1 @chapter Filtering Introduction
2 @c man begin FILTERING INTRODUCTION
4 Filtering in FFmpeg is enabled through the libavfilter library.
6 In libavfilter, a filter can have multiple inputs and multiple
8 To illustrate the sorts of things that are possible, we consider the
13 input --> split ---------------------> overlay --> output
16 +-----> crop --> vflip -------+
19 This filtergraph splits the input stream in two streams, then sends one
20 stream through the crop filter and the vflip filter, before merging it
21 back with the other stream by overlaying it on top. You can use the
22 following command to achieve this:
25 ffmpeg -i INPUT -vf "split [main][tmp]; [tmp] crop=iw:ih/2:0:0, vflip [flip]; [main][flip] overlay=0:H/2" OUTPUT
28 The result will be that the top half of the video is mirrored
29 onto the bottom half of the output video.
31 Filters in the same linear chain are separated by commas, and distinct
32 linear chains of filters are separated by semicolons. In our example,
33 @var{crop,vflip} are in one linear chain, @var{split} and
34 @var{overlay} are separately in another. The points where the linear
35 chains join are labelled by names enclosed in square brackets. In the
36 example, the split filter generates two outputs that are associated to
37 the labels @var{[main]} and @var{[tmp]}.
39 The stream sent to the second output of @var{split}, labelled as
40 @var{[tmp]}, is processed through the @var{crop} filter, which crops
41 away the lower half part of the video, and then vertically flipped. The
42 @var{overlay} filter takes in input the first unchanged output of the
43 split filter (which was labelled as @var{[main]}), and overlay on its
44 lower half the output generated by the @var{crop,vflip} filterchain.
46 Some filters take in input a list of parameters: they are specified
47 after the filter name and an equal sign, and are separated from each other
50 There exist so-called @var{source filters} that do not have an
51 audio/video input, and @var{sink filters} that will not have audio/video
54 @c man end FILTERING INTRODUCTION
57 @c man begin GRAPH2DOT
59 The @file{graph2dot} program included in the FFmpeg @file{tools}
60 directory can be used to parse a filtergraph description and issue a
61 corresponding textual representation in the dot language.
68 to see how to use @file{graph2dot}.
70 You can then pass the dot description to the @file{dot} program (from
71 the graphviz suite of programs) and obtain a graphical representation
74 For example the sequence of commands:
76 echo @var{GRAPH_DESCRIPTION} | \
77 tools/graph2dot -o graph.tmp && \
78 dot -Tpng graph.tmp -o graph.png && \
82 can be used to create and display an image representing the graph
83 described by the @var{GRAPH_DESCRIPTION} string. Note that this string must be
84 a complete self-contained graph, with its inputs and outputs explicitly defined.
85 For example if your command line is of the form:
87 ffmpeg -i infile -vf scale=640:360 outfile
89 your @var{GRAPH_DESCRIPTION} string will need to be of the form:
91 nullsrc,scale=640:360,nullsink
93 you may also need to set the @var{nullsrc} parameters and add a @var{format}
94 filter in order to simulate a specific input file.
98 @chapter Filtergraph description
99 @c man begin FILTERGRAPH DESCRIPTION
101 A filtergraph is a directed graph of connected filters. It can contain
102 cycles, and there can be multiple links between a pair of
103 filters. Each link has one input pad on one side connecting it to one
104 filter from which it takes its input, and one output pad on the other
105 side connecting it to one filter accepting its output.
107 Each filter in a filtergraph is an instance of a filter class
108 registered in the application, which defines the features and the
109 number of input and output pads of the filter.
111 A filter with no input pads is called a "source", and a filter with no
112 output pads is called a "sink".
114 @anchor{Filtergraph syntax}
115 @section Filtergraph syntax
117 A filtergraph has a textual representation, which is recognized by the
118 @option{-filter}/@option{-vf}/@option{-af} and
119 @option{-filter_complex} options in @command{ffmpeg} and
120 @option{-vf}/@option{-af} in @command{ffplay}, and by the
121 @code{avfilter_graph_parse_ptr()} function defined in
122 @file{libavfilter/avfilter.h}.
124 A filterchain consists of a sequence of connected filters, each one
125 connected to the previous one in the sequence. A filterchain is
126 represented by a list of ","-separated filter descriptions.
128 A filtergraph consists of a sequence of filterchains. A sequence of
129 filterchains is represented by a list of ";"-separated filterchain
132 A filter is represented by a string of the form:
133 [@var{in_link_1}]...[@var{in_link_N}]@var{filter_name}@@@var{id}=@var{arguments}[@var{out_link_1}]...[@var{out_link_M}]
135 @var{filter_name} is the name of the filter class of which the
136 described filter is an instance of, and has to be the name of one of
137 the filter classes registered in the program optionally followed by "@@@var{id}".
138 The name of the filter class is optionally followed by a string
141 @var{arguments} is a string which contains the parameters used to
142 initialize the filter instance. It may have one of two forms:
146 A ':'-separated list of @var{key=value} pairs.
149 A ':'-separated list of @var{value}. In this case, the keys are assumed to be
150 the option names in the order they are declared. E.g. the @code{fade} filter
151 declares three options in this order -- @option{type}, @option{start_frame} and
152 @option{nb_frames}. Then the parameter list @var{in:0:30} means that the value
153 @var{in} is assigned to the option @option{type}, @var{0} to
154 @option{start_frame} and @var{30} to @option{nb_frames}.
157 A ':'-separated list of mixed direct @var{value} and long @var{key=value}
158 pairs. The direct @var{value} must precede the @var{key=value} pairs, and
159 follow the same constraints order of the previous point. The following
160 @var{key=value} pairs can be set in any preferred order.
164 If the option value itself is a list of items (e.g. the @code{format} filter
165 takes a list of pixel formats), the items in the list are usually separated by
168 The list of arguments can be quoted using the character @samp{'} as initial
169 and ending mark, and the character @samp{\} for escaping the characters
170 within the quoted text; otherwise the argument string is considered
171 terminated when the next special character (belonging to the set
172 @samp{[]=;,}) is encountered.
174 The name and arguments of the filter are optionally preceded and
175 followed by a list of link labels.
176 A link label allows one to name a link and associate it to a filter output
177 or input pad. The preceding labels @var{in_link_1}
178 ... @var{in_link_N}, are associated to the filter input pads,
179 the following labels @var{out_link_1} ... @var{out_link_M}, are
180 associated to the output pads.
182 When two link labels with the same name are found in the
183 filtergraph, a link between the corresponding input and output pad is
186 If an output pad is not labelled, it is linked by default to the first
187 unlabelled input pad of the next filter in the filterchain.
188 For example in the filterchain
190 nullsrc, split[L1], [L2]overlay, nullsink
192 the split filter instance has two output pads, and the overlay filter
193 instance two input pads. The first output pad of split is labelled
194 "L1", the first input pad of overlay is labelled "L2", and the second
195 output pad of split is linked to the second input pad of overlay,
196 which are both unlabelled.
198 In a filter description, if the input label of the first filter is not
199 specified, "in" is assumed; if the output label of the last filter is not
200 specified, "out" is assumed.
202 In a complete filterchain all the unlabelled filter input and output
203 pads must be connected. A filtergraph is considered valid if all the
204 filter input and output pads of all the filterchains are connected.
206 Libavfilter will automatically insert @ref{scale} filters where format
207 conversion is required. It is possible to specify swscale flags
208 for those automatically inserted scalers by prepending
209 @code{sws_flags=@var{flags};}
210 to the filtergraph description.
212 Here is a BNF description of the filtergraph syntax:
214 @var{NAME} ::= sequence of alphanumeric characters and '_'
215 @var{FILTER_NAME} ::= @var{NAME}["@@"@var{NAME}]
216 @var{LINKLABEL} ::= "[" @var{NAME} "]"
217 @var{LINKLABELS} ::= @var{LINKLABEL} [@var{LINKLABELS}]
218 @var{FILTER_ARGUMENTS} ::= sequence of chars (possibly quoted)
219 @var{FILTER} ::= [@var{LINKLABELS}] @var{FILTER_NAME} ["=" @var{FILTER_ARGUMENTS}] [@var{LINKLABELS}]
220 @var{FILTERCHAIN} ::= @var{FILTER} [,@var{FILTERCHAIN}]
221 @var{FILTERGRAPH} ::= [sws_flags=@var{flags};] @var{FILTERCHAIN} [;@var{FILTERGRAPH}]
224 @anchor{filtergraph escaping}
225 @section Notes on filtergraph escaping
227 Filtergraph description composition entails several levels of
228 escaping. See @ref{quoting_and_escaping,,the "Quoting and escaping"
229 section in the ffmpeg-utils(1) manual,ffmpeg-utils} for more
230 information about the employed escaping procedure.
232 A first level escaping affects the content of each filter option
233 value, which may contain the special character @code{:} used to
234 separate values, or one of the escaping characters @code{\'}.
236 A second level escaping affects the whole filter description, which
237 may contain the escaping characters @code{\'} or the special
238 characters @code{[],;} used by the filtergraph description.
240 Finally, when you specify a filtergraph on a shell commandline, you
241 need to perform a third level escaping for the shell special
242 characters contained within it.
244 For example, consider the following string to be embedded in
245 the @ref{drawtext} filter description @option{text} value:
247 this is a 'string': may contain one, or more, special characters
250 This string contains the @code{'} special escaping character, and the
251 @code{:} special character, so it needs to be escaped in this way:
253 text=this is a \'string\'\: may contain one, or more, special characters
256 A second level of escaping is required when embedding the filter
257 description in a filtergraph description, in order to escape all the
258 filtergraph special characters. Thus the example above becomes:
260 drawtext=text=this is a \\\'string\\\'\\: may contain one\, or more\, special characters
262 (note that in addition to the @code{\'} escaping special characters,
263 also @code{,} needs to be escaped).
265 Finally an additional level of escaping is needed when writing the
266 filtergraph description in a shell command, which depends on the
267 escaping rules of the adopted shell. For example, assuming that
268 @code{\} is special and needs to be escaped with another @code{\}, the
269 previous string will finally result in:
271 -vf "drawtext=text=this is a \\\\\\'string\\\\\\'\\\\: may contain one\\, or more\\, special characters"
274 @chapter Timeline editing
276 Some filters support a generic @option{enable} option. For the filters
277 supporting timeline editing, this option can be set to an expression which is
278 evaluated before sending a frame to the filter. If the evaluation is non-zero,
279 the filter will be enabled, otherwise the frame will be sent unchanged to the
280 next filter in the filtergraph.
282 The expression accepts the following values:
285 timestamp expressed in seconds, NAN if the input timestamp is unknown
288 sequential number of the input frame, starting from 0
291 the position in the file of the input frame, NAN if unknown
295 width and height of the input frame if video
298 Additionally, these filters support an @option{enable} command that can be used
299 to re-define the expression.
301 Like any other filtering option, the @option{enable} option follows the same
304 For example, to enable a blur filter (@ref{smartblur}) from 10 seconds to 3
305 minutes, and a @ref{curves} filter starting at 3 seconds:
307 smartblur = enable='between(t,10,3*60)',
308 curves = enable='gte(t,3)' : preset=cross_process
311 See @code{ffmpeg -filters} to view which filters have timeline support.
313 @c man end FILTERGRAPH DESCRIPTION
316 @chapter Changing options at runtime with a command
318 Some options can be changed during the operation of the filter using
319 a command. These options are marked 'T' on the output of
320 @command{ffmpeg} @option{-h filter=<name of filter>}.
321 The name of the command is the name of the option and the argument is
325 @chapter Options for filters with several inputs (framesync)
326 @c man begin OPTIONS FOR FILTERS WITH SEVERAL INPUTS
328 Some filters with several inputs support a common set of options.
329 These options can only be set by name, not with the short notation.
333 The action to take when EOF is encountered on the secondary input; it accepts
334 one of the following values:
338 Repeat the last frame (the default).
342 Pass the main input through.
346 If set to 1, force the output to terminate when the shortest input
347 terminates. Default value is 0.
350 If set to 1, force the filter to extend the last frame of secondary streams
351 until the end of the primary stream. A value of 0 disables this behavior.
355 @c man end OPTIONS FOR FILTERS WITH SEVERAL INPUTS
357 @chapter Audio Filters
358 @c man begin AUDIO FILTERS
360 When you configure your FFmpeg build, you can disable any of the
361 existing filters using @code{--disable-filters}.
362 The configure output will show the audio filters included in your
365 Below is a description of the currently available audio filters.
369 A compressor is mainly used to reduce the dynamic range of a signal.
370 Especially modern music is mostly compressed at a high ratio to
371 improve the overall loudness. It's done to get the highest attention
372 of a listener, "fatten" the sound and bring more "power" to the track.
373 If a signal is compressed too much it may sound dull or "dead"
374 afterwards or it may start to "pump" (which could be a powerful effect
375 but can also destroy a track completely).
376 The right compression is the key to reach a professional sound and is
377 the high art of mixing and mastering. Because of its complex settings
378 it may take a long time to get the right feeling for this kind of effect.
380 Compression is done by detecting the volume above a chosen level
381 @code{threshold} and dividing it by the factor set with @code{ratio}.
382 So if you set the threshold to -12dB and your signal reaches -6dB a ratio
383 of 2:1 will result in a signal at -9dB. Because an exact manipulation of
384 the signal would cause distortion of the waveform the reduction can be
385 levelled over the time. This is done by setting "Attack" and "Release".
386 @code{attack} determines how long the signal has to rise above the threshold
387 before any reduction will occur and @code{release} sets the time the signal
388 has to fall below the threshold to reduce the reduction again. Shorter signals
389 than the chosen attack time will be left untouched.
390 The overall reduction of the signal can be made up afterwards with the
391 @code{makeup} setting. So compressing the peaks of a signal about 6dB and
392 raising the makeup to this level results in a signal twice as loud than the
393 source. To gain a softer entry in the compression the @code{knee} flattens the
394 hard edge at the threshold in the range of the chosen decibels.
396 The filter accepts the following options:
400 Set input gain. Default is 1. Range is between 0.015625 and 64.
403 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
404 Default is @code{downward}.
407 If a signal of stream rises above this level it will affect the gain
409 By default it is 0.125. Range is between 0.00097563 and 1.
412 Set a ratio by which the signal is reduced. 1:2 means that if the level
413 rose 4dB above the threshold, it will be only 2dB above after the reduction.
414 Default is 2. Range is between 1 and 20.
417 Amount of milliseconds the signal has to rise above the threshold before gain
418 reduction starts. Default is 20. Range is between 0.01 and 2000.
421 Amount of milliseconds the signal has to fall below the threshold before
422 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
425 Set the amount by how much signal will be amplified after processing.
426 Default is 1. Range is from 1 to 64.
429 Curve the sharp knee around the threshold to enter gain reduction more softly.
430 Default is 2.82843. Range is between 1 and 8.
433 Choose if the @code{average} level between all channels of input stream
434 or the louder(@code{maximum}) channel of input stream affects the
435 reduction. Default is @code{average}.
438 Should the exact signal be taken in case of @code{peak} or an RMS one in case
439 of @code{rms}. Default is @code{rms} which is mostly smoother.
442 How much to use compressed signal in output. Default is 1.
443 Range is between 0 and 1.
448 This filter supports the all above options as @ref{commands}.
451 Simple audio dynamic range compression/expansion filter.
453 The filter accepts the following options:
457 Set contrast. Default is 33. Allowed range is between 0 and 100.
462 Copy the input audio source unchanged to the output. This is mainly useful for
467 Apply cross fade from one input audio stream to another input audio stream.
468 The cross fade is applied for specified duration near the end of first stream.
470 The filter accepts the following options:
474 Specify the number of samples for which the cross fade effect has to last.
475 At the end of the cross fade effect the first input audio will be completely
476 silent. Default is 44100.
479 Specify the duration of the cross fade effect. See
480 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
481 for the accepted syntax.
482 By default the duration is determined by @var{nb_samples}.
483 If set this option is used instead of @var{nb_samples}.
486 Should first stream end overlap with second stream start. Default is enabled.
489 Set curve for cross fade transition for first stream.
492 Set curve for cross fade transition for second stream.
494 For description of available curve types see @ref{afade} filter description.
501 Cross fade from one input to another:
503 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
507 Cross fade from one input to another but without overlapping:
509 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
514 Split audio stream into several bands.
516 This filter splits audio stream into two or more frequency ranges.
517 Summing all streams back will give flat output.
519 The filter accepts the following options:
523 Set split frequencies. Those must be positive and increasing.
526 Set filter order for each band split. This controls filter roll-off or steepness
527 of filter transfer function.
528 Available values are:
553 Default is @var{4th}.
556 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
559 Set output gain for each band. Default value is 1 for all bands.
566 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
567 each band will be in separate stream:
569 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
573 Same as above, but with higher filter order:
575 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
579 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
581 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=8th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
587 Reduce audio bit resolution.
589 This filter is bit crusher with enhanced functionality. A bit crusher
590 is used to audibly reduce number of bits an audio signal is sampled
591 with. This doesn't change the bit depth at all, it just produces the
592 effect. Material reduced in bit depth sounds more harsh and "digital".
593 This filter is able to even round to continuous values instead of discrete
595 Additionally it has a D/C offset which results in different crushing of
596 the lower and the upper half of the signal.
597 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
599 Another feature of this filter is the logarithmic mode.
600 This setting switches from linear distances between bits to logarithmic ones.
601 The result is a much more "natural" sounding crusher which doesn't gate low
602 signals for example. The human ear has a logarithmic perception,
603 so this kind of crushing is much more pleasant.
604 Logarithmic crushing is also able to get anti-aliased.
606 The filter accepts the following options:
622 Can be linear: @code{lin} or logarithmic: @code{log}.
631 Set sample reduction.
634 Enable LFO. By default disabled.
645 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
649 Remove impulsive noise from input audio.
651 Samples detected as impulsive noise are replaced by interpolated samples using
652 autoregressive modelling.
656 Set window size, in milliseconds. Allowed range is from @code{10} to
657 @code{100}. Default value is @code{55} milliseconds.
658 This sets size of window which will be processed at once.
661 Set window overlap, in percentage of window size. Allowed range is from
662 @code{50} to @code{95}. Default value is @code{75} percent.
663 Setting this to a very high value increases impulsive noise removal but makes
664 whole process much slower.
667 Set autoregression order, in percentage of window size. Allowed range is from
668 @code{0} to @code{25}. Default value is @code{2} percent. This option also
669 controls quality of interpolated samples using neighbour good samples.
672 Set threshold value. Allowed range is from @code{1} to @code{100}.
673 Default value is @code{2}.
674 This controls the strength of impulsive noise which is going to be removed.
675 The lower value, the more samples will be detected as impulsive noise.
678 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
679 @code{10}. Default value is @code{2}.
680 If any two samples detected as noise are spaced less than this value then any
681 sample between those two samples will be also detected as noise.
686 It accepts the following values:
689 Select overlap-add method. Even not interpolated samples are slightly
690 changed with this method.
693 Select overlap-save method. Not interpolated samples remain unchanged.
696 Default value is @code{a}.
700 Remove clipped samples from input audio.
702 Samples detected as clipped are replaced by interpolated samples using
703 autoregressive modelling.
707 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
708 Default value is @code{55} milliseconds.
709 This sets size of window which will be processed at once.
712 Set window overlap, in percentage of window size. Allowed range is from @code{50}
713 to @code{95}. Default value is @code{75} percent.
716 Set autoregression order, in percentage of window size. Allowed range is from
717 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
718 quality of interpolated samples using neighbour good samples.
721 Set threshold value. Allowed range is from @code{1} to @code{100}.
722 Default value is @code{10}. Higher values make clip detection less aggressive.
725 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
726 Default value is @code{1000}. Higher values make clip detection less aggressive.
731 It accepts the following values:
734 Select overlap-add method. Even not interpolated samples are slightly changed
738 Select overlap-save method. Not interpolated samples remain unchanged.
741 Default value is @code{a}.
746 Delay one or more audio channels.
748 Samples in delayed channel are filled with silence.
750 The filter accepts the following option:
754 Set list of delays in milliseconds for each channel separated by '|'.
755 Unused delays will be silently ignored. If number of given delays is
756 smaller than number of channels all remaining channels will not be delayed.
757 If you want to delay exact number of samples, append 'S' to number.
758 If you want instead to delay in seconds, append 's' to number.
761 Use last set delay for all remaining channels. By default is disabled.
762 This option if enabled changes how option @code{delays} is interpreted.
769 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
770 the second channel (and any other channels that may be present) unchanged.
776 Delay second channel by 500 samples, the third channel by 700 samples and leave
777 the first channel (and any other channels that may be present) unchanged.
783 Delay all channels by same number of samples:
785 adelay=delays=64S:all=1
790 Remedy denormals in audio by adding extremely low-level noise.
792 This filter shall be placed before any filter that can produce denormals.
794 A description of the accepted parameters follows.
798 Set level of added noise in dB. Default is @code{-351}.
799 Allowed range is from -451 to -90.
802 Set type of added noise.
815 Default is @code{dc}.
820 This filter supports the all above options as @ref{commands}.
822 @section aderivative, aintegral
824 Compute derivative/integral of audio stream.
826 Applying both filters one after another produces original audio.
830 Apply echoing to the input audio.
832 Echoes are reflected sound and can occur naturally amongst mountains
833 (and sometimes large buildings) when talking or shouting; digital echo
834 effects emulate this behaviour and are often used to help fill out the
835 sound of a single instrument or vocal. The time difference between the
836 original signal and the reflection is the @code{delay}, and the
837 loudness of the reflected signal is the @code{decay}.
838 Multiple echoes can have different delays and decays.
840 A description of the accepted parameters follows.
844 Set input gain of reflected signal. Default is @code{0.6}.
847 Set output gain of reflected signal. Default is @code{0.3}.
850 Set list of time intervals in milliseconds between original signal and reflections
851 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
852 Default is @code{1000}.
855 Set list of loudness of reflected signals separated by '|'.
856 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
857 Default is @code{0.5}.
864 Make it sound as if there are twice as many instruments as are actually playing:
866 aecho=0.8:0.88:60:0.4
870 If delay is very short, then it sounds like a (metallic) robot playing music:
876 A longer delay will sound like an open air concert in the mountains:
878 aecho=0.8:0.9:1000:0.3
882 Same as above but with one more mountain:
884 aecho=0.8:0.9:1000|1800:0.3|0.25
889 Audio emphasis filter creates or restores material directly taken from LPs or
890 emphased CDs with different filter curves. E.g. to store music on vinyl the
891 signal has to be altered by a filter first to even out the disadvantages of
892 this recording medium.
893 Once the material is played back the inverse filter has to be applied to
894 restore the distortion of the frequency response.
896 The filter accepts the following options:
906 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
907 use @code{production} mode. Default is @code{reproduction} mode.
910 Set filter type. Selects medium. Can be one of the following:
922 select Compact Disc (CD).
928 select 50µs (FM-KF).
930 select 75µs (FM-KF).
936 This filter supports the all above options as @ref{commands}.
940 Modify an audio signal according to the specified expressions.
942 This filter accepts one or more expressions (one for each channel),
943 which are evaluated and used to modify a corresponding audio signal.
945 It accepts the following parameters:
949 Set the '|'-separated expressions list for each separate channel. If
950 the number of input channels is greater than the number of
951 expressions, the last specified expression is used for the remaining
954 @item channel_layout, c
955 Set output channel layout. If not specified, the channel layout is
956 specified by the number of expressions. If set to @samp{same}, it will
957 use by default the same input channel layout.
960 Each expression in @var{exprs} can contain the following constants and functions:
964 channel number of the current expression
967 number of the evaluated sample, starting from 0
973 time of the evaluated sample expressed in seconds
976 @item nb_out_channels
977 input and output number of channels
980 the value of input channel with number @var{CH}
983 Note: this filter is slow. For faster processing you should use a
992 aeval=val(ch)/2:c=same
996 Invert phase of the second channel:
1005 Apply fade-in/out effect to input audio.
1007 A description of the accepted parameters follows.
1011 Specify the effect type, can be either @code{in} for fade-in, or
1012 @code{out} for a fade-out effect. Default is @code{in}.
1014 @item start_sample, ss
1015 Specify the number of the start sample for starting to apply the fade
1016 effect. Default is 0.
1018 @item nb_samples, ns
1019 Specify the number of samples for which the fade effect has to last. At
1020 the end of the fade-in effect the output audio will have the same
1021 volume as the input audio, at the end of the fade-out transition
1022 the output audio will be silence. Default is 44100.
1024 @item start_time, st
1025 Specify the start time of the fade effect. Default is 0.
1026 The value must be specified as a time duration; see
1027 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1028 for the accepted syntax.
1029 If set this option is used instead of @var{start_sample}.
1032 Specify the duration of the fade effect. See
1033 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1034 for the accepted syntax.
1035 At the end of the fade-in effect the output audio will have the same
1036 volume as the input audio, at the end of the fade-out transition
1037 the output audio will be silence.
1038 By default the duration is determined by @var{nb_samples}.
1039 If set this option is used instead of @var{nb_samples}.
1042 Set curve for fade transition.
1044 It accepts the following values:
1047 select triangular, linear slope (default)
1049 select quarter of sine wave
1051 select half of sine wave
1053 select exponential sine wave
1057 select inverted parabola
1071 select inverted quarter of sine wave
1073 select inverted half of sine wave
1075 select double-exponential seat
1077 select double-exponential sigmoid
1079 select logistic sigmoid
1081 select sine cardinal function
1083 select inverted sine cardinal function
1089 @subsection Examples
1093 Fade in first 15 seconds of audio:
1095 afade=t=in:ss=0:d=15
1099 Fade out last 25 seconds of a 900 seconds audio:
1101 afade=t=out:st=875:d=25
1106 Denoise audio samples with FFT.
1108 A description of the accepted parameters follows.
1112 Set the noise reduction in dB, allowed range is 0.01 to 97.
1113 Default value is 12 dB.
1116 Set the noise floor in dB, allowed range is -80 to -20.
1117 Default value is -50 dB.
1122 It accepts the following values:
1131 Select shellac noise.
1134 Select custom noise, defined in @code{bn} option.
1136 Default value is white noise.
1140 Set custom band noise for every one of 15 bands.
1141 Bands are separated by ' ' or '|'.
1144 Set the residual floor in dB, allowed range is -80 to -20.
1145 Default value is -38 dB.
1148 Enable noise tracking. By default is disabled.
1149 With this enabled, noise floor is automatically adjusted.
1152 Enable residual tracking. By default is disabled.
1155 Set the output mode.
1157 It accepts the following values:
1160 Pass input unchanged.
1163 Pass noise filtered out.
1168 Default value is @var{o}.
1172 @subsection Commands
1174 This filter supports the following commands:
1176 @item sample_noise, sn
1177 Start or stop measuring noise profile.
1178 Syntax for the command is : "start" or "stop" string.
1179 After measuring noise profile is stopped it will be
1180 automatically applied in filtering.
1182 @item noise_reduction, nr
1183 Change noise reduction. Argument is single float number.
1184 Syntax for the command is : "@var{noise_reduction}"
1186 @item noise_floor, nf
1187 Change noise floor. Argument is single float number.
1188 Syntax for the command is : "@var{noise_floor}"
1190 @item output_mode, om
1191 Change output mode operation.
1192 Syntax for the command is : "i", "o" or "n" string.
1196 Apply arbitrary expressions to samples in frequency domain.
1200 Set frequency domain real expression for each separate channel separated
1201 by '|'. Default is "re".
1202 If the number of input channels is greater than the number of
1203 expressions, the last specified expression is used for the remaining
1207 Set frequency domain imaginary expression for each separate channel
1208 separated by '|'. Default is "im".
1210 Each expression in @var{real} and @var{imag} can contain the following
1211 constants and functions:
1218 current frequency bin number
1221 number of available bins
1224 channel number of the current expression
1233 current real part of frequency bin of current channel
1236 current imaginary part of frequency bin of current channel
1239 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1242 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1246 Set window size. Allowed range is from 16 to 131072.
1247 Default is @code{4096}
1250 Set window function. Default is @code{hann}.
1253 Set window overlap. If set to 1, the recommended overlap for selected
1254 window function will be picked. Default is @code{0.75}.
1257 @subsection Examples
1261 Leave almost only low frequencies in audio:
1263 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1267 Apply robotize effect:
1269 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1273 Apply whisper effect:
1275 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"
1282 Apply an arbitrary Finite Impulse Response filter.
1284 This filter is designed for applying long FIR filters,
1285 up to 60 seconds long.
1287 It can be used as component for digital crossover filters,
1288 room equalization, cross talk cancellation, wavefield synthesis,
1289 auralization, ambiophonics, ambisonics and spatialization.
1291 This filter uses the streams higher than first one as FIR coefficients.
1292 If the non-first stream holds a single channel, it will be used
1293 for all input channels in the first stream, otherwise
1294 the number of channels in the non-first stream must be same as
1295 the number of channels in the first stream.
1297 It accepts the following parameters:
1301 Set dry gain. This sets input gain.
1304 Set wet gain. This sets final output gain.
1307 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1310 Enable applying gain measured from power of IR.
1312 Set which approach to use for auto gain measurement.
1316 Do not apply any gain.
1319 select peak gain, very conservative approach. This is default value.
1322 select DC gain, limited application.
1325 select gain to noise approach, this is most popular one.
1329 Set gain to be applied to IR coefficients before filtering.
1330 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1333 Set format of IR stream. Can be @code{mono} or @code{input}.
1334 Default is @code{input}.
1337 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1338 Allowed range is 0.1 to 60 seconds.
1341 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1342 By default it is disabled.
1345 Set for which IR channel to display frequency response. By default is first channel
1346 displayed. This option is used only when @var{response} is enabled.
1349 Set video stream size. This option is used only when @var{response} is enabled.
1352 Set video stream frame rate. This option is used only when @var{response} is enabled.
1355 Set minimal partition size used for convolution. Default is @var{8192}.
1356 Allowed range is from @var{1} to @var{32768}.
1357 Lower values decreases latency at cost of higher CPU usage.
1360 Set maximal partition size used for convolution. Default is @var{8192}.
1361 Allowed range is from @var{8} to @var{32768}.
1362 Lower values may increase CPU usage.
1365 Set number of input impulse responses streams which will be switchable at runtime.
1366 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1369 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1370 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1371 This option can be changed at runtime via @ref{commands}.
1374 @subsection Examples
1378 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1380 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1387 Set output format constraints for the input audio. The framework will
1388 negotiate the most appropriate format to minimize conversions.
1390 It accepts the following parameters:
1393 @item sample_fmts, f
1394 A '|'-separated list of requested sample formats.
1396 @item sample_rates, r
1397 A '|'-separated list of requested sample rates.
1399 @item channel_layouts, cl
1400 A '|'-separated list of requested channel layouts.
1402 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1403 for the required syntax.
1406 If a parameter is omitted, all values are allowed.
1408 Force the output to either unsigned 8-bit or signed 16-bit stereo
1410 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1414 Apply frequency shift to input audio samples.
1416 The filter accepts the following options:
1420 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1421 Default value is 0.0.
1424 @subsection Commands
1426 This filter supports the above option as @ref{commands}.
1430 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1431 processing reduces disturbing noise between useful signals.
1433 Gating is done by detecting the volume below a chosen level @var{threshold}
1434 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1435 floor is set via @var{range}. Because an exact manipulation of the signal
1436 would cause distortion of the waveform the reduction can be levelled over
1437 time. This is done by setting @var{attack} and @var{release}.
1439 @var{attack} determines how long the signal has to fall below the threshold
1440 before any reduction will occur and @var{release} sets the time the signal
1441 has to rise above the threshold to reduce the reduction again.
1442 Shorter signals than the chosen attack time will be left untouched.
1446 Set input level before filtering.
1447 Default is 1. Allowed range is from 0.015625 to 64.
1450 Set the mode of operation. Can be @code{upward} or @code{downward}.
1451 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1452 will be amplified, expanding dynamic range in upward direction.
1453 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1456 Set the level of gain reduction when the signal is below the threshold.
1457 Default is 0.06125. Allowed range is from 0 to 1.
1458 Setting this to 0 disables reduction and then filter behaves like expander.
1461 If a signal rises above this level the gain reduction is released.
1462 Default is 0.125. Allowed range is from 0 to 1.
1465 Set a ratio by which the signal is reduced.
1466 Default is 2. Allowed range is from 1 to 9000.
1469 Amount of milliseconds the signal has to rise above the threshold before gain
1471 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1474 Amount of milliseconds the signal has to fall below the threshold before the
1475 reduction is increased again. Default is 250 milliseconds.
