the following option:
20688 Set first field. Available values are:
20692 Set the frame as top-field-first.
20695 Set the frame as bottom-field-first.
20699 @subsection Examples
20703 Interlace video using @ref{select} and @ref{separatefields} filter:
20705 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20710 Apply the xBR high-quality magnification filter which is designed for pixel
20711 art. It follows a set of edge-detection rules, see
20712 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
20714 It accepts the following option:
20718 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
20719 @code{3xBR} and @code{4} for @code{4xBR}.
20720 Default is @code{3}.
20725 Apply cross fade from one input video stream to another input video stream.
20726 The cross fade is applied for specified duration.
20728 The filter accepts the following options:
20732 Set one of available transition effects:
20780 Default transition effect is fade.
20783 Set cross fade duration in seconds.
20784 Default duration is 1 second.
20787 Set cross fade start relative to first input stream in seconds.
20788 Default offset is 0.
20791 Set expression for custom transition effect.
20793 The expressions can use the following variables and functions:
20798 The coordinates of the current sample.
20802 The width and height of the image.
20805 Progress of transition effect.
20808 Currently processed plane.
20811 Return value of first input at current location and plane.
20814 Return value of second input at current location and plane.
20820 Return the value of the pixel at location (@var{x},@var{y}) of the
20821 first/second/third/fourth component of first input.
20827 Return the value of the pixel at location (@var{x},@var{y}) of the
20828 first/second/third/fourth component of second input.
20832 @subsection Examples
20836 Cross fade from one input video to another input video, with fade transition and duration of transition
20837 of 2 seconds starting at offset of 5 seconds:
20839 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20844 Pick median pixels from several input videos.
20846 The filter accepts the following options:
20850 Set number of inputs.
20851 Default is 3. Allowed range is from 3 to 255.
20852 If number of inputs is even number, than result will be mean value between two median values.
20855 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
20858 Set median percentile. Default value is @code{0.5}.
20859 Default value of @code{0.5} will pick always median values, while @code{0} will pick
20860 minimum values, and @code{1} maximum values.
20864 Stack video inputs into custom layout.
20866 All streams must be of same pixel format.
20868 The filter accepts the following options:
20872 Set number of input streams. Default is 2.
20875 Specify layout of inputs.
20876 This option requires the desired layout configuration to be explicitly set by the user.
20877 This sets position of each video input in output. Each input
20878 is separated by '|'.
20879 The first number represents the column, and the second number represents the row.
20880 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
20881 where X is video input from which to take width or height.
20882 Multiple values can be used when separated by '+'. In such
20883 case values are summed together.
20885 Note that if inputs are of different sizes gaps may appear, as not all of
20886 the output video frame will be filled. Similarly, videos can overlap each
20887 other if their position doesn't leave enough space for the full frame of
20890 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
20891 a layout must be set by the user.
20894 If set to 1, force the output to terminate when the shortest input
20895 terminates. Default value is 0.
20898 If set to valid color, all unused pixels will be filled with that color.
20899 By default fill is set to none, so it is disabled.
20902 @subsection Examples
20906 Display 4 inputs into 2x2 grid.
20910 input1(0, 0) | input3(w0, 0)
20911 input2(0, h0) | input4(w0, h0)
20915 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20918 Note that if inputs are of different sizes, gaps or overlaps may occur.
20921 Display 4 inputs into 1x4 grid.
20928 input4(0, h0+h1+h2)
20932 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20935 Note that if inputs are of different widths, unused space will appear.
20938 Display 9 inputs into 3x3 grid.
20942 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20943 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20944 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20948 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
20951 Note that if inputs are of different sizes, gaps or overlaps may occur.
20954 Display 16 inputs into 4x4 grid.
20958 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
20959 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
20960 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
20961 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
20965 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|
20966 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
20969 Note that if inputs are of different sizes, gaps or overlaps may occur.
20976 Deinterlace the input video ("yadif" means "yet another deinterlacing
20979 It accepts the following parameters:
20985 The interlacing mode to adopt. It accepts one of the following values:
20988 @item 0, send_frame
20989 Output one frame for each frame.
20990 @item 1, send_field
20991 Output one frame for each field.
20992 @item 2, send_frame_nospatial
20993 Like @code{send_frame}, but it skips the spatial interlacing check.
20994 @item 3, send_field_nospatial
20995 Like @code{send_field}, but it skips the spatial interlacing check.
20998 The default value is @code{send_frame}.
21001 The picture field parity assumed for the input interlaced video. It accepts one
21002 of the following values:
21006 Assume the top field is first.
21008 Assume the bottom field is first.
21010 Enable automatic detection of field parity.
21013 The default value is @code{auto}.
21014 If the interlacing is unknown or the decoder does not export this information,
21015 top field first will be assumed.
21018 Specify which frames to deinterlace. Accepts one of the following
21023 Deinterlace all frames.
21024 @item 1, interlaced
21025 Only deinterlace frames marked as interlaced.
21028 The default value is @code{all}.
21031 @section yadif_cuda
21033 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21034 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21037 It accepts the following parameters:
21043 The interlacing mode to adopt. It accepts one of the following values:
21046 @item 0, send_frame
21047 Output one frame for each frame.
21048 @item 1, send_field
21049 Output one frame for each field.
21050 @item 2, send_frame_nospatial
21051 Like @code{send_frame}, but it skips the spatial interlacing check.
21052 @item 3, send_field_nospatial
21053 Like @code{send_field}, but it skips the spatial interlacing check.
21056 The default value is @code{send_frame}.
21059 The picture field parity assumed for the input interlaced video. It accepts one
21060 of the following values:
21064 Assume the top field is first.
21066 Assume the bottom field is first.
21068 Enable automatic detection of field parity.
21071 The default value is @code{auto}.
21072 If the interlacing is unknown or the decoder does not export this information,
21073 top field first will be assumed.
21076 Specify which frames to deinterlace. Accepts one of the following
21081 Deinterlace all frames.
21082 @item 1, interlaced
21083 Only deinterlace frames marked as interlaced.
21086 The default value is @code{all}.
21091 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21092 The algorithm is described in
21093 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21095 It accepts the following parameters:
21099 Set the window radius. Default value is 3.
21102 Set which planes to filter. Default is only the first plane.
21105 Set blur strength. Default value is 128.
21108 @subsection Commands
21109 This filter supports same @ref{commands} as options.
21113 Apply Zoom & Pan effect.
21115 This filter accepts the following options:
21119 Set the zoom expression. Range is 1-10. Default is 1.
21123 Set the x and y expression. Default is 0.
21126 Set the duration expression in number of frames.
21127 This sets for how many number of frames effect will last for
21128 single input image.
21131 Set the output image size, default is 'hd720'.
21134 Set the output frame rate, default is '25'.
21137 Each expression can contain the following constants:
21156 Output frame count.
21159 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21161 @item out_time, time, ot
21162 The output timestamp expressed in seconds.
21166 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21167 for current input frame.
21171 'x' and 'y' of last output frame of previous input frame or 0 when there was
21172 not yet such frame (first input frame).
21175 Last calculated zoom from 'z' expression for current input frame.
21178 Last calculated zoom of last output frame of previous input frame.
21181 Number of output frames for current input frame. Calculated from 'd' expression
21182 for each input frame.
21185 number of output frames created for previous input frame
21188 Rational number: input width / input height
21191 sample aspect ratio
21194 display aspect ratio
21198 @subsection Examples
21202 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21204 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
21208 Zoom in up to 1.5x and pan always at center of picture:
21210 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21214 Same as above but without pausing:
21216 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21220 Zoom in 2x into center of picture only for the first second of the input video:
21222 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21229 Scale (resize) the input video, using the z.lib library:
21230 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21231 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21233 The zscale filter forces the output display aspect ratio to be the same
21234 as the input, by changing the output sample aspect ratio.
21236 If the input image format is different from the format requested by
21237 the next filter, the zscale filter will convert the input to the
21240 @subsection Options
21241 The filter accepts the following options.
21246 Set the output video dimension expression. Default value is the input
21249 If the @var{width} or @var{w} value is 0, the input width is used for
21250 the output. If the @var{height} or @var{h} value is 0, the input height
21251 is used for the output.
21253 If one and only one of the values is -n with n >= 1, the zscale filter
21254 will use a value that maintains the aspect ratio of the input image,
21255 calculated from the other specified dimension. After that it will,
21256 however, make sure that the calculated dimension is divisible by n and
21257 adjust the value if necessary.
21259 If both values are -n with n >= 1, the behavior will be identical to
21260 both values being set to 0 as previously detailed.
21262 See below for the list of accepted constants for use in the dimension
21266 Set the video size. For the syntax of this option, check the
21267 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21270 Set the dither type.
21272 Possible values are:
21277 @item error_diffusion
21283 Set the resize filter type.
21285 Possible values are:
21295 Default is bilinear.
21298 Set the color range.
21300 Possible values are:
21307 Default is same as input.
21310 Set the color primaries.
21312 Possible values are:
21322 Default is same as input.
21325 Set the transfer characteristics.
21327 Possible values are:
21341 Default is same as input.
21344 Set the colorspace matrix.
21346 Possible value are:
21357 Default is same as input.
21360 Set the input color range.
21362 Possible values are:
21369 Default is same as input.
21371 @item primariesin, pin
21372 Set the input color primaries.
21374 Possible values are:
21384 Default is same as input.
21386 @item transferin, tin
21387 Set the input transfer characteristics.
21389 Possible values are:
21400 Default is same as input.
21402 @item matrixin, min
21403 Set the input colorspace matrix.
21405 Possible value are:
21417 Set the output chroma location.
21419 Possible values are:
21430 @item chromalin, cin
21431 Set the input chroma location.
21433 Possible values are:
21445 Set the nominal peak luminance.
21448 The values of the @option{w} and @option{h} options are expressions
21449 containing the following constants:
21454 The input width and height
21458 These are the same as @var{in_w} and @var{in_h}.
21462 The output (scaled) width and height
21466 These are the same as @var{out_w} and @var{out_h}
21469 The same as @var{iw} / @var{ih}
21472 input sample aspect ratio
21475 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21479 horizontal and vertical input chroma subsample values. For example for the
21480 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21484 horizontal and vertical output chroma subsample values. For example for the
21485 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21488 @subsection Commands
21490 This filter supports the following commands:
21494 Set the output video dimension expression.
21495 The command accepts the same syntax of the corresponding option.
21497 If the specified expression is not valid, it is kept at its current
21501 @c man end VIDEO FILTERS
21503 @chapter OpenCL Video Filters
21504 @c man begin OPENCL VIDEO FILTERS
21506 Below is a description of the currently available OpenCL video filters.
21508 To enable compilation of these filters you need to configure FFmpeg with
21509 @code{--enable-opencl}.
21511 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21514 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21515 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21516 given device parameters.
21518 @item -filter_hw_device @var{name}
21519 Pass the hardware device called @var{name} to all filters in any filter graph.
21523 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21527 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21529 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21533 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.
21535 @section avgblur_opencl
21537 Apply average blur filter.
21539 The filter accepts the following options:
21543 Set horizontal radius size.
21544 Range is @code{[1, 1024]} and default value is @code{1}.
21547 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21550 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21553 @subsection Example
21557 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.
21559 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21563 @section boxblur_opencl
21565 Apply a boxblur algorithm to the input video.
21567 It accepts the following parameters:
21571 @item luma_radius, lr
21572 @item luma_power, lp
21573 @item chroma_radius, cr
21574 @item chroma_power, cp
21575 @item alpha_radius, ar
21576 @item alpha_power, ap
21580 A description of the accepted options follows.
21583 @item luma_radius, lr
21584 @item chroma_radius, cr
21585 @item alpha_radius, ar
21586 Set an expression for the box radius in pixels used for blurring the
21587 corresponding input plane.