1476 Allowed range is from 0.01 to 9000.
1479 Set amount of amplification of signal after processing.
1480 Default is 1. Allowed range is from 1 to 64.
1483 Curve the sharp knee around the threshold to enter gain reduction more softly.
1484 Default is 2.828427125. Allowed range is from 1 to 8.
1487 Choose if exact signal should be taken for detection or an RMS like one.
1488 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1491 Choose if the average level between all channels or the louder channel affects
1493 Default is @code{average}. Can be @code{average} or @code{maximum}.
1496 @subsection Commands
1498 This filter supports the all above options as @ref{commands}.
1502 Apply an arbitrary Infinite Impulse Response filter.
1504 It accepts the following parameters:
1508 Set B/numerator/zeros/reflection coefficients.
1511 Set A/denominator/poles/ladder coefficients.
1523 Set coefficients format.
1527 lattice-ladder function
1529 analog transfer function
1531 digital transfer function
1533 Z-plane zeros/poles, cartesian (default)
1535 Z-plane zeros/poles, polar radians
1537 Z-plane zeros/poles, polar degrees
1543 Set type of processing.
1555 Set filtering precision.
1559 double-precision floating-point (default)
1561 single-precision floating-point
1569 Normalize filter coefficients, by default is enabled.
1570 Enabling it will normalize magnitude response at DC to 0dB.
1573 How much to use filtered signal in output. Default is 1.
1574 Range is between 0 and 1.
1577 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1578 By default it is disabled.
1581 Set for which IR channel to display frequency response. By default is first channel
1582 displayed. This option is used only when @var{response} is enabled.
1585 Set video stream size. This option is used only when @var{response} is enabled.
1588 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1591 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1592 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1595 Different coefficients and gains can be provided for every channel, in such case
1596 use '|' to separate coefficients or gains. Last provided coefficients will be
1597 used for all remaining channels.
1599 @subsection Examples
1603 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1605 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
1609 Same as above but in @code{zp} format:
1611 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
1615 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1617 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1623 The limiter prevents an input signal from rising over a desired threshold.
1624 This limiter uses lookahead technology to prevent your signal from distorting.
1625 It means that there is a small delay after the signal is processed. Keep in mind
1626 that the delay it produces is the attack time you set.
1628 The filter accepts the following options:
1632 Set input gain. Default is 1.
1635 Set output gain. Default is 1.
1638 Don't let signals above this level pass the limiter. Default is 1.
1641 The limiter will reach its attenuation level in this amount of time in
1642 milliseconds. Default is 5 milliseconds.
1645 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1646 Default is 50 milliseconds.
1649 When gain reduction is always needed ASC takes care of releasing to an
1650 average reduction level rather than reaching a reduction of 0 in the release
1654 Select how much the release time is affected by ASC, 0 means nearly no changes
1655 in release time while 1 produces higher release times.
1658 Auto level output signal. Default is enabled.
1659 This normalizes audio back to 0dB if enabled.
1662 Depending on picked setting it is recommended to upsample input 2x or 4x times
1663 with @ref{aresample} before applying this filter.
1667 Apply a two-pole all-pass filter with central frequency (in Hz)
1668 @var{frequency}, and filter-width @var{width}.
1669 An all-pass filter changes the audio's frequency to phase relationship
1670 without changing its frequency to amplitude relationship.
1672 The filter accepts the following options:
1676 Set frequency in Hz.
1679 Set method to specify band-width of filter.
1694 Specify the band-width of a filter in width_type units.
1697 How much to use filtered signal in output. Default is 1.
1698 Range is between 0 and 1.
1701 Specify which channels to filter, by default all available are filtered.
1704 Normalize biquad coefficients, by default is disabled.
1705 Enabling it will normalize magnitude response at DC to 0dB.
1708 Set the filter order, can be 1 or 2. Default is 2.
1711 Set transform type of IIR filter.
1720 @subsection Commands
1722 This filter supports the following commands:
1725 Change allpass frequency.
1726 Syntax for the command is : "@var{frequency}"
1729 Change allpass width_type.
1730 Syntax for the command is : "@var{width_type}"
1733 Change allpass width.
1734 Syntax for the command is : "@var{width}"
1738 Syntax for the command is : "@var{mix}"
1745 The filter accepts the following options:
1749 Set the number of loops. Setting this value to -1 will result in infinite loops.
1753 Set maximal number of samples. Default is 0.
1756 Set first sample of loop. Default is 0.
1762 Merge two or more audio streams into a single multi-channel stream.
1764 The filter accepts the following options:
1769 Set the number of inputs. Default is 2.
1773 If the channel layouts of the inputs are disjoint, and therefore compatible,
1774 the channel layout of the output will be set accordingly and the channels
1775 will be reordered as necessary. If the channel layouts of the inputs are not
1776 disjoint, the output will have all the channels of the first input then all
1777 the channels of the second input, in that order, and the channel layout of
1778 the output will be the default value corresponding to the total number of
1781 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1782 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1783 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1784 first input, b1 is the first channel of the second input).
1786 On the other hand, if both input are in stereo, the output channels will be
1787 in the default order: a1, a2, b1, b2, and the channel layout will be
1788 arbitrarily set to 4.0, which may or may not be the expected value.
1790 All inputs must have the same sample rate, and format.
1792 If inputs do not have the same duration, the output will stop with the
1795 @subsection Examples
1799 Merge two mono files into a stereo stream:
1801 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1805 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1807 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
1813 Mixes multiple audio inputs into a single output.
1815 Note that this filter only supports float samples (the @var{amerge}
1816 and @var{pan} audio filters support many formats). If the @var{amix}
1817 input has integer samples then @ref{aresample} will be automatically
1818 inserted to perform the conversion to float samples.
1822 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1824 will mix 3 input audio streams to a single output with the same duration as the
1825 first input and a dropout transition time of 3 seconds.
1827 It accepts the following parameters:
1831 The number of inputs. If unspecified, it defaults to 2.
1834 How to determine the end-of-stream.
1838 The duration of the longest input. (default)
1841 The duration of the shortest input.
1844 The duration of the first input.
1848 @item dropout_transition
1849 The transition time, in seconds, for volume renormalization when an input
1850 stream ends. The default value is 2 seconds.
1853 Specify weight of each input audio stream as sequence.
1854 Each weight is separated by space. By default all inputs have same weight.
1857 @subsection Commands
1859 This filter supports the following commands:
1862 Syntax is same as option with same name.
1867 Multiply first audio stream with second audio stream and store result
1868 in output audio stream. Multiplication is done by multiplying each
1869 sample from first stream with sample at same position from second stream.
1871 With this element-wise multiplication one can create amplitude fades and
1872 amplitude modulations.
1874 @section anequalizer
1876 High-order parametric multiband equalizer for each channel.
1878 It accepts the following parameters:
1882 This option string is in format:
1883 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1884 Each equalizer band is separated by '|'.
1888 Set channel number to which equalization will be applied.
1889 If input doesn't have that channel the entry is ignored.
1892 Set central frequency for band.
1893 If input doesn't have that frequency the entry is ignored.
1896 Set band width in Hertz.
1899 Set band gain in dB.
1902 Set filter type for band, optional, can be:
1906 Butterworth, this is default.
1917 With this option activated frequency response of anequalizer is displayed
1921 Set video stream size. Only useful if curves option is activated.
1924 Set max gain that will be displayed. Only useful if curves option is activated.
1925 Setting this to a reasonable value makes it possible to display gain which is derived from
1926 neighbour bands which are too close to each other and thus produce higher gain
1927 when both are activated.
1930 Set frequency scale used to draw frequency response in video output.
1931 Can be linear or logarithmic. Default is logarithmic.
1934 Set color for each channel curve which is going to be displayed in video stream.
1935 This is list of color names separated by space or by '|'.
1936 Unrecognised or missing colors will be replaced by white color.
1939 @subsection Examples
1943 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1944 for first 2 channels using Chebyshev type 1 filter:
1946 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1950 @subsection Commands
1952 This filter supports the following commands:
1955 Alter existing filter parameters.
1956 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1958 @var{fN} is existing filter number, starting from 0, if no such filter is available
1960 @var{freq} set new frequency parameter.
1961 @var{width} set new width parameter in Hertz.
1962 @var{gain} set new gain parameter in dB.
1964 Full filter invocation with asendcmd may look like this:
1965 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1970 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1972 Each sample is adjusted by looking for other samples with similar contexts. This
1973 context similarity is defined by comparing their surrounding patches of size
1974 @option{p}. Patches are searched in an area of @option{r} around the sample.
1976 The filter accepts the following options:
1980 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1983 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1984 Default value is 2 milliseconds.
1987 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1988 Default value is 6 milliseconds.
1991 Set the output mode.
1993 It accepts the following values:
1996 Pass input unchanged.
1999 Pass noise filtered out.
2004 Default value is @var{o}.
2008 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
2011 @subsection Commands
2013 This filter supports the all above options as @ref{commands}.
2016 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
2018 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
2019 relate to producing the least mean square of the error signal (difference between the desired,
2020 2nd input audio stream and the actual signal, the 1st input audio stream).
2022 A description of the accepted options follows.
2035 Set the filter leakage.
2038 It accepts the following values:
2047 Pass filtered samples.
2050 Pass difference between desired and filtered samples.
2052 Default value is @var{o}.
2056 @subsection Examples
2060 One of many usages of this filter is noise reduction, input audio is filtered
2061 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2063 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2067 @subsection Commands
2069 This filter supports the same commands as options, excluding option @code{order}.
2073 Pass the audio source unchanged to the output.
2077 Pad the end of an audio stream with silence.
2079 This can be used together with @command{ffmpeg} @option{-shortest} to
2080 extend audio streams to the same length as the video stream.
2082 A description of the accepted options follows.
2086 Set silence packet size. Default value is 4096.
2089 Set the number of samples of silence to add to the end. After the
2090 value is reached, the stream is terminated. This option is mutually
2091 exclusive with @option{whole_len}.
2094 Set the minimum total number of samples in the output audio stream. If
2095 the value is longer than the input audio length, silence is added to
2096 the end, until the value is reached. This option is mutually exclusive
2097 with @option{pad_len}.
2100 Specify the duration of samples of silence to add. See
2101 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2102 for the accepted syntax. Used only if set to non-zero value.
2105 Specify the minimum total duration in the output audio stream. See
2106 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2107 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2108 the input audio length, silence is added to the end, until the value is reached.
2109 This option is mutually exclusive with @option{pad_dur}
2112 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2113 nor @option{whole_dur} option is set, the filter will add silence to the end of
2114 the input stream indefinitely.
2116 @subsection Examples
2120 Add 1024 samples of silence to the end of the input:
2126 Make sure the audio output will contain at least 10000 samples, pad
2127 the input with silence if required:
2129 apad=whole_len=10000
2133 Use @command{ffmpeg} to pad the audio input with silence, so that the
2134 video stream will always result the shortest and will be converted
2135 until the end in the output file when using the @option{shortest}
2138 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2143 Add a phasing effect to the input audio.
2145 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2146 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2148 A description of the accepted parameters follows.
2152 Set input gain. Default is 0.4.
2155 Set output gain. Default is 0.74
2158 Set delay in milliseconds. Default is 3.0.
2161 Set decay. Default is 0.4.
2164 Set modulation speed in Hz. Default is 0.5.
2167 Set modulation type. Default is triangular.
2169 It accepts the following values:
2176 @section aphaseshift
2177 Apply phase shift to input audio samples.
2179 The filter accepts the following options:
2183 Specify phase shift. Allowed range is from -1.0 to 1.0.
2184 Default value is 0.0.
2187 @subsection Commands
2189 This filter supports the above option as @ref{commands}.
2193 Audio pulsator is something between an autopanner and a tremolo.
2194 But it can produce funny stereo effects as well. Pulsator changes the volume
2195 of the left and right channel based on a LFO (low frequency oscillator) with
2196 different waveforms and shifted phases.
2197 This filter have the ability to define an offset between left and right
2198 channel. An offset of 0 means that both LFO shapes match each other.
2199 The left and right channel are altered equally - a conventional tremolo.
2200 An offset of 50% means that the shape of the right channel is exactly shifted
2201 in phase (or moved backwards about half of the frequency) - pulsator acts as
2202 an autopanner. At 1 both curves match again. Every setting in between moves the
2203 phase shift gapless between all stages and produces some "bypassing" sounds with
2204 sine and triangle waveforms. The more you set the offset near 1 (starting from
2205 the 0.5) the faster the signal passes from the left to the right speaker.
2207 The filter accepts the following options:
2211 Set input gain. By default it is 1. Range is [0.015625 - 64].
2214 Set output gain. By default it is 1. Range is [0.015625 - 64].
2217 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2218 sawup or sawdown. Default is sine.
2221 Set modulation. Define how much of original signal is affected by the LFO.
2224 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2227 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2230 Set pulse width. Default is 1. Allowed range is [0 - 2].
2233 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2236 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2240 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2244 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2245 if timing is set to hz.
2251 Resample the input audio to the specified parameters, using the
2252 libswresample library. If none are specified then the filter will
2253 automatically convert between its input and output.
2255 This filter is also able to stretch/squeeze the audio data to make it match
2256 the timestamps or to inject silence / cut out audio to make it match the
2257 timestamps, do a combination of both or do neither.
2259 The filter accepts the syntax
2260 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2261 expresses a sample rate and @var{resampler_options} is a list of
2262 @var{key}=@var{value} pairs, separated by ":". See the
2263 @ref{Resampler Options,,"Resampler Options" section in the
2264 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2265 for the complete list of supported options.
2267 @subsection Examples
2271 Resample the input audio to 44100Hz:
2277 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2278 samples per second compensation:
2280 aresample=async=1000
2286 Reverse an audio clip.
2288 Warning: This filter requires memory to buffer the entire clip, so trimming
2291 @subsection Examples
2295 Take the first 5 seconds of a clip, and reverse it.
2297 atrim=end=5,areverse
2303 Reduce noise from speech using Recurrent Neural Networks.
2305 This filter accepts the following options:
2309 Set train model file to load. This option is always required.
2312 @section asetnsamples
2314 Set the number of samples per each output audio frame.
2316 The last output packet may contain a different number of samples, as
2317 the filter will flush all the remaining samples when the input audio
2320 The filter accepts the following options:
2324 @item nb_out_samples, n
2325 Set the number of frames per each output audio frame. The number is
2326 intended as the number of samples @emph{per each channel}.
2327 Default value is 1024.
2330 If set to 1, the filter will pad the last audio frame with zeroes, so
2331 that the last frame will contain the same number of samples as the
2332 previous ones. Default value is 1.
2335 For example, to set the number of per-frame samples to 1234 and
2336 disable padding for the last frame, use:
2338 asetnsamples=n=1234:p=0
2343 Set the sample rate without altering the PCM data.
2344 This will result in a change of speed and pitch.
2346 The filter accepts the following options:
2349 @item sample_rate, r
2350 Set the output sample rate. Default is 44100 Hz.
2355 Show a line containing various information for each input audio frame.
2356 The input audio is not modified.
2358 The shown line contains a sequence of key/value pairs of the form
2359 @var{key}:@var{value}.
2361 The following values are shown in the output:
2365 The (sequential) number of the input frame, starting from 0.
2368 The presentation timestamp of the input frame, in time base units; the time base
2369 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2372 The presentation timestamp of the input frame in seconds.
2375 position of the frame in the input stream, -1 if this information in
2376 unavailable and/or meaningless (for example in case of synthetic audio)
2385 The sample rate for the audio frame.
2388 The number of samples (per channel) in the frame.
2391 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2392 audio, the data is treated as if all the planes were concatenated.
2394 @item plane_checksums
2395 A list of Adler-32 checksums for each data plane.
2399 Apply audio soft clipping.
2401 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2402 along a smooth curve, rather than the abrupt shape of hard-clipping.
2404 This filter accepts the following options:
2408 Set type of soft-clipping.
2410 It accepts the following values:
2424 Set additional parameter which controls sigmoid function.
2427 Set oversampling factor.
2430 @subsection Commands
2432 This filter supports the all above options as @ref{commands}.
2435 Automatic Speech Recognition
2437 This filter uses PocketSphinx for speech recognition. To enable
2438 compilation of this filter, you need to configure FFmpeg with
2439 @code{--enable-pocketsphinx}.
2441 It accepts the following options:
2445 Set sampling rate of input audio. Defaults is @code{16000}.
2446 This need to match speech models, otherwise one will get poor results.
2449 Set dictionary containing acoustic model files.
2452 Set pronunciation dictionary.
2455 Set language model file.
2458 Set language model set.
2461 Set which language model to use.
2464 Set output for log messages.
2467 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2472 Display time domain statistical information about the audio channels.
2473 Statistics are calculated and displayed for each audio channel and,
2474 where applicable, an overall figure is also given.
2476 It accepts the following option:
2479 Short window length in seconds, used for peak and trough RMS measurement.
2480 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2484 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2485 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2488 Available keys for each channel are:
2534 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2535 this @code{lavfi.astats.Overall.Peak_count}.
2537 For description what each key means read below.
2540 Set number of frame after which stats are going to be recalculated.
2541 Default is disabled.
2543 @item measure_perchannel
2544 Select the entries which need to be measured per channel. The metadata keys can
2545 be used as flags, default is @option{all} which measures everything.
2546 @option{none} disables all per channel measurement.
2548 @item measure_overall
2549 Select the entries which need to be measured overall. The metadata keys can
2550 be used as flags, default is @option{all} which measures everything.
2551 @option{none} disables all overall measurement.
2555 A description of each shown parameter follows:
2559 Mean amplitude displacement from zero.
2562 Minimal sample level.
2565 Maximal sample level.
2567 @item Min difference
2568 Minimal difference between two consecutive samples.
2570 @item Max difference
2571 Maximal difference between two consecutive samples.
2573 @item Mean difference
2574 Mean difference between two consecutive samples.
2575 The average of each difference between two consecutive samples.
2577 @item RMS difference
2578 Root Mean Square difference between two consecutive samples.
2582 Standard peak and RMS level measured in dBFS.
2586 Peak and trough values for RMS level measured over a short window.
2589 Standard ratio of peak to RMS level (note: not in dB).
2592 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2593 (i.e. either @var{Min level} or @var{Max level}).
2596 Number of occasions (not the number of samples) that the signal attained either
2597 @var{Min level} or @var{Max level}.
2599 @item Noise floor dB
2600 Minimum local peak measured in dBFS over a short window.
2602 @item Noise floor count
2603 Number of occasions (not the number of samples) that the signal attained
2607 Overall bit depth of audio. Number of bits used for each sample.
2610 Measured dynamic range of audio in dB.
2612 @item Zero crossings
2613 Number of points where the waveform crosses the zero level axis.
2615 @item Zero crossings rate
2616 Rate of Zero crossings and number of audio samples.
2620 Boost subwoofer frequencies.
2622 The filter accepts the following options:
2626 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2627 Default value is 0.7.
2630 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2631 Default value is 0.7.
2634 Set delay line decay gain value. Allowed range is from 0 to 1.
2635 Default value is 0.7.
2638 Set delay line feedback gain value. Allowed range is from 0 to 1.
2639 Default value is 0.9.
2642 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2643 Default value is 100.
2646 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2647 Default value is 0.5.
2650 Set delay. Allowed range is from 1 to 100.
2651 Default value is 20.
2654 @subsection Commands
2656 This filter supports the all above options as @ref{commands}.
2659 Cut super frequencies.
2661 The filter accepts the following options:
2665 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2666 Default value is 20000.
2669 Set filter order. Available values are from 3 to 20.
2670 Default value is 10.
2673 @subsection Commands
2675 This filter supports the all above options as @ref{commands}.
2681 The filter accepts exactly one parameter, the audio tempo. If not
2682 specified then the filter will assume nominal 1.0 tempo. Tempo must
2683 be in the [0.5, 100.0] range.
2685 Note that tempo greater than 2 will skip some samples rather than
2686 blend them in. If for any reason this is a concern it is always
2687 possible to daisy-chain several instances of atempo to achieve the
2688 desired product tempo.
2690 @subsection Examples
2694 Slow down audio to 80% tempo:
2700 To speed up audio to 300% tempo:
2706 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2708 atempo=sqrt(3),atempo=sqrt(3)
2712 @subsection Commands
2714 This filter supports the following commands:
2717 Change filter tempo scale factor.
2718 Syntax for the command is : "@var{tempo}"
2723 Trim the input so that the output contains one continuous subpart of the input.
2725 It accepts the following parameters:
2728 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2729 sample with the timestamp @var{start} will be the first sample in the output.
2732 Specify time of the first audio sample that will be dropped, i.e. the
2733 audio sample immediately preceding the one with the timestamp @var{end} will be
2734 the last sample in the output.
2737 Same as @var{start}, except this option sets the start timestamp in samples
2741 Same as @var{end}, except this option sets the end timestamp in samples instead
2745 The maximum duration of the output in seconds.
2748 The number of the first sample that should be output.
2751 The number of the first sample that should be dropped.
2754 @option{start}, @option{end}, and @option{duration} are expressed as time
2755 duration specifications; see
2756 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2758 Note that the first two sets of the start/end options and the @option{duration}
2759 option look at the frame timestamp, while the _sample options simply count the
2760 samples that pass through the filter. So start/end_pts and start/end_sample will
2761 give different results when the timestamps are wrong, inexact or do not start at
2762 zero. Also note that this filter does not modify the timestamps. If you wish
2763 to have the output timestamps start at zero, insert the asetpts filter after the
2766 If multiple start or end options are set, this filter tries to be greedy and
2767 keep all samples that match at least one of the specified constraints. To keep
2768 only the part that matches all the constraints at once, chain multiple atrim
2771 The defaults are such that all the input is kept. So it is possible to set e.g.
2772 just the end values to keep everything before the specified time.
2777 Drop everything except the second minute of input:
2779 ffmpeg -i INPUT -af atrim=60:120
2783 Keep only the first 1000 samples:
2785 ffmpeg -i INPUT -af atrim=end_sample=1000
2790 @section axcorrelate
2791 Calculate normalized cross-correlation between two input audio streams.
2793 Resulted samples are always between -1 and 1 inclusive.
2794 If result is 1 it means two input samples are highly correlated in that selected segment.
2795 Result 0 means they are not correlated at all.
2796 If result is -1 it means two input samples are out of phase, which means they cancel each
2799 The filter accepts the following options:
2803 Set size of segment over which cross-correlation is calculated.
2804 Default is 256. Allowed range is from 2 to 131072.
2807 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2808 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2809 are always zero and thus need much less calculations to make.
2810 This is generally not true, but is valid for typical audio streams.
2813 @subsection Examples
2817 Calculate correlation between channels in stereo audio stream:
2819 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2825 Apply a two-pole Butterworth band-pass filter with central
2826 frequency @var{frequency}, and (3dB-point) band-width width.
2827 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2828 instead of the default: constant 0dB peak gain.
2829 The filter roll off at 6dB per octave (20dB per decade).
2831 The filter accepts the following options:
2835 Set the filter's central frequency. Default is @code{3000}.
2838 Constant skirt gain if set to 1. Defaults to 0.
2841 Set method to specify band-width of filter.
2856 Specify the band-width of a filter in width_type units.
2859 How much to use filtered signal in output. Default is 1.
2860 Range is between 0 and 1.
2863 Specify which channels to filter, by default all available are filtered.
2866 Normalize biquad coefficients, by default is disabled.
2867 Enabling it will normalize magnitude response at DC to 0dB.
2870 Set transform type of IIR filter.
2879 @subsection Commands
2881 This filter supports the following commands:
2884 Change bandpass frequency.
2885 Syntax for the command is : "@var{frequency}"
2888 Change bandpass width_type.
2889 Syntax for the command is : "@var{width_type}"
2892 Change bandpass width.
2893 Syntax for the command is : "@var{width}"
2896 Change bandpass mix.
2897 Syntax for the command is : "@var{mix}"
2902 Apply a two-pole Butterworth band-reject filter with central
2903 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2904 The filter roll off at 6dB per octave (20dB per decade).
2906 The filter accepts the following options:
2910 Set the filter's central frequency. Default is @code{3000}.
2913 Set method to specify band-width of filter.
2928 Specify the band-width of a filter in width_type units.
2931 How much to use filtered signal in output. Default is 1.
2932 Range is between 0 and 1.
2935 Specify which channels to filter, by default all available are filtered.
2938 Normalize biquad coefficients, by default is disabled.
2939 Enabling it will normalize magnitude response at DC to 0dB.
2942 Set transform type of IIR filter.
2951 @subsection Commands
2953 This filter supports the following commands:
2956 Change bandreject frequency.
2957 Syntax for the command is : "@var{frequency}"
2960 Change bandreject width_type.
2961 Syntax for the command is : "@var{width_type}"
2964 Change bandreject width.
2965 Syntax for the command is : "@var{width}"
2968 Change bandreject mix.
2969 Syntax for the command is : "@var{mix}"
2972 @section bass, lowshelf
2974 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2975 shelving filter with a response similar to that of a standard
2976 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2978 The filter accepts the following options:
2982 Give the gain at 0 Hz. Its useful range is about -20
2983 (for a large cut) to +20 (for a large boost).
2984 Beware of clipping when using a positive gain.
2987 Set the filter's central frequency and so can be used
2988 to extend or reduce the frequency range to be boosted or cut.
2989 The default value is @code{100} Hz.
2992 Set method to specify band-width of filter.
3007 Determine how steep is the filter's shelf transition.
3010 How much to use filtered signal in output. Default is 1.
3011 Range is between 0 and 1.
3014 Specify which channels to filter, by default all available are filtered.
3017 Normalize biquad coefficients, by default is disabled.
3018 Enabling it will normalize magnitude response at DC to 0dB.
3021 Set transform type of IIR filter.
3030 @subsection Commands
3032 This filter supports the following commands:
3035 Change bass frequency.
3036 Syntax for the command is : "@var{frequency}"
3039 Change bass width_type.
3040 Syntax for the command is : "@var{width_type}"
3044 Syntax for the command is : "@var{width}"
3048 Syntax for the command is : "@var{gain}"
3052 Syntax for the command is : "@var{mix}"
3057 Apply a biquad IIR filter with the given coefficients.
3058 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3059 are the numerator and denominator coefficients respectively.
3060 and @var{channels}, @var{c} specify which channels to filter, by default all
3061 available are filtered.
3063 @subsection Commands
3065 This filter supports the following commands:
3073 Change biquad parameter.
3074 Syntax for the command is : "@var{value}"
3077 How much to use filtered signal in output. Default is 1.
3078 Range is between 0 and 1.
3081 Specify which channels to filter, by default all available are filtered.
3084 Normalize biquad coefficients, by default is disabled.
3085 Enabling it will normalize magnitude response at DC to 0dB.
3088 Set transform type of IIR filter.
3098 Bauer stereo to binaural transformation, which improves headphone listening of
3099 stereo audio records.
3101 To enable compilation of this filter you need to configure FFmpeg with
3102 @code{--enable-libbs2b}.
3104 It accepts the following parameters:
3108 Pre-defined crossfeed level.
3112 Default level (fcut=700, feed=50).
3115 Chu Moy circuit (fcut=700, feed=60).
3118 Jan Meier circuit (fcut=650, feed=95).
3123 Cut frequency (in Hz).
3132 Remap input channels to new locations.
3134 It accepts the following parameters:
3137 Map channels from input to output. The argument is a '|'-separated list of
3138 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3139 @var{in_channel} form. @var{in_channel} can be either the name of the input
3140 channel (e.g. FL for front left) or its index in the input channel layout.
3141 @var{out_channel} is the name of the output channel or its index in the output
3142 channel layout. If @var{out_channel} is not given then it is implicitly an
3143 index, starting with zero and increasing by one for each mapping.
3145 @item channel_layout
3146 The channel layout of the output stream.
3149 If no mapping is present, the filter will implicitly map input channels to
3150 output channels, preserving indices.
3152 @subsection Examples
3156 For example, assuming a 5.1+downmix input MOV file,
3158 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3160 will create an output WAV file tagged as stereo from the downmix channels of
3164 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3166 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3170 @section channelsplit
3172 Split each channel from an input audio stream into a separate output stream.
3174 It accepts the following parameters:
3176 @item channel_layout
3177 The channel layout of the input stream. The default is "stereo".
3179 A channel layout describing the channels to be extracted as separate output streams
3180 or "all" to extract each input channel as a separate stream. The default is "all".
3182 Choosing channels not present in channel layout in the input will result in an error.
3185 @subsection Examples
3189 For example, assuming a stereo input MP3 file,
3191 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3193 will create an output Matroska file with two audio streams, one containing only
3194 the left channel and the other the right channel.
3197 Split a 5.1 WAV file into per-channel files:
3199 ffmpeg -i in.wav -filter_complex
3200 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3201 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3202 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3207 Extract only LFE from a 5.1 WAV file:
3209 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3210 -map '[LFE]' lfe.wav
3215 Add a chorus effect to the audio.
3217 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3219 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3220 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3221 The modulation depth defines the range the modulated delay is played before or after
3222 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3223 sound tuned around the original one, like in a chorus where some vocals are slightly
3226 It accepts the following parameters:
3229 Set input gain. Default is 0.4.
3232 Set output gain. Default is 0.4.
3235 Set delays. A typical delay is around 40ms to 60ms.
3247 @subsection Examples
3253 chorus=0.7:0.9:55:0.4:0.25:2
3259 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3263 Fuller sounding chorus with three delays:
3265 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
3270 Compress or expand the audio's dynamic range.
3272 It accepts the following parameters:
3278 A list of times in seconds for each channel over which the instantaneous level
3279 of the input signal is averaged to determine its volume. @var{attacks} refers to
3280 increase of volume and @var{decays} refers to decrease of volume. For most
3281 situations, the attack time (response to the audio getting louder) should be
3282 shorter than the decay time, because the human ear is more sensitive to sudden
3283 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3284 a typical value for decay is 0.8 seconds.
3285 If specified number of attacks & decays is lower than number of channels, the last
3286 set attack/decay will be used for all remaining channels.
3289 A list of points for the transfer function, specified in dB relative to the
3290 maximum possible signal amplitude. Each key points list must be defined using
3291 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3292 @code{x0/y0 x1/y1 x2/y2 ....}
3294 The input values must be in strictly increasing order but the transfer function
3295 does not have to be monotonically rising. The point @code{0/0} is assumed but
3296 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3297 function are @code{-70/-70|-60/-20|1/0}.
3300 Set the curve radius in dB for all joints. It defaults to 0.01.
3303 Set the additional gain in dB to be applied at all points on the transfer
3304 function. This allows for easy adjustment of the overall gain.
3308 Set an initial volume, in dB, to be assumed for each channel when filtering
3309 starts. This permits the user to supply a nominal level initially, so that, for
3310 example, a very large gain is not applied to initial signal levels before the
3311 companding has begun to operate. A typical value for audio which is initially
3312 quiet is -90 dB. It defaults to 0.
3315 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3316 delayed before being fed to the volume adjuster. Specifying a delay
3317 approximately equal to the attack/decay times allows the filter to effectively
3318 operate in predictive rather than reactive mode. It defaults to 0.
3322 @subsection Examples
3326 Make music with both quiet and loud passages suitable for listening to in a
3329 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3332 Another example for audio with whisper and explosion parts:
3334 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3338 A noise gate for when the noise is at a lower level than the signal:
3340 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3344 Here is another noise gate, this time for when the noise is at a higher level
3345 than the signal (making it, in some ways, similar to squelch):
3347 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3351 2:1 compression starting at -6dB:
3353 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3357 2:1 compression starting at -9dB:
3359 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3363 2:1 compression starting at -12dB:
3365 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3369 2:1 compression starting at -18dB:
3371 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3375 3:1 compression starting at -15dB:
3377 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3383 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3389 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
3393 Hard limiter at -6dB:
3395 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3399 Hard limiter at -12dB:
3401 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3405 Hard noise gate at -35 dB:
3407 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3413 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3417 @section compensationdelay
3419 Compensation Delay Line is a metric based delay to compensate differing
3420 positions of microphones or speakers.
3422 For example, you have recorded guitar with two microphones placed in
3423 different locations. Because the front of sound wave has fixed speed in
3424 normal conditions, the phasing of microphones can vary and depends on
3425 their location and interposition. The best sound mix can be achieved when
3426 these microphones are in phase (synchronized). Note that a distance of
3427 ~30 cm between microphones makes one microphone capture the signal in
3428 antiphase to the other microphone. That makes the final mix sound moody.
3429 This filter helps to solve phasing problems by adding different delays
3430 to each microphone track and make them synchronized.
3432 The best result can be reached when you take one track as base and
3433 synchronize other tracks one by one with it.
3434 Remember that synchronization/delay tolerance depends on sample rate, too.
3435 Higher sample rates will give more tolerance.
3437 The filter accepts the following parameters:
3441 Set millimeters distance. This is compensation distance for fine tuning.
3445 Set cm distance. This is compensation distance for tightening distance setup.
3449 Set meters distance. This is compensation distance for hard distance setup.
3453 Set dry amount. Amount of unprocessed (dry) signal.
3457 Set wet amount. Amount of processed (wet) signal.
3461 Set temperature in degrees Celsius. This is the temperature of the environment.
3466 Apply headphone crossfeed filter.
3468 Crossfeed is the process of blending the left and right channels of stereo
3470 It is mainly used to reduce extreme stereo separation of low frequencies.
3472 The intent is to produce more speaker like sound to the listener.
3474 The filter accepts the following options:
3478 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3479 This sets gain of low shelf filter for side part of stereo image.
3480 Default is -6dB. Max allowed is -30db when strength is set to 1.
3483 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3484 This sets cut off frequency of low shelf filter. Default is cut off near
3485 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3488 Set curve slope of low shelf filter. Default is 0.5.
3489 Allowed range is from 0.01 to 1.
3492 Set input gain. Default is 0.9.
3495 Set output gain. Default is 1.
3498 @subsection Commands
3500 This filter supports the all above options as @ref{commands}.
3502 @section crystalizer
3503 Simple algorithm to expand audio dynamic range.
3505 The filter accepts the following options:
3509 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3510 (unchanged sound) to 10.0 (maximum effect).
3513 Enable clipping. By default is enabled.
3516 @subsection Commands
3518 This filter supports the all above options as @ref{commands}.
3521 Apply a DC shift to the audio.
3523 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3524 in the recording chain) from the audio. The effect of a DC offset is reduced
3525 headroom and hence volume. The @ref{astats} filter can be used to determine if
3526 a signal has a DC offset.
3530 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3534 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3535 used to prevent clipping.
3540 Apply de-essing to the audio samples.
3544 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3548 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3552 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3556 Set the output mode.
3558 It accepts the following values:
3561 Pass input unchanged.
3564 Pass ess filtered out.
3569 Default value is @var{o}.
3575 Measure audio dynamic range.
3577 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3578 is found in transition material. And anything less that 8 have very poor dynamics
3579 and is very compressed.
3581 The filter accepts the following options:
3585 Set window length in seconds used to split audio into segments of equal length.
3586 Default is 3 seconds.
3590 Dynamic Audio Normalizer.
3592 This filter applies a certain amount of gain to the input audio in order
3593 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3594 contrast to more "simple" normalization algorithms, the Dynamic Audio
3595 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3596 This allows for applying extra gain to the "quiet" sections of the audio
3597 while avoiding distortions or clipping the "loud" sections. In other words:
3598 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3599 sections, in the sense that the volume of each section is brought to the
3600 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3601 this goal *without* applying "dynamic range compressing". It will retain 100%
3602 of the dynamic range *within* each section of the audio file.
3606 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3607 Default is 500 milliseconds.
3608 The Dynamic Audio Normalizer processes the input audio in small chunks,
3609 referred to as frames. This is required, because a peak magnitude has no
3610 meaning for just a single sample value. Instead, we need to determine the
3611 peak magnitude for a contiguous sequence of sample values. While a "standard"
3612 normalizer would simply use the peak magnitude of the complete file, the
3613 Dynamic Audio Normalizer determines the peak magnitude individually for each
3614 frame. The length of a frame is specified in milliseconds. By default, the
3615 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3616 been found to give good results with most files.
3617 Note that the exact frame length, in number of samples, will be determined
3618 automatically, based on the sampling rate of the individual input audio file.
3621 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3622 number. Default is 31.
3623 Probably the most important parameter of the Dynamic Audio Normalizer is the
3624 @code{window size} of the Gaussian smoothing filter. The filter's window size
3625 is specified in frames, centered around the current frame. For the sake of
3626 simplicity, this must be an odd number. Consequently, the default value of 31
3627 takes into account the current frame, as well as the 15 preceding frames and
3628 the 15 subsequent frames. Using a larger window results in a stronger
3629 smoothing effect and thus in less gain variation, i.e. slower gain
3630 adaptation. Conversely, using a smaller window results in a weaker smoothing
3631 effect and thus in more gain variation, i.e. faster gain adaptation.
3632 In other words, the more you increase this value, the more the Dynamic Audio
3633 Normalizer will behave like a "traditional" normalization filter. On the
3634 contrary, the more you decrease this value, the more the Dynamic Audio
3635 Normalizer will behave like a dynamic range compressor.
3638 Set the target peak value. This specifies the highest permissible magnitude
3639 level for the normalized audio input. This filter will try to approach the
3640 target peak magnitude as closely as possible, but at the same time it also
3641 makes sure that the normalized signal will never exceed the peak magnitude.
3642 A frame's maximum local gain factor is imposed directly by the target peak
3643 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3644 It is not recommended to go above this value.
3647 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3648 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3649 factor for each input frame, i.e. the maximum gain factor that does not
3650 result in clipping or distortion. The maximum gain factor is determined by
3651 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3652 additionally bounds the frame's maximum gain factor by a predetermined
3653 (global) maximum gain factor. This is done in order to avoid excessive gain
3654 factors in "silent" or almost silent frames. By default, the maximum gain
3655 factor is 10.0, For most inputs the default value should be sufficient and
3656 it usually is not recommended to increase this value. Though, for input
3657 with an extremely low overall volume level, it may be necessary to allow even
3658 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3659 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3660 Instead, a "sigmoid" threshold function will be applied. This way, the
3661 gain factors will smoothly approach the threshold value, but never exceed that
3665 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3666 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3667 This means that the maximum local gain factor for each frame is defined
3668 (only) by the frame's highest magnitude sample. This way, the samples can
3669 be amplified as much as possible without exceeding the maximum signal
3670 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3671 Normalizer can also take into account the frame's root mean square,
3672 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3673 determine the power of a time-varying signal. It is therefore considered
3674 that the RMS is a better approximation of the "perceived loudness" than
3675 just looking at the signal's peak magnitude. Consequently, by adjusting all
3676 frames to a constant RMS value, a uniform "perceived loudness" can be
3677 established. If a target RMS value has been specified, a frame's local gain
3678 factor is defined as the factor that would result in exactly that RMS value.
3679 Note, however, that the maximum local gain factor is still restricted by the
3680 frame's highest magnitude sample, in order to prevent clipping.
3683 Enable channels coupling. By default is enabled.
3684 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3685 amount. This means the same gain factor will be applied to all channels, i.e.
3686 the maximum possible gain factor is determined by the "loudest" channel.
3687 However, in some recordings, it may happen that the volume of the different
3688 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3689 In this case, this option can be used to disable the channel coupling. This way,
3690 the gain factor will be determined independently for each channel, depending
3691 only on the individual channel's highest magnitude sample. This allows for
3692 harmonizing the volume of the different channels.
3695 Enable DC bias correction. By default is disabled.
3696 An audio signal (in the time domain) is a sequence of sample values.
3697 In the Dynamic Audio Normalizer these sample values are represented in the
3698 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3699 audio signal, or "waveform", should be centered around the zero point.
3700 That means if we calculate the mean value of all samples in a file, or in a
3701 single frame, then the result should be 0.0 or at least very close to that
3702 value. If, however, there is a significant deviation of the mean value from
3703 0.0, in either positive or negative direction, this is referred to as a
3704 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3705 Audio Normalizer provides optional DC bias correction.
3706 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3707 the mean value, or "DC correction" offset, of each input frame and subtract
3708 that value from all of the frame's sample values which ensures those samples
3709 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3710 boundaries, the DC correction offset values will be interpolated smoothly
3711 between neighbouring frames.
3713 @item altboundary, b
3714 Enable alternative boundary mode. By default is disabled.
3715 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3716 around each frame. This includes the preceding frames as well as the
3717 subsequent frames. However, for the "boundary" frames, located at the very
3718 beginning and at the very end of the audio file, not all neighbouring
3719 frames are available. In particular, for the first few frames in the audio
3720 file, the preceding frames are not known. And, similarly, for the last few
3721 frames in the audio file, the subsequent frames are not known. Thus, the
3722 question arises which gain factors should be assumed for the missing frames
3723 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3724 to deal with this situation. The default boundary mode assumes a gain factor
3725 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3726 "fade out" at the beginning and at the end of the input, respectively.
3729 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3730 By default, the Dynamic Audio Normalizer does not apply "traditional"
3731 compression. This means that signal peaks will not be pruned and thus the
3732 full dynamic range will be retained within each local neighbourhood. However,
3733 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3734 normalization algorithm with a more "traditional" compression.
3735 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3736 (thresholding) function. If (and only if) the compression feature is enabled,
3737 all input frames will be processed by a soft knee thresholding function prior
3738 to the actual normalization process. Put simply, the thresholding function is
3739 going to prune all samples whose magnitude exceeds a certain threshold value.
3740 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3741 value. Instead, the threshold value will be adjusted for each individual
3743 In general, smaller parameters result in stronger compression, and vice versa.
3744 Values below 3.0 are not recommended, because audible distortion may appear.
3747 Set the target threshold value. This specifies the lowest permissible
3748 magnitude level for the audio input which will be normalized.
3749 If input frame volume is above this value frame will be normalized.
3750 Otherwise frame may not be normalized at all. The default value is set
3751 to 0, which means all input frames will be normalized.
3752 This option is mostly useful if digital noise is not wanted to be amplified.
3755 @subsection Commands
3757 This filter supports the all above options as @ref{commands}.
3761 Make audio easier to listen to on headphones.
3763 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3764 so that when listened to on headphones the stereo image is moved from
3765 inside your head (standard for headphones) to outside and in front of
3766 the listener (standard for speakers).
3772 Apply a two-pole peaking equalisation (EQ) filter. With this
3773 filter, the signal-level at and around a selected frequency can
3774 be increased or decreased, whilst (unlike bandpass and bandreject
3775 filters) that at all other frequencies is unchanged.
3777 In order to produce complex equalisation curves, this filter can
3778 be given several times, each with a different central frequency.
3780 The filter accepts the following options:
3784 Set the filter's central frequency in Hz.
3787 Set method to specify band-width of filter.
3802 Specify the band-width of a filter in width_type units.
3805 Set the required gain or attenuation in dB.
3806 Beware of clipping when using a positive gain.
3809 How much to use filtered signal in output. Default is 1.
3810 Range is between 0 and 1.
3813 Specify which channels to filter, by default all available are filtered.
3816 Normalize biquad coefficients, by default is disabled.
3817 Enabling it will normalize magnitude response at DC to 0dB.
3820 Set transform type of IIR filter.
3829 @subsection Examples
3832 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3834 equalizer=f=1000:t=h:width=200:g=-10
3838 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3840 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3844 @subsection Commands
3846 This filter supports the following commands:
3849 Change equalizer frequency.
3850 Syntax for the command is : "@var{frequency}"
3853 Change equalizer width_type.
3854 Syntax for the command is : "@var{width_type}"
3857 Change equalizer width.
3858 Syntax for the command is : "@var{width}"
3861 Change equalizer gain.
3862 Syntax for the command is : "@var{gain}"
3865 Change equalizer mix.
3866 Syntax for the command is : "@var{mix}"
3869 @section extrastereo
3871 Linearly increases the difference between left and right channels which
3872 adds some sort of "live" effect to playback.
3874 The filter accepts the following options:
3878 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3879 (average of both channels), with 1.0 sound will be unchanged, with
3880 -1.0 left and right channels will be swapped.
3883 Enable clipping. By default is enabled.
3886 @subsection Commands
3888 This filter supports the all above options as @ref{commands}.
3890 @section firequalizer
3891 Apply FIR Equalization using arbitrary frequency response.
3893 The filter accepts the following option:
3897 Set gain curve equation (in dB). The expression can contain variables:
3900 the evaluated frequency
3904 channel number, set to 0 when multichannels evaluation is disabled
3906 channel id, see libavutil/channel_layout.h, set to the first channel id when
3907 multichannels evaluation is disabled
3911 channel_layout, see libavutil/channel_layout.h
3916 @item gain_interpolate(f)
3917 interpolate gain on frequency f based on gain_entry
3918 @item cubic_interpolate(f)
3919 same as gain_interpolate, but smoother
3921 This option is also available as command. Default is @code{gain_interpolate(f)}.
3924 Set gain entry for gain_interpolate function. The expression can
3928 store gain entry at frequency f with value g
3930 This option is also available as command.
3933 Set filter delay in seconds. Higher value means more accurate.
3934 Default is @code{0.01}.
3937 Set filter accuracy in Hz. Lower value means more accurate.
3938 Default is @code{5}.
3941 Set window function. Acceptable values are:
3944 rectangular window, useful when gain curve is already smooth
3946 hann window (default)
3952 3-terms continuous 1st derivative nuttall window
3954 minimum 3-terms discontinuous nuttall window
3956 4-terms continuous 1st derivative nuttall window
3958 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3960 blackman-harris window
3966 If enabled, use fixed number of audio samples. This improves speed when
3967 filtering with large delay. Default is disabled.
3970 Enable multichannels evaluation on gain. Default is disabled.
3973 Enable zero phase mode by subtracting timestamp to compensate delay.
3974 Default is disabled.
3977 Set scale used by gain. Acceptable values are:
3980 linear frequency, linear gain
3982 linear frequency, logarithmic (in dB) gain (default)
3984 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3986 logarithmic frequency, logarithmic gain
3990 Set file for dumping, suitable for gnuplot.
3993 Set scale for dumpfile. Acceptable values are same with scale option.
3997 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3998 Default is disabled.
4001 Enable minimum phase impulse response. Default is disabled.
4004 @subsection Examples
4009 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4012 lowpass at 1000 Hz with gain_entry:
4014 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4017 custom equalization:
4019 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4022 higher delay with zero phase to compensate delay:
4024 firequalizer=delay=0.1:fixed=on:zero_phase=on
4027 lowpass on left channel, highpass on right channel:
4029 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4030 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4035 Apply a flanging effect to the audio.
4037 The filter accepts the following options:
4041 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4044 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4047 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4051 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4052 Default value is 71.
4055 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4058 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4059 Default value is @var{sinusoidal}.
4062 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4063 Default value is 25.
4066 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4067 Default is @var{linear}.
4071 Apply Haas effect to audio.
4073 Note that this makes most sense to apply on mono signals.
4074 With this filter applied to mono signals it give some directionality and
4075 stretches its stereo image.
4077 The filter accepts the following options:
4081 Set input level. By default is @var{1}, or 0dB
4084 Set output level. By default is @var{1}, or 0dB.
4087 Set gain applied to side part of signal. By default is @var{1}.
4090 Set kind of middle source. Can be one of the following:
4100 Pick middle part signal of stereo image.
4103 Pick side part signal of stereo image.
4107 Change middle phase. By default is disabled.
4110 Set left channel delay. By default is @var{2.05} milliseconds.
4113 Set left channel balance. By default is @var{-1}.
4116 Set left channel gain. By default is @var{1}.
4119 Change left phase. By default is disabled.
4122 Set right channel delay. By defaults is @var{2.12} milliseconds.
4125 Set right channel balance. By default is @var{1}.
4128 Set right channel gain. By default is @var{1}.
4131 Change right phase. By default is enabled.
4136 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4137 embedded HDCD codes is expanded into a 20-bit PCM stream.
4139 The filter supports the Peak Extend and Low-level Gain Adjustment features
4140 of HDCD, and detects the Transient Filter flag.
4143 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4146 When using the filter with wav, note the default encoding for wav is 16-bit,
4147 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4148 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4150 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4151 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4154 The filter accepts the following options:
4157 @item disable_autoconvert
4158 Disable any automatic format conversion or resampling in the filter graph.
4160 @item process_stereo
4161 Process the stereo channels together. If target_gain does not match between
4162 channels, consider it invalid and use the last valid target_gain.
4165 Set the code detect timer period in ms.
4168 Always extend peaks above -3dBFS even if PE isn't signaled.
4171 Replace audio with a solid tone and adjust the amplitude to signal some
4172 specific aspect of the decoding process. The output file can be loaded in
4173 an audio editor alongside the original to aid analysis.
4175 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4182 Gain adjustment level at each sample
4184 Samples where peak extend occurs
4186 Samples where the code detect timer is active
4188 Samples where the target gain does not match between channels
4194 Apply head-related transfer functions (HRTFs) to create virtual
4195 loudspeakers around the user for binaural listening via headphones.
4196 The HRIRs are provided via additional streams, for each channel
4197 one stereo input stream is needed.
4199 The filter accepts the following options:
4203 Set mapping of input streams for convolution.
4204 The argument is a '|'-separated list of channel names in order as they
4205 are given as additional stream inputs for filter.
4206 This also specify number of input streams. Number of input streams
4207 must be not less than number of channels in first stream plus one.
4210 Set gain applied to audio. Value is in dB. Default is 0.
4213 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4214 processing audio in time domain which is slow.
4215 @var{freq} is processing audio in frequency domain which is fast.
4216 Default is @var{freq}.
4219 Set custom gain for LFE channels. Value is in dB. Default is 0.
4222 Set size of frame in number of samples which will be processed at once.
4223 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4226 Set format of hrir stream.
4227 Default value is @var{stereo}. Alternative value is @var{multich}.
4228 If value is set to @var{stereo}, number of additional streams should
4229 be greater or equal to number of input channels in first input stream.
4230 Also each additional stream should have stereo number of channels.
4231 If value is set to @var{multich}, number of additional streams should
4232 be exactly one. Also number of input channels of additional stream
4233 should be equal or greater than twice number of channels of first input
4237 @subsection Examples
4241 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4242 each amovie filter use stereo file with IR coefficients as input.
4243 The files give coefficients for each position of virtual loudspeaker:
4246 -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"
4251 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4252 but now in @var{multich} @var{hrir} format.
4254 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"
4261 Apply a high-pass filter with 3dB point frequency.
4262 The filter can be either single-pole, or double-pole (the default).
4263 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4265 The filter accepts the following options:
4269 Set frequency in Hz. Default is 3000.
4272 Set number of poles. Default is 2.
4275 Set method to specify band-width of filter.
4290 Specify the band-width of a filter in width_type units.
4291 Applies only to double-pole filter.
4292 The default is 0.707q and gives a Butterworth response.
4295 How much to use filtered signal in output. Default is 1.
4296 Range is between 0 and 1.
4299 Specify which channels to filter, by default all available are filtered.
4302 Normalize biquad coefficients, by default is disabled.
4303 Enabling it will normalize magnitude response at DC to 0dB.
4306 Set transform type of IIR filter.
4315 @subsection Commands
4317 This filter supports the following commands:
4320 Change highpass frequency.
4321 Syntax for the command is : "@var{frequency}"
4324 Change highpass width_type.
4325 Syntax for the command is : "@var{width_type}"
4328 Change highpass width.
4329 Syntax for the command is : "@var{width}"
4332 Change highpass mix.
4333 Syntax for the command is : "@var{mix}"
4338 Join multiple input streams into one multi-channel stream.
4340 It accepts the following parameters:
4344 The number of input streams. It defaults to 2.
4346 @item channel_layout
4347 The desired output channel layout. It defaults to stereo.
4350 Map channels from inputs to output. The argument is a '|'-separated list of
4351 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4352 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4353 can be either the name of the input channel (e.g. FL for front left) or its
4354 index in the specified input stream. @var{out_channel} is the name of the output
4358 The filter will attempt to guess the mappings when they are not specified
4359 explicitly. It does so by first trying to find an unused matching input channel
4360 and if that fails it picks the first unused input channel.
4362 Join 3 inputs (with properly set channel layouts):
4364 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4367 Build a 5.1 output from 6 single-channel streams:
4369 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4370 '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'
4376 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4378 To enable compilation of this filter you need to configure FFmpeg with
4379 @code{--enable-ladspa}.
4383 Specifies the name of LADSPA plugin library to load. If the environment
4384 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4385 each one of the directories specified by the colon separated list in
4386 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4387 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4388 @file{/usr/lib/ladspa/}.
4391 Specifies the plugin within the library. Some libraries contain only
4392 one plugin, but others contain many of them. If this is not set filter
4393 will list all available plugins within the specified library.
4396 Set the '|' separated list of controls which are zero or more floating point
4397 values that determine the behavior of the loaded plugin (for example delay,
4399 Controls need to be defined using the following syntax:
4400 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4401 @var{valuei} is the value set on the @var{i}-th control.
4402 Alternatively they can be also defined using the following syntax:
4403 @var{value0}|@var{value1}|@var{value2}|..., where
4404 @var{valuei} is the value set on the @var{i}-th control.
4405 If @option{controls} is set to @code{help}, all available controls and
4406 their valid ranges are printed.
4408 @item sample_rate, s
4409 Specify the sample rate, default to 44100. Only used if plugin have
4413 Set the number of samples per channel per each output frame, default
4414 is 1024. Only used if plugin have zero inputs.
4417 Set the minimum duration of the sourced audio. See
4418 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4419 for the accepted syntax.
4420 Note that the resulting duration may be greater than the specified duration,
4421 as the generated audio is always cut at the end of a complete frame.
4422 If not specified, or the expressed duration is negative, the audio is
4423 supposed to be generated forever.
4424 Only used if plugin have zero inputs.
4427 Enable latency compensation, by default is disabled.
4428 Only used if plugin have inputs.
4431 @subsection Examples
4435 List all available plugins within amp (LADSPA example plugin) library:
4441 List all available controls and their valid ranges for @code{vcf_notch}
4442 plugin from @code{VCF} library:
4444 ladspa=f=vcf:p=vcf_notch:c=help
4448 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4451 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4455 Add reverberation to the audio using TAP-plugins
4456 (Tom's Audio Processing plugins):
4458 ladspa=file=tap_reverb:tap_reverb
4462 Generate white noise, with 0.2 amplitude:
4464 ladspa=file=cmt:noise_source_white:c=c0=.2
4468 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4469 @code{C* Audio Plugin Suite} (CAPS) library:
4471 ladspa=file=caps:Click:c=c1=20'
4475 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4477 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4481 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4482 @code{SWH Plugins} collection:
4484 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4488 Attenuate low frequencies using Multiband EQ from Steve Harris
4489 @code{SWH Plugins} collection:
4491 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4495 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4498 ladspa=caps:Narrower
4502 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4504 ladspa=caps:White:.2
4508 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4510 ladspa=caps:Fractal:c=c1=1
4514 Dynamic volume normalization using @code{VLevel} plugin:
4516 ladspa=vlevel-ladspa:vlevel_mono
4520 @subsection Commands
4522 This filter supports the following commands:
4525 Modify the @var{N}-th control value.
4527 If the specified value is not valid, it is ignored and prior one is kept.
4532 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4533 Support for both single pass (livestreams, files) and double pass (files) modes.
4534 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4535 detect true peaks, the audio stream will be upsampled to 192 kHz.
4536 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4538 The filter accepts the following options:
4542 Set integrated loudness target.
4543 Range is -70.0 - -5.0. Default value is -24.0.
4546 Set loudness range target.
4547 Range is 1.0 - 20.0. Default value is 7.0.
4550 Set maximum true peak.
4551 Range is -9.0 - +0.0. Default value is -2.0.
4553 @item measured_I, measured_i
4554 Measured IL of input file.
4555 Range is -99.0 - +0.0.
4557 @item measured_LRA, measured_lra
4558 Measured LRA of input file.
4559 Range is 0.0 - 99.0.
4561 @item measured_TP, measured_tp
4562 Measured true peak of input file.
4563 Range is -99.0 - +99.0.
4565 @item measured_thresh
4566 Measured threshold of input file.
4567 Range is -99.0 - +0.0.
4570 Set offset gain. Gain is applied before the true-peak limiter.
4571 Range is -99.0 - +99.0. Default is +0.0.
4574 Normalize by linearly scaling the source audio.
4575 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4576 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4577 be lower than source LRA and the change in integrated loudness shouldn't
4578 result in a true peak which exceeds the target TP. If any of these
4579 conditions aren't met, normalization mode will revert to @var{dynamic}.
4580 Options are @code{true} or @code{false}. Default is @code{true}.
4583 Treat mono input files as "dual-mono". If a mono file is intended for playback
4584 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4585 If set to @code{true}, this option will compensate for this effect.
4586 Multi-channel input files are not affected by this option.
4587 Options are true or false. Default is false.
4590 Set print format for stats. Options are summary, json, or none.
4591 Default value is none.
4596 Apply a low-pass filter with 3dB point frequency.
4597 The filter can be either single-pole or double-pole (the default).
4598 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4600 The filter accepts the following options:
4604 Set frequency in Hz. Default is 500.
4607 Set number of poles. Default is 2.
4610 Set method to specify band-width of filter.
4625 Specify the band-width of a filter in width_type units.
4626 Applies only to double-pole filter.
4627 The default is 0.707q and gives a Butterworth response.
4630 How much to use filtered signal in output. Default is 1.
4631 Range is between 0 and 1.
4634 Specify which channels to filter, by default all available are filtered.
4637 Normalize biquad coefficients, by default is disabled.
4638 Enabling it will normalize magnitude response at DC to 0dB.
4641 Set transform type of IIR filter.
4650 @subsection Examples
4653 Lowpass only LFE channel, it LFE is not present it does nothing:
4659 @subsection Commands
4661 This filter supports the following commands:
4664 Change lowpass frequency.
4665 Syntax for the command is : "@var{frequency}"
4668 Change lowpass width_type.
4669 Syntax for the command is : "@var{width_type}"
4672 Change lowpass width.
4673 Syntax for the command is : "@var{width}"
4677 Syntax for the command is : "@var{mix}"
4682 Load a LV2 (LADSPA Version 2) plugin.
4684 To enable compilation of this filter you need to configure FFmpeg with
4685 @code{--enable-lv2}.
4689 Specifies the plugin URI. You may need to escape ':'.
4692 Set the '|' separated list of controls which are zero or more floating point
4693 values that determine the behavior of the loaded plugin (for example delay,
4695 If @option{controls} is set to @code{help}, all available controls and
4696 their valid ranges are printed.
4698 @item sample_rate, s
4699 Specify the sample rate, default to 44100. Only used if plugin have
4703 Set the number of samples per channel per each output frame, default
4704 is 1024. Only used if plugin have zero inputs.
4707 Set the minimum duration of the sourced audio. See
4708 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4709 for the accepted syntax.
4710 Note that the resulting duration may be greater than the specified duration,
4711 as the generated audio is always cut at the end of a complete frame.
4712 If not specified, or the expressed duration is negative, the audio is
4713 supposed to be generated forever.
4714 Only used if plugin have zero inputs.
4717 @subsection Examples
4721 Apply bass enhancer plugin from Calf:
4723 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4727 Apply vinyl plugin from Calf:
4729 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4733 Apply bit crusher plugin from ArtyFX:
4735 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4740 Multiband Compress or expand the audio's dynamic range.
4742 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4743 This is akin to the crossover of a loudspeaker, and results in flat frequency
4744 response when absent compander action.
4746 It accepts the following parameters:
4750 This option syntax is:
4751 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4752 For explanation of each item refer to compand filter documentation.
4758 Mix channels with specific gain levels. The filter accepts the output
4759 channel layout followed by a set of channels definitions.
4761 This filter is also designed to efficiently remap the channels of an audio
4764 The filter accepts parameters of the form:
4765 "@var{l}|@var{outdef}|@var{outdef}|..."
4769 output channel layout or number of channels
4772 output channel specification, of the form:
4773 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4776 output channel to define, either a channel name (FL, FR, etc.) or a channel
4777 number (c0, c1, etc.)
4780 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4783 input channel to use, see out_name for details; it is not possible to mix
4784 named and numbered input channels
4787 If the `=' in a channel specification is replaced by `<', then the gains for
4788 that specification will be renormalized so that the total is 1, thus
4789 avoiding clipping noise.
4791 @subsection Mixing examples
4793 For example, if you want to down-mix from stereo to mono, but with a bigger
4794 factor for the left channel:
4796 pan=1c|c0=0.9*c0+0.1*c1
4799 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4800 7-channels surround:
4802 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4805 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4806 that should be preferred (see "-ac" option) unless you have very specific
4809 @subsection Remapping examples
4811 The channel remapping will be effective if, and only if:
4814 @item gain coefficients are zeroes or ones,
4815 @item only one input per channel output,
4818 If all these conditions are satisfied, the filter will notify the user ("Pure
4819 channel mapping detected"), and use an optimized and lossless method to do the
4822 For example, if you have a 5.1 source and want a stereo audio stream by
4823 dropping the extra channels:
4825 pan="stereo| c0=FL | c1=FR"
4828 Given the same source, you can also switch front left and front right channels
4829 and keep the input channel layout:
4831 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4834 If the input is a stereo audio stream, you can mute the front left channel (and
4835 still keep the stereo channel layout) with:
4840 Still with a stereo audio stream input, you can copy the right channel in both
4841 front left and right:
4843 pan="stereo| c0=FR | c1=FR"
4848 ReplayGain scanner filter. This filter takes an audio stream as an input and
4849 outputs it unchanged.
4850 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4854 Convert the audio sample format, sample rate and channel layout. It is
4855 not meant to be used directly.
4858 Apply time-stretching and pitch-shifting with librubberband.
4860 To enable compilation of this filter, you need to configure FFmpeg with
4861 @code{--enable-librubberband}.
4863 The filter accepts the following options:
4867 Set tempo scale factor.
4870 Set pitch scale factor.
4873 Set transients detector.
4874 Possible values are:
4883 Possible values are:
4892 Possible values are:
4899 Set processing window size.
4900 Possible values are:
4909 Possible values are:
4916 Enable formant preservation when shift pitching.
4917 Possible values are:
4925 Possible values are:
4934 Possible values are:
4941 @subsection Commands
4943 This filter supports the following commands:
4946 Change filter tempo scale factor.
4947 Syntax for the command is : "@var{tempo}"
4950 Change filter pitch scale factor.
4951 Syntax for the command is : "@var{pitch}"
4954 @section sidechaincompress
4956 This filter acts like normal compressor but has the ability to compress
4957 detected signal using second input signal.
4958 It needs two input streams and returns one output stream.
4959 First input stream will be processed depending on second stream signal.
4960 The filtered signal then can be filtered with other filters in later stages of
4961 processing. See @ref{pan} and @ref{amerge} filter.
4963 The filter accepts the following options:
4967 Set input gain. Default is 1. Range is between 0.015625 and 64.
4970 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4971 Default is @code{downward}.
4974 If a signal of second stream raises above this level it will affect the gain
4975 reduction of first stream.
4976 By default is 0.125. Range is between 0.00097563 and 1.
4979 Set a ratio about which the signal is reduced. 1:2 means that if the level
4980 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4981 Default is 2. Range is between 1 and 20.
4984 Amount of milliseconds the signal has to rise above the threshold before gain
4985 reduction starts. Default is 20. Range is between 0.01 and 2000.
4988 Amount of milliseconds the signal has to fall below the threshold before
4989 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4992 Set the amount by how much signal will be amplified after processing.
4993 Default is 1. Range is from 1 to 64.
4996 Curve the sharp knee around the threshold to enter gain reduction more softly.
4997 Default is 2.82843. Range is between 1 and 8.
5000 Choose if the @code{average} level between all channels of side-chain stream
5001 or the louder(@code{maximum}) channel of side-chain stream affects the
5002 reduction. Default is @code{average}.
5005 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5006 of @code{rms}. Default is @code{rms} which is mainly smoother.
5009 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5012 How much to use compressed signal in output. Default is 1.
5013 Range is between 0 and 1.
5016 @subsection Commands
5018 This filter supports the all above options as @ref{commands}.
5020 @subsection Examples
5024 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5025 depending on the signal of 2nd input and later compressed signal to be
5026 merged with 2nd input:
5028 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5032 @section sidechaingate
5034 A sidechain gate acts like a normal (wideband) gate but has the ability to
5035 filter the detected signal before sending it to the gain reduction stage.
5036 Normally a gate uses the full range signal to detect a level above the
5038 For example: If you cut all lower frequencies from your sidechain signal
5039 the gate will decrease the volume of your track only if not enough highs
5040 appear. With this technique you are able to reduce the resonation of a
5041 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5043 It needs two input streams and returns one output stream.
5044 First input stream will be processed depending on second stream signal.
5046 The filter accepts the following options:
5050 Set input level before filtering.
5051 Default is 1. Allowed range is from 0.015625 to 64.
5054 Set the mode of operation. Can be @code{upward} or @code{downward}.
5055 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5056 will be amplified, expanding dynamic range in upward direction.
5057 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5060 Set the level of gain reduction when the signal is below the threshold.
5061 Default is 0.06125. Allowed range is from 0 to 1.
5062 Setting this to 0 disables reduction and then filter behaves like expander.
5065 If a signal rises above this level the gain reduction is released.
5066 Default is 0.125. Allowed range is from 0 to 1.
5069 Set a ratio about which the signal is reduced.
5070 Default is 2. Allowed range is from 1 to 9000.
5073 Amount of milliseconds the signal has to rise above the threshold before gain
5075 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5078 Amount of milliseconds the signal has to fall below the threshold before the
5079 reduction is increased again. Default is 250 milliseconds.