21589 The radius value must be a non-negative number, and must not be
21590 greater than the value of the expression @code{min(w,h)/2} for the
21591 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21594 Default value for @option{luma_radius} is "2". If not specified,
21595 @option{chroma_radius} and @option{alpha_radius} default to the
21596 corresponding value set for @option{luma_radius}.
21598 The expressions can contain the following constants:
21602 The input width and height in pixels.
21606 The input chroma image width and height in pixels.
21610 The horizontal and vertical chroma subsample values. For example, for the
21611 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21614 @item luma_power, lp
21615 @item chroma_power, cp
21616 @item alpha_power, ap
21617 Specify how many times the boxblur filter is applied to the
21618 corresponding plane.
21620 Default value for @option{luma_power} is 2. If not specified,
21621 @option{chroma_power} and @option{alpha_power} default to the
21622 corresponding value set for @option{luma_power}.
21624 A value of 0 will disable the effect.
21627 @subsection Examples
21629 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.
21633 Apply a boxblur filter with the luma, chroma, and alpha radius
21634 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.
21636 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21637 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21641 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.
21643 For the luma plane, a 2x2 box radius will be run once.
21645 For the chroma plane, a 4x4 box radius will be run 5 times.
21647 For the alpha plane, a 3x3 box radius will be run 7 times.
21649 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21653 @section colorkey_opencl
21654 RGB colorspace color keying.
21656 The filter accepts the following options:
21660 The color which will be replaced with transparency.
21663 Similarity percentage with the key color.
21665 0.01 matches only the exact key color, while 1.0 matches everything.
21670 0.0 makes pixels either fully transparent, or not transparent at all.
21672 Higher values result in semi-transparent pixels, with a higher transparency
21673 the more similar the pixels color is to the key color.
21676 @subsection Examples
21680 Make every semi-green pixel in the input transparent with some slight blending:
21682 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
21686 @section convolution_opencl
21688 Apply convolution of 3x3, 5x5, 7x7 matrix.
21690 The filter accepts the following options:
21697 Set matrix for each plane.
21698 Matrix is sequence of 9, 25 or 49 signed numbers.
21699 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
21705 Set multiplier for calculated value for each plane.
21706 If unset or 0, it will be sum of all matrix elements.
21707 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
21713 Set bias for each plane. This value is added to the result of the multiplication.
21714 Useful for making the overall image brighter or darker.
21715 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
21719 @subsection Examples
21725 -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
21731 -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
21735 Apply edge enhance:
21737 -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
21743 -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
21747 Apply laplacian edge detector which includes diagonals:
21749 -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
21755 -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
21759 @section erosion_opencl
21761 Apply erosion effect to the video.
21763 This filter replaces the pixel by the local(3x3) minimum.
21765 It accepts the following options:
21772 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21773 If @code{0}, plane will remain unchanged.
21776 Flag which specifies the pixel to refer to.
21777 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21779 Flags to local 3x3 coordinates region centered on @code{x}:
21788 @subsection Example
21792 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.
21794 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21798 @section deshake_opencl
21799 Feature-point based video stabilization filter.
21801 The filter accepts the following options:
21805 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
21808 Whether or not additional debug info should be displayed, both in the processed output and in the console.
21810 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
21812 Viewing point matches in the output video is only supported for RGB input.
21814 Defaults to @code{0}.
21816 @item adaptive_crop
21817 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
21819 Defaults to @code{1}.
21821 @item refine_features
21822 Whether or not feature points should be refined at a sub-pixel level.
21824 This can be turned off for a slight performance gain at the cost of precision.
21826 Defaults to @code{1}.
21828 @item smooth_strength
21829 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
21831 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
21833 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
21835 Defaults to @code{0.0}.
21837 @item smooth_window_multiplier
21838 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
21840 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
21842 Acceptable values range from @code{0.1} to @code{10.0}.
21844 Larger values increase the amount of motion data available for determining how to smooth the camera path,
21845 potentially improving smoothness, but also increase latency and memory usage.
21847 Defaults to @code{2.0}.
21851 @subsection Examples
21855 Stabilize a video with a fixed, medium smoothing strength:
21857 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21861 Stabilize a video with debugging (both in console and in rendered video):
21863 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21867 @section dilation_opencl
21869 Apply dilation effect to the video.
21871 This filter replaces the pixel by the local(3x3) maximum.
21873 It accepts the following options:
21880 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21881 If @code{0}, plane will remain unchanged.
21884 Flag which specifies the pixel to refer to.
21885 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21887 Flags to local 3x3 coordinates region centered on @code{x}:
21896 @subsection Example
21900 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.
21902 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21906 @section nlmeans_opencl
21908 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
21910 @section overlay_opencl
21912 Overlay one video on top of another.
21914 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
21915 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
21917 The filter accepts the following options:
21922 Set the x coordinate of the overlaid video on the main video.
21923 Default value is @code{0}.
21926 Set the y coordinate of the overlaid video on the main video.
21927 Default value is @code{0}.
21931 @subsection Examples
21935 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
21937 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21940 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
21942 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21947 @section pad_opencl
21949 Add paddings to the input image, and place the original input at the
21950 provided @var{x}, @var{y} coordinates.
21952 It accepts the following options:
21957 Specify an expression for the size of the output image with the
21958 paddings added. If the value for @var{width} or @var{height} is 0, the
21959 corresponding input size is used for the output.
21961 The @var{width} expression can reference the value set by the
21962 @var{height} expression, and vice versa.
21964 The default value of @var{width} and @var{height} is 0.
21968 Specify the offsets to place the input image at within the padded area,
21969 with respect to the top/left border of the output image.
21971 The @var{x} expression can reference the value set by the @var{y}
21972 expression, and vice versa.
21974 The default value of @var{x} and @var{y} is 0.
21976 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
21977 so the input image is centered on the padded area.
21980 Specify the color of the padded area. For the syntax of this option,
21981 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
21982 manual,ffmpeg-utils}.
21985 Pad to an aspect instead to a resolution.
21988 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
21989 options are expressions containing the following constants:
21994 The input video width and height.
21998 These are the same as @var{in_w} and @var{in_h}.
22002 The output width and height (the size of the padded area), as
22003 specified by the @var{width} and @var{height} expressions.
22007 These are the same as @var{out_w} and @var{out_h}.
22011 The x and y offsets as specified by the @var{x} and @var{y}
22012 expressions, or NAN if not yet specified.
22015 same as @var{iw} / @var{ih}
22018 input sample aspect ratio
22021 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22024 @section prewitt_opencl
22026 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22028 The filter accepts the following option:
22032 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22035 Set value which will be multiplied with filtered result.
22036 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22039 Set value which will be added to filtered result.
22040 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22043 @subsection Example
22047 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22049 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22053 @anchor{program_opencl}
22054 @section program_opencl
22056 Filter video using an OpenCL program.
22061 OpenCL program source file.
22064 Kernel name in program.
22067 Number of inputs to the filter. Defaults to 1.
22070 Size of output frames. Defaults to the same as the first input.
22074 The @code{program_opencl} filter also supports the @ref{framesync} options.
22076 The program source file must contain a kernel function with the given name,
22077 which will be run once for each plane of the output. Each run on a plane
22078 gets enqueued as a separate 2D global NDRange with one work-item for each
22079 pixel to be generated. The global ID offset for each work-item is therefore
22080 the coordinates of a pixel in the destination image.
22082 The kernel function needs to take the following arguments:
22085 Destination image, @var{__write_only image2d_t}.
22087 This image will become the output; the kernel should write all of it.
22089 Frame index, @var{unsigned int}.
22091 This is a counter starting from zero and increasing by one for each frame.
22093 Source images, @var{__read_only image2d_t}.
22095 These are the most recent images on each input. The kernel may read from
22096 them to generate the output, but they can't be written to.
22103 Copy the input to the output (output must be the same size as the input).
22105 __kernel void copy(__write_only image2d_t destination,
22106 unsigned int index,
22107 __read_only image2d_t source)
22109 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22111 int2 location = (int2)(get_global_id(0), get_global_id(1));
22113 float4 value = read_imagef(source, sampler, location);
22115 write_imagef(destination, location, value);
22120 Apply a simple transformation, rotating the input by an amount increasing
22121 with the index counter. Pixel values are linearly interpolated by the
22122 sampler, and the output need not have the same dimensions as the input.
22124 __kernel void rotate_image(__write_only image2d_t dst,
22125 unsigned int index,
22126 __read_only image2d_t src)
22128 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22129 CLK_FILTER_LINEAR);
22131 float angle = (float)index / 100.0f;
22133 float2 dst_dim = convert_float2(get_image_dim(dst));
22134 float2 src_dim = convert_float2(get_image_dim(src));
22136 float2 dst_cen = dst_dim / 2.0f;
22137 float2 src_cen = src_dim / 2.0f;
22139 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22141 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22143 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22144 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22146 src_pos = src_pos * src_dim / dst_dim;
22148 float2 src_loc = src_pos + src_cen;
22150 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22151 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22152 write_imagef(dst, dst_loc, 0.5f);
22154 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22159 Blend two inputs together, with the amount of each input used varying
22160 with the index counter.
22162 __kernel void blend_images(__write_only image2d_t dst,
22163 unsigned int index,
22164 __read_only image2d_t src1,
22165 __read_only image2d_t src2)
22167 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22168 CLK_FILTER_LINEAR);
22170 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22172 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22173 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22174 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22176 float4 val1 = read_imagef(src1, sampler, src1_loc);
22177 float4 val2 = read_imagef(src2, sampler, src2_loc);
22179 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22185 @section roberts_opencl
22186 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22188 The filter accepts the following option:
22192 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22195 Set value which will be multiplied with filtered result.
22196 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22199 Set value which will be added to filtered result.
22200 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22203 @subsection Example
22207 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22209 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22213 @section sobel_opencl
22215 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22217 The filter accepts the following option:
22221 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22224 Set value which will be multiplied with filtered result.
22225 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22228 Set value which will be added to filtered result.
22229 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22232 @subsection Example
22236 Apply sobel operator with scale set to 2 and delta set to 10
22238 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22242 @section tonemap_opencl
22244 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22246 It accepts the following parameters:
22250 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22253 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22256 Apply desaturation for highlights that exceed this level of brightness. The
22257 higher the parameter, the more color information will be preserved. This
22258 setting helps prevent unnaturally blown-out colors for super-highlights, by
22259 (smoothly) turning into white instead. This makes images feel more natural,
22260 at the cost of reducing information about out-of-range colors.
22262 The default value is 0.5, and the algorithm here is a little different from
22263 the cpu version tonemap currently. A setting of 0.0 disables this option.
22266 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22267 is used to detect whether the scene has changed or not. If the distance between
22268 the current frame average brightness and the current running average exceeds
22269 a threshold value, we would re-calculate scene average and peak brightness.
22270 The default value is 0.2.
22273 Specify the output pixel format.
22275 Currently supported formats are:
22282 Set the output color range.
22284 Possible values are:
22290 Default is same as input.
22293 Set the output color primaries.
22295 Possible values are:
22301 Default is same as input.
22304 Set the output transfer characteristics.
22306 Possible values are:
22315 Set the output colorspace matrix.
22317 Possible value are:
22323 Default is same as input.
22327 @subsection Example
22331 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22333 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22337 @section unsharp_opencl
22339 Sharpen or blur the input video.
22341 It accepts the following parameters:
22344 @item luma_msize_x, lx
22345 Set the luma matrix horizontal size.
22346 Range is @code{[1, 23]} and default value is @code{5}.
22348 @item luma_msize_y, ly
22349 Set the luma matrix vertical size.
22350 Range is @code{[1, 23]} and default value is @code{5}.
22352 @item luma_amount, la
22353 Set the luma effect strength.
22354 Range is @code{[-10, 10]} and default value is @code{1.0}.
22356 Negative values will blur the input video, while positive values will
22357 sharpen it, a value of zero will disable the effect.
22359 @item chroma_msize_x, cx
22360 Set the chroma matrix horizontal size.
22361 Range is @code{[1, 23]} and default value is @code{5}.
22363 @item chroma_msize_y, cy
22364 Set the chroma matrix vertical size.
22365 Range is @code{[1, 23]} and default value is @code{5}.
22367 @item chroma_amount, ca
22368 Set the chroma effect strength.
22369 Range is @code{[-10, 10]} and default value is @code{0.0}.
22371 Negative values will blur the input video, while positive values will
22372 sharpen it, a value of zero will disable the effect.
22376 All parameters are optional and default to the equivalent of the
22377 string '5:5:1.0:5:5:0.0'.
22379 @subsection Examples
22383 Apply strong luma sharpen effect:
22385 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22389 Apply a strong blur of both luma and chroma parameters:
22391 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22395 @section xfade_opencl
22397 Cross fade two videos with custom transition effect by using OpenCL.
22399 It accepts the following options:
22403 Set one of possible transition effects.
22407 Select custom transition effect, the actual transition description
22408 will be picked from source and kernel options.
22420 Default transition is fade.
22424 OpenCL program source file for custom transition.
22427 Set name of kernel to use for custom transition from program source file.
22430 Set duration of video transition.
22433 Set time of start of transition relative to first video.
22436 The program source file must contain a kernel function with the given name,
22437 which will be run once for each plane of the output. Each run on a plane
22438 gets enqueued as a separate 2D global NDRange with one work-item for each
22439 pixel to be generated. The global ID offset for each work-item is therefore
22440 the coordinates of a pixel in the destination image.
22442 The kernel function needs to take the following arguments:
22445 Destination image, @var{__write_only image2d_t}.
22447 This image will become the output; the kernel should write all of it.
22450 First Source image, @var{__read_only image2d_t}.
22451 Second Source image, @var{__read_only image2d_t}.
22453 These are the most recent images on each input. The kernel may read from
22454 them to generate the output, but they can't be written to.
22457 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22464 Apply dots curtain transition effect:
22466 __kernel void blend_images(__write_only image2d_t dst,
22467 __read_only image2d_t src1,
22468 __read_only image2d_t src2,
22471 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22472 CLK_FILTER_LINEAR);
22473 int2 p = (int2)(get_global_id(0), get_global_id(1));
22474 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22475 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22478 float2 dots = (float2)(20.0, 20.0);
22479 float2 center = (float2)(0,0);
22482 float4 val1 = read_imagef(src1, sampler, p);
22483 float4 val2 = read_imagef(src2, sampler, p);
22484 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22486 write_imagef(dst, p, next ? val1 : val2);
22492 @c man end OPENCL VIDEO FILTERS
22494 @chapter VAAPI Video Filters
22495 @c man begin VAAPI VIDEO FILTERS
22497 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22499 To enable compilation of these filters you need to configure FFmpeg with
22500 @code{--enable-vaapi}.
22502 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}
22504 @section tonemap_vaapi
22506 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22507 It maps the dynamic range of HDR10 content to the SDR content.
22508 It currently only accepts HDR10 as input.
22510 It accepts the following parameters:
22514 Specify the output pixel format.
22516 Currently supported formats are:
22525 Set the output color primaries.
22527 Default is same as input.
22530 Set the output transfer characteristics.
22535 Set the output colorspace matrix.
22537 Default is same as input.
22541 @subsection Example
22545 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22547 tonemap_vaapi=format=p010:t=bt2020-10
22551 @c man end VAAPI VIDEO FILTERS
22553 @chapter Video Sources
22554 @c man begin VIDEO SOURCES
22556 Below is a description of the currently available video sources.
22560 Buffer video frames, and make them available to the filter chain.
22562 This source is mainly intended for a programmatic use, in particular
22563 through the interface defined in @file{libavfilter/buffersrc.h}.
22565 It accepts the following parameters:
22570 Specify the size (width and height) of the buffered video frames. For the
22571 syntax of this option, check the
22572 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22575 The input video width.
22578 The input video height.
22581 A string representing the pixel format of the buffered video frames.
22582 It may be a number corresponding to a pixel format, or a pixel format
22586 Specify the timebase assumed by the timestamps of the buffered frames.
22589 Specify the frame rate expected for the video stream.
22591 @item pixel_aspect, sar
22592 The sample (pixel) aspect ratio of the input video.
22595 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22596 to the filtergraph description to specify swscale flags for automatically
22597 inserted scalers. See @ref{Filtergraph syntax}.
22599 @item hw_frames_ctx
22600 When using a hardware pixel format, this should be a reference to an
22601 AVHWFramesContext describing input frames.
22606 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22609 will instruct the source to accept video frames with size 320x240 and
22610 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22611 square pixels (1:1 sample aspect ratio).
22612 Since the pixel format with name "yuv410p" corresponds to the number 6
22613 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22614 this example corresponds to:
22616 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22619 Alternatively, the options can be specified as a flat string, but this
22620 syntax is deprecated:
22622 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22626 Create a pattern generated by an elementary cellular automaton.
22628 The initial state of the cellular automaton can be defined through the
22629 @option{filename} and @option{pattern} options. If such options are
22630 not specified an initial state is created randomly.
22632 At each new frame a new row in the video is filled with the result of
22633 the cellular automaton next generation. The behavior when the whole
22634 frame is filled is defined by the @option{scroll} option.
22636 This source accepts the following options:
22640 Read the initial cellular automaton state, i.e. the starting row, from
22641 the specified file.
22642 In the file, each non-whitespace character is considered an alive
22643 cell, a newline will terminate the row, and further characters in the
22644 file will be ignored.
22647 Read the initial cellular automaton state, i.e. the starting row, from
22648 the specified string.
22650 Each non-whitespace character in the string is considered an alive
22651 cell, a newline will terminate the row, and further characters in the
22652 string will be ignored.
22655 Set the video rate, that is the number of frames generated per second.
22658 @item random_fill_ratio, ratio
22659 Set the random fill ratio for the initial cellular automaton row. It
22660 is a floating point number value ranging from 0 to 1, defaults to
22663 This option is ignored when a file or a pattern is specified.
22665 @item random_seed, seed
22666 Set the seed for filling randomly the initial row, must be an integer
22667 included between 0 and UINT32_MAX. If not specified, or if explicitly
22668 set to -1, the filter will try to use a good random seed on a best
22672 Set the cellular automaton rule, it is a number ranging from 0 to 255.
22673 Default value is 110.
22676 Set the size of the output video. For the syntax of this option, check the
22677 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22679 If @option{filename} or @option{pattern} is specified, the size is set
22680 by default to the width of the specified initial state row, and the
22681 height is set to @var{width} * PHI.
22683 If @option{size} is set, it must contain the width of the specified
22684 pattern string, and the specified pattern will be centered in the
22687 If a filename or a pattern string is not specified, the size value
22688 defaults to "320x518" (used for a randomly generated initial state).
22691 If set to 1, scroll the output upward when all the rows in the output
22692 have been already filled. If set to 0, the new generated row will be
22693 written over the top row just after the bottom row is filled.
22696 @item start_full, full
22697 If set to 1, completely fill the output with generated rows before
22698 outputting the first frame.
22699 This is the default behavior, for disabling set the value to 0.
22702 If set to 1, stitch the left and right row edges together.
22703 This is the default behavior, for disabling set the value to 0.
22706 @subsection Examples
22710 Read the initial state from @file{pattern}, and specify an output of
22713 cellauto=f=pattern:s=200x400
22717 Generate a random initial row with a width of 200 cells, with a fill
22720 cellauto=ratio=2/3:s=200x200
22724 Create a pattern generated by rule 18 starting by a single alive cell
22725 centered on an initial row with width 100:
22727 cellauto=p=@@:s=100x400:full=0:rule=18
22731 Specify a more elaborated initial pattern:
22733 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
22738 @anchor{coreimagesrc}
22739 @section coreimagesrc
22740 Video source generated on GPU using Apple's CoreImage API on OSX.
22742 This video source is a specialized version of the @ref{coreimage} video filter.
22743 Use a core image generator at the beginning of the applied filterchain to
22744 generate the content.
22746 The coreimagesrc video source accepts the following options:
22748 @item list_generators
22749 List all available generators along with all their respective options as well as
22750 possible minimum and maximum values along with the default values.
22752 list_generators=true
22756 Specify the size of the sourced video. For the syntax of this option, check the
22757 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22758 The default value is @code{320x240}.
22761 Specify the frame rate of the sourced video, as the number of frames
22762 generated per second. It has to be a string in the format
22763 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22764 number or a valid video frame rate abbreviation. The default value is
22768 Set the sample aspect ratio of the sourced video.
22771 Set the duration of the sourced video. See
22772 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22773 for the accepted syntax.
22775 If not specified, or the expressed duration is negative, the video is
22776 supposed to be generated forever.
22779 Additionally, all options of the @ref{coreimage} video filter are accepted.
22780 A complete filterchain can be used for further processing of the
22781 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
22782 and examples for details.
22784 @subsection Examples
22789 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22790 given as complete and escaped command-line for Apple's standard bash shell:
22792 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
22794 This example is equivalent to the QRCode example of @ref{coreimage} without the
22795 need for a nullsrc video source.
22800 Generate several gradients.
22804 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22805 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22808 Set frame rate, expressed as number of frames per second. Default
22811 @item c0, c1, c2, c3, c4, c5, c6, c7
22812 Set 8 colors. Default values for colors is to pick random one.
22814 @item x0, y0, y0, y1
22815 Set gradient line source and destination points. If negative or out of range, random ones
22819 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
22822 Set seed for picking gradient line points.
22825 Set the duration of the sourced video. See
22826 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22827 for the accepted syntax.
22829 If not specified, or the expressed duration is negative, the video is
22830 supposed to be generated forever.
22833 Set speed of gradients rotation.
22837 @section mandelbrot
22839 Generate a Mandelbrot set fractal, and progressively zoom towards the
22840 point specified with @var{start_x} and @var{start_y}.
22842 This source accepts the following options:
22847 Set the terminal pts value. Default value is 400.
22850 Set the terminal scale value.
22851 Must be a floating point value. Default value is 0.3.
22854 Set the inner coloring mode, that is the algorithm used to draw the
22855 Mandelbrot fractal internal region.
22857 It shall assume one of the following values:
22862 Show time until convergence.
22864 Set color based on point closest to the origin of the iterations.
22869 Default value is @var{mincol}.
22872 Set the bailout value. Default value is 10.0.
22875 Set the maximum of iterations performed by the rendering
22876 algorithm. Default value is 7189.
22879 Set outer coloring mode.
22880 It shall assume one of following values:
22882 @item iteration_count
22883 Set iteration count mode.
22884 @item normalized_iteration_count
22885 set normalized iteration count mode.
22887 Default value is @var{normalized_iteration_count}.
22890 Set frame rate, expressed as number of frames per second. Default
22894 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22895 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22898 Set the initial scale value. Default value is 3.0.
22901 Set the initial x position. Must be a floating point value between
22902 -100 and 100. Default value is -0.743643887037158704752191506114774.
22905 Set the initial y position. Must be a floating point value between
22906 -100 and 100. Default value is -0.131825904205311970493132056385139.
22911 Generate various test patterns, as generated by the MPlayer test filter.
22913 The size of the generated video is fixed, and is 256x256.
22914 This source is useful in particular for testing encoding features.
22916 This source accepts the following options:
22921 Specify the frame rate of the sourced video, as the number of frames
22922 generated per second. It has to be a string in the format
22923 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22924 number or a valid video frame rate abbreviation. The default value is
22928 Set the duration of the sourced video. See
22929 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22930 for the accepted syntax.
22932 If not specified, or the expressed duration is negative, the video is
22933 supposed to be generated forever.
22937 Set the number or the name of the test to perform. Supported tests are:
22951 @item max_frames, m
22952 Set the maximum number of frames generated for each test, default value is 30.
22956 Default value is "all", which will cycle through the list of all tests.
22961 mptestsrc=t=dc_luma
22964 will generate a "dc_luma" test pattern.