5080 Allowed range is from 0.01 to 9000.
5083 Set amount of amplification of signal after processing.
5084 Default is 1. Allowed range is from 1 to 64.
5087 Curve the sharp knee around the threshold to enter gain reduction more softly.
5088 Default is 2.828427125. Allowed range is from 1 to 8.
5091 Choose if exact signal should be taken for detection or an RMS like one.
5092 Default is rms. Can be peak or rms.
5095 Choose if the average level between all channels or the louder channel affects
5097 Default is average. Can be average or maximum.
5100 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5103 @subsection Commands
5105 This filter supports the all above options as @ref{commands}.
5107 @section silencedetect
5109 Detect silence in an audio stream.
5111 This filter logs a message when it detects that the input audio volume is less
5112 or equal to a noise tolerance value for a duration greater or equal to the
5113 minimum detected noise duration.
5115 The printed times and duration are expressed in seconds. The
5116 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5117 is set on the first frame whose timestamp equals or exceeds the detection
5118 duration and it contains the timestamp of the first frame of the silence.
5120 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5121 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5122 keys are set on the first frame after the silence. If @option{mono} is
5123 enabled, and each channel is evaluated separately, the @code{.X}
5124 suffixed keys are used, and @code{X} corresponds to the channel number.
5126 The filter accepts the following options:
5130 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5131 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5134 Set silence duration until notification (default is 2 seconds). See
5135 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5136 for the accepted syntax.
5139 Process each channel separately, instead of combined. By default is disabled.
5142 @subsection Examples
5146 Detect 5 seconds of silence with -50dB noise tolerance:
5148 silencedetect=n=-50dB:d=5
5152 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5153 tolerance in @file{silence.mp3}:
5155 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5159 @section silenceremove
5161 Remove silence from the beginning, middle or end of the audio.
5163 The filter accepts the following options:
5167 This value is used to indicate if audio should be trimmed at beginning of
5168 the audio. A value of zero indicates no silence should be trimmed from the
5169 beginning. When specifying a non-zero value, it trims audio up until it
5170 finds non-silence. Normally, when trimming silence from beginning of audio
5171 the @var{start_periods} will be @code{1} but it can be increased to higher
5172 values to trim all audio up to specific count of non-silence periods.
5173 Default value is @code{0}.
5175 @item start_duration
5176 Specify the amount of time that non-silence must be detected before it stops
5177 trimming audio. By increasing the duration, bursts of noises can be treated
5178 as silence and trimmed off. Default value is @code{0}.
5180 @item start_threshold
5181 This indicates what sample value should be treated as silence. For digital
5182 audio, a value of @code{0} may be fine but for audio recorded from analog,
5183 you may wish to increase the value to account for background noise.
5184 Can be specified in dB (in case "dB" is appended to the specified value)
5185 or amplitude ratio. Default value is @code{0}.
5188 Specify max duration of silence at beginning that will be kept after
5189 trimming. Default is 0, which is equal to trimming all samples detected
5193 Specify mode of detection of silence end in start of multi-channel audio.
5194 Can be @var{any} or @var{all}. Default is @var{any}.
5195 With @var{any}, any sample that is detected as non-silence will cause
5196 stopped trimming of silence.
5197 With @var{all}, only if all channels are detected as non-silence will cause
5198 stopped trimming of silence.
5201 Set the count for trimming silence from the end of audio.
5202 To remove silence from the middle of a file, specify a @var{stop_periods}
5203 that is negative. This value is then treated as a positive value and is
5204 used to indicate the effect should restart processing as specified by
5205 @var{start_periods}, making it suitable for removing periods of silence
5206 in the middle of the audio.
5207 Default value is @code{0}.
5210 Specify a duration of silence that must exist before audio is not copied any
5211 more. By specifying a higher duration, silence that is wanted can be left in
5213 Default value is @code{0}.
5215 @item stop_threshold
5216 This is the same as @option{start_threshold} but for trimming silence from
5218 Can be specified in dB (in case "dB" is appended to the specified value)
5219 or amplitude ratio. Default value is @code{0}.
5222 Specify max duration of silence at end that will be kept after
5223 trimming. Default is 0, which is equal to trimming all samples detected
5227 Specify mode of detection of silence start in end of multi-channel audio.
5228 Can be @var{any} or @var{all}. Default is @var{any}.
5229 With @var{any}, any sample that is detected as non-silence will cause
5230 stopped trimming of silence.
5231 With @var{all}, only if all channels are detected as non-silence will cause
5232 stopped trimming of silence.
5235 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5236 and works better with digital silence which is exactly 0.
5237 Default value is @code{rms}.
5240 Set duration in number of seconds used to calculate size of window in number
5241 of samples for detecting silence.
5242 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5245 @subsection Examples
5249 The following example shows how this filter can be used to start a recording
5250 that does not contain the delay at the start which usually occurs between
5251 pressing the record button and the start of the performance:
5253 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5257 Trim all silence encountered from beginning to end where there is more than 1
5258 second of silence in audio:
5260 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5264 Trim all digital silence samples, using peak detection, from beginning to end
5265 where there is more than 0 samples of digital silence in audio and digital
5266 silence is detected in all channels at same positions in stream:
5268 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5274 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5275 loudspeakers around the user for binaural listening via headphones (audio
5276 formats up to 9 channels supported).
5277 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5278 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5279 Austrian Academy of Sciences.
5281 To enable compilation of this filter you need to configure FFmpeg with
5282 @code{--enable-libmysofa}.
5284 The filter accepts the following options:
5288 Set the SOFA file used for rendering.
5291 Set gain applied to audio. Value is in dB. Default is 0.
5294 Set rotation of virtual loudspeakers in deg. Default is 0.
5297 Set elevation of virtual speakers in deg. Default is 0.
5300 Set distance in meters between loudspeakers and the listener with near-field
5301 HRTFs. Default is 1.
5304 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5305 processing audio in time domain which is slow.
5306 @var{freq} is processing audio in frequency domain which is fast.
5307 Default is @var{freq}.
5310 Set custom positions of virtual loudspeakers. Syntax for this option is:
5311 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5312 Each virtual loudspeaker is described with short channel name following with
5313 azimuth and elevation in degrees.
5314 Each virtual loudspeaker description is separated by '|'.
5315 For example to override front left and front right channel positions use:
5316 'speakers=FL 45 15|FR 345 15'.
5317 Descriptions with unrecognised channel names are ignored.
5320 Set custom gain for LFE channels. Value is in dB. Default is 0.
5323 Set custom frame size in number of samples. Default is 1024.
5324 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5325 is set to @var{freq}.
5328 Should all IRs be normalized upon importing SOFA file.
5329 By default is enabled.
5332 Should nearest IRs be interpolated with neighbor IRs if exact position
5333 does not match. By default is disabled.
5336 Minphase all IRs upon loading of SOFA file. By default is disabled.
5339 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5342 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5345 @subsection Examples
5349 Using ClubFritz6 sofa file:
5351 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5355 Using ClubFritz12 sofa file and bigger radius with small rotation:
5357 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5361 Similar as above but with custom speaker positions for front left, front right, back left and back right
5362 and also with custom gain:
5364 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5371 This filter expands or compresses each half-cycle of audio samples
5372 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5373 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5375 The filter accepts the following options:
5379 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5380 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5383 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5384 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5385 would be such that local peak value reaches target peak value but never to surpass it and that
5386 ratio between new and previous peak value does not surpass this option value.
5388 @item compression, c
5389 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5390 This option controls maximum local half-cycle of samples compression. This option is used
5391 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5392 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5393 that peak's half-cycle will be compressed by current compression factor.
5396 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5397 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5398 Any half-cycle samples with their local peak value below or same as this option value will be
5399 compressed by current compression factor, otherwise, if greater than threshold value they will be
5400 expanded with expansion factor so that it could reach peak target value but never surpass it.
5403 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5404 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5405 each new half-cycle until it reaches @option{expansion} value.
5406 Setting this options too high may lead to distortions.
5409 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5410 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5411 each new half-cycle until it reaches @option{compression} value.
5414 Specify which channels to filter, by default all available channels are filtered.
5417 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5418 option. When enabled any half-cycle of samples with their local peak value below or same as
5419 @option{threshold} option will be expanded otherwise it will be compressed.
5422 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5423 When disabled each filtered channel gain calculation is independent, otherwise when this option
5424 is enabled the minimum of all possible gains for each filtered channel is used.
5427 @subsection Commands
5429 This filter supports the all above options as @ref{commands}.
5431 @section stereotools
5433 This filter has some handy utilities to manage stereo signals, for converting
5434 M/S stereo recordings to L/R signal while having control over the parameters
5435 or spreading the stereo image of master track.
5437 The filter accepts the following options:
5441 Set input level before filtering for both channels. Defaults is 1.
5442 Allowed range is from 0.015625 to 64.
5445 Set output level after filtering for both channels. Defaults is 1.
5446 Allowed range is from 0.015625 to 64.
5449 Set input balance between both channels. Default is 0.
5450 Allowed range is from -1 to 1.
5453 Set output balance between both channels. Default is 0.
5454 Allowed range is from -1 to 1.
5457 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5458 clipping. Disabled by default.
5461 Mute the left channel. Disabled by default.
5464 Mute the right channel. Disabled by default.
5467 Change the phase of the left channel. Disabled by default.
5470 Change the phase of the right channel. Disabled by default.
5473 Set stereo mode. Available values are:
5477 Left/Right to Left/Right, this is default.
5480 Left/Right to Mid/Side.
5483 Mid/Side to Left/Right.
5486 Left/Right to Left/Left.
5489 Left/Right to Right/Right.
5492 Left/Right to Left + Right.
5495 Left/Right to Right/Left.
5498 Mid/Side to Left/Left.
5501 Mid/Side to Right/Right.
5505 Set level of side signal. Default is 1.
5506 Allowed range is from 0.015625 to 64.
5509 Set balance of side signal. Default is 0.
5510 Allowed range is from -1 to 1.
5513 Set level of the middle signal. Default is 1.
5514 Allowed range is from 0.015625 to 64.
5517 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5520 Set stereo base between mono and inversed channels. Default is 0.
5521 Allowed range is from -1 to 1.
5524 Set delay in milliseconds how much to delay left from right channel and
5525 vice versa. Default is 0. Allowed range is from -20 to 20.
5528 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5531 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5533 @item bmode_in, bmode_out
5534 Set balance mode for balance_in/balance_out option.
5536 Can be one of the following:
5540 Classic balance mode. Attenuate one channel at time.
5541 Gain is raised up to 1.
5544 Similar as classic mode above but gain is raised up to 2.
5547 Equal power distribution, from -6dB to +6dB range.
5551 @subsection Examples
5555 Apply karaoke like effect:
5557 stereotools=mlev=0.015625
5561 Convert M/S signal to L/R:
5563 "stereotools=mode=ms>lr"
5567 @section stereowiden
5569 This filter enhance the stereo effect by suppressing signal common to both
5570 channels and by delaying the signal of left into right and vice versa,
5571 thereby widening the stereo effect.
5573 The filter accepts the following options:
5577 Time in milliseconds of the delay of left signal into right and vice versa.
5578 Default is 20 milliseconds.
5581 Amount of gain in delayed signal into right and vice versa. Gives a delay
5582 effect of left signal in right output and vice versa which gives widening
5583 effect. Default is 0.3.
5586 Cross feed of left into right with inverted phase. This helps in suppressing
5587 the mono. If the value is 1 it will cancel all the signal common to both
5588 channels. Default is 0.3.
5591 Set level of input signal of original channel. Default is 0.8.
5594 @subsection Commands
5596 This filter supports the all above options except @code{delay} as @ref{commands}.
5598 @section superequalizer
5599 Apply 18 band equalizer.
5601 The filter accepts the following options:
5608 Set 131Hz band gain.
5610 Set 185Hz band gain.
5612 Set 262Hz band gain.
5614 Set 370Hz band gain.
5616 Set 523Hz band gain.
5618 Set 740Hz band gain.
5620 Set 1047Hz band gain.
5622 Set 1480Hz band gain.
5624 Set 2093Hz band gain.
5626 Set 2960Hz band gain.
5628 Set 4186Hz band gain.
5630 Set 5920Hz band gain.
5632 Set 8372Hz band gain.
5634 Set 11840Hz band gain.
5636 Set 16744Hz band gain.
5638 Set 20000Hz band gain.
5642 Apply audio surround upmix filter.
5644 This filter allows to produce multichannel output from audio stream.
5646 The filter accepts the following options:
5650 Set output channel layout. By default, this is @var{5.1}.
5652 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5653 for the required syntax.
5656 Set input channel layout. By default, this is @var{stereo}.
5658 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5659 for the required syntax.
5662 Set input volume level. By default, this is @var{1}.
5665 Set output volume level. By default, this is @var{1}.
5668 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5671 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5674 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5677 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5678 In @var{add} mode, LFE channel is created from input audio and added to output.
5679 In @var{sub} mode, LFE channel is created from input audio and added to output but
5680 also all non-LFE output channels are subtracted with output LFE channel.
5683 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5684 Default is @var{90}.
5687 Set front center input volume. By default, this is @var{1}.
5690 Set front center output volume. By default, this is @var{1}.
5693 Set front left input volume. By default, this is @var{1}.
5696 Set front left output volume. By default, this is @var{1}.
5699 Set front right input volume. By default, this is @var{1}.
5702 Set front right output volume. By default, this is @var{1}.
5705 Set side left input volume. By default, this is @var{1}.
5708 Set side left output volume. By default, this is @var{1}.
5711 Set side right input volume. By default, this is @var{1}.
5714 Set side right output volume. By default, this is @var{1}.
5717 Set back left input volume. By default, this is @var{1}.
5720 Set back left output volume. By default, this is @var{1}.
5723 Set back right input volume. By default, this is @var{1}.
5726 Set back right output volume. By default, this is @var{1}.
5729 Set back center input volume. By default, this is @var{1}.
5732 Set back center output volume. By default, this is @var{1}.
5735 Set LFE input volume. By default, this is @var{1}.
5738 Set LFE output volume. By default, this is @var{1}.
5741 Set spread usage of stereo image across X axis for all channels.
5744 Set spread usage of stereo image across Y axis for all channels.
5746 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5747 Set spread usage of stereo image across X axis for each channel.
5749 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5750 Set spread usage of stereo image across Y axis for each channel.
5753 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5756 Set window function.
5758 It accepts the following values:
5781 Default is @code{hann}.
5784 Set window overlap. If set to 1, the recommended overlap for selected
5785 window function will be picked. Default is @code{0.5}.
5788 @section treble, highshelf
5790 Boost or cut treble (upper) frequencies of the audio using a two-pole
5791 shelving filter with a response similar to that of a standard
5792 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5794 The filter accepts the following options:
5798 Give the gain at whichever is the lower of ~22 kHz and the
5799 Nyquist frequency. Its useful range is about -20 (for a large cut)
5800 to +20 (for a large boost). Beware of clipping when using a positive gain.
5803 Set the filter's central frequency and so can be used
5804 to extend or reduce the frequency range to be boosted or cut.
5805 The default value is @code{3000} Hz.
5808 Set method to specify band-width of filter.
5823 Determine how steep is the filter's shelf transition.
5826 How much to use filtered signal in output. Default is 1.
5827 Range is between 0 and 1.
5830 Specify which channels to filter, by default all available are filtered.
5833 Normalize biquad coefficients, by default is disabled.
5834 Enabling it will normalize magnitude response at DC to 0dB.
5837 Set transform type of IIR filter.
5846 @subsection Commands
5848 This filter supports the following commands:
5851 Change treble frequency.
5852 Syntax for the command is : "@var{frequency}"
5855 Change treble width_type.
5856 Syntax for the command is : "@var{width_type}"
5859 Change treble width.
5860 Syntax for the command is : "@var{width}"
5864 Syntax for the command is : "@var{gain}"
5868 Syntax for the command is : "@var{mix}"
5873 Sinusoidal amplitude modulation.
5875 The filter accepts the following options:
5879 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5880 (20 Hz or lower) will result in a tremolo effect.
5881 This filter may also be used as a ring modulator by specifying
5882 a modulation frequency higher than 20 Hz.
5883 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5886 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5887 Default value is 0.5.
5892 Sinusoidal phase modulation.
5894 The filter accepts the following options:
5898 Modulation frequency in Hertz.
5899 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5902 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5903 Default value is 0.5.
5908 Adjust the input audio volume.
5910 It accepts the following parameters:
5914 Set audio volume expression.
5916 Output values are clipped to the maximum value.
5918 The output audio volume is given by the relation:
5920 @var{output_volume} = @var{volume} * @var{input_volume}
5923 The default value for @var{volume} is "1.0".
5926 This parameter represents the mathematical precision.
5928 It determines which input sample formats will be allowed, which affects the
5929 precision of the volume scaling.
5933 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5935 32-bit floating-point; this limits input sample format to FLT. (default)
5937 64-bit floating-point; this limits input sample format to DBL.
5941 Choose the behaviour on encountering ReplayGain side data in input frames.
5945 Remove ReplayGain side data, ignoring its contents (the default).
5948 Ignore ReplayGain side data, but leave it in the frame.
5951 Prefer the track gain, if present.
5954 Prefer the album gain, if present.
5957 @item replaygain_preamp
5958 Pre-amplification gain in dB to apply to the selected replaygain gain.
5960 Default value for @var{replaygain_preamp} is 0.0.
5962 @item replaygain_noclip
5963 Prevent clipping by limiting the gain applied.
5965 Default value for @var{replaygain_noclip} is 1.
5968 Set when the volume expression is evaluated.
5970 It accepts the following values:
5973 only evaluate expression once during the filter initialization, or
5974 when the @samp{volume} command is sent
5977 evaluate expression for each incoming frame
5980 Default value is @samp{once}.
5983 The volume expression can contain the following parameters.
5987 frame number (starting at zero)
5990 @item nb_consumed_samples
5991 number of samples consumed by the filter
5993 number of samples in the current frame
5995 original frame position in the file
6001 PTS at start of stream
6003 time at start of stream
6009 last set volume value
6012 Note that when @option{eval} is set to @samp{once} only the
6013 @var{sample_rate} and @var{tb} variables are available, all other
6014 variables will evaluate to NAN.
6016 @subsection Commands
6018 This filter supports the following commands:
6021 Modify the volume expression.
6022 The command accepts the same syntax of the corresponding option.
6024 If the specified expression is not valid, it is kept at its current
6028 @subsection Examples
6032 Halve the input audio volume:
6036 volume=volume=-6.0206dB
6039 In all the above example the named key for @option{volume} can be
6040 omitted, for example like in:
6046 Increase input audio power by 6 decibels using fixed-point precision:
6048 volume=volume=6dB:precision=fixed
6052 Fade volume after time 10 with an annihilation period of 5 seconds:
6054 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6058 @section volumedetect
6060 Detect the volume of the input video.
6062 The filter has no parameters. The input is not modified. Statistics about
6063 the volume will be printed in the log when the input stream end is reached.
6065 In particular it will show the mean volume (root mean square), maximum
6066 volume (on a per-sample basis), and the beginning of a histogram of the
6067 registered volume values (from the maximum value to a cumulated 1/1000 of
6070 All volumes are in decibels relative to the maximum PCM value.
6072 @subsection Examples
6074 Here is an excerpt of the output:
6076 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6077 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6078 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6079 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6080 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6081 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6082 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6083 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6084 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6090 The mean square energy is approximately -27 dB, or 10^-2.7.
6092 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6094 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6097 In other words, raising the volume by +4 dB does not cause any clipping,
6098 raising it by +5 dB causes clipping for 6 samples, etc.
6100 @c man end AUDIO FILTERS
6102 @chapter Audio Sources
6103 @c man begin AUDIO SOURCES
6105 Below is a description of the currently available audio sources.
6109 Buffer audio frames, and make them available to the filter chain.
6111 This source is mainly intended for a programmatic use, in particular
6112 through the interface defined in @file{libavfilter/buffersrc.h}.
6114 It accepts the following parameters:
6118 The timebase which will be used for timestamps of submitted frames. It must be
6119 either a floating-point number or in @var{numerator}/@var{denominator} form.
6122 The sample rate of the incoming audio buffers.
6125 The sample format of the incoming audio buffers.
6126 Either a sample format name or its corresponding integer representation from
6127 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6129 @item channel_layout
6130 The channel layout of the incoming audio buffers.
6131 Either a channel layout name from channel_layout_map in
6132 @file{libavutil/channel_layout.c} or its corresponding integer representation
6133 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6136 The number of channels of the incoming audio buffers.
6137 If both @var{channels} and @var{channel_layout} are specified, then they
6142 @subsection Examples
6145 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6148 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6149 Since the sample format with name "s16p" corresponds to the number
6150 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6153 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6158 Generate an audio signal specified by an expression.
6160 This source accepts in input one or more expressions (one for each
6161 channel), which are evaluated and used to generate a corresponding
6164 This source accepts the following options:
6168 Set the '|'-separated expressions list for each separate channel. In case the
6169 @option{channel_layout} option is not specified, the selected channel layout
6170 depends on the number of provided expressions. Otherwise the last
6171 specified expression is applied to the remaining output channels.
6173 @item channel_layout, c
6174 Set the channel layout. The number of channels in the specified layout
6175 must be equal to the number of specified expressions.
6178 Set the minimum duration of the sourced audio. See
6179 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6180 for the accepted syntax.
6181 Note that the resulting duration may be greater than the specified
6182 duration, as the generated audio is always cut at the end of a
6185 If not specified, or the expressed duration is negative, the audio is
6186 supposed to be generated forever.
6189 Set the number of samples per channel per each output frame,
6192 @item sample_rate, s
6193 Specify the sample rate, default to 44100.
6196 Each expression in @var{exprs} can contain the following constants:
6200 number of the evaluated sample, starting from 0
6203 time of the evaluated sample expressed in seconds, starting from 0
6210 @subsection Examples
6220 Generate a sin signal with frequency of 440 Hz, set sample rate to
6223 aevalsrc="sin(440*2*PI*t):s=8000"
6227 Generate a two channels signal, specify the channel layout (Front
6228 Center + Back Center) explicitly:
6230 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6234 Generate white noise:
6236 aevalsrc="-2+random(0)"
6240 Generate an amplitude modulated signal:
6242 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6246 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6248 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6255 Generate a FIR coefficients using frequency sampling method.
6257 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6259 The filter accepts the following options:
6263 Set number of filter coefficents in output audio stream.
6264 Default value is 1025.
6267 Set frequency points from where magnitude and phase are set.
6268 This must be in non decreasing order, and first element must be 0, while last element
6269 must be 1. Elements are separated by white spaces.
6272 Set magnitude value for every frequency point set by @option{frequency}.
6273 Number of values must be same as number of frequency points.
6274 Values are separated by white spaces.
6277 Set phase value for every frequency point set by @option{frequency}.
6278 Number of values must be same as number of frequency points.
6279 Values are separated by white spaces.
6281 @item sample_rate, r
6282 Set sample rate, default is 44100.
6285 Set number of samples per each frame. Default is 1024.
6288 Set window function. Default is blackman.
6293 The null audio source, return unprocessed audio frames. It is mainly useful
6294 as a template and to be employed in analysis / debugging tools, or as
6295 the source for filters which ignore the input data (for example the sox
6298 This source accepts the following options:
6302 @item channel_layout, cl
6304 Specifies the channel layout, and can be either an integer or a string
6305 representing a channel layout. The default value of @var{channel_layout}
6308 Check the channel_layout_map definition in
6309 @file{libavutil/channel_layout.c} for the mapping between strings and
6310 channel layout values.
6312 @item sample_rate, r
6313 Specifies the sample rate, and defaults to 44100.
6316 Set the number of samples per requested frames.
6319 Set the duration of the sourced audio. See
6320 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6321 for the accepted syntax.
6323 If not specified, or the expressed duration is negative, the audio is
6324 supposed to be generated forever.
6327 @subsection Examples
6331 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6333 anullsrc=r=48000:cl=4
6337 Do the same operation with a more obvious syntax:
6339 anullsrc=r=48000:cl=mono
6343 All the parameters need to be explicitly defined.
6347 Synthesize a voice utterance using the libflite library.
6349 To enable compilation of this filter you need to configure FFmpeg with
6350 @code{--enable-libflite}.
6352 Note that versions of the flite library prior to 2.0 are not thread-safe.
6354 The filter accepts the following options:
6359 If set to 1, list the names of the available voices and exit
6360 immediately. Default value is 0.
6363 Set the maximum number of samples per frame. Default value is 512.
6366 Set the filename containing the text to speak.
6369 Set the text to speak.
6372 Set the voice to use for the speech synthesis. Default value is
6373 @code{kal}. See also the @var{list_voices} option.
6376 @subsection Examples
6380 Read from file @file{speech.txt}, and synthesize the text using the
6381 standard flite voice:
6383 flite=textfile=speech.txt
6387 Read the specified text selecting the @code{slt} voice:
6389 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6393 Input text to ffmpeg:
6395 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6399 Make @file{ffplay} speak the specified text, using @code{flite} and
6400 the @code{lavfi} device:
6402 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6406 For more information about libflite, check:
6407 @url{http://www.festvox.org/flite/}
6411 Generate a noise audio signal.
6413 The filter accepts the following options:
6416 @item sample_rate, r
6417 Specify the sample rate. Default value is 48000 Hz.
6420 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6424 Specify the duration of the generated audio stream. Not specifying this option
6425 results in noise with an infinite length.
6427 @item color, colour, c
6428 Specify the color of noise. Available noise colors are white, pink, brown,
6429 blue, violet and velvet. Default color is white.
6432 Specify a value used to seed the PRNG.
6435 Set the number of samples per each output frame, default is 1024.
6438 @subsection Examples
6443 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6445 anoisesrc=d=60:c=pink:r=44100:a=0.5
6451 Generate odd-tap Hilbert transform FIR coefficients.
6453 The resulting stream can be used with @ref{afir} filter for phase-shifting
6454 the signal by 90 degrees.
6456 This is used in many matrix coding schemes and for analytic signal generation.
6457 The process is often written as a multiplication by i (or j), the imaginary unit.
6459 The filter accepts the following options:
6463 @item sample_rate, s
6464 Set sample rate, default is 44100.
6467 Set length of FIR filter, default is 22051.
6470 Set number of samples per each frame.
6473 Set window function to be used when generating FIR coefficients.
6478 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6480 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6482 The filter accepts the following options:
6485 @item sample_rate, r
6486 Set sample rate, default is 44100.
6489 Set number of samples per each frame. Default is 1024.
6492 Set high-pass frequency. Default is 0.
6495 Set low-pass frequency. Default is 0.
6496 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6497 is higher than 0 then filter will create band-pass filter coefficients,
6498 otherwise band-reject filter coefficients.
6501 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6504 Set Kaiser window beta.
6507 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6510 Enable rounding, by default is disabled.
6513 Set number of taps for high-pass filter.
6516 Set number of taps for low-pass filter.
6521 Generate an audio signal made of a sine wave with amplitude 1/8.
6523 The audio signal is bit-exact.
6525 The filter accepts the following options:
6530 Set the carrier frequency. Default is 440 Hz.
6532 @item beep_factor, b
6533 Enable a periodic beep every second with frequency @var{beep_factor} times
6534 the carrier frequency. Default is 0, meaning the beep is disabled.
6536 @item sample_rate, r
6537 Specify the sample rate, default is 44100.
6540 Specify the duration of the generated audio stream.
6542 @item samples_per_frame
6543 Set the number of samples per output frame.
6545 The expression can contain the following constants:
6549 The (sequential) number of the output audio frame, starting from 0.
6552 The PTS (Presentation TimeStamp) of the output audio frame,
6553 expressed in @var{TB} units.
6556 The PTS of the output audio frame, expressed in seconds.
6559 The timebase of the output audio frames.
6562 Default is @code{1024}.
6565 @subsection Examples
6570 Generate a simple 440 Hz sine wave:
6576 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6580 sine=frequency=220:beep_factor=4:duration=5
6584 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6587 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6591 @c man end AUDIO SOURCES
6593 @chapter Audio Sinks
6594 @c man begin AUDIO SINKS
6596 Below is a description of the currently available audio sinks.
6598 @section abuffersink
6600 Buffer audio frames, and make them available to the end of filter chain.
6602 This sink is mainly intended for programmatic use, in particular
6603 through the interface defined in @file{libavfilter/buffersink.h}
6604 or the options system.
6606 It accepts a pointer to an AVABufferSinkContext structure, which
6607 defines the incoming buffers' formats, to be passed as the opaque
6608 parameter to @code{avfilter_init_filter} for initialization.
6611 Null audio sink; do absolutely nothing with the input audio. It is
6612 mainly useful as a template and for use in analysis / debugging
6615 @c man end AUDIO SINKS
6617 @chapter Video Filters
6618 @c man begin VIDEO FILTERS
6620 When you configure your FFmpeg build, you can disable any of the
6621 existing filters using @code{--disable-filters}.
6622 The configure output will show the video filters included in your
6625 Below is a description of the currently available video filters.
6629 Mark a region of interest in a video frame.
6631 The frame data is passed through unchanged, but metadata is attached
6632 to the frame indicating regions of interest which can affect the
6633 behaviour of later encoding. Multiple regions can be marked by
6634 applying the filter multiple times.
6638 Region distance in pixels from the left edge of the frame.
6640 Region distance in pixels from the top edge of the frame.
6642 Region width in pixels.
6644 Region height in pixels.
6646 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6647 and may contain the following variables:
6650 Width of the input frame.
6652 Height of the input frame.
6656 Quantisation offset to apply within the region.
6658 This must be a real value in the range -1 to +1. A value of zero
6659 indicates no quality change. A negative value asks for better quality
6660 (less quantisation), while a positive value asks for worse quality
6661 (greater quantisation).
6663 The range is calibrated so that the extreme values indicate the
6664 largest possible offset - if the rest of the frame is encoded with the
6665 worst possible quality, an offset of -1 indicates that this region
6666 should be encoded with the best possible quality anyway. Intermediate
6667 values are then interpolated in some codec-dependent way.
6669 For example, in 10-bit H.264 the quantisation parameter varies between
6670 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6671 this region should be encoded with a QP around one-tenth of the full
6672 range better than the rest of the frame. So, if most of the frame
6673 were to be encoded with a QP of around 30, this region would get a QP
6674 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6675 An extreme value of -1 would indicate that this region should be
6676 encoded with the best possible quality regardless of the treatment of
6677 the rest of the frame - that is, should be encoded at a QP of -12.
6679 If set to true, remove any existing regions of interest marked on the
6680 frame before adding the new one.
6683 @subsection Examples
6687 Mark the centre quarter of the frame as interesting.
6689 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6692 Mark the 100-pixel-wide region on the left edge of the frame as very
6693 uninteresting (to be encoded at much lower quality than the rest of
6696 addroi=0:0:100:ih:+1/5
6700 @section alphaextract
6702 Extract the alpha component from the input as a grayscale video. This
6703 is especially useful with the @var{alphamerge} filter.
6707 Add or replace the alpha component of the primary input with the
6708 grayscale value of a second input. This is intended for use with
6709 @var{alphaextract} to allow the transmission or storage of frame
6710 sequences that have alpha in a format that doesn't support an alpha
6713 For example, to reconstruct full frames from a normal YUV-encoded video
6714 and a separate video created with @var{alphaextract}, you might use:
6716 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6721 Amplify differences between current pixel and pixels of adjacent frames in
6722 same pixel location.
6724 This filter accepts the following options:
6728 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6729 For example radius of 3 will instruct filter to calculate average of 7 frames.
6732 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6735 Set threshold for difference amplification. Any difference greater or equal to
6736 this value will not alter source pixel. Default is 10.
6737 Allowed range is from 0 to 65535.
6740 Set tolerance for difference amplification. Any difference lower to
6741 this value will not alter source pixel. Default is 0.
6742 Allowed range is from 0 to 65535.
6745 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6746 This option controls maximum possible value that will decrease source pixel value.
6749 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6750 This option controls maximum possible value that will increase source pixel value.
6753 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6756 @subsection Commands
6758 This filter supports the following @ref{commands} that corresponds to option of same name:
6770 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6771 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6772 Substation Alpha) subtitles files.