22966 @section frei0r_src
22968 Provide a frei0r source.
22970 To enable compilation of this filter you need to install the frei0r
22971 header and configure FFmpeg with @code{--enable-frei0r}.
22973 This source accepts the following parameters:
22978 The size of the video to generate. For the syntax of this option, check the
22979 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22982 The framerate of the generated video. It may be a string of the form
22983 @var{num}/@var{den} or a frame rate abbreviation.
22986 The name to the frei0r source to load. For more information regarding frei0r and
22987 how to set the parameters, read the @ref{frei0r} section in the video filters
22990 @item filter_params
22991 A '|'-separated list of parameters to pass to the frei0r source.
22995 For example, to generate a frei0r partik0l source with size 200x200
22996 and frame rate 10 which is overlaid on the overlay filter main input:
22998 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23003 Generate a life pattern.
23005 This source is based on a generalization of John Conway's life game.
23007 The sourced input represents a life grid, each pixel represents a cell
23008 which can be in one of two possible states, alive or dead. Every cell
23009 interacts with its eight neighbours, which are the cells that are
23010 horizontally, vertically, or diagonally adjacent.
23012 At each interaction the grid evolves according to the adopted rule,
23013 which specifies the number of neighbor alive cells which will make a
23014 cell stay alive or born. The @option{rule} option allows one to specify
23017 This source accepts the following options:
23021 Set the file from which to read the initial grid state. In the file,
23022 each non-whitespace character is considered an alive cell, and newline
23023 is used to delimit the end of each row.
23025 If this option is not specified, the initial grid is generated
23029 Set the video rate, that is the number of frames generated per second.
23032 @item random_fill_ratio, ratio
23033 Set the random fill ratio for the initial random grid. It is a
23034 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23035 It is ignored when a file is specified.
23037 @item random_seed, seed
23038 Set the seed for filling the initial random grid, must be an integer
23039 included between 0 and UINT32_MAX. If not specified, or if explicitly
23040 set to -1, the filter will try to use a good random seed on a best
23046 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23047 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23048 @var{NS} specifies the number of alive neighbor cells which make a
23049 live cell stay alive, and @var{NB} the number of alive neighbor cells
23050 which make a dead cell to become alive (i.e. to "born").
23051 "s" and "b" can be used in place of "S" and "B", respectively.
23053 Alternatively a rule can be specified by an 18-bits integer. The 9
23054 high order bits are used to encode the next cell state if it is alive
23055 for each number of neighbor alive cells, the low order bits specify
23056 the rule for "borning" new cells. Higher order bits encode for an
23057 higher number of neighbor cells.
23058 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23059 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23061 Default value is "S23/B3", which is the original Conway's game of life
23062 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23063 cells, and will born a new cell if there are three alive cells around
23067 Set the size of the output video. For the syntax of this option, check the
23068 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23070 If @option{filename} is specified, the size is set by default to the
23071 same size of the input file. If @option{size} is set, it must contain
23072 the size specified in the input file, and the initial grid defined in
23073 that file is centered in the larger resulting area.
23075 If a filename is not specified, the size value defaults to "320x240"
23076 (used for a randomly generated initial grid).
23079 If set to 1, stitch the left and right grid edges together, and the
23080 top and bottom edges also. Defaults to 1.
23083 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23084 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23085 value from 0 to 255.
23088 Set the color of living (or new born) cells.
23091 Set the color of dead cells. If @option{mold} is set, this is the first color
23092 used to represent a dead cell.
23095 Set mold color, for definitely dead and moldy cells.
23097 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23098 ffmpeg-utils manual,ffmpeg-utils}.
23101 @subsection Examples
23105 Read a grid from @file{pattern}, and center it on a grid of size
23108 life=f=pattern:s=300x300
23112 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23114 life=ratio=2/3:s=200x200
23118 Specify a custom rule for evolving a randomly generated grid:
23124 Full example with slow death effect (mold) using @command{ffplay}:
23126 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23133 @anchor{haldclutsrc}
23136 @anchor{pal100bars}
23137 @anchor{rgbtestsrc}
23139 @anchor{smptehdbars}
23142 @anchor{yuvtestsrc}
23143 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23145 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23147 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23149 The @code{color} source provides an uniformly colored input.
23151 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23152 @ref{haldclut} filter.
23154 The @code{nullsrc} source returns unprocessed video frames. It is
23155 mainly useful to be employed in analysis / debugging tools, or as the
23156 source for filters which ignore the input data.
23158 The @code{pal75bars} source generates a color bars pattern, based on
23159 EBU PAL recommendations with 75% color levels.
23161 The @code{pal100bars} source generates a color bars pattern, based on
23162 EBU PAL recommendations with 100% color levels.
23164 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23165 detecting RGB vs BGR issues. You should see a red, green and blue
23166 stripe from top to bottom.
23168 The @code{smptebars} source generates a color bars pattern, based on
23169 the SMPTE Engineering Guideline EG 1-1990.
23171 The @code{smptehdbars} source generates a color bars pattern, based on
23172 the SMPTE RP 219-2002.
23174 The @code{testsrc} source generates a test video pattern, showing a
23175 color pattern, a scrolling gradient and a timestamp. This is mainly
23176 intended for testing purposes.
23178 The @code{testsrc2} source is similar to testsrc, but supports more
23179 pixel formats instead of just @code{rgb24}. This allows using it as an
23180 input for other tests without requiring a format conversion.
23182 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23183 see a y, cb and cr stripe from top to bottom.
23185 The sources accept the following parameters:
23190 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23191 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23192 pixels to be used as identity matrix for 3D lookup tables. Each component is
23193 coded on a @code{1/(N*N)} scale.
23196 Specify the color of the source, only available in the @code{color}
23197 source. For the syntax of this option, check the
23198 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23201 Specify the size of the sourced video. For the syntax of this option, check the
23202 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23203 The default value is @code{320x240}.
23205 This option is not available with the @code{allrgb}, @code{allyuv}, and
23206 @code{haldclutsrc} filters.
23209 Specify the frame rate of the sourced video, as the number of frames
23210 generated per second. It has to be a string in the format
23211 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23212 number or a valid video frame rate abbreviation. The default value is
23216 Set the duration of the sourced video. See
23217 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23218 for the accepted syntax.
23220 If not specified, or the expressed duration is negative, the video is
23221 supposed to be generated forever.
23223 Since the frame rate is used as time base, all frames including the last one
23224 will have their full duration. If the specified duration is not a multiple
23225 of the frame duration, it will be rounded up.
23228 Set the sample aspect ratio of the sourced video.
23231 Specify the alpha (opacity) of the background, only available in the
23232 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23233 255 (fully opaque, the default).
23236 Set the number of decimals to show in the timestamp, only available in the
23237 @code{testsrc} source.
23239 The displayed timestamp value will correspond to the original
23240 timestamp value multiplied by the power of 10 of the specified
23241 value. Default value is 0.
23244 @subsection Examples
23248 Generate a video with a duration of 5.3 seconds, with size
23249 176x144 and a frame rate of 10 frames per second:
23251 testsrc=duration=5.3:size=qcif:rate=10
23255 The following graph description will generate a red source
23256 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23259 color=c=red@@0.2:s=qcif:r=10
23263 If the input content is to be ignored, @code{nullsrc} can be used. The
23264 following command generates noise in the luminance plane by employing
23265 the @code{geq} filter:
23267 nullsrc=s=256x256, geq=random(1)*255:128:128
23271 @subsection Commands
23273 The @code{color} source supports the following commands:
23277 Set the color of the created image. Accepts the same syntax of the
23278 corresponding @option{color} option.
23283 Generate video using an OpenCL program.
23288 OpenCL program source file.
23291 Kernel name in program.
23294 Size of frames to generate. This must be set.
23297 Pixel format to use for the generated frames. This must be set.
23300 Number of frames generated every second. Default value is '25'.
23304 For details of how the program loading works, see the @ref{program_opencl}
23311 Generate a colour ramp by setting pixel values from the position of the pixel
23312 in the output image. (Note that this will work with all pixel formats, but
23313 the generated output will not be the same.)
23315 __kernel void ramp(__write_only image2d_t dst,
23316 unsigned int index)
23318 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23321 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23323 write_imagef(dst, loc, val);
23328 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23330 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23331 unsigned int index)
23333 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23335 float4 value = 0.0f;
23336 int x = loc.x + index;
23337 int y = loc.y + index;
23338 while (x > 0 || y > 0) {
23339 if (x % 3 == 1 && y % 3 == 1) {
23347 write_imagef(dst, loc, value);
23353 @section sierpinski
23355 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23357 This source accepts the following options:
23361 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23362 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23365 Set frame rate, expressed as number of frames per second. Default
23369 Set seed which is used for random panning.
23372 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23375 Set fractal type, can be default @code{carpet} or @code{triangle}.
23378 @c man end VIDEO SOURCES
23380 @chapter Video Sinks
23381 @c man begin VIDEO SINKS
23383 Below is a description of the currently available video sinks.
23385 @section buffersink
23387 Buffer video frames, and make them available to the end of the filter
23390 This sink is mainly intended for programmatic use, in particular
23391 through the interface defined in @file{libavfilter/buffersink.h}
23392 or the options system.
23394 It accepts a pointer to an AVBufferSinkContext structure, which
23395 defines the incoming buffers' formats, to be passed as the opaque
23396 parameter to @code{avfilter_init_filter} for initialization.
23400 Null video sink: do absolutely nothing with the input video. It is
23401 mainly useful as a template and for use in analysis / debugging
23404 @c man end VIDEO SINKS
23406 @chapter Multimedia Filters
23407 @c man begin MULTIMEDIA FILTERS
23409 Below is a description of the currently available multimedia filters.
23413 Convert input audio to a video output, displaying the audio bit scope.
23415 The filter accepts the following options:
23419 Set frame rate, expressed as number of frames per second. Default
23423 Specify the video size for the output. For the syntax of this option, check the
23424 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23425 Default value is @code{1024x256}.
23428 Specify list of colors separated by space or by '|' which will be used to
23429 draw channels. Unrecognized or missing colors will be replaced
23433 @section adrawgraph
23434 Draw a graph using input audio metadata.
23436 See @ref{drawgraph}
23438 @section agraphmonitor
23440 See @ref{graphmonitor}.
23442 @section ahistogram
23444 Convert input audio to a video output, displaying the volume histogram.
23446 The filter accepts the following options:
23450 Specify how histogram is calculated.
23452 It accepts the following values:
23455 Use single histogram for all channels.
23457 Use separate histogram for each channel.
23459 Default is @code{single}.
23462 Set frame rate, expressed as number of frames per second. Default
23466 Specify the video size for the output. For the syntax of this option, check the
23467 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23468 Default value is @code{hd720}.
23473 It accepts the following values:
23484 reverse logarithmic
23486 Default is @code{log}.
23489 Set amplitude scale.
23491 It accepts the following values:
23498 Default is @code{log}.
23501 Set how much frames to accumulate in histogram.
23502 Default is 1. Setting this to -1 accumulates all frames.
23505 Set histogram ratio of window height.
23508 Set sonogram sliding.
23510 It accepts the following values:
23513 replace old rows with new ones.
23515 scroll from top to bottom.
23517 Default is @code{replace}.
23520 @section aphasemeter
23522 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23523 representing mean phase of current audio frame. A video output can also be produced and is
23524 enabled by default. The audio is passed through as first output.
23526 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23527 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23528 and @code{1} means channels are in phase.
23530 The filter accepts the following options, all related to its video output:
23534 Set the output frame rate. Default value is @code{25}.
23537 Set the video size for the output. For the syntax of this option, check the
23538 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23539 Default value is @code{800x400}.
23544 Specify the red, green, blue contrast. Default values are @code{2},
23545 @code{7} and @code{1}.
23546 Allowed range is @code{[0, 255]}.
23549 Set color which will be used for drawing median phase. If color is
23550 @code{none} which is default, no median phase value will be drawn.
23553 Enable video output. Default is enabled.
23556 @subsection phasing detection
23558 The filter also detects out of phase and mono sequences in stereo streams.
23559 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23561 The filter accepts the following options for this detection:
23565 Enable mono and out of phase detection. Default is disabled.
23568 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23569 Allowed range is @code{[0, 1]}.
23572 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23573 Allowed range is @code{[90, 180]}.
23576 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23579 @subsection Examples
23583 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23585 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23589 @section avectorscope
23591 Convert input audio to a video output, representing the audio vector
23594 The filter is used to measure the difference between channels of stereo
23595 audio stream. A monaural signal, consisting of identical left and right
23596 signal, results in straight vertical line. Any stereo separation is visible
23597 as a deviation from this line, creating a Lissajous figure.
23598 If the straight (or deviation from it) but horizontal line appears this
23599 indicates that the left and right channels are out of phase.
23601 The filter accepts the following options:
23605 Set the vectorscope mode.
23607 Available values are:
23610 Lissajous rotated by 45 degrees.
23613 Same as above but not rotated.
23616 Shape resembling half of circle.
23619 Default value is @samp{lissajous}.
23622 Set the video size for the output. For the syntax of this option, check the
23623 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23624 Default value is @code{400x400}.
23627 Set the output frame rate. Default value is @code{25}.
23633 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23634 @code{160}, @code{80} and @code{255}.
23635 Allowed range is @code{[0, 255]}.
23641 Specify the red, green, blue and alpha fade. Default values are @code{15},
23642 @code{10}, @code{5} and @code{5}.
23643 Allowed range is @code{[0, 255]}.
23646 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23647 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23650 Set the vectorscope drawing mode.
23652 Available values are:
23655 Draw dot for each sample.
23658 Draw line between previous and current sample.
23661 Default value is @samp{dot}.
23664 Specify amplitude scale of audio samples.
23666 Available values are:
23682 Swap left channel axis with right channel axis.
23692 Mirror only x axis.
23695 Mirror only y axis.
23703 @subsection Examples
23707 Complete example using @command{ffplay}:
23709 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23710 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
23714 @section bench, abench
23716 Benchmark part of a filtergraph.
23718 The filter accepts the following options:
23722 Start or stop a timer.
23724 Available values are:
23727 Get the current time, set it as frame metadata (using the key
23728 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
23731 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
23732 the input frame metadata to get the time difference. Time difference, average,
23733 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
23734 @code{min}) are then printed. The timestamps are expressed in seconds.
23738 @subsection Examples
23742 Benchmark @ref{selectivecolor} filter:
23744 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
23750 Concatenate audio and video streams, joining them together one after the
23753 The filter works on segments of synchronized video and audio streams. All
23754 segments must have the same number of streams of each type, and that will
23755 also be the number of streams at output.
23757 The filter accepts the following options:
23762 Set the number of segments. Default is 2.
23765 Set the number of output video streams, that is also the number of video
23766 streams in each segment. Default is 1.
23769 Set the number of output audio streams, that is also the number of audio
23770 streams in each segment. Default is 0.
23773 Activate unsafe mode: do not fail if segments have a different format.
23777 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
23778 @var{a} audio outputs.
23780 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
23781 segment, in the same order as the outputs, then the inputs for the second
23784 Related streams do not always have exactly the same duration, for various
23785 reasons including codec frame size or sloppy authoring. For that reason,
23786 related synchronized streams (e.g. a video and its audio track) should be
23787 concatenated at once. The concat filter will use the duration of the longest
23788 stream in each segment (except the last one), and if necessary pad shorter
23789 audio streams with silence.
23791 For this filter to work correctly, all segments must start at timestamp 0.
23793 All corresponding streams must have the same parameters in all segments; the
23794 filtering system will automatically select a common pixel format for video
23795 streams, and a common sample format, sample rate and channel layout for
23796 audio streams, but other settings, such as resolution, must be converted
23797 explicitly by the user.
23799 Different frame rates are acceptable but will result in variable frame rate
23800 at output; be sure to configure the output file to handle it.
23802 @subsection Examples
23806 Concatenate an opening, an episode and an ending, all in bilingual version
23807 (video in stream 0, audio in streams 1 and 2):
23809 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23810 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23811 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23812 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23816 Concatenate two parts, handling audio and video separately, using the
23817 (a)movie sources, and adjusting the resolution:
23819 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23820 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23821 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23823 Note that a desync will happen at the stitch if the audio and video streams
23824 do not have exactly the same duration in the first file.
23828 @subsection Commands
23830 This filter supports the following commands:
23833 Close the current segment and step to the next one
23839 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
23840 level. By default, it logs a message at a frequency of 10Hz with the
23841 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
23842 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
23844 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
23845 sample format is double-precision floating point. The input stream will be converted to
23846 this specification, if needed. Users may need to insert aformat and/or aresample filters
23847 after this filter to obtain the original parameters.
23849 The filter also has a video output (see the @var{video} option) with a real
23850 time graph to observe the loudness evolution. The graphic contains the logged
23851 message mentioned above, so it is not printed anymore when this option is set,
23852 unless the verbose logging is set. The main graphing area contains the
23853 short-term loudness (3 seconds of analysis), and the gauge on the right is for
23854 the momentary loudness (400 milliseconds), but can optionally be configured
23855 to instead display short-term loudness (see @var{gauge}).
23857 The green area marks a +/- 1LU target range around the target loudness
23858 (-23LUFS by default, unless modified through @var{target}).
23860 More information about the Loudness Recommendation EBU R128 on
23861 @url{http://tech.ebu.ch/loudness}.
23863 The filter accepts the following options:
23868 Activate the video output. The audio stream is passed unchanged whether this
23869 option is set or no. The video stream will be the first output stream if
23870 activated. Default is @code{0}.
23873 Set the video size. This option is for video only. For the syntax of this
23875 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23876 Default and minimum resolution is @code{640x480}.
23879 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
23880 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
23881 other integer value between this range is allowed.
23884 Set metadata injection. If set to @code{1}, the audio input will be segmented
23885 into 100ms output frames, each of them containing various loudness information
23886 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
23888 Default is @code{0}.
23891 Force the frame logging level.
23893 Available values are:
23896 information logging level
23898 verbose logging level
23901 By default, the logging level is set to @var{info}. If the @option{video} or
23902 the @option{metadata} options are set, it switches to @var{verbose}.
23907 Available modes can be cumulated (the option is a @code{flag} type). Possible
23911 Disable any peak mode (default).
23913 Enable sample-peak mode.
23915 Simple peak mode looking for the higher sample value. It logs a message
23916 for sample-peak (identified by @code{SPK}).
23918 Enable true-peak mode.
23920 If enabled, the peak lookup is done on an over-sampled version of the input
23921 stream for better peak accuracy. It logs a message for true-peak.
23922 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
23923 This mode requires a build with @code{libswresample}.
23927 Treat mono input files as "dual mono". If a mono file is intended for playback
23928 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
23929 If set to @code{true}, this option will compensate for this effect.
23930 Multi-channel input files are not affected by this option.
23933 Set a specific pan law to be used for the measurement of dual mono files.
23934 This parameter is optional, and has a default value of -3.01dB.
23937 Set a specific target level (in LUFS) used as relative zero in the visualization.
23938 This parameter is optional and has a default value of -23LUFS as specified
23939 by EBU R128. However, material published online may prefer a level of -16LUFS
23940 (e.g. for use with podcasts or video platforms).
23943 Set the value displayed by the gauge. Valid values are @code{momentary} and s
23944 @code{shortterm}. By default the momentary value will be used, but in certain
23945 scenarios it may be more useful to observe the short term value instead (e.g.
23949 Sets the display scale for the loudness. Valid parameters are @code{absolute}
23950 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
23951 video output, not the summary or continuous log output.
23954 @subsection Examples
23958 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
23960 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
23964 Run an analysis with @command{ffmpeg}:
23966 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
23970 @section interleave, ainterleave
23972 Temporally interleave frames from several inputs.
23974 @code{interleave} works with video inputs, @code{ainterleave} with audio.
23976 These filters read frames from several inputs and send the oldest
23977 queued frame to the output.
23979 Input streams must have well defined, monotonically increasing frame
23982 In order to submit one frame to output, these filters need to enqueue
23983 at least one frame for each input, so they cannot work in case one
23984 input is not yet terminated and will not receive incoming frames.
23986 For example consider the case when one input is a @code{select} filter
23987 which always drops input frames. The @code{interleave} filter will keep
23988 reading from that input, but it will never be able to send new frames
23989 to output until the input sends an end-of-stream signal.
23991 Also, depending on inputs synchronization, the filters will drop
23992 frames in case one input receives more frames than the other ones, and
23993 the queue is already filled.
23995 These filters accept the following options:
23999 Set the number of different inputs, it is 2 by default.
24002 How to determine the end-of-stream.
24006 The duration of the longest input. (default)
24009 The duration of the shortest input.
24012 The duration of the first input.
24017 @subsection Examples
24021 Interleave frames belonging to different streams using @command{ffmpeg}:
24023 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24027 Add flickering blur effect:
24029 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24033 @section metadata, ametadata
24035 Manipulate frame metadata.
24037 This filter accepts the following options:
24041 Set mode of operation of the filter.
24043 Can be one of the following:
24047 If both @code{value} and @code{key} is set, select frames
24048 which have such metadata. If only @code{key} is set, select
24049 every frame that has such key in metadata.
24052 Add new metadata @code{key} and @code{value}. If key is already available
24056 Modify value of already present key.
24059 If @code{value} is set, delete only keys that have such value.
24060 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24064 Print key and its value if metadata was found. If @code{key} is not set print all
24065 metadata values available in frame.
24069 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24072 Set metadata value which will be used. This option is mandatory for
24073 @code{modify} and @code{add} mode.
24076 Which function to use when comparing metadata value and @code{value}.
24078 Can be one of following:
24082 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24085 Values are interpreted as strings, returns true if metadata value starts with
24086 the @code{value} option string.
24089 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24092 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24095 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24098 Values are interpreted as floats, returns true if expression from option @code{expr}
24102 Values are interpreted as strings, returns true if metadata value ends with
24103 the @code{value} option string.
24107 Set expression which is used when @code{function} is set to @code{expr}.
24108 The expression is evaluated through the eval API and can contain the following
24113 Float representation of @code{value} from metadata key.
24116 Float representation of @code{value} as supplied by user in @code{value} option.
24120 If specified in @code{print} mode, output is written to the named file. Instead of
24121 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24122 for standard output. If @code{file} option is not set, output is written to the log
24123 with AV_LOG_INFO loglevel.
24126 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24130 @subsection Examples
24134 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24137 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24140 Print silencedetect output to file @file{metadata.txt}.
24142 silencedetect,ametadata=mode=print:file=metadata.txt
24145 Direct all metadata to a pipe with file descriptor 4.
24147 metadata=mode=print:file='pipe\:4'
24151 @section perms, aperms
24153 Set read/write permissions for the output frames.
24155 These filters are mainly aimed at developers to test direct path in the
24156 following filter in the filtergraph.
24158 The filters accept the following options:
24162 Select the permissions mode.
24164 It accepts the following values:
24167 Do nothing. This is the default.
24169 Set all the output frames read-only.
24171 Set all the output frames directly writable.
24173 Make the frame read-only if writable, and writable if read-only.
24175 Set each output frame read-only or writable randomly.
24179 Set the seed for the @var{random} mode, must be an integer included between
24180 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24181 @code{-1}, the filter will try to use a good random seed on a best effort
24185 Note: in case of auto-inserted filter between the permission filter and the
24186 following one, the permission might not be received as expected in that
24187 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24188 perms/aperms filter can avoid this problem.
24190 @section realtime, arealtime
24192 Slow down filtering to match real time approximately.
24194 These filters will pause the filtering for a variable amount of time to
24195 match the output rate with the input timestamps.