6774 This filter accepts the following option in addition to the common options from
6775 the @ref{subtitles} filter:
6779 Set the shaping engine
6781 Available values are:
6784 The default libass shaping engine, which is the best available.
6786 Fast, font-agnostic shaper that can do only substitutions
6788 Slower shaper using OpenType for substitutions and positioning
6791 The default is @code{auto}.
6795 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6797 The filter accepts the following options:
6801 Set threshold A for 1st plane. Default is 0.02.
6802 Valid range is 0 to 0.3.
6805 Set threshold B for 1st plane. Default is 0.04.
6806 Valid range is 0 to 5.
6809 Set threshold A for 2nd plane. Default is 0.02.
6810 Valid range is 0 to 0.3.
6813 Set threshold B for 2nd plane. Default is 0.04.
6814 Valid range is 0 to 5.
6817 Set threshold A for 3rd plane. Default is 0.02.
6818 Valid range is 0 to 0.3.
6821 Set threshold B for 3rd plane. Default is 0.04.
6822 Valid range is 0 to 5.
6824 Threshold A is designed to react on abrupt changes in the input signal and
6825 threshold B is designed to react on continuous changes in the input signal.
6828 Set number of frames filter will use for averaging. Default is 9. Must be odd
6829 number in range [5, 129].
6832 Set what planes of frame filter will use for averaging. Default is all.
6835 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6836 Alternatively can be set to @code{s} serial.
6838 Parallel can be faster then serial, while other way around is never true.
6839 Parallel will abort early on first change being greater then thresholds, while serial
6840 will continue processing other side of frames if they are equal or below thresholds.
6843 @subsection Commands
6844 This filter supports same @ref{commands} as options except option @code{s}.
6845 The command accepts the same syntax of the corresponding option.
6849 Apply average blur filter.
6851 The filter accepts the following options:
6855 Set horizontal radius size.
6858 Set which planes to filter. By default all planes are filtered.
6861 Set vertical radius size, if zero it will be same as @code{sizeX}.
6862 Default is @code{0}.
6865 @subsection Commands
6866 This filter supports same commands as options.
6867 The command accepts the same syntax of the corresponding option.
6869 If the specified expression is not valid, it is kept at its current
6874 Compute the bounding box for the non-black pixels in the input frame
6877 This filter computes the bounding box containing all the pixels with a
6878 luminance value greater than the minimum allowed value.
6879 The parameters describing the bounding box are printed on the filter
6882 The filter accepts the following option:
6886 Set the minimal luminance value. Default is @code{16}.
6890 Apply bilateral filter, spatial smoothing while preserving edges.
6892 The filter accepts the following options:
6895 Set sigma of gaussian function to calculate spatial weight.
6896 Allowed range is 0 to 512. Default is 0.1.
6899 Set sigma of gaussian function to calculate range weight.
6900 Allowed range is 0 to 1. Default is 0.1.
6903 Set planes to filter. Default is first only.
6906 @section bitplanenoise
6908 Show and measure bit plane noise.
6910 The filter accepts the following options:
6914 Set which plane to analyze. Default is @code{1}.
6917 Filter out noisy pixels from @code{bitplane} set above.
6918 Default is disabled.
6921 @section blackdetect
6923 Detect video intervals that are (almost) completely black. Can be
6924 useful to detect chapter transitions, commercials, or invalid
6927 The filter outputs its detection analysis to both the log as well as
6928 frame metadata. If a black segment of at least the specified minimum
6929 duration is found, a line with the start and end timestamps as well
6930 as duration is printed to the log with level @code{info}. In addition,
6931 a log line with level @code{debug} is printed per frame showing the
6932 black amount detected for that frame.
6934 The filter also attaches metadata to the first frame of a black
6935 segment with key @code{lavfi.black_start} and to the first frame
6936 after the black segment ends with key @code{lavfi.black_end}. The
6937 value is the frame's timestamp. This metadata is added regardless
6938 of the minimum duration specified.
6940 The filter accepts the following options:
6943 @item black_min_duration, d
6944 Set the minimum detected black duration expressed in seconds. It must
6945 be a non-negative floating point number.
6947 Default value is 2.0.
6949 @item picture_black_ratio_th, pic_th
6950 Set the threshold for considering a picture "black".
6951 Express the minimum value for the ratio:
6953 @var{nb_black_pixels} / @var{nb_pixels}
6956 for which a picture is considered black.
6957 Default value is 0.98.
6959 @item pixel_black_th, pix_th
6960 Set the threshold for considering a pixel "black".
6962 The threshold expresses the maximum pixel luminance value for which a
6963 pixel is considered "black". The provided value is scaled according to
6964 the following equation:
6966 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6969 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6970 the input video format, the range is [0-255] for YUV full-range
6971 formats and [16-235] for YUV non full-range formats.
6973 Default value is 0.10.
6976 The following example sets the maximum pixel threshold to the minimum
6977 value, and detects only black intervals of 2 or more seconds:
6979 blackdetect=d=2:pix_th=0.00
6984 Detect frames that are (almost) completely black. Can be useful to
6985 detect chapter transitions or commercials. Output lines consist of
6986 the frame number of the detected frame, the percentage of blackness,
6987 the position in the file if known or -1 and the timestamp in seconds.
6989 In order to display the output lines, you need to set the loglevel at
6990 least to the AV_LOG_INFO value.
6992 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6993 The value represents the percentage of pixels in the picture that
6994 are below the threshold value.
6996 It accepts the following parameters:
7001 The percentage of the pixels that have to be below the threshold; it defaults to
7004 @item threshold, thresh
7005 The threshold below which a pixel value is considered black; it defaults to
7013 Blend two video frames into each other.
7015 The @code{blend} filter takes two input streams and outputs one
7016 stream, the first input is the "top" layer and second input is
7017 "bottom" layer. By default, the output terminates when the longest input terminates.
7019 The @code{tblend} (time blend) filter takes two consecutive frames
7020 from one single stream, and outputs the result obtained by blending
7021 the new frame on top of the old frame.
7023 A description of the accepted options follows.
7031 Set blend mode for specific pixel component or all pixel components in case
7032 of @var{all_mode}. Default value is @code{normal}.
7034 Available values for component modes are:
7076 Set blend opacity for specific pixel component or all pixel components in case
7077 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7084 Set blend expression for specific pixel component or all pixel components in case
7085 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7087 The expressions can use the following variables:
7091 The sequential number of the filtered frame, starting from @code{0}.
7095 the coordinates of the current sample
7099 the width and height of currently filtered plane
7103 Width and height scale for the plane being filtered. It is the
7104 ratio between the dimensions of the current plane to the luma plane,
7105 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7106 the luma plane and @code{0.5,0.5} for the chroma planes.
7109 Time of the current frame, expressed in seconds.
7112 Value of pixel component at current location for first video frame (top layer).
7115 Value of pixel component at current location for second video frame (bottom layer).
7119 The @code{blend} filter also supports the @ref{framesync} options.
7121 @subsection Examples
7125 Apply transition from bottom layer to top layer in first 10 seconds:
7127 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7131 Apply linear horizontal transition from top layer to bottom layer:
7133 blend=all_expr='A*(X/W)+B*(1-X/W)'
7137 Apply 1x1 checkerboard effect:
7139 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7143 Apply uncover left effect:
7145 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7149 Apply uncover down effect:
7151 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7155 Apply uncover up-left effect:
7157 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7161 Split diagonally video and shows top and bottom layer on each side:
7163 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7167 Display differences between the current and the previous frame:
7169 tblend=all_mode=grainextract
7175 Denoise frames using Block-Matching 3D algorithm.
7177 The filter accepts the following options.
7181 Set denoising strength. Default value is 1.
7182 Allowed range is from 0 to 999.9.
7183 The denoising algorithm is very sensitive to sigma, so adjust it
7184 according to the source.
7187 Set local patch size. This sets dimensions in 2D.
7190 Set sliding step for processing blocks. Default value is 4.
7191 Allowed range is from 1 to 64.
7192 Smaller values allows processing more reference blocks and is slower.
7195 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7196 When set to 1, no block matching is done. Larger values allows more blocks
7198 Allowed range is from 1 to 256.
7201 Set radius for search block matching. Default is 9.
7202 Allowed range is from 1 to INT32_MAX.
7205 Set step between two search locations for block matching. Default is 1.
7206 Allowed range is from 1 to 64. Smaller is slower.
7209 Set threshold of mean square error for block matching. Valid range is 0 to
7213 Set thresholding parameter for hard thresholding in 3D transformed domain.
7214 Larger values results in stronger hard-thresholding filtering in frequency
7218 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7219 Default is @code{basic}.
7222 If enabled, filter will use 2nd stream for block matching.
7223 Default is disabled for @code{basic} value of @var{estim} option,
7224 and always enabled if value of @var{estim} is @code{final}.
7227 Set planes to filter. Default is all available except alpha.
7230 @subsection Examples
7234 Basic filtering with bm3d:
7236 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7240 Same as above, but filtering only luma:
7242 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7246 Same as above, but with both estimation modes:
7248 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
7252 Same as above, but prefilter with @ref{nlmeans} filter instead:
7254 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
7260 Apply a boxblur algorithm to the input video.
7262 It accepts the following parameters:
7266 @item luma_radius, lr
7267 @item luma_power, lp
7268 @item chroma_radius, cr
7269 @item chroma_power, cp
7270 @item alpha_radius, ar
7271 @item alpha_power, ap
7275 A description of the accepted options follows.
7278 @item luma_radius, lr
7279 @item chroma_radius, cr
7280 @item alpha_radius, ar
7281 Set an expression for the box radius in pixels used for blurring the
7282 corresponding input plane.
7284 The radius value must be a non-negative number, and must not be
7285 greater than the value of the expression @code{min(w,h)/2} for the
7286 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7289 Default value for @option{luma_radius} is "2". If not specified,
7290 @option{chroma_radius} and @option{alpha_radius} default to the
7291 corresponding value set for @option{luma_radius}.
7293 The expressions can contain the following constants:
7297 The input width and height in pixels.
7301 The input chroma image width and height in pixels.
7305 The horizontal and vertical chroma subsample values. For example, for the
7306 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7309 @item luma_power, lp
7310 @item chroma_power, cp
7311 @item alpha_power, ap
7312 Specify how many times the boxblur filter is applied to the
7313 corresponding plane.
7315 Default value for @option{luma_power} is 2. If not specified,
7316 @option{chroma_power} and @option{alpha_power} default to the
7317 corresponding value set for @option{luma_power}.
7319 A value of 0 will disable the effect.
7322 @subsection Examples
7326 Apply a boxblur filter with the luma, chroma, and alpha radii
7329 boxblur=luma_radius=2:luma_power=1
7334 Set the luma radius to 2, and alpha and chroma radius to 0:
7336 boxblur=2:1:cr=0:ar=0
7340 Set the luma and chroma radii to a fraction of the video dimension:
7342 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7348 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7349 Deinterlacing Filter").
7351 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7352 interpolation algorithms.
7353 It accepts the following parameters:
7357 The interlacing mode to adopt. It accepts one of the following values:
7361 Output one frame for each frame.
7363 Output one frame for each field.
7366 The default value is @code{send_field}.
7369 The picture field parity assumed for the input interlaced video. It accepts one
7370 of the following values:
7374 Assume the top field is first.
7376 Assume the bottom field is first.
7378 Enable automatic detection of field parity.
7381 The default value is @code{auto}.
7382 If the interlacing is unknown or the decoder does not export this information,
7383 top field first will be assumed.
7386 Specify which frames to deinterlace. Accepts one of the following
7391 Deinterlace all frames.
7393 Only deinterlace frames marked as interlaced.
7396 The default value is @code{all}.
7401 Apply Contrast Adaptive Sharpen filter to video stream.
7403 The filter accepts the following options:
7407 Set the sharpening strength. Default value is 0.
7410 Set planes to filter. Default value is to filter all
7411 planes except alpha plane.
7415 Remove all color information for all colors except for certain one.
7417 The filter accepts the following options:
7421 The color which will not be replaced with neutral chroma.
7424 Similarity percentage with the above color.
7425 0.01 matches only the exact key color, while 1.0 matches everything.
7429 0.0 makes pixels either fully gray, or not gray at all.
7430 Higher values result in more preserved color.
7433 Signals that the color passed is already in YUV instead of RGB.
7435 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7436 This can be used to pass exact YUV values as hexadecimal numbers.
7439 @subsection Commands
7440 This filter supports same @ref{commands} as options.
7441 The command accepts the same syntax of the corresponding option.
7443 If the specified expression is not valid, it is kept at its current
7447 YUV colorspace color/chroma keying.
7449 The filter accepts the following options:
7453 The color which will be replaced with transparency.
7456 Similarity percentage with the key color.
7458 0.01 matches only the exact key color, while 1.0 matches everything.
7463 0.0 makes pixels either fully transparent, or not transparent at all.
7465 Higher values result in semi-transparent pixels, with a higher transparency
7466 the more similar the pixels color is to the key color.
7469 Signals that the color passed is already in YUV instead of RGB.
7471 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7472 This can be used to pass exact YUV values as hexadecimal numbers.
7475 @subsection Commands
7476 This filter supports same @ref{commands} as options.
7477 The command accepts the same syntax of the corresponding option.
7479 If the specified expression is not valid, it is kept at its current
7482 @subsection Examples
7486 Make every green pixel in the input image transparent:
7488 ffmpeg -i input.png -vf chromakey=green out.png
7492 Overlay a greenscreen-video on top of a static black background.
7494 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
7499 Reduce chrominance noise.
7501 The filter accepts the following options:
7505 Set threshold for averaging chrominance values.
7506 Sum of absolute difference of U and V pixel components or current
7507 pixel and neighbour pixels lower than this threshold will be used in
7508 averaging. Luma component is left unchanged and is copied to output.
7509 Default value is 30. Allowed range is from 1 to 200.
7512 Set horizontal radius of rectangle used for averaging.
7513 Allowed range is from 1 to 100. Default value is 5.
7516 Set vertical radius of rectangle used for averaging.
7517 Allowed range is from 1 to 100. Default value is 5.
7520 Set horizontal step when averaging. Default value is 1.
7521 Allowed range is from 1 to 50.
7522 Mostly useful to speed-up filtering.
7525 Set vertical step when averaging. Default value is 1.
7526 Allowed range is from 1 to 50.
7527 Mostly useful to speed-up filtering.
7530 @subsection Commands
7531 This filter supports same @ref{commands} as options.
7532 The command accepts the same syntax of the corresponding option.
7534 @section chromashift
7535 Shift chroma pixels horizontally and/or vertically.
7537 The filter accepts the following options:
7540 Set amount to shift chroma-blue horizontally.
7542 Set amount to shift chroma-blue vertically.
7544 Set amount to shift chroma-red horizontally.
7546 Set amount to shift chroma-red vertically.
7548 Set edge mode, can be @var{smear}, default, or @var{warp}.
7551 @subsection Commands
7553 This filter supports the all above options as @ref{commands}.
7557 Display CIE color diagram with pixels overlaid onto it.
7559 The filter accepts the following options:
7574 @item uhdtv, rec2020
7588 Set what gamuts to draw.
7590 See @code{system} option for available values.
7593 Set ciescope size, by default set to 512.
7596 Set intensity used to map input pixel values to CIE diagram.
7599 Set contrast used to draw tongue colors that are out of active color system gamut.
7602 Correct gamma displayed on scope, by default enabled.
7605 Show white point on CIE diagram, by default disabled.
7608 Set input gamma. Used only with XYZ input color space.
7613 Visualize information exported by some codecs.
7615 Some codecs can export information through frames using side-data or other
7616 means. For example, some MPEG based codecs export motion vectors through the
7617 @var{export_mvs} flag in the codec @option{flags2} option.
7619 The filter accepts the following option:
7623 Set motion vectors to visualize.
7625 Available flags for @var{mv} are:
7629 forward predicted MVs of P-frames
7631 forward predicted MVs of B-frames
7633 backward predicted MVs of B-frames
7637 Display quantization parameters using the chroma planes.
7640 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7642 Available flags for @var{mv_type} are:
7646 forward predicted MVs
7648 backward predicted MVs
7651 @item frame_type, ft
7652 Set frame type to visualize motion vectors of.
7654 Available flags for @var{frame_type} are:
7658 intra-coded frames (I-frames)
7660 predicted frames (P-frames)
7662 bi-directionally predicted frames (B-frames)
7666 @subsection Examples
7670 Visualize forward predicted MVs of all frames using @command{ffplay}:
7672 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7676 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7678 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7682 @section colorbalance
7683 Modify intensity of primary colors (red, green and blue) of input frames.
7685 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7686 regions for the red-cyan, green-magenta or blue-yellow balance.
7688 A positive adjustment value shifts the balance towards the primary color, a negative
7689 value towards the complementary color.
7691 The filter accepts the following options:
7697 Adjust red, green and blue shadows (darkest pixels).
7702 Adjust red, green and blue midtones (medium pixels).
7707 Adjust red, green and blue highlights (brightest pixels).
7709 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7712 Preserve lightness when changing color balance. Default is disabled.
7715 @subsection Examples
7719 Add red color cast to shadows:
7725 @subsection Commands
7727 This filter supports the all above options as @ref{commands}.
7729 @section colorchannelmixer
7731 Adjust video input frames by re-mixing color channels.
7733 This filter modifies a color channel by adding the values associated to
7734 the other channels of the same pixels. For example if the value to
7735 modify is red, the output value will be:
7737 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7740 The filter accepts the following options:
7747 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7748 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7754 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7755 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7761 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7762 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7768 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7769 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7771 Allowed ranges for options are @code{[-2.0, 2.0]}.
7774 @subsection Examples
7778 Convert source to grayscale:
7780 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7783 Simulate sepia tones:
7785 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7789 @subsection Commands
7791 This filter supports the all above options as @ref{commands}.
7794 RGB colorspace color keying.
7796 The filter accepts the following options:
7800 The color which will be replaced with transparency.
7803 Similarity percentage with the key color.
7805 0.01 matches only the exact key color, while 1.0 matches everything.
7810 0.0 makes pixels either fully transparent, or not transparent at all.
7812 Higher values result in semi-transparent pixels, with a higher transparency
7813 the more similar the pixels color is to the key color.
7816 @subsection Examples
7820 Make every green pixel in the input image transparent:
7822 ffmpeg -i input.png -vf colorkey=green out.png
7826 Overlay a greenscreen-video on top of a static background image.
7828 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
7832 @subsection Commands
7833 This filter supports same @ref{commands} as options.
7834 The command accepts the same syntax of the corresponding option.
7836 If the specified expression is not valid, it is kept at its current
7840 Remove all color information for all RGB colors except for certain one.
7842 The filter accepts the following options:
7846 The color which will not be replaced with neutral gray.
7849 Similarity percentage with the above color.
7850 0.01 matches only the exact key color, while 1.0 matches everything.
7853 Blend percentage. 0.0 makes pixels fully gray.
7854 Higher values result in more preserved color.
7857 @subsection Commands
7858 This filter supports same @ref{commands} as options.
7859 The command accepts the same syntax of the corresponding option.
7861 If the specified expression is not valid, it is kept at its current
7864 @section colorlevels
7866 Adjust video input frames using levels.
7868 The filter accepts the following options:
7875 Adjust red, green, blue and alpha input black point.
7876 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7882 Adjust red, green, blue and alpha input white point.
7883 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7885 Input levels are used to lighten highlights (bright tones), darken shadows
7886 (dark tones), change the balance of bright and dark tones.
7892 Adjust red, green, blue and alpha output black point.
7893 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7899 Adjust red, green, blue and alpha output white point.
7900 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7902 Output levels allows manual selection of a constrained output level range.
7905 @subsection Examples
7909 Make video output darker:
7911 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7917 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7921 Make video output lighter:
7923 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7927 Increase brightness:
7929 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7933 @subsection Commands
7935 This filter supports the all above options as @ref{commands}.
7937 @section colormatrix
7939 Convert color matrix.
7941 The filter accepts the following options:
7946 Specify the source and destination color matrix. Both values must be
7949 The accepted values are:
7977 For example to convert from BT.601 to SMPTE-240M, use the command:
7979 colormatrix=bt601:smpte240m
7984 Convert colorspace, transfer characteristics or color primaries.
7985 Input video needs to have an even size.
7987 The filter accepts the following options:
7992 Specify all color properties at once.
7994 The accepted values are:
8024 Specify output colorspace.
8026 The accepted values are:
8035 BT.470BG or BT.601-6 625
8038 SMPTE-170M or BT.601-6 525
8047 BT.2020 with non-constant luminance
8053 Specify output transfer characteristics.
8055 The accepted values are:
8067 Constant gamma of 2.2
8070 Constant gamma of 2.8
8073 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8091 BT.2020 for 10-bits content
8094 BT.2020 for 12-bits content
8100 Specify output color primaries.
8102 The accepted values are:
8111 BT.470BG or BT.601-6 625
8114 SMPTE-170M or BT.601-6 525
8138 Specify output color range.
8140 The accepted values are:
8143 TV (restricted) range
8146 MPEG (restricted) range
8157 Specify output color format.
8159 The accepted values are:
8162 YUV 4:2:0 planar 8-bits
8165 YUV 4:2:0 planar 10-bits
8168 YUV 4:2:0 planar 12-bits
8171 YUV 4:2:2 planar 8-bits
8174 YUV 4:2:2 planar 10-bits
8177 YUV 4:2:2 planar 12-bits
8180 YUV 4:4:4 planar 8-bits
8183 YUV 4:4:4 planar 10-bits
8186 YUV 4:4:4 planar 12-bits
8191 Do a fast conversion, which skips gamma/primary correction. This will take
8192 significantly less CPU, but will be mathematically incorrect. To get output
8193 compatible with that produced by the colormatrix filter, use fast=1.
8196 Specify dithering mode.
8198 The accepted values are:
8204 Floyd-Steinberg dithering
8208 Whitepoint adaptation mode.
8210 The accepted values are:
8213 Bradford whitepoint adaptation
8216 von Kries whitepoint adaptation
8219 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8223 Override all input properties at once. Same accepted values as @ref{all}.
8226 Override input colorspace. Same accepted values as @ref{space}.
8229 Override input color primaries. Same accepted values as @ref{primaries}.
8232 Override input transfer characteristics. Same accepted values as @ref{trc}.
8235 Override input color range. Same accepted values as @ref{range}.
8239 The filter converts the transfer characteristics, color space and color
8240 primaries to the specified user values. The output value, if not specified,
8241 is set to a default value based on the "all" property. If that property is
8242 also not specified, the filter will log an error. The output color range and
8243 format default to the same value as the input color range and format. The
8244 input transfer characteristics, color space, color primaries and color range
8245 should be set on the input data. If any of these are missing, the filter will
8246 log an error and no conversion will take place.
8248 For example to convert the input to SMPTE-240M, use the command:
8250 colorspace=smpte240m
8253 @section convolution
8255 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8257 The filter accepts the following options:
8264 Set matrix for each plane.
8265 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8266 and from 1 to 49 odd number of signed integers in @var{row} mode.
8272 Set multiplier for calculated value for each plane.
8273 If unset or 0, it will be sum of all matrix elements.
8279 Set bias for each plane. This value is added to the result of the multiplication.
8280 Useful for making the overall image brighter or darker. Default is 0.0.
8286 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8287 Default is @var{square}.
8290 @subsection Examples
8296 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"
8302 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"
8308 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"
8314 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"
8318 Apply laplacian edge detector which includes diagonals:
8320 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"
8326 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"
8332 Apply 2D convolution of video stream in frequency domain using second stream
8335 The filter accepts the following options:
8339 Set which planes to process.
8342 Set which impulse video frames will be processed, can be @var{first}
8343 or @var{all}. Default is @var{all}.
8346 The @code{convolve} filter also supports the @ref{framesync} options.
8350 Copy the input video source unchanged to the output. This is mainly useful for
8355 Video filtering on GPU using Apple's CoreImage API on OSX.
8357 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8358 processed by video hardware. However, software-based OpenGL implementations
8359 exist which means there is no guarantee for hardware processing. It depends on
8362 There are many filters and image generators provided by Apple that come with a
8363 large variety of options. The filter has to be referenced by its name along
8366 The coreimage filter accepts the following options:
8369 List all available filters and generators along with all their respective
8370 options as well as possible minimum and maximum values along with the default
8377 Specify all filters by their respective name and options.
8378 Use @var{list_filters} to determine all valid filter names and options.
8379 Numerical options are specified by a float value and are automatically clamped
8380 to their respective value range. Vector and color options have to be specified
8381 by a list of space separated float values. Character escaping has to be done.
8382 A special option name @code{default} is available to use default options for a
8385 It is required to specify either @code{default} or at least one of the filter options.
8386 All omitted options are used with their default values.
8387 The syntax of the filter string is as follows:
8389 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8393 Specify a rectangle where the output of the filter chain is copied into the
8394 input image. It is given by a list of space separated float values:
8396 output_rect=x\ y\ width\ height
8398 If not given, the output rectangle equals the dimensions of the input image.
8399 The output rectangle is automatically cropped at the borders of the input
8400 image. Negative values are valid for each component.
8402 output_rect=25\ 25\ 100\ 100
8406 Several filters can be chained for successive processing without GPU-HOST
8407 transfers allowing for fast processing of complex filter chains.
8408 Currently, only filters with zero (generators) or exactly one (filters) input
8409 image and one output image are supported. Also, transition filters are not yet
8412 Some filters generate output images with additional padding depending on the
8413 respective filter kernel. The padding is automatically removed to ensure the
8414 filter output has the same size as the input image.
8416 For image generators, the size of the output image is determined by the
8417 previous output image of the filter chain or the input image of the whole
8418 filterchain, respectively. The generators do not use the pixel information of
8419 this image to generate their output. However, the generated output is
8420 blended onto this image, resulting in partial or complete coverage of the
8423 The @ref{coreimagesrc} video source can be used for generating input images
8424 which are directly fed into the filter chain. By using it, providing input
8425 images by another video source or an input video is not required.
8427 @subsection Examples
8432 List all filters available:
8434 coreimage=list_filters=true
8438 Use the CIBoxBlur filter with default options to blur an image:
8440 coreimage=filter=CIBoxBlur@@default
8444 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8445 its center at 100x100 and a radius of 50 pixels:
8447 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8451 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8452 given as complete and escaped command-line for Apple's standard bash shell:
8454 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8460 Cover a rectangular object
8462 It accepts the following options:
8466 Filepath of the optional cover image, needs to be in yuv420.
8471 It accepts the following values:
8474 cover it by the supplied image
8476 cover it by interpolating the surrounding pixels
8479 Default value is @var{blur}.
8482 @subsection Examples
8486 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8488 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8494 Crop the input video to given dimensions.
8496 It accepts the following parameters:
8500 The width of the output video. It defaults to @code{iw}.
8501 This expression is evaluated only once during the filter
8502 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8505 The height of the output video. It defaults to @code{ih}.
8506 This expression is evaluated only once during the filter
8507 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8510 The horizontal position, in the input video, of the left edge of the output
8511 video. It defaults to @code{(in_w-out_w)/2}.
8512 This expression is evaluated per-frame.
8515 The vertical position, in the input video, of the top edge of the output video.
8516 It defaults to @code{(in_h-out_h)/2}.
8517 This expression is evaluated per-frame.
8520 If set to 1 will force the output display aspect ratio
8521 to be the same of the input, by changing the output sample aspect
8522 ratio. It defaults to 0.
8525 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8526 width/height/x/y as specified and will not be rounded to nearest smaller value.
8530 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8531 expressions containing the following constants:
8536 The computed values for @var{x} and @var{y}. They are evaluated for
8541 The input width and height.
8545 These are the same as @var{in_w} and @var{in_h}.
8549 The output (cropped) width and height.
8553 These are the same as @var{out_w} and @var{out_h}.
8556 same as @var{iw} / @var{ih}
8559 input sample aspect ratio
8562 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8566 horizontal and vertical chroma subsample values. For example for the
8567 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8570 The number of the input frame, starting from 0.
8573 the position in the file of the input frame, NAN if unknown
8576 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8580 The expression for @var{out_w} may depend on the value of @var{out_h},
8581 and the expression for @var{out_h} may depend on @var{out_w}, but they
8582 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8583 evaluated after @var{out_w} and @var{out_h}.
8585 The @var{x} and @var{y} parameters specify the expressions for the
8586 position of the top-left corner of the output (non-cropped) area. They
8587 are evaluated for each frame. If the evaluated value is not valid, it
8588 is approximated to the nearest valid value.
8590 The expression for @var{x} may depend on @var{y}, and the expression
8591 for @var{y} may depend on @var{x}.
8593 @subsection Examples
8597 Crop area with size 100x100 at position (12,34).
8602 Using named options, the example above becomes:
8604 crop=w=100:h=100:x=12:y=34
8608 Crop the central input area with size 100x100:
8614 Crop the central input area with size 2/3 of the input video:
8616 crop=2/3*in_w:2/3*in_h
8620 Crop the input video central square:
8627 Delimit the rectangle with the top-left corner placed at position
8628 100:100 and the right-bottom corner corresponding to the right-bottom
8629 corner of the input image.
8631 crop=in_w-100:in_h-100:100:100
8635 Crop 10 pixels from the left and right borders, and 20 pixels from
8636 the top and bottom borders
8638 crop=in_w-2*10:in_h-2*20
8642 Keep only the bottom right quarter of the input image:
8644 crop=in_w/2:in_h/2:in_w/2:in_h/2
8648 Crop height for getting Greek harmony:
8650 crop=in_w:1/PHI*in_w
8654 Apply trembling effect:
8656 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)
8660 Apply erratic camera effect depending on timestamp:
8662 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)"
8666 Set x depending on the value of y:
8668 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8672 @subsection Commands
8674 This filter supports the following commands:
8680 Set width/height of the output video and the horizontal/vertical position
8682 The command accepts the same syntax of the corresponding option.
8684 If the specified expression is not valid, it is kept at its current
8690 Auto-detect the crop size.
8692 It calculates the necessary cropping parameters and prints the
8693 recommended parameters via the logging system. The detected dimensions
8694 correspond to the non-black area of the input video.
8696 It accepts the following parameters:
8701 Set higher black value threshold, which can be optionally specified
8702 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8703 value greater to the set value is considered non-black. It defaults to 24.
8704 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8705 on the bitdepth of the pixel format.
8708 The value which the width/height should be divisible by. It defaults to
8709 16. The offset is automatically adjusted to center the video. Use 2 to
8710 get only even dimensions (needed for 4:2:2 video). 16 is best when
8711 encoding to most video codecs.
8713 @item reset_count, reset
8714 Set the counter that determines after how many frames cropdetect will
8715 reset the previously detected largest video area and start over to
8716 detect the current optimal crop area. Default value is 0.
8718 This can be useful when channel logos distort the video area. 0
8719 indicates 'never reset', and returns the largest area encountered during
8726 Delay video filtering until a given wallclock timestamp. The filter first
8727 passes on @option{preroll} amount of frames, then it buffers at most
8728 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8729 it forwards the buffered frames and also any subsequent frames coming in its
8732 The filter can be used synchronize the output of multiple ffmpeg processes for
8733 realtime output devices like decklink. By putting the delay in the filtering
8734 chain and pre-buffering frames the process can pass on data to output almost
8735 immediately after the target wallclock timestamp is reached.
8737 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8743 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8746 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8749 The maximum duration of content to buffer before waiting for the cue expressed
8750 in seconds. Default is 0.
8757 Apply color adjustments using curves.
8759 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8760 component (red, green and blue) has its values defined by @var{N} key points
8761 tied from each other using a smooth curve. The x-axis represents the pixel
8762 values from the input frame, and the y-axis the new pixel values to be set for
8765 By default, a component curve is defined by the two points @var{(0;0)} and
8766 @var{(1;1)}. This creates a straight line where each original pixel value is
8767 "adjusted" to its own value, which means no change to the image.
8769 The filter allows you to redefine these two points and add some more. A new
8770 curve (using a natural cubic spline interpolation) will be define to pass
8771 smoothly through all these new coordinates. The new defined points needs to be
8772 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8773 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8774 the vector spaces, the values will be clipped accordingly.
8776 The filter accepts the following options:
8780 Select one of the available color presets. This option can be used in addition
8781 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8782 options takes priority on the preset values.
8783 Available presets are:
8786 @item color_negative
8789 @item increase_contrast
8791 @item linear_contrast
8792 @item medium_contrast
8794 @item strong_contrast
8797 Default is @code{none}.
8799 Set the master key points. These points will define a second pass mapping. It
8800 is sometimes called a "luminance" or "value" mapping. It can be used with
8801 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8802 post-processing LUT.
8804 Set the key points for the red component.
8806 Set the key points for the green component.
8808 Set the key points for the blue component.
8810 Set the key points for all components (not including master).
8811 Can be used in addition to the other key points component
8812 options. In this case, the unset component(s) will fallback on this
8813 @option{all} setting.
8815 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8817 Save Gnuplot script of the curves in specified file.