24196 They are similar to the @option{re} option to @code{ffmpeg}.
24198 They accept the following options:
24202 Time limit for the pauses. Any pause longer than that will be considered
24203 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24205 Speed factor for processing. The value must be a float larger than zero.
24206 Values larger than 1.0 will result in faster than realtime processing,
24207 smaller will slow processing down. The @var{limit} is automatically adapted
24208 accordingly. Default is 1.0.
24210 A processing speed faster than what is possible without these filters cannot
24215 @section select, aselect
24217 Select frames to pass in output.
24219 This filter accepts the following options:
24224 Set expression, which is evaluated for each input frame.
24226 If the expression is evaluated to zero, the frame is discarded.
24228 If the evaluation result is negative or NaN, the frame is sent to the
24229 first output; otherwise it is sent to the output with index
24230 @code{ceil(val)-1}, assuming that the input index starts from 0.
24232 For example a value of @code{1.2} corresponds to the output with index
24233 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24236 Set the number of outputs. The output to which to send the selected
24237 frame is based on the result of the evaluation. Default value is 1.
24240 The expression can contain the following constants:
24244 The (sequential) number of the filtered frame, starting from 0.
24247 The (sequential) number of the selected frame, starting from 0.
24249 @item prev_selected_n
24250 The sequential number of the last selected frame. It's NAN if undefined.
24253 The timebase of the input timestamps.
24256 The PTS (Presentation TimeStamp) of the filtered video frame,
24257 expressed in @var{TB} units. It's NAN if undefined.
24260 The PTS of the filtered video frame,
24261 expressed in seconds. It's NAN if undefined.
24264 The PTS of the previously filtered video frame. It's NAN if undefined.
24266 @item prev_selected_pts
24267 The PTS of the last previously filtered video frame. It's NAN if undefined.
24269 @item prev_selected_t
24270 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24273 The PTS of the first video frame in the video. It's NAN if undefined.
24276 The time of the first video frame in the video. It's NAN if undefined.
24278 @item pict_type @emph{(video only)}
24279 The type of the filtered frame. It can assume one of the following
24291 @item interlace_type @emph{(video only)}
24292 The frame interlace type. It can assume one of the following values:
24295 The frame is progressive (not interlaced).
24297 The frame is top-field-first.
24299 The frame is bottom-field-first.
24302 @item consumed_sample_n @emph{(audio only)}
24303 the number of selected samples before the current frame
24305 @item samples_n @emph{(audio only)}
24306 the number of samples in the current frame
24308 @item sample_rate @emph{(audio only)}
24309 the input sample rate
24312 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24315 the position in the file of the filtered frame, -1 if the information
24316 is not available (e.g. for synthetic video)
24318 @item scene @emph{(video only)}
24319 value between 0 and 1 to indicate a new scene; a low value reflects a low
24320 probability for the current frame to introduce a new scene, while a higher
24321 value means the current frame is more likely to be one (see the example below)
24323 @item concatdec_select
24324 The concat demuxer can select only part of a concat input file by setting an
24325 inpoint and an outpoint, but the output packets may not be entirely contained
24326 in the selected interval. By using this variable, it is possible to skip frames
24327 generated by the concat demuxer which are not exactly contained in the selected
24330 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24331 and the @var{lavf.concat.duration} packet metadata values which are also
24332 present in the decoded frames.
24334 The @var{concatdec_select} variable is -1 if the frame pts is at least
24335 start_time and either the duration metadata is missing or the frame pts is less
24336 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24339 That basically means that an input frame is selected if its pts is within the
24340 interval set by the concat demuxer.
24344 The default value of the select expression is "1".
24346 @subsection Examples
24350 Select all frames in input:
24355 The example above is the same as:
24367 Select only I-frames:
24369 select='eq(pict_type\,I)'
24373 Select one frame every 100:
24375 select='not(mod(n\,100))'
24379 Select only frames contained in the 10-20 time interval:
24381 select=between(t\,10\,20)
24385 Select only I-frames contained in the 10-20 time interval:
24387 select=between(t\,10\,20)*eq(pict_type\,I)
24391 Select frames with a minimum distance of 10 seconds:
24393 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24397 Use aselect to select only audio frames with samples number > 100:
24399 aselect='gt(samples_n\,100)'
24403 Create a mosaic of the first scenes:
24405 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24408 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24412 Send even and odd frames to separate outputs, and compose them:
24414 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24418 Select useful frames from an ffconcat file which is using inpoints and
24419 outpoints but where the source files are not intra frame only.
24421 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24425 @section sendcmd, asendcmd
24427 Send commands to filters in the filtergraph.
24429 These filters read commands to be sent to other filters in the
24432 @code{sendcmd} must be inserted between two video filters,
24433 @code{asendcmd} must be inserted between two audio filters, but apart
24434 from that they act the same way.
24436 The specification of commands can be provided in the filter arguments
24437 with the @var{commands} option, or in a file specified by the
24438 @var{filename} option.
24440 These filters accept the following options:
24443 Set the commands to be read and sent to the other filters.
24445 Set the filename of the commands to be read and sent to the other
24449 @subsection Commands syntax
24451 A commands description consists of a sequence of interval
24452 specifications, comprising a list of commands to be executed when a
24453 particular event related to that interval occurs. The occurring event
24454 is typically the current frame time entering or leaving a given time
24457 An interval is specified by the following syntax:
24459 @var{START}[-@var{END}] @var{COMMANDS};
24462 The time interval is specified by the @var{START} and @var{END} times.
24463 @var{END} is optional and defaults to the maximum time.
24465 The current frame time is considered within the specified interval if
24466 it is included in the interval [@var{START}, @var{END}), that is when
24467 the time is greater or equal to @var{START} and is lesser than
24470 @var{COMMANDS} consists of a sequence of one or more command
24471 specifications, separated by ",", relating to that interval. The
24472 syntax of a command specification is given by:
24474 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24477 @var{FLAGS} is optional and specifies the type of events relating to
24478 the time interval which enable sending the specified command, and must
24479 be a non-null sequence of identifier flags separated by "+" or "|" and
24480 enclosed between "[" and "]".
24482 The following flags are recognized:
24485 The command is sent when the current frame timestamp enters the
24486 specified interval. In other words, the command is sent when the
24487 previous frame timestamp was not in the given interval, and the
24491 The command is sent when the current frame timestamp leaves the
24492 specified interval. In other words, the command is sent when the
24493 previous frame timestamp was in the given interval, and the
24497 The command @var{ARG} is interpreted as expression and result of
24498 expression is passed as @var{ARG}.
24500 The expression is evaluated through the eval API and can contain the following
24505 Original position in the file of the frame, or undefined if undefined
24506 for the current frame.
24509 The presentation timestamp in input.
24512 The count of the input frame for video or audio, starting from 0.
24515 The time in seconds of the current frame.
24518 The start time in seconds of the current command interval.
24521 The end time in seconds of the current command interval.
24524 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24529 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24532 @var{TARGET} specifies the target of the command, usually the name of
24533 the filter class or a specific filter instance name.
24535 @var{COMMAND} specifies the name of the command for the target filter.
24537 @var{ARG} is optional and specifies the optional list of argument for
24538 the given @var{COMMAND}.
24540 Between one interval specification and another, whitespaces, or
24541 sequences of characters starting with @code{#} until the end of line,
24542 are ignored and can be used to annotate comments.
24544 A simplified BNF description of the commands specification syntax
24547 @var{COMMAND_FLAG} ::= "enter" | "leave"
24548 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24549 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24550 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24551 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24552 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24555 @subsection Examples
24559 Specify audio tempo change at second 4:
24561 asendcmd=c='4.0 atempo tempo 1.5',atempo
24565 Target a specific filter instance:
24567 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24571 Specify a list of drawtext and hue commands in a file.
24573 # show text in the interval 5-10
24574 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24575 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24577 # desaturate the image in the interval 15-20
24578 15.0-20.0 [enter] hue s 0,
24579 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24581 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24583 # apply an exponential saturation fade-out effect, starting from time 25
24584 25 [enter] hue s exp(25-t)
24587 A filtergraph allowing to read and process the above command list
24588 stored in a file @file{test.cmd}, can be specified with:
24590 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24595 @section setpts, asetpts
24597 Change the PTS (presentation timestamp) of the input frames.
24599 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24601 This filter accepts the following options:
24606 The expression which is evaluated for each frame to construct its timestamp.
24610 The expression is evaluated through the eval API and can contain the following
24614 @item FRAME_RATE, FR
24615 frame rate, only defined for constant frame-rate video
24618 The presentation timestamp in input
24621 The count of the input frame for video or the number of consumed samples,
24622 not including the current frame for audio, starting from 0.
24624 @item NB_CONSUMED_SAMPLES
24625 The number of consumed samples, not including the current frame (only
24628 @item NB_SAMPLES, S
24629 The number of samples in the current frame (only audio)
24631 @item SAMPLE_RATE, SR
24632 The audio sample rate.
24635 The PTS of the first frame.
24638 the time in seconds of the first frame
24641 State whether the current frame is interlaced.
24644 the time in seconds of the current frame
24647 original position in the file of the frame, or undefined if undefined
24648 for the current frame
24651 The previous input PTS.
24654 previous input time in seconds
24657 The previous output PTS.
24660 previous output time in seconds
24663 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
24667 The wallclock (RTC) time at the start of the movie in microseconds.
24670 The timebase of the input timestamps.
24674 @subsection Examples
24678 Start counting PTS from zero
24680 setpts=PTS-STARTPTS
24684 Apply fast motion effect:
24690 Apply slow motion effect:
24696 Set fixed rate of 25 frames per second:
24702 Set fixed rate 25 fps with some jitter:
24704 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
24708 Apply an offset of 10 seconds to the input PTS:
24714 Generate timestamps from a "live source" and rebase onto the current timebase:
24716 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
24720 Generate timestamps by counting samples:
24729 Force color range for the output video frame.
24731 The @code{setrange} filter marks the color range property for the
24732 output frames. It does not change the input frame, but only sets the
24733 corresponding property, which affects how the frame is treated by
24736 The filter accepts the following options:
24741 Available values are:
24745 Keep the same color range property.
24747 @item unspecified, unknown
24748 Set the color range as unspecified.
24750 @item limited, tv, mpeg
24751 Set the color range as limited.
24753 @item full, pc, jpeg
24754 Set the color range as full.
24758 @section settb, asettb
24760 Set the timebase to use for the output frames timestamps.
24761 It is mainly useful for testing timebase configuration.
24763 It accepts the following parameters:
24768 The expression which is evaluated into the output timebase.
24772 The value for @option{tb} is an arithmetic expression representing a
24773 rational. The expression can contain the constants "AVTB" (the default
24774 timebase), "intb" (the input timebase) and "sr" (the sample rate,
24775 audio only). Default value is "intb".
24777 @subsection Examples
24781 Set the timebase to 1/25:
24787 Set the timebase to 1/10:
24793 Set the timebase to 1001/1000:
24799 Set the timebase to 2*intb:
24805 Set the default timebase value:
24812 Convert input audio to a video output representing frequency spectrum
24813 logarithmically using Brown-Puckette constant Q transform algorithm with
24814 direct frequency domain coefficient calculation (but the transform itself
24815 is not really constant Q, instead the Q factor is actually variable/clamped),
24816 with musical tone scale, from E0 to D#10.
24818 The filter accepts the following options:
24822 Specify the video size for the output. It must be even. For the syntax of this option,
24823 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24824 Default value is @code{1920x1080}.
24827 Set the output frame rate. Default value is @code{25}.
24830 Set the bargraph height. It must be even. Default value is @code{-1} which
24831 computes the bargraph height automatically.
24834 Set the axis height. It must be even. Default value is @code{-1} which computes
24835 the axis height automatically.
24838 Set the sonogram height. It must be even. Default value is @code{-1} which
24839 computes the sonogram height automatically.
24842 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
24843 instead. Default value is @code{1}.