8820 To avoid some filtergraph syntax conflicts, each key points list need to be
8821 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8823 @subsection Examples
8827 Increase slightly the middle level of blue:
8829 curves=blue='0/0 0.5/0.58 1/1'
8835 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'
8837 Here we obtain the following coordinates for each components:
8840 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8842 @code{(0;0) (0.50;0.48) (1;1)}
8844 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8848 The previous example can also be achieved with the associated built-in preset:
8850 curves=preset=vintage
8860 Use a Photoshop preset and redefine the points of the green component:
8862 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8866 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8867 and @command{gnuplot}:
8869 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8870 gnuplot -p /tmp/curves.plt
8876 Video data analysis filter.
8878 This filter shows hexadecimal pixel values of part of video.
8880 The filter accepts the following options:
8884 Set output video size.
8887 Set x offset from where to pick pixels.
8890 Set y offset from where to pick pixels.
8893 Set scope mode, can be one of the following:
8896 Draw hexadecimal pixel values with white color on black background.
8899 Draw hexadecimal pixel values with input video pixel color on black
8903 Draw hexadecimal pixel values on color background picked from input video,
8904 the text color is picked in such way so its always visible.
8908 Draw rows and columns numbers on left and top of video.
8911 Set background opacity.
8914 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
8918 Apply Directional blur filter.
8920 The filter accepts the following options:
8924 Set angle of directional blur. Default is @code{45}.
8927 Set radius of directional blur. Default is @code{5}.
8930 Set which planes to filter. By default all planes are filtered.
8933 @subsection Commands
8934 This filter supports same @ref{commands} as options.
8935 The command accepts the same syntax of the corresponding option.
8937 If the specified expression is not valid, it is kept at its current
8942 Denoise frames using 2D DCT (frequency domain filtering).
8944 This filter is not designed for real time.
8946 The filter accepts the following options:
8950 Set the noise sigma constant.
8952 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8953 coefficient (absolute value) below this threshold with be dropped.
8955 If you need a more advanced filtering, see @option{expr}.
8957 Default is @code{0}.
8960 Set number overlapping pixels for each block. Since the filter can be slow, you
8961 may want to reduce this value, at the cost of a less effective filter and the
8962 risk of various artefacts.
8964 If the overlapping value doesn't permit processing the whole input width or
8965 height, a warning will be displayed and according borders won't be denoised.
8967 Default value is @var{blocksize}-1, which is the best possible setting.
8970 Set the coefficient factor expression.
8972 For each coefficient of a DCT block, this expression will be evaluated as a
8973 multiplier value for the coefficient.
8975 If this is option is set, the @option{sigma} option will be ignored.
8977 The absolute value of the coefficient can be accessed through the @var{c}
8981 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8982 @var{blocksize}, which is the width and height of the processed blocks.
8984 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8985 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8986 on the speed processing. Also, a larger block size does not necessarily means a
8990 @subsection Examples
8992 Apply a denoise with a @option{sigma} of @code{4.5}:
8997 The same operation can be achieved using the expression system:
8999 dctdnoiz=e='gte(c, 4.5*3)'
9002 Violent denoise using a block size of @code{16x16}:
9009 Remove banding artifacts from input video.
9010 It works by replacing banded pixels with average value of referenced pixels.
9012 The filter accepts the following options:
9019 Set banding detection threshold for each plane. Default is 0.02.
9020 Valid range is 0.00003 to 0.5.
9021 If difference between current pixel and reference pixel is less than threshold,
9022 it will be considered as banded.
9025 Banding detection range in pixels. Default is 16. If positive, random number
9026 in range 0 to set value will be used. If negative, exact absolute value
9028 The range defines square of four pixels around current pixel.
9031 Set direction in radians from which four pixel will be compared. If positive,
9032 random direction from 0 to set direction will be picked. If negative, exact of
9033 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9034 will pick only pixels on same row and -PI/2 will pick only pixels on same
9038 If enabled, current pixel is compared with average value of all four
9039 surrounding pixels. The default is enabled. If disabled current pixel is
9040 compared with all four surrounding pixels. The pixel is considered banded
9041 if only all four differences with surrounding pixels are less than threshold.
9044 If enabled, current pixel is changed if and only if all pixel components are banded,
9045 e.g. banding detection threshold is triggered for all color components.
9046 The default is disabled.
9051 Remove blocking artifacts from input video.
9053 The filter accepts the following options:
9057 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9058 This controls what kind of deblocking is applied.
9061 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9067 Set blocking detection thresholds. Allowed range is 0 to 1.
9068 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9069 Using higher threshold gives more deblocking strength.
9070 Setting @var{alpha} controls threshold detection at exact edge of block.
9071 Remaining options controls threshold detection near the edge. Each one for
9072 below/above or left/right. Setting any of those to @var{0} disables
9076 Set planes to filter. Default is to filter all available planes.
9079 @subsection Examples
9083 Deblock using weak filter and block size of 4 pixels.
9085 deblock=filter=weak:block=4
9089 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9090 deblocking more edges.
9092 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9096 Similar as above, but filter only first plane.
9098 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9102 Similar as above, but filter only second and third plane.
9104 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9111 Drop duplicated frames at regular intervals.
9113 The filter accepts the following options:
9117 Set the number of frames from which one will be dropped. Setting this to
9118 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9119 Default is @code{5}.
9122 Set the threshold for duplicate detection. If the difference metric for a frame
9123 is less than or equal to this value, then it is declared as duplicate. Default
9127 Set scene change threshold. Default is @code{15}.
9131 Set the size of the x and y-axis blocks used during metric calculations.
9132 Larger blocks give better noise suppression, but also give worse detection of
9133 small movements. Must be a power of two. Default is @code{32}.
9136 Mark main input as a pre-processed input and activate clean source input
9137 stream. This allows the input to be pre-processed with various filters to help
9138 the metrics calculation while keeping the frame selection lossless. When set to
9139 @code{1}, the first stream is for the pre-processed input, and the second
9140 stream is the clean source from where the kept frames are chosen. Default is
9144 Set whether or not chroma is considered in the metric calculations. Default is
9150 Apply 2D deconvolution of video stream in frequency domain using second stream
9153 The filter accepts the following options:
9157 Set which planes to process.
9160 Set which impulse video frames will be processed, can be @var{first}
9161 or @var{all}. Default is @var{all}.
9164 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9165 and height are not same and not power of 2 or if stream prior to convolving
9169 The @code{deconvolve} filter also supports the @ref{framesync} options.
9173 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9175 It accepts the following options:
9179 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9180 @var{rainbows} for cross-color reduction.
9183 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9186 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9189 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9192 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9197 Apply deflate effect to the video.
9199 This filter replaces the pixel by the local(3x3) average by taking into account
9200 only values lower than the pixel.
9202 It accepts the following options:
9209 Limit the maximum change for each plane, default is 65535.
9210 If 0, plane will remain unchanged.
9213 @subsection Commands
9215 This filter supports the all above options as @ref{commands}.
9219 Remove temporal frame luminance variations.
9221 It accepts the following options:
9225 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9228 Set averaging mode to smooth temporal luminance variations.
9230 Available values are:
9255 Do not actually modify frame. Useful when one only wants metadata.
9260 Remove judder produced by partially interlaced telecined content.
9262 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9263 source was partially telecined content then the output of @code{pullup,dejudder}
9264 will have a variable frame rate. May change the recorded frame rate of the
9265 container. Aside from that change, this filter will not affect constant frame
9268 The option available in this filter is:
9272 Specify the length of the window over which the judder repeats.
9274 Accepts any integer greater than 1. Useful values are:
9278 If the original was telecined from 24 to 30 fps (Film to NTSC).
9281 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9284 If a mixture of the two.
9287 The default is @samp{4}.
9292 Suppress a TV station logo by a simple interpolation of the surrounding
9293 pixels. Just set a rectangle covering the logo and watch it disappear
9294 (and sometimes something even uglier appear - your mileage may vary).
9296 It accepts the following parameters:
9301 Specify the top left corner coordinates of the logo. They must be
9306 Specify the width and height of the logo to clear. They must be
9310 Specify the thickness of the fuzzy edge of the rectangle (added to
9311 @var{w} and @var{h}). The default value is 1. This option is
9312 deprecated, setting higher values should no longer be necessary and
9316 When set to 1, a green rectangle is drawn on the screen to simplify
9317 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9318 The default value is 0.
9320 The rectangle is drawn on the outermost pixels which will be (partly)
9321 replaced with interpolated values. The values of the next pixels
9322 immediately outside this rectangle in each direction will be used to
9323 compute the interpolated pixel values inside the rectangle.
9327 @subsection Examples
9331 Set a rectangle covering the area with top left corner coordinates 0,0
9332 and size 100x77, and a band of size 10:
9334 delogo=x=0:y=0:w=100:h=77:band=10
9342 Remove the rain in the input image/video by applying the derain methods based on
9343 convolutional neural networks. Supported models:
9347 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9348 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9351 Training as well as model generation scripts are provided in
9352 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9354 Native model files (.model) can be generated from TensorFlow model
9355 files (.pb) by using tools/python/convert.py
9357 The filter accepts the following options:
9361 Specify which filter to use. This option accepts the following values:
9365 Derain filter. To conduct derain filter, you need to use a derain model.
9368 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9370 Default value is @samp{derain}.
9373 Specify which DNN backend to use for model loading and execution. This option accepts
9374 the following values:
9378 Native implementation of DNN loading and execution.
9381 TensorFlow backend. To enable this backend you
9382 need to install the TensorFlow for C library (see
9383 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9384 @code{--enable-libtensorflow}
9386 Default value is @samp{native}.
9389 Set path to model file specifying network architecture and its parameters.
9390 Note that different backends use different file formats. TensorFlow and native
9391 backend can load files for only its format.
9394 It can also be finished with @ref{dnn_processing} filter.
9398 Attempt to fix small changes in horizontal and/or vertical shift. This
9399 filter helps remove camera shake from hand-holding a camera, bumping a
9400 tripod, moving on a vehicle, etc.
9402 The filter accepts the following options:
9410 Specify a rectangular area where to limit the search for motion
9412 If desired the search for motion vectors can be limited to a
9413 rectangular area of the frame defined by its top left corner, width
9414 and height. These parameters have the same meaning as the drawbox
9415 filter which can be used to visualise the position of the bounding
9418 This is useful when simultaneous movement of subjects within the frame
9419 might be confused for camera motion by the motion vector search.
9421 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9422 then the full frame is used. This allows later options to be set
9423 without specifying the bounding box for the motion vector search.
9425 Default - search the whole frame.
9429 Specify the maximum extent of movement in x and y directions in the
9430 range 0-64 pixels. Default 16.
9433 Specify how to generate pixels to fill blanks at the edge of the
9434 frame. Available values are:
9437 Fill zeroes at blank locations
9439 Original image at blank locations
9441 Extruded edge value at blank locations
9443 Mirrored edge at blank locations
9445 Default value is @samp{mirror}.
9448 Specify the blocksize to use for motion search. Range 4-128 pixels,
9452 Specify the contrast threshold for blocks. Only blocks with more than
9453 the specified contrast (difference between darkest and lightest
9454 pixels) will be considered. Range 1-255, default 125.
9457 Specify the search strategy. Available values are:
9460 Set exhaustive search
9462 Set less exhaustive search.
9464 Default value is @samp{exhaustive}.
9467 If set then a detailed log of the motion search is written to the
9474 Remove unwanted contamination of foreground colors, caused by reflected color of
9475 greenscreen or bluescreen.
9477 This filter accepts the following options:
9481 Set what type of despill to use.
9484 Set how spillmap will be generated.
9487 Set how much to get rid of still remaining spill.
9490 Controls amount of red in spill area.
9493 Controls amount of green in spill area.
9494 Should be -1 for greenscreen.
9497 Controls amount of blue in spill area.
9498 Should be -1 for bluescreen.
9501 Controls brightness of spill area, preserving colors.
9504 Modify alpha from generated spillmap.
9507 @subsection Commands
9509 This filter supports the all above options as @ref{commands}.
9513 Apply an exact inverse of the telecine operation. It requires a predefined
9514 pattern specified using the pattern option which must be the same as that passed
9515 to the telecine filter.
9517 This filter accepts the following options:
9526 The default value is @code{top}.
9530 A string of numbers representing the pulldown pattern you wish to apply.
9531 The default value is @code{23}.
9534 A number representing position of the first frame with respect to the telecine
9535 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9540 Apply dilation effect to the video.
9542 This filter replaces the pixel by the local(3x3) maximum.
9544 It accepts the following options:
9551 Limit the maximum change for each plane, default is 65535.
9552 If 0, plane will remain unchanged.
9555 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9558 Flags to local 3x3 coordinates maps like this:
9565 @subsection Commands
9567 This filter supports the all above options as @ref{commands}.
9571 Displace pixels as indicated by second and third input stream.
9573 It takes three input streams and outputs one stream, the first input is the
9574 source, and second and third input are displacement maps.
9576 The second input specifies how much to displace pixels along the
9577 x-axis, while the third input specifies how much to displace pixels
9579 If one of displacement map streams terminates, last frame from that
9580 displacement map will be used.
9582 Note that once generated, displacements maps can be reused over and over again.
9584 A description of the accepted options follows.
9588 Set displace behavior for pixels that are out of range.
9590 Available values are:
9593 Missing pixels are replaced by black pixels.
9596 Adjacent pixels will spread out to replace missing pixels.
9599 Out of range pixels are wrapped so they point to pixels of other side.
9602 Out of range pixels will be replaced with mirrored pixels.
9604 Default is @samp{smear}.
9608 @subsection Examples
9612 Add ripple effect to rgb input of video size hd720:
9614 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
9618 Add wave effect to rgb input of video size hd720:
9620 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
9624 @anchor{dnn_processing}
9625 @section dnn_processing
9627 Do image processing with deep neural networks. It works together with another filter
9628 which converts the pixel format of the Frame to what the dnn network requires.
9630 The filter accepts the following options:
9634 Specify which DNN backend to use for model loading and execution. This option accepts
9635 the following values:
9639 Native implementation of DNN loading and execution.
9642 TensorFlow backend. To enable this backend you
9643 need to install the TensorFlow for C library (see
9644 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9645 @code{--enable-libtensorflow}
9648 OpenVINO backend. To enable this backend you
9649 need to build and install the OpenVINO for C library (see
9650 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9651 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9652 be needed if the header files and libraries are not installed into system path)
9656 Default value is @samp{native}.
9659 Set path to model file specifying network architecture and its parameters.
9660 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9661 backend can load files for only its format.
9663 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9666 Set the input name of the dnn network.
9669 Set the output name of the dnn network.
9673 @subsection Examples
9677 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9679 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9683 Halve the pixel value of the frame with format gray32f:
9685 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
9689 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9691 ./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
9695 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9697 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9704 Draw a colored box on the input image.
9706 It accepts the following parameters:
9711 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9715 The expressions which specify the width and height of the box; if 0 they are interpreted as
9716 the input width and height. It defaults to 0.
9719 Specify the color of the box to write. For the general syntax of this option,
9720 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9721 value @code{invert} is used, the box edge color is the same as the
9722 video with inverted luma.
9725 The expression which sets the thickness of the box edge.
9726 A value of @code{fill} will create a filled box. Default value is @code{3}.
9728 See below for the list of accepted constants.
9731 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9732 will overwrite the video's color and alpha pixels.
9733 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9736 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9737 following constants:
9741 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9745 horizontal and vertical chroma subsample values. For example for the
9746 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9750 The input width and height.
9753 The input sample aspect ratio.
9757 The x and y offset coordinates where the box is drawn.
9761 The width and height of the drawn box.
9764 The thickness of the drawn box.
9766 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9767 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9771 @subsection Examples
9775 Draw a black box around the edge of the input image:
9781 Draw a box with color red and an opacity of 50%:
9783 drawbox=10:20:200:60:red@@0.5
9786 The previous example can be specified as:
9788 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9792 Fill the box with pink color:
9794 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9798 Draw a 2-pixel red 2.40:1 mask:
9800 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
9804 @subsection Commands
9805 This filter supports same commands as options.
9806 The command accepts the same syntax of the corresponding option.
9808 If the specified expression is not valid, it is kept at its current
9813 Draw a graph using input video metadata.
9815 It accepts the following parameters:
9819 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9822 Set 1st foreground color expression.
9825 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9828 Set 2nd foreground color expression.
9831 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9834 Set 3rd foreground color expression.
9837 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9840 Set 4th foreground color expression.
9843 Set minimal value of metadata value.
9846 Set maximal value of metadata value.
9849 Set graph background color. Default is white.
9854 Available values for mode is:
9861 Default is @code{line}.
9866 Available values for slide is:
9869 Draw new frame when right border is reached.
9872 Replace old columns with new ones.
9875 Scroll from right to left.
9878 Scroll from left to right.
9881 Draw single picture.
9884 Default is @code{frame}.
9887 Set size of graph video. For the syntax of this option, check the
9888 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9889 The default value is @code{900x256}.
9892 Set the output frame rate. Default value is @code{25}.
9894 The foreground color expressions can use the following variables:
9897 Minimal value of metadata value.
9900 Maximal value of metadata value.
9903 Current metadata key value.
9906 The color is defined as 0xAABBGGRR.
9909 Example using metadata from @ref{signalstats} filter:
9911 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9914 Example using metadata from @ref{ebur128} filter:
9916 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9921 Draw a grid on the input image.
9923 It accepts the following parameters:
9928 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9932 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9933 input width and height, respectively, minus @code{thickness}, so image gets
9934 framed. Default to 0.
9937 Specify the color of the grid. For the general syntax of this option,
9938 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9939 value @code{invert} is used, the grid color is the same as the
9940 video with inverted luma.
9943 The expression which sets the thickness of the grid line. Default value is @code{1}.
9945 See below for the list of accepted constants.
9948 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9949 will overwrite the video's color and alpha pixels.
9950 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9953 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9954 following constants:
9958 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9962 horizontal and vertical chroma subsample values. For example for the
9963 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9967 The input grid cell width and height.
9970 The input sample aspect ratio.
9974 The x and y coordinates of some point of grid intersection (meant to configure offset).
9978 The width and height of the drawn cell.
9981 The thickness of the drawn cell.
9983 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9984 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9988 @subsection Examples
9992 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9994 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9998 Draw a white 3x3 grid with an opacity of 50%:
10000 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10004 @subsection Commands
10005 This filter supports same commands as options.
10006 The command accepts the same syntax of the corresponding option.
10008 If the specified expression is not valid, it is kept at its current
10014 Draw a text string or text from a specified file on top of a video, using the
10015 libfreetype library.
10017 To enable compilation of this filter, you need to configure FFmpeg with
10018 @code{--enable-libfreetype}.
10019 To enable default font fallback and the @var{font} option you need to
10020 configure FFmpeg with @code{--enable-libfontconfig}.
10021 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10022 @code{--enable-libfribidi}.
10026 It accepts the following parameters:
10031 Used to draw a box around text using the background color.
10032 The value must be either 1 (enable) or 0 (disable).
10033 The default value of @var{box} is 0.
10036 Set the width of the border to be drawn around the box using @var{boxcolor}.
10037 The default value of @var{boxborderw} is 0.
10040 The color to be used for drawing box around text. For the syntax of this
10041 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10043 The default value of @var{boxcolor} is "white".
10046 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10047 The default value of @var{line_spacing} is 0.
10050 Set the width of the border to be drawn around the text using @var{bordercolor}.
10051 The default value of @var{borderw} is 0.
10054 Set the color to be used for drawing border around text. For the syntax of this
10055 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10057 The default value of @var{bordercolor} is "black".
10060 Select how the @var{text} is expanded. Can be either @code{none},
10061 @code{strftime} (deprecated) or
10062 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10066 Set a start time for the count. Value is in microseconds. Only applied
10067 in the deprecated strftime expansion mode. To emulate in normal expansion
10068 mode use the @code{pts} function, supplying the start time (in seconds)
10069 as the second argument.
10072 If true, check and fix text coords to avoid clipping.
10075 The color to be used for drawing fonts. For the syntax of this option, check
10076 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10078 The default value of @var{fontcolor} is "black".
10080 @item fontcolor_expr
10081 String which is expanded the same way as @var{text} to obtain dynamic
10082 @var{fontcolor} value. By default this option has empty value and is not
10083 processed. When this option is set, it overrides @var{fontcolor} option.
10086 The font family to be used for drawing text. By default Sans.
10089 The font file to be used for drawing text. The path must be included.
10090 This parameter is mandatory if the fontconfig support is disabled.
10093 Draw the text applying alpha blending. The value can
10094 be a number between 0.0 and 1.0.
10095 The expression accepts the same variables @var{x, y} as well.
10096 The default value is 1.
10097 Please see @var{fontcolor_expr}.
10100 The font size to be used for drawing text.
10101 The default value of @var{fontsize} is 16.
10104 If set to 1, attempt to shape the text (for example, reverse the order of
10105 right-to-left text and join Arabic characters) before drawing it.
10106 Otherwise, just draw the text exactly as given.
10107 By default 1 (if supported).
10109 @item ft_load_flags
10110 The flags to be used for loading the fonts.
10112 The flags map the corresponding flags supported by libfreetype, and are
10113 a combination of the following values:
10120 @item vertical_layout
10121 @item force_autohint
10124 @item ignore_global_advance_width
10126 @item ignore_transform
10128 @item linear_design
10132 Default value is "default".
10134 For more information consult the documentation for the FT_LOAD_*
10138 The color to be used for drawing a shadow behind the drawn text. For the
10139 syntax of this option, check the @ref{color syntax,,"Color" section in the
10140 ffmpeg-utils manual,ffmpeg-utils}.
10142 The default value of @var{shadowcolor} is "black".
10146 The x and y offsets for the text shadow position with respect to the
10147 position of the text. They can be either positive or negative
10148 values. The default value for both is "0".
10151 The starting frame number for the n/frame_num variable. The default value
10155 The size in number of spaces to use for rendering the tab.
10156 Default value is 4.
10159 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10160 format. It can be used with or without text parameter. @var{timecode_rate}
10161 option must be specified.
10163 @item timecode_rate, rate, r
10164 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10165 integer. Minimum value is "1".
10166 Drop-frame timecode is supported for frame rates 30 & 60.
10169 If set to 1, the output of the timecode option will wrap around at 24 hours.
10170 Default is 0 (disabled).
10173 The text string to be drawn. The text must be a sequence of UTF-8
10174 encoded characters.
10175 This parameter is mandatory if no file is specified with the parameter
10179 A text file containing text to be drawn. The text must be a sequence
10180 of UTF-8 encoded characters.
10182 This parameter is mandatory if no text string is specified with the
10183 parameter @var{text}.
10185 If both @var{text} and @var{textfile} are specified, an error is thrown.
10188 If set to 1, the @var{textfile} will be reloaded before each frame.
10189 Be sure to update it atomically, or it may be read partially, or even fail.
10193 The expressions which specify the offsets where text will be drawn
10194 within the video frame. They are relative to the top/left border of the
10197 The default value of @var{x} and @var{y} is "0".
10199 See below for the list of accepted constants and functions.
10202 The parameters for @var{x} and @var{y} are expressions containing the
10203 following constants and functions:
10207 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10211 horizontal and vertical chroma subsample values. For example for the
10212 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10215 the height of each text line
10223 @item max_glyph_a, ascent
10224 the maximum distance from the baseline to the highest/upper grid
10225 coordinate used to place a glyph outline point, for all the rendered
10227 It is a positive value, due to the grid's orientation with the Y axis
10230 @item max_glyph_d, descent
10231 the maximum distance from the baseline to the lowest grid coordinate
10232 used to place a glyph outline point, for all the rendered glyphs.
10233 This is a negative value, due to the grid's orientation, with the Y axis
10237 maximum glyph height, that is the maximum height for all the glyphs
10238 contained in the rendered text, it is equivalent to @var{ascent} -
10242 maximum glyph width, that is the maximum width for all the glyphs
10243 contained in the rendered text
10246 the number of input frame, starting from 0
10248 @item rand(min, max)
10249 return a random number included between @var{min} and @var{max}
10252 The input sample aspect ratio.
10255 timestamp expressed in seconds, NAN if the input timestamp is unknown
10258 the height of the rendered text
10261 the width of the rendered text
10265 the x and y offset coordinates where the text is drawn.
10267 These parameters allow the @var{x} and @var{y} expressions to refer
10268 to each other, so you can for example specify @code{y=x/dar}.
10271 A one character description of the current frame's picture type.
10274 The current packet's position in the input file or stream
10275 (in bytes, from the start of the input). A value of -1 indicates
10276 this info is not available.
10279 The current packet's duration, in seconds.
10282 The current packet's size (in bytes).
10285 @anchor{drawtext_expansion}
10286 @subsection Text expansion
10288 If @option{expansion} is set to @code{strftime},
10289 the filter recognizes strftime() sequences in the provided text and
10290 expands them accordingly. Check the documentation of strftime(). This
10291 feature is deprecated.
10293 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10295 If @option{expansion} is set to @code{normal} (which is the default),
10296 the following expansion mechanism is used.
10298 The backslash character @samp{\}, followed by any character, always expands to
10299 the second character.
10301 Sequences of the form @code{%@{...@}} are expanded. The text between the
10302 braces is a function name, possibly followed by arguments separated by ':'.
10303 If the arguments contain special characters or delimiters (':' or '@}'),
10304 they should be escaped.
10306 Note that they probably must also be escaped as the value for the
10307 @option{text} option in the filter argument string and as the filter
10308 argument in the filtergraph description, and possibly also for the shell,
10309 that makes up to four levels of escaping; using a text file avoids these
10312 The following functions are available:
10317 The expression evaluation result.
10319 It must take one argument specifying the expression to be evaluated,
10320 which accepts the same constants and functions as the @var{x} and
10321 @var{y} values. Note that not all constants should be used, for
10322 example the text size is not known when evaluating the expression, so
10323 the constants @var{text_w} and @var{text_h} will have an undefined
10326 @item expr_int_format, eif
10327 Evaluate the expression's value and output as formatted integer.
10329 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10330 The second argument specifies the output format. Allowed values are @samp{x},
10331 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10332 @code{printf} function.
10333 The third parameter is optional and sets the number of positions taken by the output.
10334 It can be used to add padding with zeros from the left.
10337 The time at which the filter is running, expressed in UTC.
10338 It can accept an argument: a strftime() format string.
10341 The time at which the filter is running, expressed in the local time zone.
10342 It can accept an argument: a strftime() format string.
10345 Frame metadata. Takes one or two arguments.
10347 The first argument is mandatory and specifies the metadata key.
10349 The second argument is optional and specifies a default value, used when the
10350 metadata key is not found or empty.
10352 Available metadata can be identified by inspecting entries
10353 starting with TAG included within each frame section
10354 printed by running @code{ffprobe -show_frames}.
10356 String metadata generated in filters leading to
10357 the drawtext filter are also available.
10360 The frame number, starting from 0.
10363 A one character description of the current picture type.
10366 The timestamp of the current frame.
10367 It can take up to three arguments.
10369 The first argument is the format of the timestamp; it defaults to @code{flt}
10370 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10371 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10372 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10373 @code{localtime} stands for the timestamp of the frame formatted as
10374 local time zone time.
10376 The second argument is an offset added to the timestamp.
10378 If the format is set to @code{hms}, a third argument @code{24HH} may be
10379 supplied to present the hour part of the formatted timestamp in 24h format
10382 If the format is set to @code{localtime} or @code{gmtime},
10383 a third argument may be supplied: a strftime() format string.
10384 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10387 @subsection Commands
10389 This filter supports altering parameters via commands:
10392 Alter existing filter parameters.
10394 Syntax for the argument is the same as for filter invocation, e.g.
10397 fontsize=56:fontcolor=green:text='Hello World'
10400 Full filter invocation with sendcmd would look like this:
10403 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10407 If the entire argument can't be parsed or applied as valid values then the filter will
10408 continue with its existing parameters.
10410 @subsection Examples
10414 Draw "Test Text" with font FreeSerif, using the default values for the
10415 optional parameters.
10418 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10422 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10423 and y=50 (counting from the top-left corner of the screen), text is
10424 yellow with a red box around it. Both the text and the box have an
10428 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10429 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10432 Note that the double quotes are not necessary if spaces are not used
10433 within the parameter list.
10436 Show the text at the center of the video frame:
10438 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10442 Show the text at a random position, switching to a new position every 30 seconds:
10444 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)"
10448 Show a text line sliding from right to left in the last row of the video
10449 frame. The file @file{LONG_LINE} is assumed to contain a single line
10452 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10456 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10458 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10462 Draw a single green letter "g", at the center of the input video.
10463 The glyph baseline is placed at half screen height.
10465 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10469 Show text for 1 second every 3 seconds:
10471 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10475 Use fontconfig to set the font. Note that the colons need to be escaped.
10477 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10481 Draw "Test Text" with font size dependent on height of the video.
10483 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10487 Print the date of a real-time encoding (see strftime(3)):
10489 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10493 Show text fading in and out (appearing/disappearing):
10496 DS=1.0 # display start
10497 DE=10.0 # display end
10498 FID=1.5 # fade in duration
10499 FOD=5 # fade out duration
10500 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 @}"
10504 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10505 and the @option{fontsize} value are included in the @option{y} offset.
10507 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10508 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10512 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10513 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10514 must have option @option{-export_path_metadata 1} for the special metadata fields
10515 to be available for filters.
10517 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10522 For more information about libfreetype, check:
10523 @url{http://www.freetype.org/}.
10525 For more information about fontconfig, check:
10526 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10528 For more information about libfribidi, check:
10529 @url{http://fribidi.org/}.
10531 @section edgedetect
10533 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10535 The filter accepts the following options:
10540 Set low and high threshold values used by the Canny thresholding
10543 The high threshold selects the "strong" edge pixels, which are then
10544 connected through 8-connectivity with the "weak" edge pixels selected
10545 by the low threshold.
10547 @var{low} and @var{high} threshold values must be chosen in the range
10548 [0,1], and @var{low} should be lesser or equal to @var{high}.
10550 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10554 Define the drawing mode.
10558 Draw white/gray wires on black background.
10561 Mix the colors to create a paint/cartoon effect.
10564 Apply Canny edge detector on all selected planes.
10566 Default value is @var{wires}.
10569 Select planes for filtering. By default all available planes are filtered.
10572 @subsection Examples
10576 Standard edge detection with custom values for the hysteresis thresholding:
10578 edgedetect=low=0.1:high=0.4
10582 Painting effect without thresholding:
10584 edgedetect=mode=colormix:high=0
10590 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10592 For each input image, the filter will compute the optimal mapping from
10593 the input to the output given the codebook length, that is the number
10594 of distinct output colors.
10596 This filter accepts the following options.
10599 @item codebook_length, l
10600 Set codebook length. The value must be a positive integer, and
10601 represents the number of distinct output colors. Default value is 256.
10604 Set the maximum number of iterations to apply for computing the optimal
10605 mapping. The higher the value the better the result and the higher the
10606 computation time. Default value is 1.
10609 Set a random seed, must be an integer included between 0 and
10610 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10611 will try to use a good random seed on a best effort basis.
10614 Set pal8 output pixel format. This option does not work with codebook
10615 length greater than 256.
10620 Measure graylevel entropy in histogram of color channels of video frames.
10622 It accepts the following parameters:
10626 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10628 @var{diff} mode measures entropy of histogram delta values, absolute differences
10629 between neighbour histogram values.
10633 Set brightness, contrast, saturation and approximate gamma adjustment.
10635 The filter accepts the following options:
10639 Set the contrast expression. The value must be a float value in range
10640 @code{-1000.0} to @code{1000.0}. The default value is "1".
10643 Set the brightness expression. The value must be a float value in
10644 range @code{-1.0} to @code{1.0}. The default value is "0".
10647 Set the saturation expression. The value must be a float in
10648 range @code{0.0} to @code{3.0}. The default value is "1".
10651 Set the gamma expression. The value must be a float in range
10652 @code{0.1} to @code{10.0}. The default value is "1".
10655 Set the gamma expression for red. The value must be a float in
10656 range @code{0.1} to @code{10.0}. The default value is "1".
10659 Set the gamma expression for green. The value must be a float in range
10660 @code{0.1} to @code{10.0}. The default value is "1".
10663 Set the gamma expression for blue. The value must be a float in range
10664 @code{0.1} to @code{10.0}. The default value is "1".
10667 Set the gamma weight expression. It can be used to reduce the effect
10668 of a high gamma value on bright image areas, e.g. keep them from
10669 getting overamplified and just plain white. The value must be a float
10670 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10671 gamma correction all the way down while @code{1.0} leaves it at its
10672 full strength. Default is "1".
10675 Set when the expressions for brightness, contrast, saturation and
10676 gamma expressions are evaluated.