24845 @item sono_v, volume
24846 Specify the sonogram volume expression. It can contain variables:
24849 the @var{bar_v} evaluated expression
24850 @item frequency, freq, f
24851 the frequency where it is evaluated
24852 @item timeclamp, tc
24853 the value of @var{timeclamp} option
24857 @item a_weighting(f)
24858 A-weighting of equal loudness
24859 @item b_weighting(f)
24860 B-weighting of equal loudness
24861 @item c_weighting(f)
24862 C-weighting of equal loudness.
24864 Default value is @code{16}.
24866 @item bar_v, volume2
24867 Specify the bargraph volume expression. It can contain variables:
24870 the @var{sono_v} evaluated expression
24871 @item frequency, freq, f
24872 the frequency where it is evaluated
24873 @item timeclamp, tc
24874 the value of @var{timeclamp} option
24878 @item a_weighting(f)
24879 A-weighting of equal loudness
24880 @item b_weighting(f)
24881 B-weighting of equal loudness
24882 @item c_weighting(f)
24883 C-weighting of equal loudness.
24885 Default value is @code{sono_v}.
24887 @item sono_g, gamma
24888 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
24889 higher gamma makes the spectrum having more range. Default value is @code{3}.
24890 Acceptable range is @code{[1, 7]}.
24892 @item bar_g, gamma2
24893 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
24897 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
24898 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
24900 @item timeclamp, tc
24901 Specify the transform timeclamp. At low frequency, there is trade-off between
24902 accuracy in time domain and frequency domain. If timeclamp is lower,
24903 event in time domain is represented more accurately (such as fast bass drum),
24904 otherwise event in frequency domain is represented more accurately
24905 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
24908 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
24909 limits future samples by applying asymmetric windowing in time domain, useful
24910 when low latency is required. Accepted range is @code{[0, 1]}.
24913 Specify the transform base frequency. Default value is @code{20.01523126408007475},
24914 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
24917 Specify the transform end frequency. Default value is @code{20495.59681441799654},
24918 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
24921 This option is deprecated and ignored.
24924 Specify the transform length in time domain. Use this option to control accuracy
24925 trade-off between time domain and frequency domain at every frequency sample.
24926 It can contain variables:
24928 @item frequency, freq, f
24929 the frequency where it is evaluated
24930 @item timeclamp, tc
24931 the value of @var{timeclamp} option.
24933 Default value is @code{384*tc/(384+tc*f)}.
24936 Specify the transform count for every video frame. Default value is @code{6}.
24937 Acceptable range is @code{[1, 30]}.
24940 Specify the transform count for every single pixel. Default value is @code{0},
24941 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
24944 Specify font file for use with freetype to draw the axis. If not specified,
24945 use embedded font. Note that drawing with font file or embedded font is not
24946 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
24950 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
24951 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
24955 Specify font color expression. This is arithmetic expression that should return
24956 integer value 0xRRGGBB. It can contain variables:
24958 @item frequency, freq, f
24959 the frequency where it is evaluated
24960 @item timeclamp, tc
24961 the value of @var{timeclamp} option
24966 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
24967 @item r(x), g(x), b(x)
24968 red, green, and blue value of intensity x.
24970 Default value is @code{st(0, (midi(f)-59.5)/12);
24971 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
24972 r(1-ld(1)) + b(ld(1))}.
24975 Specify image file to draw the axis. This option override @var{fontfile} and
24976 @var{fontcolor} option.
24979 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
24980 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
24981 Default value is @code{1}.
24984 Set colorspace. The accepted values are:
24987 Unspecified (default)
24996 BT.470BG or BT.601-6 625
24999 SMPTE-170M or BT.601-6 525
25005 BT.2020 with non-constant luminance
25010 Set spectrogram color scheme. This is list of floating point values with format
25011 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25012 The default is @code{1|0.5|0|0|0.5|1}.
25016 @subsection Examples
25020 Playing audio while showing the spectrum:
25022 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25026 Same as above, but with frame rate 30 fps:
25028 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25032 Playing at 1280x720:
25034 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25038 Disable sonogram display:
25044 A1 and its harmonics: A1, A2, (near)E3, A3:
25046 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),
25047 asplit[a][out1]; [a] showcqt [out0]'
25051 Same as above, but with more accuracy in frequency domain:
25053 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),
25054 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25060 bar_v=10:sono_v=bar_v*a_weighting(f)
25064 Custom gamma, now spectrum is linear to the amplitude.
25070 Custom tlength equation:
25072 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)))'
25076 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25078 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25082 Custom font using fontconfig:
25084 font='Courier New,Monospace,mono|bold'
25088 Custom frequency range with custom axis using image file:
25090 axisfile=myaxis.png:basefreq=40:endfreq=10000
25096 Convert input audio to video output representing the audio power spectrum.
25097 Audio amplitude is on Y-axis while frequency is on X-axis.
25099 The filter accepts the following options:
25103 Specify size of video. For the syntax of this option, check the
25104 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25105 Default is @code{1024x512}.
25109 This set how each frequency bin will be represented.
25111 It accepts the following values:
25117 Default is @code{bar}.
25120 Set amplitude scale.
25122 It accepts the following values:
25136 Default is @code{log}.
25139 Set frequency scale.
25141 It accepts the following values:
25150 Reverse logarithmic scale.
25152 Default is @code{lin}.
25155 Set window size. Allowed range is from 16 to 65536.
25157 Default is @code{2048}
25160 Set windowing function.
25162 It accepts the following values:
25185 Default is @code{hanning}.
25188 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25189 which means optimal overlap for selected window function will be picked.
25192 Set time averaging. Setting this to 0 will display current maximal peaks.
25193 Default is @code{1}, which means time averaging is disabled.
25196 Specify list of colors separated by space or by '|' which will be used to
25197 draw channel frequencies. Unrecognized or missing colors will be replaced
25201 Set channel display mode.
25203 It accepts the following values:
25208 Default is @code{combined}.
25211 Set minimum amplitude used in @code{log} amplitude scaler.
25214 Set data display mode.
25216 It accepts the following values:
25221 Default is @code{magnitude}.
25224 @section showspatial
25226 Convert stereo input audio to a video output, representing the spatial relationship
25227 between two channels.
25229 The filter accepts the following options:
25233 Specify the video size for the output. For the syntax of this option, check the
25234 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25235 Default value is @code{512x512}.
25238 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25241 Set window function.
25243 It accepts the following values:
25268 Default value is @code{hann}.
25271 Set ratio of overlap window. Default value is @code{0.5}.
25272 When value is @code{1} overlap is set to recommended size for specific
25273 window function currently used.
25276 @anchor{showspectrum}
25277 @section showspectrum
25279 Convert input audio to a video output, representing the audio frequency
25282 The filter accepts the following options:
25286 Specify the video size for the output. For the syntax of this option, check the
25287 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25288 Default value is @code{640x512}.
25291 Specify how the spectrum should slide along the window.
25293 It accepts the following values:
25296 the samples start again on the left when they reach the right
25298 the samples scroll from right to left
25300 frames are only produced when the samples reach the right
25302 the samples scroll from left to right
25305 Default value is @code{replace}.
25308 Specify display mode.
25310 It accepts the following values:
25313 all channels are displayed in the same row
25315 all channels are displayed in separate rows
25318 Default value is @samp{combined}.
25321 Specify display color mode.
25323 It accepts the following values:
25326 each channel is displayed in a separate color
25328 each channel is displayed using the same color scheme
25330 each channel is displayed using the rainbow color scheme
25332 each channel is displayed using the moreland color scheme
25334 each channel is displayed using the nebulae color scheme
25336 each channel is displayed using the fire color scheme
25338 each channel is displayed using the fiery color scheme
25340 each channel is displayed using the fruit color scheme
25342 each channel is displayed using the cool color scheme
25344 each channel is displayed using the magma color scheme
25346 each channel is displayed using the green color scheme
25348 each channel is displayed using the viridis color scheme
25350 each channel is displayed using the plasma color scheme
25352 each channel is displayed using the cividis color scheme
25354 each channel is displayed using the terrain color scheme
25357 Default value is @samp{channel}.
25360 Specify scale used for calculating intensity color values.
25362 It accepts the following values:
25367 square root, default
25378 Default value is @samp{sqrt}.
25381 Specify frequency scale.
25383 It accepts the following values:
25391 Default value is @samp{lin}.
25394 Set saturation modifier for displayed colors. Negative values provide
25395 alternative color scheme. @code{0} is no saturation at all.
25396 Saturation must be in [-10.0, 10.0] range.
25397 Default value is @code{1}.
25400 Set window function.
25402 It accepts the following values:
25427 Default value is @code{hann}.
25430 Set orientation of time vs frequency axis. Can be @code{vertical} or
25431 @code{horizontal}. Default is @code{vertical}.
25434 Set ratio of overlap window. Default value is @code{0}.
25435 When value is @code{1} overlap is set to recommended size for specific
25436 window function currently used.
25439 Set scale gain for calculating intensity color values.
25440 Default value is @code{1}.
25443 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25446 Set color rotation, must be in [-1.0, 1.0] range.
25447 Default value is @code{0}.
25450 Set start frequency from which to display spectrogram. Default is @code{0}.
25453 Set stop frequency to which to display spectrogram. Default is @code{0}.
25456 Set upper frame rate limit. Default is @code{auto}, unlimited.
25459 Draw time and frequency axes and legends. Default is disabled.
25462 The usage is very similar to the showwaves filter; see the examples in that
25465 @subsection Examples
25469 Large window with logarithmic color scaling:
25471 showspectrum=s=1280x480:scale=log
25475 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25477 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25478 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25482 @section showspectrumpic
25484 Convert input audio to a single video frame, representing the audio frequency
25487 The filter accepts the following options:
25491 Specify the video size for the output. For the syntax of this option, check the
25492 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25493 Default value is @code{4096x2048}.
25496 Specify display mode.
25498 It accepts the following values:
25501 all channels are displayed in the same row
25503 all channels are displayed in separate rows
25505 Default value is @samp{combined}.
25508 Specify display color mode.
25510 It accepts the following values:
25513 each channel is displayed in a separate color
25515 each channel is displayed using the same color scheme
25517 each channel is displayed using the rainbow color scheme
25519 each channel is displayed using the moreland color scheme
25521 each channel is displayed using the nebulae color scheme
25523 each channel is displayed using the fire color scheme
25525 each channel is displayed using the fiery color scheme
25527 each channel is displayed using the fruit color scheme
25529 each channel is displayed using the cool color scheme
25531 each channel is displayed using the magma color scheme
25533 each channel is displayed using the green color scheme
25535 each channel is displayed using the viridis color scheme
25537 each channel is displayed using the plasma color scheme
25539 each channel is displayed using the cividis color scheme
25541 each channel is displayed using the terrain color scheme
25543 Default value is @samp{intensity}.
25546 Specify scale used for calculating intensity color values.
25548 It accepts the following values:
25553 square root, default
25563 Default value is @samp{log}.
25566 Specify frequency scale.
25568 It accepts the following values:
25576 Default value is @samp{lin}.
25579 Set saturation modifier for displayed colors. Negative values provide
25580 alternative color scheme. @code{0} is no saturation at all.
25581 Saturation must be in [-10.0, 10.0] range.
25582 Default value is @code{1}.
25585 Set window function.
25587 It accepts the following values:
25611 Default value is @code{hann}.
25614 Set orientation of time vs frequency axis. Can be @code{vertical} or
25615 @code{horizontal}. Default is @code{vertical}.