10678 It accepts the following values:
10681 only evaluate expressions once during the filter initialization or
10682 when a command is processed
10685 evaluate expressions for each incoming frame
10688 Default value is @samp{init}.
10691 The expressions accept the following parameters:
10694 frame count of the input frame starting from 0
10697 byte position of the corresponding packet in the input file, NAN if
10701 frame rate of the input video, NAN if the input frame rate is unknown
10704 timestamp expressed in seconds, NAN if the input timestamp is unknown
10707 @subsection Commands
10708 The filter supports the following commands:
10712 Set the contrast expression.
10715 Set the brightness expression.
10718 Set the saturation expression.
10721 Set the gamma expression.
10724 Set the gamma_r expression.
10727 Set gamma_g expression.
10730 Set gamma_b expression.
10733 Set gamma_weight expression.
10735 The command accepts the same syntax of the corresponding option.
10737 If the specified expression is not valid, it is kept at its current
10744 Apply erosion effect to the video.
10746 This filter replaces the pixel by the local(3x3) minimum.
10748 It accepts the following options:
10755 Limit the maximum change for each plane, default is 65535.
10756 If 0, plane will remain unchanged.
10759 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10762 Flags to local 3x3 coordinates maps like this:
10769 @subsection Commands
10771 This filter supports the all above options as @ref{commands}.
10773 @section extractplanes
10775 Extract color channel components from input video stream into
10776 separate grayscale video streams.
10778 The filter accepts the following option:
10782 Set plane(s) to extract.
10784 Available values for planes are:
10795 Choosing planes not available in the input will result in an error.
10796 That means you cannot select @code{r}, @code{g}, @code{b} planes
10797 with @code{y}, @code{u}, @code{v} planes at same time.
10800 @subsection Examples
10804 Extract luma, u and v color channel component from input video frame
10805 into 3 grayscale outputs:
10807 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
10813 Apply a fade-in/out effect to the input video.
10815 It accepts the following parameters:
10819 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10821 Default is @code{in}.
10823 @item start_frame, s
10824 Specify the number of the frame to start applying the fade
10825 effect at. Default is 0.
10828 The number of frames that the fade effect lasts. At the end of the
10829 fade-in effect, the output video will have the same intensity as the input video.
10830 At the end of the fade-out transition, the output video will be filled with the
10831 selected @option{color}.
10835 If set to 1, fade only alpha channel, if one exists on the input.
10836 Default value is 0.
10838 @item start_time, st
10839 Specify the timestamp (in seconds) of the frame to start to apply the fade
10840 effect. If both start_frame and start_time are specified, the fade will start at
10841 whichever comes last. Default is 0.
10844 The number of seconds for which the fade effect has to last. At the end of the
10845 fade-in effect the output video will have the same intensity as the input video,
10846 at the end of the fade-out transition the output video will be filled with the
10847 selected @option{color}.
10848 If both duration and nb_frames are specified, duration is used. Default is 0
10849 (nb_frames is used by default).
10852 Specify the color of the fade. Default is "black".
10855 @subsection Examples
10859 Fade in the first 30 frames of video:
10864 The command above is equivalent to:
10870 Fade out the last 45 frames of a 200-frame video:
10873 fade=type=out:start_frame=155:nb_frames=45
10877 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10879 fade=in:0:25, fade=out:975:25
10883 Make the first 5 frames yellow, then fade in from frame 5-24:
10885 fade=in:5:20:color=yellow
10889 Fade in alpha over first 25 frames of video:
10891 fade=in:0:25:alpha=1
10895 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10897 fade=t=in:st=5.5:d=0.5
10903 Denoise frames using 3D FFT (frequency domain filtering).
10905 The filter accepts the following options:
10909 Set the noise sigma constant. This sets denoising strength.
10910 Default value is 1. Allowed range is from 0 to 30.
10911 Using very high sigma with low overlap may give blocking artifacts.
10914 Set amount of denoising. By default all detected noise is reduced.
10915 Default value is 1. Allowed range is from 0 to 1.
10918 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10919 Actual size of block in pixels is 2 to power of @var{block}, so by default
10920 block size in pixels is 2^4 which is 16.
10923 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10926 Set number of previous frames to use for denoising. By default is set to 0.
10929 Set number of next frames to to use for denoising. By default is set to 0.
10932 Set planes which will be filtered, by default are all available filtered
10937 Apply arbitrary expressions to samples in frequency domain
10941 Adjust the dc value (gain) of the luma plane of the image. The filter
10942 accepts an integer value in range @code{0} to @code{1000}. The default
10943 value is set to @code{0}.
10946 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10947 filter accepts an integer value in range @code{0} to @code{1000}. The
10948 default value is set to @code{0}.
10951 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10952 filter accepts an integer value in range @code{0} to @code{1000}. The
10953 default value is set to @code{0}.
10956 Set the frequency domain weight expression for the luma plane.
10959 Set the frequency domain weight expression for the 1st chroma plane.
10962 Set the frequency domain weight expression for the 2nd chroma plane.
10965 Set when the expressions are evaluated.
10967 It accepts the following values:
10970 Only evaluate expressions once during the filter initialization.
10973 Evaluate expressions for each incoming frame.
10976 Default value is @samp{init}.
10978 The filter accepts the following variables:
10981 The coordinates of the current sample.
10985 The width and height of the image.
10988 The number of input frame, starting from 0.
10991 @subsection Examples
10997 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11003 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11009 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11015 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11022 Extract a single field from an interlaced image using stride
11023 arithmetic to avoid wasting CPU time. The output frames are marked as
11026 The filter accepts the following options:
11030 Specify whether to extract the top (if the value is @code{0} or
11031 @code{top}) or the bottom field (if the value is @code{1} or
11037 Create new frames by copying the top and bottom fields from surrounding frames
11038 supplied as numbers by the hint file.
11042 Set file containing hints: absolute/relative frame numbers.
11044 There must be one line for each frame in a clip. Each line must contain two
11045 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11046 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11047 is current frame number for @code{absolute} mode or out of [-1, 1] range
11048 for @code{relative} mode. First number tells from which frame to pick up top
11049 field and second number tells from which frame to pick up bottom field.
11051 If optionally followed by @code{+} output frame will be marked as interlaced,
11052 else if followed by @code{-} output frame will be marked as progressive, else
11053 it will be marked same as input frame.
11054 If optionally followed by @code{t} output frame will use only top field, or in
11055 case of @code{b} it will use only bottom field.
11056 If line starts with @code{#} or @code{;} that line is skipped.
11059 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11062 Example of first several lines of @code{hint} file for @code{relative} mode:
11064 0,0 - # first frame
11065 1,0 - # second frame, use third's frame top field and second's frame bottom field
11066 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11081 @section fieldmatch
11083 Field matching filter for inverse telecine. It is meant to reconstruct the
11084 progressive frames from a telecined stream. The filter does not drop duplicated
11085 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11086 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11088 The separation of the field matching and the decimation is notably motivated by
11089 the possibility of inserting a de-interlacing filter fallback between the two.
11090 If the source has mixed telecined and real interlaced content,
11091 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11092 But these remaining combed frames will be marked as interlaced, and thus can be
11093 de-interlaced by a later filter such as @ref{yadif} before decimation.
11095 In addition to the various configuration options, @code{fieldmatch} can take an
11096 optional second stream, activated through the @option{ppsrc} option. If
11097 enabled, the frames reconstruction will be based on the fields and frames from
11098 this second stream. This allows the first input to be pre-processed in order to
11099 help the various algorithms of the filter, while keeping the output lossless
11100 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11101 or brightness/contrast adjustments can help.
11103 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11104 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11105 which @code{fieldmatch} is based on. While the semantic and usage are very
11106 close, some behaviour and options names can differ.
11108 The @ref{decimate} filter currently only works for constant frame rate input.
11109 If your input has mixed telecined (30fps) and progressive content with a lower
11110 framerate like 24fps use the following filterchain to produce the necessary cfr
11111 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11113 The filter accepts the following options:
11117 Specify the assumed field order of the input stream. Available values are:
11121 Auto detect parity (use FFmpeg's internal parity value).
11123 Assume bottom field first.
11125 Assume top field first.
11128 Note that it is sometimes recommended not to trust the parity announced by the
11131 Default value is @var{auto}.
11134 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11135 sense that it won't risk creating jerkiness due to duplicate frames when
11136 possible, but if there are bad edits or blended fields it will end up
11137 outputting combed frames when a good match might actually exist. On the other
11138 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11139 but will almost always find a good frame if there is one. The other values are
11140 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11141 jerkiness and creating duplicate frames versus finding good matches in sections
11142 with bad edits, orphaned fields, blended fields, etc.
11144 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11146 Available values are:
11150 2-way matching (p/c)
11152 2-way matching, and trying 3rd match if still combed (p/c + n)
11154 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11156 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11157 still combed (p/c + n + u/b)
11159 3-way matching (p/c/n)
11161 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11162 detected as combed (p/c/n + u/b)
11165 The parenthesis at the end indicate the matches that would be used for that
11166 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11169 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11172 Default value is @var{pc_n}.
11175 Mark the main input stream as a pre-processed input, and enable the secondary
11176 input stream as the clean source to pick the fields from. See the filter
11177 introduction for more details. It is similar to the @option{clip2} feature from
11180 Default value is @code{0} (disabled).
11183 Set the field to match from. It is recommended to set this to the same value as
11184 @option{order} unless you experience matching failures with that setting. In
11185 certain circumstances changing the field that is used to match from can have a
11186 large impact on matching performance. Available values are:
11190 Automatic (same value as @option{order}).
11192 Match from the bottom field.
11194 Match from the top field.
11197 Default value is @var{auto}.
11200 Set whether or not chroma is included during the match comparisons. In most
11201 cases it is recommended to leave this enabled. You should set this to @code{0}
11202 only if your clip has bad chroma problems such as heavy rainbowing or other
11203 artifacts. Setting this to @code{0} could also be used to speed things up at
11204 the cost of some accuracy.
11206 Default value is @code{1}.
11210 These define an exclusion band which excludes the lines between @option{y0} and
11211 @option{y1} from being included in the field matching decision. An exclusion
11212 band can be used to ignore subtitles, a logo, or other things that may
11213 interfere with the matching. @option{y0} sets the starting scan line and
11214 @option{y1} sets the ending line; all lines in between @option{y0} and
11215 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11216 @option{y0} and @option{y1} to the same value will disable the feature.
11217 @option{y0} and @option{y1} defaults to @code{0}.
11220 Set the scene change detection threshold as a percentage of maximum change on
11221 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11222 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11223 @option{scthresh} is @code{[0.0, 100.0]}.
11225 Default value is @code{12.0}.
11228 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11229 account the combed scores of matches when deciding what match to use as the
11230 final match. Available values are:
11234 No final matching based on combed scores.
11236 Combed scores are only used when a scene change is detected.
11238 Use combed scores all the time.
11241 Default is @var{sc}.
11244 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11245 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11246 Available values are:
11250 No forced calculation.
11252 Force p/c/n calculations.
11254 Force p/c/n/u/b calculations.
11257 Default value is @var{none}.
11260 This is the area combing threshold used for combed frame detection. This
11261 essentially controls how "strong" or "visible" combing must be to be detected.
11262 Larger values mean combing must be more visible and smaller values mean combing
11263 can be less visible or strong and still be detected. Valid settings are from
11264 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11265 be detected as combed). This is basically a pixel difference value. A good
11266 range is @code{[8, 12]}.
11268 Default value is @code{9}.
11271 Sets whether or not chroma is considered in the combed frame decision. Only
11272 disable this if your source has chroma problems (rainbowing, etc.) that are
11273 causing problems for the combed frame detection with chroma enabled. Actually,
11274 using @option{chroma}=@var{0} is usually more reliable, except for the case
11275 where there is chroma only combing in the source.
11277 Default value is @code{0}.
11281 Respectively set the x-axis and y-axis size of the window used during combed
11282 frame detection. This has to do with the size of the area in which
11283 @option{combpel} pixels are required to be detected as combed for a frame to be
11284 declared combed. See the @option{combpel} parameter description for more info.
11285 Possible values are any number that is a power of 2 starting at 4 and going up
11288 Default value is @code{16}.
11291 The number of combed pixels inside any of the @option{blocky} by
11292 @option{blockx} size blocks on the frame for the frame to be detected as
11293 combed. While @option{cthresh} controls how "visible" the combing must be, this
11294 setting controls "how much" combing there must be in any localized area (a
11295 window defined by the @option{blockx} and @option{blocky} settings) on the
11296 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11297 which point no frames will ever be detected as combed). This setting is known
11298 as @option{MI} in TFM/VFM vocabulary.
11300 Default value is @code{80}.
11303 @anchor{p/c/n/u/b meaning}
11304 @subsection p/c/n/u/b meaning
11306 @subsubsection p/c/n
11308 We assume the following telecined stream:
11311 Top fields: 1 2 2 3 4
11312 Bottom fields: 1 2 3 4 4
11315 The numbers correspond to the progressive frame the fields relate to. Here, the
11316 first two frames are progressive, the 3rd and 4th are combed, and so on.
11318 When @code{fieldmatch} is configured to run a matching from bottom
11319 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11324 B 1 2 3 4 4 <-- matching reference
11333 As a result of the field matching, we can see that some frames get duplicated.
11334 To perform a complete inverse telecine, you need to rely on a decimation filter
11335 after this operation. See for instance the @ref{decimate} filter.
11337 The same operation now matching from top fields (@option{field}=@var{top})
11342 T 1 2 2 3 4 <-- matching reference
11352 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11353 basically, they refer to the frame and field of the opposite parity:
11356 @item @var{p} matches the field of the opposite parity in the previous frame
11357 @item @var{c} matches the field of the opposite parity in the current frame
11358 @item @var{n} matches the field of the opposite parity in the next frame
11363 The @var{u} and @var{b} matching are a bit special in the sense that they match
11364 from the opposite parity flag. In the following examples, we assume that we are
11365 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11366 'x' is placed above and below each matched fields.
11368 With bottom matching (@option{field}=@var{bottom}):
11373 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11374 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11382 With top matching (@option{field}=@var{top}):
11387 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11388 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11396 @subsection Examples
11398 Simple IVTC of a top field first telecined stream:
11400 fieldmatch=order=tff:combmatch=none, decimate
11403 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11405 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11408 @section fieldorder
11410 Transform the field order of the input video.
11412 It accepts the following parameters:
11417 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11418 for bottom field first.
11421 The default value is @samp{tff}.
11423 The transformation is done by shifting the picture content up or down
11424 by one line, and filling the remaining line with appropriate picture content.
11425 This method is consistent with most broadcast field order converters.
11427 If the input video is not flagged as being interlaced, or it is already
11428 flagged as being of the required output field order, then this filter does
11429 not alter the incoming video.
11431 It is very useful when converting to or from PAL DV material,
11432 which is bottom field first.
11436 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11439 @section fifo, afifo
11441 Buffer input images and send them when they are requested.
11443 It is mainly useful when auto-inserted by the libavfilter
11446 It does not take parameters.
11448 @section fillborders
11450 Fill borders of the input video, without changing video stream dimensions.
11451 Sometimes video can have garbage at the four edges and you may not want to
11452 crop video input to keep size multiple of some number.
11454 This filter accepts the following options:
11458 Number of pixels to fill from left border.
11461 Number of pixels to fill from right border.
11464 Number of pixels to fill from top border.
11467 Number of pixels to fill from bottom border.
11472 It accepts the following values:
11475 fill pixels using outermost pixels
11478 fill pixels using mirroring
11481 fill pixels with constant value
11484 Default is @var{smear}.
11487 Set color for pixels in fixed mode. Default is @var{black}.
11490 @subsection Commands
11491 This filter supports same @ref{commands} as options.
11492 The command accepts the same syntax of the corresponding option.
11494 If the specified expression is not valid, it is kept at its current
11499 Find a rectangular object
11501 It accepts the following options:
11505 Filepath of the object image, needs to be in gray8.
11508 Detection threshold, default is 0.5.
11511 Number of mipmaps, default is 3.
11513 @item xmin, ymin, xmax, ymax
11514 Specifies the rectangle in which to search.
11517 @subsection Examples
11521 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11523 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11529 Flood area with values of same pixel components with another values.
11531 It accepts the following options:
11534 Set pixel x coordinate.
11537 Set pixel y coordinate.
11540 Set source #0 component value.
11543 Set source #1 component value.
11546 Set source #2 component value.
11549 Set source #3 component value.
11552 Set destination #0 component value.
11555 Set destination #1 component value.
11558 Set destination #2 component value.
11561 Set destination #3 component value.
11567 Convert the input video to one of the specified pixel formats.
11568 Libavfilter will try to pick one that is suitable as input to
11571 It accepts the following parameters:
11575 A '|'-separated list of pixel format names, such as
11576 "pix_fmts=yuv420p|monow|rgb24".
11580 @subsection Examples
11584 Convert the input video to the @var{yuv420p} format
11586 format=pix_fmts=yuv420p
11589 Convert the input video to any of the formats in the list
11591 format=pix_fmts=yuv420p|yuv444p|yuv410p
11598 Convert the video to specified constant frame rate by duplicating or dropping
11599 frames as necessary.
11601 It accepts the following parameters:
11605 The desired output frame rate. The default is @code{25}.
11608 Assume the first PTS should be the given value, in seconds. This allows for
11609 padding/trimming at the start of stream. By default, no assumption is made
11610 about the first frame's expected PTS, so no padding or trimming is done.
11611 For example, this could be set to 0 to pad the beginning with duplicates of
11612 the first frame if a video stream starts after the audio stream or to trim any
11613 frames with a negative PTS.
11616 Timestamp (PTS) rounding method.
11618 Possible values are:
11625 round towards -infinity
11627 round towards +infinity
11631 The default is @code{near}.
11634 Action performed when reading the last frame.
11636 Possible values are:
11639 Use same timestamp rounding method as used for other frames.
11641 Pass through last frame if input duration has not been reached yet.
11643 The default is @code{round}.
11647 Alternatively, the options can be specified as a flat string:
11648 @var{fps}[:@var{start_time}[:@var{round}]].
11650 See also the @ref{setpts} filter.
11652 @subsection Examples
11656 A typical usage in order to set the fps to 25:
11662 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11664 fps=fps=film:round=near
11670 Pack two different video streams into a stereoscopic video, setting proper
11671 metadata on supported codecs. The two views should have the same size and
11672 framerate and processing will stop when the shorter video ends. Please note
11673 that you may conveniently adjust view properties with the @ref{scale} and
11676 It accepts the following parameters:
11680 The desired packing format. Supported values are:
11685 The views are next to each other (default).
11688 The views are on top of each other.
11691 The views are packed by line.
11694 The views are packed by column.
11697 The views are temporally interleaved.
11706 # Convert left and right views into a frame-sequential video
11707 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11709 # Convert views into a side-by-side video with the same output resolution as the input
11710 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
11715 Change the frame rate by interpolating new video output frames from the source
11718 This filter is not designed to function correctly with interlaced media. If
11719 you wish to change the frame rate of interlaced media then you are required
11720 to deinterlace before this filter and re-interlace after this filter.
11722 A description of the accepted options follows.
11726 Specify the output frames per second. This option can also be specified
11727 as a value alone. The default is @code{50}.
11730 Specify the start of a range where the output frame will be created as a
11731 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11732 the default is @code{15}.
11735 Specify the end of a range where the output frame will be created as a
11736 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11737 the default is @code{240}.
11740 Specify the level at which a scene change is detected as a value between
11741 0 and 100 to indicate a new scene; a low value reflects a low
11742 probability for the current frame to introduce a new scene, while a higher
11743 value means the current frame is more likely to be one.
11744 The default is @code{8.2}.
11747 Specify flags influencing the filter process.
11749 Available value for @var{flags} is:
11752 @item scene_change_detect, scd
11753 Enable scene change detection using the value of the option @var{scene}.
11754 This flag is enabled by default.
11760 Select one frame every N-th frame.
11762 This filter accepts the following option:
11765 Select frame after every @code{step} frames.
11766 Allowed values are positive integers higher than 0. Default value is @code{1}.
11769 @section freezedetect
11771 Detect frozen video.
11773 This filter logs a message and sets frame metadata when it detects that the
11774 input video has no significant change in content during a specified duration.
11775 Video freeze detection calculates the mean average absolute difference of all
11776 the components of video frames and compares it to a noise floor.
11778 The printed times and duration are expressed in seconds. The
11779 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11780 whose timestamp equals or exceeds the detection duration and it contains the
11781 timestamp of the first frame of the freeze. The
11782 @code{lavfi.freezedetect.freeze_duration} and
11783 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11786 The filter accepts the following options:
11790 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11791 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11795 Set freeze duration until notification (default is 2 seconds).
11798 @section freezeframes
11800 Freeze video frames.
11802 This filter freezes video frames using frame from 2nd input.
11804 The filter accepts the following options:
11808 Set number of first frame from which to start freeze.
11811 Set number of last frame from which to end freeze.
11814 Set number of frame from 2nd input which will be used instead of replaced frames.
11820 Apply a frei0r effect to the input video.
11822 To enable the compilation of this filter, you need to install the frei0r
11823 header and configure FFmpeg with @code{--enable-frei0r}.
11825 It accepts the following parameters:
11830 The name of the frei0r effect to load. If the environment variable
11831 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11832 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11833 Otherwise, the standard frei0r paths are searched, in this order:
11834 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11835 @file{/usr/lib/frei0r-1/}.
11837 @item filter_params
11838 A '|'-separated list of parameters to pass to the frei0r effect.
11842 A frei0r effect parameter can be a boolean (its value is either
11843 "y" or "n"), a double, a color (specified as
11844 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11845 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11846 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11847 a position (specified as @var{X}/@var{Y}, where
11848 @var{X} and @var{Y} are floating point numbers) and/or a string.
11850 The number and types of parameters depend on the loaded effect. If an
11851 effect parameter is not specified, the default value is set.
11853 @subsection Examples
11857 Apply the distort0r effect, setting the first two double parameters:
11859 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11863 Apply the colordistance effect, taking a color as the first parameter:
11865 frei0r=colordistance:0.2/0.3/0.4
11866 frei0r=colordistance:violet
11867 frei0r=colordistance:0x112233
11871 Apply the perspective effect, specifying the top left and top right image
11874 frei0r=perspective:0.2/0.2|0.8/0.2
11878 For more information, see
11879 @url{http://frei0r.dyne.org}
11881 @subsection Commands
11883 This filter supports the @option{filter_params} option as @ref{commands}.
11887 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11889 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11890 processing filter, one of them is performed once per block, not per pixel.
11891 This allows for much higher speed.
11893 The filter accepts the following options:
11897 Set quality. This option defines the number of levels for averaging. It accepts
11898 an integer in the range 4-5. Default value is @code{4}.
11901 Force a constant quantization parameter. It accepts an integer in range 0-63.
11902 If not set, the filter will use the QP from the video stream (if available).
11905 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11906 more details but also more artifacts, while higher values make the image smoother
11907 but also blurrier. Default value is @code{0} − PSNR optimal.
11909 @item use_bframe_qp
11910 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11911 option may cause flicker since the B-Frames have often larger QP. Default is
11912 @code{0} (not enabled).
11918 Apply Gaussian blur filter.
11920 The filter accepts the following options:
11924 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11927 Set number of steps for Gaussian approximation. Default is @code{1}.
11930 Set which planes to filter. By default all planes are filtered.
11933 Set vertical sigma, if negative it will be same as @code{sigma}.
11934 Default is @code{-1}.
11937 @subsection Commands
11938 This filter supports same commands as options.
11939 The command accepts the same syntax of the corresponding option.
11941 If the specified expression is not valid, it is kept at its current
11946 Apply generic equation to each pixel.
11948 The filter accepts the following options:
11951 @item lum_expr, lum
11952 Set the luminance expression.
11954 Set the chrominance blue expression.
11956 Set the chrominance red expression.
11957 @item alpha_expr, a
11958 Set the alpha expression.
11960 Set the red expression.
11961 @item green_expr, g
11962 Set the green expression.
11964 Set the blue expression.
11967 The colorspace is selected according to the specified options. If one
11968 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11969 options is specified, the filter will automatically select a YCbCr
11970 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11971 @option{blue_expr} options is specified, it will select an RGB
11974 If one of the chrominance expression is not defined, it falls back on the other
11975 one. If no alpha expression is specified it will evaluate to opaque value.
11976 If none of chrominance expressions are specified, they will evaluate
11977 to the luminance expression.
11979 The expressions can use the following variables and functions:
11983 The sequential number of the filtered frame, starting from @code{0}.
11987 The coordinates of the current sample.
11991 The width and height of the image.
11995 Width and height scale depending on the currently filtered plane. It is the
11996 ratio between the corresponding luma plane number of pixels and the current
11997 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11998 @code{0.5,0.5} for chroma planes.
12001 Time of the current frame, expressed in seconds.
12004 Return the value of the pixel at location (@var{x},@var{y}) of the current
12008 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12012 Return the value of the pixel at location (@var{x},@var{y}) of the
12013 blue-difference chroma plane. Return 0 if there is no such plane.
12016 Return the value of the pixel at location (@var{x},@var{y}) of the
12017 red-difference chroma plane. Return 0 if there is no such plane.
12022 Return the value of the pixel at location (@var{x},@var{y}) of the
12023 red/green/blue component. Return 0 if there is no such component.
12026 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12027 plane. Return 0 if there is no such plane.
12029 @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)
12030 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12031 sums of samples within a rectangle. See the functions without the sum postfix.
12033 @item interpolation
12034 Set one of interpolation methods:
12039 Default is bilinear.
12042 For functions, if @var{x} and @var{y} are outside the area, the value will be
12043 automatically clipped to the closer edge.
12045 Please note that this filter can use multiple threads in which case each slice
12046 will have its own expression state. If you want to use only a single expression
12047 state because your expressions depend on previous state then you should limit
12048 the number of filter threads to 1.
12050 @subsection Examples
12054 Flip the image horizontally:
12060 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12061 wavelength of 100 pixels:
12063 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12067 Generate a fancy enigmatic moving light:
12069 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
12073 Generate a quick emboss effect:
12075 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12079 Modify RGB components depending on pixel position:
12081 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12085 Create a radial gradient that is the same size as the input (also see
12086 the @ref{vignette} filter):
12088 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12094 Fix the banding artifacts that are sometimes introduced into nearly flat
12095 regions by truncation to 8-bit color depth.
12096 Interpolate the gradients that should go where the bands are, and
12099 It is designed for playback only. Do not use it prior to
12100 lossy compression, because compression tends to lose the dither and
12101 bring back the bands.
12103 It accepts the following parameters:
12108 The maximum amount by which the filter will change any one pixel. This is also
12109 the threshold for detecting nearly flat regions. Acceptable values range from
12110 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12114 The neighborhood to fit the gradient to. A larger radius makes for smoother
12115 gradients, but also prevents the filter from modifying the pixels near detailed
12116 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12117 values will be clipped to the valid range.
12121 Alternatively, the options can be specified as a flat string:
12122 @var{strength}[:@var{radius}]
12124 @subsection Examples
12128 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12134 Specify radius, omitting the strength (which will fall-back to the default
12142 @anchor{graphmonitor}
12143 @section graphmonitor
12144 Show various filtergraph stats.
12146 With this filter one can debug complete filtergraph.
12147 Especially issues with links filling with queued frames.
12149 The filter accepts the following options:
12153 Set video output size. Default is @var{hd720}.
12156 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12159 Set output mode, can be @var{fulll} or @var{compact}.
12160 In @var{compact} mode only filters with some queued frames have displayed stats.
12163 Set flags which enable which stats are shown in video.
12165 Available values for flags are:
12168 Display number of queued frames in each link.
12170 @item frame_count_in
12171 Display number of frames taken from filter.
12173 @item frame_count_out
12174 Display number of frames given out from filter.
12177 Display current filtered frame pts.
12180 Display current filtered frame time.
12183 Display time base for filter link.
12186 Display used format for filter link.
12189 Display video size or number of audio channels in case of audio used by filter link.
12192 Display video frame rate or sample rate in case of audio used by filter link.
12195 Display link output status.
12199 Set upper limit for video rate of output stream, Default value is @var{25}.
12200 This guarantee that output video frame rate will not be higher than this value.
12204 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12205 and corrects the scene colors accordingly.
12207 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12209 The filter accepts the following options:
12213 The order of differentiation to be applied on the scene. Must be chosen in the range
12214 [0,2] and default value is 1.
12217 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12218 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12219 max value instead of calculating Minkowski distance.
12222 The standard deviation of Gaussian blur to be applied on the scene. Must be
12223 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12224 can't be equal to 0 if @var{difford} is greater than 0.
12227 @subsection Examples
12233 greyedge=difford=1:minknorm=5:sigma=2
12239 greyedge=difford=1:minknorm=0:sigma=2
12247 Apply a Hald CLUT to a video stream.
12249 First input is the video stream to process, and second one is the Hald CLUT.
12250 The Hald CLUT input can be a simple picture or a complete video stream.
12252 The filter accepts the following options:
12256 Force termination when the shortest input terminates. Default is @code{0}.
12258 Continue applying the last CLUT after the end of the stream. A value of
12259 @code{0} disable the filter after the last frame of the CLUT is reached.
12260 Default is @code{1}.
12263 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12264 filters share the same internals).
12266 This filter also supports the @ref{framesync} options.
12268 More information about the Hald CLUT can be found on Eskil Steenberg's website
12269 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12271 @subsection Workflow examples
12273 @subsubsection Hald CLUT video stream
12275 Generate an identity Hald CLUT stream altered with various effects:
12277 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
12280 Note: make sure you use a lossless codec.
12282 Then use it with @code{haldclut} to apply it on some random stream:
12284 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12287 The Hald CLUT will be applied to the 10 first seconds (duration of
12288 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12289 to the remaining frames of the @code{mandelbrot} stream.
12291 @subsubsection Hald CLUT with preview
12293 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12294 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12295 biggest possible square starting at the top left of the picture. The remaining
12296 padding pixels (bottom or right) will be ignored. This area can be used to add
12297 a preview of the Hald CLUT.
12299 Typically, the following generated Hald CLUT will be supported by the
12300 @code{haldclut} filter:
12303 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12304 pad=iw+320 [padded_clut];
12305 smptebars=s=320x256, split [a][b];
12306 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12307 [main][b] overlay=W-320" -frames:v 1 clut.png
12310 It contains the original and a preview of the effect of the CLUT: SMPTE color
12311 bars are displayed on the right-top, and below the same color bars processed by
12314 Then, the effect of this Hald CLUT can be visualized with:
12316 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12321 Flip the input video horizontally.
12323 For example, to horizontally flip the input video with @command{ffmpeg}:
12325 ffmpeg -i in.avi -vf "hflip" out.avi
12329 This filter applies a global color histogram equalization on a
12332 It can be used to correct video that has a compressed range of pixel
12333 intensities. The filter redistributes the pixel intensities to
12334 equalize their distribution across the intensity range. It may be
12335 viewed as an "automatically adjusting contrast filter". This filter is
12336 useful only for correcting degraded or poorly captured source
12339 The filter accepts the following options:
12343 Determine the amount of equalization to be applied. As the strength
12344 is reduced, the distribution of pixel intensities more-and-more
12345 approaches that of the input frame. The value must be a float number
12346 in the range [0,1] and defaults to 0.200.
12349 Set the maximum intensity that can generated and scale the output
12350 values appropriately. The strength should be set as desired and then
12351 the intensity can be limited if needed to avoid washing-out. The value
12352 must be a float number in the range [0,1] and defaults to 0.210.
12355 Set the antibanding level. If enabled the filter will randomly vary
12356 the luminance of output pixels by a small amount to avoid banding of
12357 the histogram. Possible values are @code{none}, @code{weak} or
12358 @code{strong}. It defaults to @code{none}.
12364 Compute and draw a color distribution histogram for the input video.
12366 The computed histogram is a representation of the color component
12367 distribution in an image.
12369 Standard histogram displays the color components distribution in an image.
12370 Displays color graph for each color component. Shows distribution of
12371 the Y, U, V, A or R, G, B components, depending on input format, in the
12372 current frame. Below each graph a color component scale meter is shown.
12374 The filter accepts the following options:
12378 Set height of level. Default value is @code{200}.
12379 Allowed range is [50, 2048].
12382 Set height of color scale. Default value is @code{12}.
12383 Allowed range is [0, 40].
12387 It accepts the following values:
12390 Per color component graphs are placed below each other.
12393 Per color component graphs are placed side by side.
12396 Presents information identical to that in the @code{parade}, except
12397 that the graphs representing color components are superimposed directly
12400 Default is @code{stack}.
12403 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12404 Default is @code{linear}.