25618 Set scale gain for calculating intensity color values.
25619 Default value is @code{1}.
25622 Draw time and frequency axes and legends. Default is enabled.
25625 Set color rotation, must be in [-1.0, 1.0] range.
25626 Default value is @code{0}.
25629 Set start frequency from which to display spectrogram. Default is @code{0}.
25632 Set stop frequency to which to display spectrogram. Default is @code{0}.
25635 @subsection Examples
25639 Extract an audio spectrogram of a whole audio track
25640 in a 1024x1024 picture using @command{ffmpeg}:
25642 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25646 @section showvolume
25648 Convert input audio volume to a video output.
25650 The filter accepts the following options:
25657 Set border width, allowed range is [0, 5]. Default is 1.
25660 Set channel width, allowed range is [80, 8192]. Default is 400.
25663 Set channel height, allowed range is [1, 900]. Default is 20.
25666 Set fade, allowed range is [0, 1]. Default is 0.95.
25669 Set volume color expression.
25671 The expression can use the following variables:
25675 Current max volume of channel in dB.
25681 Current channel number, starting from 0.
25685 If set, displays channel names. Default is enabled.
25688 If set, displays volume values. Default is enabled.
25691 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
25692 default is @code{h}.
25695 Set step size, allowed range is [0, 5]. Default is 0, which means
25699 Set background opacity, allowed range is [0, 1]. Default is 0.
25702 Set metering mode, can be peak: @code{p} or rms: @code{r},
25703 default is @code{p}.
25706 Set display scale, can be linear: @code{lin} or log: @code{log},
25707 default is @code{lin}.
25711 If set to > 0., display a line for the max level
25712 in the previous seconds.
25713 default is disabled: @code{0.}
25716 The color of the max line. Use when @code{dm} option is set to > 0.
25717 default is: @code{orange}
25722 Convert input audio to a video output, representing the samples waves.
25724 The filter accepts the following options:
25728 Specify the video size for the output. For the syntax of this option, check the
25729 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25730 Default value is @code{600x240}.
25735 Available values are:
25738 Draw a point for each sample.
25741 Draw a vertical line for each sample.
25744 Draw a point for each sample and a line between them.
25747 Draw a centered vertical line for each sample.
25750 Default value is @code{point}.
25753 Set the number of samples which are printed on the same column. A
25754 larger value will decrease the frame rate. Must be a positive
25755 integer. This option can be set only if the value for @var{rate}
25756 is not explicitly specified.
25759 Set the (approximate) output frame rate. This is done by setting the
25760 option @var{n}. Default value is "25".
25762 @item split_channels
25763 Set if channels should be drawn separately or overlap. Default value is 0.
25766 Set colors separated by '|' which are going to be used for drawing of each channel.
25769 Set amplitude scale.
25771 Available values are:
25789 Set the draw mode. This is mostly useful to set for high @var{n}.
25791 Available values are:
25794 Scale pixel values for each drawn sample.
25797 Draw every sample directly.
25800 Default value is @code{scale}.
25803 @subsection Examples
25807 Output the input file audio and the corresponding video representation
25810 amovie=a.mp3,asplit[out0],showwaves[out1]
25814 Create a synthetic signal and show it with showwaves, forcing a
25815 frame rate of 30 frames per second:
25817 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
25821 @section showwavespic
25823 Convert input audio to a single video frame, representing the samples waves.
25825 The filter accepts the following options:
25829 Specify the video size for the output. For the syntax of this option, check the
25830 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25831 Default value is @code{600x240}.
25833 @item split_channels
25834 Set if channels should be drawn separately or overlap. Default value is 0.
25837 Set colors separated by '|' which are going to be used for drawing of each channel.
25840 Set amplitude scale.
25842 Available values are:
25862 Available values are:
25865 Scale pixel values for each drawn sample.
25868 Draw every sample directly.
25871 Default value is @code{scale}.
25874 Set the filter mode.
25876 Available values are:
25879 Use average samples values for each drawn sample.
25882 Use peak samples values for each drawn sample.
25885 Default value is @code{average}.
25888 @subsection Examples
25892 Extract a channel split representation of the wave form of a whole audio track
25893 in a 1024x800 picture using @command{ffmpeg}:
25895 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25899 @section sidedata, asidedata
25901 Delete frame side data, or select frames based on it.
25903 This filter accepts the following options:
25907 Set mode of operation of the filter.
25909 Can be one of the following:
25913 Select every frame with side data of @code{type}.
25916 Delete side data of @code{type}. If @code{type} is not set, delete all side
25922 Set side data type used with all modes. Must be set for @code{select} mode. For
25923 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
25924 in @file{libavutil/frame.h}. For example, to choose
25925 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
25929 @section spectrumsynth
25931 Synthesize audio from 2 input video spectrums, first input stream represents
25932 magnitude across time and second represents phase across time.
25933 The filter will transform from frequency domain as displayed in videos back
25934 to time domain as presented in audio output.
25936 This filter is primarily created for reversing processed @ref{showspectrum}
25937 filter outputs, but can synthesize sound from other spectrograms too.
25938 But in such case results are going to be poor if the phase data is not
25939 available, because in such cases phase data need to be recreated, usually
25940 it's just recreated from random noise.
25941 For best results use gray only output (@code{channel} color mode in
25942 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
25943 @code{lin} scale for phase video. To produce phase, for 2nd video, use
25944 @code{data} option. Inputs videos should generally use @code{fullframe}
25945 slide mode as that saves resources needed for decoding video.
25947 The filter accepts the following options:
25951 Specify sample rate of output audio, the sample rate of audio from which
25952 spectrum was generated may differ.
25955 Set number of channels represented in input video spectrums.
25958 Set scale which was used when generating magnitude input spectrum.
25959 Can be @code{lin} or @code{log}. Default is @code{log}.
25962 Set slide which was used when generating inputs spectrums.
25963 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
25964 Default is @code{fullframe}.
25967 Set window function used for resynthesis.
25970 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25971 which means optimal overlap for selected window function will be picked.
25974 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
25975 Default is @code{vertical}.
25978 @subsection Examples
25982 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
25983 then resynthesize videos back to audio with spectrumsynth:
25985 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
25986 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
25987 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
25991 @section split, asplit
25993 Split input into several identical outputs.
25995 @code{asplit} works with audio input, @code{split} with video.
25997 The filter accepts a single parameter which specifies the number of outputs. If
25998 unspecified, it defaults to 2.
26000 @subsection Examples
26004 Create two separate outputs from the same input:
26006 [in] split [out0][out1]
26010 To create 3 or more outputs, you need to specify the number of
26013 [in] asplit=3 [out0][out1][out2]
26017 Create two separate outputs from the same input, one cropped and
26020 [in] split [splitout1][splitout2];
26021 [splitout1] crop=100:100:0:0 [cropout];
26022 [splitout2] pad=200:200:100:100 [padout];
26026 Create 5 copies of the input audio with @command{ffmpeg}:
26028 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26034 Receive commands sent through a libzmq client, and forward them to
26035 filters in the filtergraph.
26037 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26038 must be inserted between two video filters, @code{azmq} between two
26039 audio filters. Both are capable to send messages to any filter type.
26041 To enable these filters you need to install the libzmq library and
26042 headers and configure FFmpeg with @code{--enable-libzmq}.
26044 For more information about libzmq see:
26045 @url{http://www.zeromq.org/}
26047 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26048 receives messages sent through a network interface defined by the
26049 @option{bind_address} (or the abbreviation "@option{b}") option.
26050 Default value of this option is @file{tcp://localhost:5555}. You may
26051 want to alter this value to your needs, but do not forget to escape any
26052 ':' signs (see @ref{filtergraph escaping}).
26054 The received message must be in the form:
26056 @var{TARGET} @var{COMMAND} [@var{ARG}]
26059 @var{TARGET} specifies the target of the command, usually the name of
26060 the filter class or a specific filter instance name. The default
26061 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26062 but you can override this by using the @samp{filter_name@@id} syntax
26063 (see @ref{Filtergraph syntax}).
26065 @var{COMMAND} specifies the name of the command for the target filter.
26067 @var{ARG} is optional and specifies the optional argument list for the
26068 given @var{COMMAND}.
26070 Upon reception, the message is processed and the corresponding command
26071 is injected into the filtergraph. Depending on the result, the filter
26072 will send a reply to the client, adopting the format:
26074 @var{ERROR_CODE} @var{ERROR_REASON}
26078 @var{MESSAGE} is optional.
26080 @subsection Examples
26082 Look at @file{tools/zmqsend} for an example of a zmq client which can
26083 be used to send commands processed by these filters.
26085 Consider the following filtergraph generated by @command{ffplay}.
26086 In this example the last overlay filter has an instance name. All other
26087 filters will have default instance names.
26090 ffplay -dumpgraph 1 -f lavfi "
26091 color=s=100x100:c=red [l];
26092 color=s=100x100:c=blue [r];
26093 nullsrc=s=200x100, zmq [bg];
26094 [bg][l] overlay [bg+l];
26095 [bg+l][r] overlay@@my=x=100 "
26098 To change the color of the left side of the video, the following
26099 command can be used:
26101 echo Parsed_color_0 c yellow | tools/zmqsend
26104 To change the right side:
26106 echo Parsed_color_1 c pink | tools/zmqsend
26109 To change the position of the right side:
26111 echo overlay@@my x 150 | tools/zmqsend
26115 @c man end MULTIMEDIA FILTERS
26117 @chapter Multimedia Sources
26118 @c man begin MULTIMEDIA SOURCES
26120 Below is a description of the currently available multimedia sources.
26124 This is the same as @ref{movie} source, except it selects an audio
26130 Read audio and/or video stream(s) from a movie container.
26132 It accepts the following parameters:
26136 The name of the resource to read (not necessarily a file; it can also be a
26137 device or a stream accessed through some protocol).
26139 @item format_name, f
26140 Specifies the format assumed for the movie to read, and can be either
26141 the name of a container or an input device. If not specified, the
26142 format is guessed from @var{movie_name} or by probing.
26144 @item seek_point, sp
26145 Specifies the seek point in seconds. The frames will be output
26146 starting from this seek point. The parameter is evaluated with
26147 @code{av_strtod}, so the numerical value may be suffixed by an IS
26148 postfix. The default value is "0".
26151 Specifies the streams to read. Several streams can be specified,
26152 separated by "+". The source will then have as many outputs, in the
26153 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26154 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26155 respectively the default (best suited) video and audio stream. Default
26156 is "dv", or "da" if the filter is called as "amovie".
26158 @item stream_index, si
26159 Specifies the index of the video stream to read. If the value is -1,
26160 the most suitable video stream will be automatically selected. The default
26161 value is "-1". Deprecated. If the filter is called "amovie", it will select
26162 audio instead of video.
26165 Specifies how many times to read the stream in sequence.
26166 If the value is 0, the stream will be looped infinitely.
26167 Default value is "1".
26169 Note that when the movie is looped the source timestamps are not
26170 changed, so it will generate non monotonically increasing timestamps.
26172 @item discontinuity
26173 Specifies the time difference between frames above which the point is
26174 considered a timestamp discontinuity which is removed by adjusting the later
26178 It allows overlaying a second video on top of the main input of
26179 a filtergraph, as shown in this graph:
26181 input -----------> deltapts0 --> overlay --> output
26184 movie --> scale--> deltapts1 -------+
26186 @subsection Examples
26190 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26191 on top of the input labelled "in":
26193 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26194 [in] setpts=PTS-STARTPTS [main];
26195 [main][over] overlay=16:16 [out]
26199 Read from a video4linux2 device, and overlay it on top of the input
26202 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26203 [in] setpts=PTS-STARTPTS [main];
26204 [main][over] overlay=16:16 [out]
26208 Read the first video stream and the audio stream with id 0x81 from
26209 dvd.vob; the video is connected to the pad named "video" and the audio is
26210 connected to the pad named "audio":
26212 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26216 @subsection Commands
26218 Both movie and amovie support the following commands:
26221 Perform seek using "av_seek_frame".
26222 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26225 @var{stream_index}: If stream_index is -1, a default
26226 stream is selected, and @var{timestamp} is automatically converted
26227 from AV_TIME_BASE units to the stream specific time_base.
26229 @var{timestamp}: Timestamp in AVStream.time_base units
26230 or, if no stream is specified, in AV_TIME_BASE units.
26232 @var{flags}: Flags which select direction and seeking mode.
26236 Get movie duration in AV_TIME_BASE units.
26240 @c man end MULTIMEDIA SOURCES