12407 Set what color components to display.
12408 Default is @code{7}.
12411 Set foreground opacity. Default is @code{0.7}.
12414 Set background opacity. Default is @code{0.5}.
12417 @subsection Examples
12422 Calculate and draw histogram:
12424 ffplay -i input -vf histogram
12432 This is a high precision/quality 3d denoise filter. It aims to reduce
12433 image noise, producing smooth images and making still images really
12434 still. It should enhance compressibility.
12436 It accepts the following optional parameters:
12440 A non-negative floating point number which specifies spatial luma strength.
12441 It defaults to 4.0.
12443 @item chroma_spatial
12444 A non-negative floating point number which specifies spatial chroma strength.
12445 It defaults to 3.0*@var{luma_spatial}/4.0.
12448 A floating point number which specifies luma temporal strength. It defaults to
12449 6.0*@var{luma_spatial}/4.0.
12452 A floating point number which specifies chroma temporal strength. It defaults to
12453 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12456 @subsection Commands
12457 This filter supports same @ref{commands} as options.
12458 The command accepts the same syntax of the corresponding option.
12460 If the specified expression is not valid, it is kept at its current
12463 @anchor{hwdownload}
12464 @section hwdownload
12466 Download hardware frames to system memory.
12468 The input must be in hardware frames, and the output a non-hardware format.
12469 Not all formats will be supported on the output - it may be necessary to insert
12470 an additional @option{format} filter immediately following in the graph to get
12471 the output in a supported format.
12475 Map hardware frames to system memory or to another device.
12477 This filter has several different modes of operation; which one is used depends
12478 on the input and output formats:
12481 Hardware frame input, normal frame output
12483 Map the input frames to system memory and pass them to the output. If the
12484 original hardware frame is later required (for example, after overlaying
12485 something else on part of it), the @option{hwmap} filter can be used again
12486 in the next mode to retrieve it.
12488 Normal frame input, hardware frame output
12490 If the input is actually a software-mapped hardware frame, then unmap it -
12491 that is, return the original hardware frame.
12493 Otherwise, a device must be provided. Create new hardware surfaces on that
12494 device for the output, then map them back to the software format at the input
12495 and give those frames to the preceding filter. This will then act like the
12496 @option{hwupload} filter, but may be able to avoid an additional copy when
12497 the input is already in a compatible format.
12499 Hardware frame input and output
12501 A device must be supplied for the output, either directly or with the
12502 @option{derive_device} option. The input and output devices must be of
12503 different types and compatible - the exact meaning of this is
12504 system-dependent, but typically it means that they must refer to the same
12505 underlying hardware context (for example, refer to the same graphics card).
12507 If the input frames were originally created on the output device, then unmap
12508 to retrieve the original frames.
12510 Otherwise, map the frames to the output device - create new hardware frames
12511 on the output corresponding to the frames on the input.
12514 The following additional parameters are accepted:
12518 Set the frame mapping mode. Some combination of:
12521 The mapped frame should be readable.
12523 The mapped frame should be writeable.
12525 The mapping will always overwrite the entire frame.
12527 This may improve performance in some cases, as the original contents of the
12528 frame need not be loaded.
12530 The mapping must not involve any copying.
12532 Indirect mappings to copies of frames are created in some cases where either
12533 direct mapping is not possible or it would have unexpected properties.
12534 Setting this flag ensures that the mapping is direct and will fail if that is
12537 Defaults to @var{read+write} if not specified.
12539 @item derive_device @var{type}
12540 Rather than using the device supplied at initialisation, instead derive a new
12541 device of type @var{type} from the device the input frames exist on.
12544 In a hardware to hardware mapping, map in reverse - create frames in the sink
12545 and map them back to the source. This may be necessary in some cases where
12546 a mapping in one direction is required but only the opposite direction is
12547 supported by the devices being used.
12549 This option is dangerous - it may break the preceding filter in undefined
12550 ways if there are any additional constraints on that filter's output.
12551 Do not use it without fully understanding the implications of its use.
12557 Upload system memory frames to hardware surfaces.
12559 The device to upload to must be supplied when the filter is initialised. If
12560 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12561 option or with the @option{derive_device} option. The input and output devices
12562 must be of different types and compatible - the exact meaning of this is
12563 system-dependent, but typically it means that they must refer to the same
12564 underlying hardware context (for example, refer to the same graphics card).
12566 The following additional parameters are accepted:
12569 @item derive_device @var{type}
12570 Rather than using the device supplied at initialisation, instead derive a new
12571 device of type @var{type} from the device the input frames exist on.
12574 @anchor{hwupload_cuda}
12575 @section hwupload_cuda
12577 Upload system memory frames to a CUDA device.
12579 It accepts the following optional parameters:
12583 The number of the CUDA device to use
12588 Apply a high-quality magnification filter designed for pixel art. This filter
12589 was originally created by Maxim Stepin.
12591 It accepts the following option:
12595 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12596 @code{hq3x} and @code{4} for @code{hq4x}.
12597 Default is @code{3}.
12601 Stack input videos horizontally.
12603 All streams must be of same pixel format and of same height.
12605 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12606 to create same output.
12608 The filter accepts the following option:
12612 Set number of input streams. Default is 2.
12615 If set to 1, force the output to terminate when the shortest input
12616 terminates. Default value is 0.
12621 Modify the hue and/or the saturation of the input.
12623 It accepts the following parameters:
12627 Specify the hue angle as a number of degrees. It accepts an expression,
12628 and defaults to "0".
12631 Specify the saturation in the [-10,10] range. It accepts an expression and
12635 Specify the hue angle as a number of radians. It accepts an
12636 expression, and defaults to "0".
12639 Specify the brightness in the [-10,10] range. It accepts an expression and
12643 @option{h} and @option{H} are mutually exclusive, and can't be
12644 specified at the same time.
12646 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12647 expressions containing the following constants:
12651 frame count of the input frame starting from 0
12654 presentation timestamp of the input frame expressed in time base units
12657 frame rate of the input video, NAN if the input frame rate is unknown
12660 timestamp expressed in seconds, NAN if the input timestamp is unknown
12663 time base of the input video
12666 @subsection Examples
12670 Set the hue to 90 degrees and the saturation to 1.0:
12676 Same command but expressing the hue in radians:
12682 Rotate hue and make the saturation swing between 0
12683 and 2 over a period of 1 second:
12685 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12689 Apply a 3 seconds saturation fade-in effect starting at 0:
12691 hue="s=min(t/3\,1)"
12694 The general fade-in expression can be written as:
12696 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12700 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12702 hue="s=max(0\, min(1\, (8-t)/3))"
12705 The general fade-out expression can be written as:
12707 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12712 @subsection Commands
12714 This filter supports the following commands:
12720 Modify the hue and/or the saturation and/or brightness of the input video.
12721 The command accepts the same syntax of the corresponding option.
12723 If the specified expression is not valid, it is kept at its current
12727 @section hysteresis
12729 Grow first stream into second stream by connecting components.
12730 This makes it possible to build more robust edge masks.
12732 This filter accepts the following options:
12736 Set which planes will be processed as bitmap, unprocessed planes will be
12737 copied from first stream.
12738 By default value 0xf, all planes will be processed.
12741 Set threshold which is used in filtering. If pixel component value is higher than
12742 this value filter algorithm for connecting components is activated.
12743 By default value is 0.
12746 The @code{hysteresis} filter also supports the @ref{framesync} options.
12750 Detect video interlacing type.
12752 This filter tries to detect if the input frames are interlaced, progressive,
12753 top or bottom field first. It will also try to detect fields that are
12754 repeated between adjacent frames (a sign of telecine).
12756 Single frame detection considers only immediately adjacent frames when classifying each frame.
12757 Multiple frame detection incorporates the classification history of previous frames.
12759 The filter will log these metadata values:
12762 @item single.current_frame
12763 Detected type of current frame using single-frame detection. One of:
12764 ``tff'' (top field first), ``bff'' (bottom field first),
12765 ``progressive'', or ``undetermined''
12768 Cumulative number of frames detected as top field first using single-frame detection.
12771 Cumulative number of frames detected as top field first using multiple-frame detection.
12774 Cumulative number of frames detected as bottom field first using single-frame detection.
12776 @item multiple.current_frame
12777 Detected type of current frame using multiple-frame detection. One of:
12778 ``tff'' (top field first), ``bff'' (bottom field first),
12779 ``progressive'', or ``undetermined''
12782 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12784 @item single.progressive
12785 Cumulative number of frames detected as progressive using single-frame detection.
12787 @item multiple.progressive
12788 Cumulative number of frames detected as progressive using multiple-frame detection.
12790 @item single.undetermined
12791 Cumulative number of frames that could not be classified using single-frame detection.
12793 @item multiple.undetermined
12794 Cumulative number of frames that could not be classified using multiple-frame detection.
12796 @item repeated.current_frame
12797 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12799 @item repeated.neither
12800 Cumulative number of frames with no repeated field.
12803 Cumulative number of frames with the top field repeated from the previous frame's top field.
12805 @item repeated.bottom
12806 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12809 The filter accepts the following options:
12813 Set interlacing threshold.
12815 Set progressive threshold.
12817 Threshold for repeated field detection.
12819 Number of frames after which a given frame's contribution to the
12820 statistics is halved (i.e., it contributes only 0.5 to its
12821 classification). The default of 0 means that all frames seen are given
12822 full weight of 1.0 forever.
12823 @item analyze_interlaced_flag
12824 When this is not 0 then idet will use the specified number of frames to determine
12825 if the interlaced flag is accurate, it will not count undetermined frames.
12826 If the flag is found to be accurate it will be used without any further
12827 computations, if it is found to be inaccurate it will be cleared without any
12828 further computations. This allows inserting the idet filter as a low computational
12829 method to clean up the interlaced flag
12834 Deinterleave or interleave fields.
12836 This filter allows one to process interlaced images fields without
12837 deinterlacing them. Deinterleaving splits the input frame into 2
12838 fields (so called half pictures). Odd lines are moved to the top
12839 half of the output image, even lines to the bottom half.
12840 You can process (filter) them independently and then re-interleave them.
12842 The filter accepts the following options:
12846 @item chroma_mode, c
12847 @item alpha_mode, a
12848 Available values for @var{luma_mode}, @var{chroma_mode} and
12849 @var{alpha_mode} are:
12855 @item deinterleave, d
12856 Deinterleave fields, placing one above the other.
12858 @item interleave, i
12859 Interleave fields. Reverse the effect of deinterleaving.
12861 Default value is @code{none}.
12863 @item luma_swap, ls
12864 @item chroma_swap, cs
12865 @item alpha_swap, as
12866 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12869 @subsection Commands
12871 This filter supports the all above options as @ref{commands}.
12875 Apply inflate effect to the video.
12877 This filter replaces the pixel by the local(3x3) average by taking into account
12878 only values higher than the pixel.
12880 It accepts the following options:
12887 Limit the maximum change for each plane, default is 65535.
12888 If 0, plane will remain unchanged.
12891 @subsection Commands
12893 This filter supports the all above options as @ref{commands}.
12897 Simple interlacing filter from progressive contents. This interleaves upper (or
12898 lower) lines from odd frames with lower (or upper) lines from even frames,
12899 halving the frame rate and preserving image height.
12902 Original Original New Frame
12903 Frame 'j' Frame 'j+1' (tff)
12904 ========== =========== ==================
12905 Line 0 --------------------> Frame 'j' Line 0
12906 Line 1 Line 1 ----> Frame 'j+1' Line 1
12907 Line 2 ---------------------> Frame 'j' Line 2
12908 Line 3 Line 3 ----> Frame 'j+1' Line 3
12910 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12913 It accepts the following optional parameters:
12917 This determines whether the interlaced frame is taken from the even
12918 (tff - default) or odd (bff) lines of the progressive frame.
12921 Vertical lowpass filter to avoid twitter interlacing and
12922 reduce moire patterns.
12926 Disable vertical lowpass filter
12929 Enable linear filter (default)
12932 Enable complex filter. This will slightly less reduce twitter and moire
12933 but better retain detail and subjective sharpness impression.
12940 Deinterlace input video by applying Donald Graft's adaptive kernel
12941 deinterling. Work on interlaced parts of a video to produce
12942 progressive frames.
12944 The description of the accepted parameters follows.
12948 Set the threshold which affects the filter's tolerance when
12949 determining if a pixel line must be processed. It must be an integer
12950 in the range [0,255] and defaults to 10. A value of 0 will result in
12951 applying the process on every pixels.
12954 Paint pixels exceeding the threshold value to white if set to 1.
12958 Set the fields order. Swap fields if set to 1, leave fields alone if
12962 Enable additional sharpening if set to 1. Default is 0.
12965 Enable twoway sharpening if set to 1. Default is 0.
12968 @subsection Examples
12972 Apply default values:
12974 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12978 Enable additional sharpening:
12984 Paint processed pixels in white:
12992 Slowly update darker pixels.
12994 This filter makes short flashes of light appear longer.
12995 This filter accepts the following options:
12999 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13002 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13005 @section lenscorrection
13007 Correct radial lens distortion
13009 This filter can be used to correct for radial distortion as can result from the use
13010 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13011 one can use tools available for example as part of opencv or simply trial-and-error.
13012 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13013 and extract the k1 and k2 coefficients from the resulting matrix.
13015 Note that effectively the same filter is available in the open-source tools Krita and
13016 Digikam from the KDE project.
13018 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13019 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13020 brightness distribution, so you may want to use both filters together in certain
13021 cases, though you will have to take care of ordering, i.e. whether vignetting should
13022 be applied before or after lens correction.
13024 @subsection Options
13026 The filter accepts the following options:
13030 Relative x-coordinate of the focal point of the image, and thereby the center of the
13031 distortion. This value has a range [0,1] and is expressed as fractions of the image
13032 width. Default is 0.5.
13034 Relative y-coordinate of the focal point of the image, and thereby the center of the
13035 distortion. This value has a range [0,1] and is expressed as fractions of the image
13036 height. Default is 0.5.
13038 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13039 no correction. Default is 0.
13041 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13042 0 means no correction. Default is 0.
13045 The formula that generates the correction is:
13047 @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)
13049 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13050 distances from the focal point in the source and target images, respectively.
13054 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13056 The @code{lensfun} filter requires the camera make, camera model, and lens model
13057 to apply the lens correction. The filter will load the lensfun database and
13058 query it to find the corresponding camera and lens entries in the database. As
13059 long as these entries can be found with the given options, the filter can
13060 perform corrections on frames. Note that incomplete strings will result in the
13061 filter choosing the best match with the given options, and the filter will
13062 output the chosen camera and lens models (logged with level "info"). You must
13063 provide the make, camera model, and lens model as they are required.
13065 The filter accepts the following options:
13069 The make of the camera (for example, "Canon"). This option is required.
13072 The model of the camera (for example, "Canon EOS 100D"). This option is
13076 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13077 option is required.
13080 The type of correction to apply. The following values are valid options:
13084 Enables fixing lens vignetting.
13087 Enables fixing lens geometry. This is the default.
13090 Enables fixing chromatic aberrations.
13093 Enables fixing lens vignetting and lens geometry.
13096 Enables fixing lens vignetting and chromatic aberrations.
13099 Enables fixing both lens geometry and chromatic aberrations.
13102 Enables all possible corrections.
13106 The focal length of the image/video (zoom; expected constant for video). For
13107 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13108 range should be chosen when using that lens. Default 18.
13111 The aperture of the image/video (expected constant for video). Note that
13112 aperture is only used for vignetting correction. Default 3.5.
13114 @item focus_distance
13115 The focus distance of the image/video (expected constant for video). Note that
13116 focus distance is only used for vignetting and only slightly affects the
13117 vignetting correction process. If unknown, leave it at the default value (which
13121 The scale factor which is applied after transformation. After correction the
13122 video is no longer necessarily rectangular. This parameter controls how much of
13123 the resulting image is visible. The value 0 means that a value will be chosen
13124 automatically such that there is little or no unmapped area in the output
13125 image. 1.0 means that no additional scaling is done. Lower values may result
13126 in more of the corrected image being visible, while higher values may avoid
13127 unmapped areas in the output.
13129 @item target_geometry
13130 The target geometry of the output image/video. The following values are valid
13134 @item rectilinear (default)
13137 @item equirectangular
13138 @item fisheye_orthographic
13139 @item fisheye_stereographic
13140 @item fisheye_equisolid
13141 @item fisheye_thoby
13144 Apply the reverse of image correction (instead of correcting distortion, apply
13147 @item interpolation
13148 The type of interpolation used when correcting distortion. The following values
13153 @item linear (default)
13158 @subsection Examples
13162 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13163 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13167 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
13171 Apply the same as before, but only for the first 5 seconds of video.
13174 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
13181 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13182 score between two input videos.
13184 The obtained VMAF score is printed through the logging system.
13186 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13187 After installing the library it can be enabled using:
13188 @code{./configure --enable-libvmaf}.
13189 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13191 The filter has following options:
13195 Set the model path which is to be used for SVM.
13196 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13199 Set the file path to be used to store logs.
13202 Set the format of the log file (csv, json or xml).
13204 @item enable_transform
13205 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13206 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13207 Default value: @code{false}
13210 Invokes the phone model which will generate VMAF scores higher than in the
13211 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13212 Default value: @code{false}
13215 Enables computing psnr along with vmaf.
13216 Default value: @code{false}
13219 Enables computing ssim along with vmaf.
13220 Default value: @code{false}
13223 Enables computing ms_ssim along with vmaf.
13224 Default value: @code{false}
13227 Set the pool method to be used for computing vmaf.
13228 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13231 Set number of threads to be used when computing vmaf.
13232 Default value: @code{0}, which makes use of all available logical processors.
13235 Set interval for frame subsampling used when computing vmaf.
13236 Default value: @code{1}
13238 @item enable_conf_interval
13239 Enables confidence interval.
13240 Default value: @code{false}
13243 This filter also supports the @ref{framesync} options.
13245 @subsection Examples
13248 On the below examples the input file @file{main.mpg} being processed is
13249 compared with the reference file @file{ref.mpg}.
13252 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13256 Example with options:
13258 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13262 Example with options and different containers:
13264 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 -
13270 Limits the pixel components values to the specified range [min, max].
13272 The filter accepts the following options:
13276 Lower bound. Defaults to the lowest allowed value for the input.
13279 Upper bound. Defaults to the highest allowed value for the input.
13282 Specify which planes will be processed. Defaults to all available.
13289 The filter accepts the following options:
13293 Set the number of loops. Setting this value to -1 will result in infinite loops.
13297 Set maximal size in number of frames. Default is 0.
13300 Set first frame of loop. Default is 0.
13303 @subsection Examples
13307 Loop single first frame infinitely:
13309 loop=loop=-1:size=1:start=0
13313 Loop single first frame 10 times:
13315 loop=loop=10:size=1:start=0
13319 Loop 10 first frames 5 times:
13321 loop=loop=5:size=10:start=0
13327 Apply a 1D LUT to an input video.
13329 The filter accepts the following options:
13333 Set the 1D LUT file name.
13335 Currently supported formats:
13344 Select interpolation mode.
13346 Available values are:
13350 Use values from the nearest defined point.
13352 Interpolate values using the linear interpolation.
13354 Interpolate values using the cosine interpolation.
13356 Interpolate values using the cubic interpolation.
13358 Interpolate values using the spline interpolation.
13365 Apply a 3D LUT to an input video.
13367 The filter accepts the following options:
13371 Set the 3D LUT file name.
13373 Currently supported formats:
13387 Select interpolation mode.
13389 Available values are:
13393 Use values from the nearest defined point.
13395 Interpolate values using the 8 points defining a cube.
13397 Interpolate values using a tetrahedron.
13403 Turn certain luma values into transparency.
13405 The filter accepts the following options:
13409 Set the luma which will be used as base for transparency.
13410 Default value is @code{0}.
13413 Set the range of luma values to be keyed out.
13414 Default value is @code{0.01}.
13417 Set the range of softness. Default value is @code{0}.
13418 Use this to control gradual transition from zero to full transparency.
13421 @subsection Commands
13422 This filter supports same @ref{commands} as options.
13423 The command accepts the same syntax of the corresponding option.
13425 If the specified expression is not valid, it is kept at its current
13428 @section lut, lutrgb, lutyuv
13430 Compute a look-up table for binding each pixel component input value
13431 to an output value, and apply it to the input video.
13433 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13434 to an RGB input video.
13436 These filters accept the following parameters:
13439 set first pixel component expression
13441 set second pixel component expression
13443 set third pixel component expression
13445 set fourth pixel component expression, corresponds to the alpha component
13448 set red component expression
13450 set green component expression
13452 set blue component expression
13454 alpha component expression
13457 set Y/luminance component expression
13459 set U/Cb component expression
13461 set V/Cr component expression
13464 Each of them specifies the expression to use for computing the lookup table for
13465 the corresponding pixel component values.
13467 The exact component associated to each of the @var{c*} options depends on the
13470 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13471 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13473 The expressions can contain the following constants and functions:
13478 The input width and height.
13481 The input value for the pixel component.
13484 The input value, clipped to the @var{minval}-@var{maxval} range.
13487 The maximum value for the pixel component.
13490 The minimum value for the pixel component.
13493 The negated value for the pixel component value, clipped to the
13494 @var{minval}-@var{maxval} range; it corresponds to the expression
13495 "maxval-clipval+minval".
13498 The computed value in @var{val}, clipped to the
13499 @var{minval}-@var{maxval} range.
13501 @item gammaval(gamma)
13502 The computed gamma correction value of the pixel component value,
13503 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13505 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13509 All expressions default to "val".
13511 @subsection Examples
13515 Negate input video:
13517 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13518 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13521 The above is the same as:
13523 lutrgb="r=negval:g=negval:b=negval"
13524 lutyuv="y=negval:u=negval:v=negval"
13534 Remove chroma components, turning the video into a graytone image:
13536 lutyuv="u=128:v=128"
13540 Apply a luma burning effect:
13546 Remove green and blue components:
13552 Set a constant alpha channel value on input:
13554 format=rgba,lutrgb=a="maxval-minval/2"
13558 Correct luminance gamma by a factor of 0.5:
13560 lutyuv=y=gammaval(0.5)
13564 Discard least significant bits of luma:
13566 lutyuv=y='bitand(val, 128+64+32)'
13570 Technicolor like effect:
13572 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13576 @section lut2, tlut2
13578 The @code{lut2} filter takes two input streams and outputs one
13581 The @code{tlut2} (time lut2) filter takes two consecutive frames
13582 from one single stream.
13584 This filter accepts the following parameters:
13587 set first pixel component expression
13589 set second pixel component expression
13591 set third pixel component expression
13593 set fourth pixel component expression, corresponds to the alpha component
13596 set output bit depth, only available for @code{lut2} filter. By default is 0,
13597 which means bit depth is automatically picked from first input format.
13600 The @code{lut2} filter also supports the @ref{framesync} options.
13602 Each of them specifies the expression to use for computing the lookup table for
13603 the corresponding pixel component values.
13605 The exact component associated to each of the @var{c*} options depends on the
13608 The expressions can contain the following constants:
13613 The input width and height.
13616 The first input value for the pixel component.
13619 The second input value for the pixel component.
13622 The first input video bit depth.
13625 The second input video bit depth.
13628 All expressions default to "x".
13630 @subsection Examples
13634 Highlight differences between two RGB video streams:
13636 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)'
13640 Highlight differences between two YUV video streams:
13642 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)'
13646 Show max difference between two video streams:
13648 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)))'
13652 @section maskedclamp
13654 Clamp the first input stream with the second input and third input stream.
13656 Returns the value of first stream to be between second input
13657 stream - @code{undershoot} and third input stream + @code{overshoot}.
13659 This filter accepts the following options:
13662 Default value is @code{0}.
13665 Default value is @code{0}.
13668 Set which planes will be processed as bitmap, unprocessed planes will be
13669 copied from first stream.
13670 By default value 0xf, all planes will be processed.
13675 Merge the second and third input stream into output stream using absolute differences
13676 between second input stream and first input stream and absolute difference between
13677 third input stream and first input stream. The picked value will be from second input
13678 stream if second absolute difference is greater than first one or from third input stream
13681 This filter accepts the following options:
13684 Set which planes will be processed as bitmap, unprocessed planes will be
13685 copied from first stream.
13686 By default value 0xf, all planes will be processed.
13689 @section maskedmerge
13691 Merge the first input stream with the second input stream using per pixel
13692 weights in the third input stream.
13694 A value of 0 in the third stream pixel component means that pixel component
13695 from first stream is returned unchanged, while maximum value (eg. 255 for
13696 8-bit videos) means that pixel component from second stream is returned
13697 unchanged. Intermediate values define the amount of merging between both
13698 input stream's pixel components.
13700 This filter accepts the following options:
13703 Set which planes will be processed as bitmap, unprocessed planes will be
13704 copied from first stream.
13705 By default value 0xf, all planes will be processed.
13710 Merge the second and third input stream into output stream using absolute differences
13711 between second input stream and first input stream and absolute difference between
13712 third input stream and first input stream. The picked value will be from second input
13713 stream if second absolute difference is less than first one or from third input stream
13716 This filter accepts the following options:
13719 Set which planes will be processed as bitmap, unprocessed planes will be
13720 copied from first stream.
13721 By default value 0xf, all planes will be processed.
13724 @section maskedthreshold
13725 Pick pixels comparing absolute difference of two video streams with fixed
13728 If absolute difference between pixel component of first and second video
13729 stream is equal or lower than user supplied threshold than pixel component
13730 from first video stream is picked, otherwise pixel component from second
13731 video stream is picked.
13733 This filter accepts the following options:
13736 Set threshold used when picking pixels from absolute difference from two input
13740 Set which planes will be processed as bitmap, unprocessed planes will be
13741 copied from second stream.
13742 By default value 0xf, all planes will be processed.
13746 Create mask from input video.
13748 For example it is useful to create motion masks after @code{tblend} filter.
13750 This filter accepts the following options:
13754 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13757 Set high threshold. Any pixel component higher than this value will be set to max value
13758 allowed for current pixel format.
13761 Set planes to filter, by default all available planes are filtered.
13764 Fill all frame pixels with this value.
13767 Set max average pixel value for frame. If sum of all pixel components is higher that this
13768 average, output frame will be completely filled with value set by @var{fill} option.
13769 Typically useful for scene changes when used in combination with @code{tblend} filter.
13774 Apply motion-compensation deinterlacing.
13776 It needs one field per frame as input and must thus be used together
13777 with yadif=1/3 or equivalent.
13779 This filter accepts the following options:
13782 Set the deinterlacing mode.
13784 It accepts one of the following values:
13789 use iterative motion estimation
13791 like @samp{slow}, but use multiple reference frames.
13793 Default value is @samp{fast}.
13796 Set the picture field parity assumed for the input video. It must be
13797 one of the following values:
13801 assume top field first
13803 assume bottom field first
13806 Default value is @samp{bff}.
13809 Set per-block quantization parameter (QP) used by the internal
13812 Higher values should result in a smoother motion vector field but less
13813 optimal individual vectors. Default value is 1.
13818 Pick median pixel from certain rectangle defined by radius.
13820 This filter accepts the following options:
13824 Set horizontal radius size. Default value is @code{1}.
13825 Allowed range is integer from 1 to 127.
13828 Set which planes to process. Default is @code{15}, which is all available planes.
13831 Set vertical radius size. Default value is @code{0}.
13832 Allowed range is integer from 0 to 127.
13833 If it is 0, value will be picked from horizontal @code{radius} option.
13836 Set median percentile. Default value is @code{0.5}.
13837 Default value of @code{0.5} will pick always median values, while @code{0} will pick
13838 minimum values, and @code{1} maximum values.
13841 @subsection Commands
13842 This filter supports same @ref{commands} as options.
13843 The command accepts the same syntax of the corresponding option.
13845 If the specified expression is not valid, it is kept at its current
13848 @section mergeplanes
13850 Merge color channel components from several video streams.
13852 The filter accepts up to 4 input streams, and merge selected input
13853 planes to the output video.
13855 This filter accepts the following options:
13858 Set input to output plane mapping. Default is @code{0}.
13860 The mappings is specified as a bitmap. It should be specified as a
13861 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13862 mapping for the first plane of the output stream. 'A' sets the number of
13863 the input stream to use (from 0 to 3), and 'a' the plane number of the
13864 corresponding input to use (from 0 to 3). The rest of the mappings is
13865 similar, 'Bb' describes the mapping for the output stream second
13866 plane, 'Cc' describes the mapping for the output stream third plane and
13867 'Dd' describes the mapping for the output stream fourth plane.
13870 Set output pixel format. Default is @code{yuva444p}.
13873 @subsection Examples
13877 Merge three gray video streams of same width and height into single video stream:
13879 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13883 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13885 [a0][a1]mergeplanes=0x00010210:yuva444p
13889 Swap Y and A plane in yuva444p stream:
13891 format=yuva444p,mergeplanes=0x03010200:yuva444p
13895 Swap U and V plane in yuv420p stream:
13897 format=yuv420p,mergeplanes=0x000201:yuv420p
13901 Cast a rgb24 clip to yuv444p:
13903 format=rgb24,mergeplanes=0x000102:yuv444p
13909 Estimate and export motion vectors using block matching algorithms.
13910 Motion vectors are stored in frame side data to be used by other filters.
13912 This filter accepts the following options:
13915 Specify the motion estimation method. Accepts one of the following values:
13919 Exhaustive search algorithm.
13921 Three step search algorithm.
13923 Two dimensional logarithmic search algorithm.
13925 New three step search algorithm.
13927 Four step search algorithm.
13929 Diamond search algorithm.
13931 Hexagon-based search algorithm.
13933 Enhanced predictive zonal search algorithm.
13935 Uneven multi-hexagon search algorithm.
13937 Default value is @samp{esa}.
13940 Macroblock size. Default @code{16}.
13943 Search parameter. Default @code{7}.
13946 @section midequalizer
13948 Apply Midway Image Equalization effect using two video streams.
13950 Midway Image Equalization adjusts a pair of images to have the same
13951 histogram, while maintaining their dynamics as much as possible. It's
13952 useful for e.g. matching exposures from a pair of stereo cameras.
13954 This filter has two inputs and one output, which must be of same pixel format, but
13955 may be of different sizes. The output of filter is first input adjusted with
13956 midway histogram of both inputs.
13958 This filter accepts the following option:
13962 Set which planes to process. Default is @code{15}, which is all available planes.
13965 @section minterpolate
13967 Convert the video to specified frame rate using motion interpolation.
13969 This filter accepts the following options:
13972 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}.
13975 Motion interpolation mode. Following values are accepted:
13978 Duplicate previous or next frame for interpolating new ones.
13980 Blend source frames. Interpolated frame is mean of previous and next frames.
13982 Motion compensated interpolation. Following options are effective when this mode is selected:
13986 Motion compensation mode. Following values are accepted:
13989 Overlapped block motion compensation.
13991 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13993 Default mode is @samp{obmc}.
13996 Motion estimation mode. Following values are accepted:
13999 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14001 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14003 Default mode is @samp{bilat}.
14006 The algorithm to be used for motion estimation. Following values are accepted:
14009 Exhaustive search algorithm.
14011 Three step search algorithm.
14013 Two dimensional logarithmic search algorithm.
14015 New three step search algorithm.
14017 Four step search algorithm.
14019 Diamond search algorithm.
14021 Hexagon-based search algorithm.
14023 Enhanced predictive zonal search algorithm.
14025 Uneven multi-hexagon search algorithm.
14027 Default algorithm is @samp{epzs}.
14030 Macroblock size. Default @code{16}.
14033 Motion estimation search parameter. Default @code{32}.
14036 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).
14041 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:
14044 Disable scene change detection.
14046 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14048 Default method is @samp{fdiff}.
14050 @item scd_threshold
14051 Scene change detection threshold. Default is @code{10.}.
14056 Mix several video input streams into one video stream.
14058 A description of the accepted options follows.
14062 The number of inputs. If unspecified, it defaults to 2.
14065 Specify weight of each input video stream as sequence.
14066 Each weight is separated by space. If number of weights
14067 is smaller than number of @var{frames} last specified
14068 weight will be used for all remaining unset weights.
14071 Specify scale, if it is set it will be multiplied with sum
14072 of each weight multiplied with pixel values to give final destination
14073 pixel value. By default @var{scale} is auto scaled to sum of weights.
14076 Specify how end of stream is determined.
14079 The duration of the longest input. (default)
14082 The duration of the shortest input.
14085 The duration of the first input.
14089 @section mpdecimate
14091 Drop frames that do not differ greatly from the previous frame in
14092 order to reduce frame rate.
14094 The main use of this filter is for very-low-bitrate encoding
14095 (e.g. streaming over dialup modem), but it could in theory be used for
14096 fixing movies that were inverse-telecined incorrectly.
14098 A description of the accepted options follows.
14102 Set the maximum number of consecutive frames which can be dropped (if
14103 positive), or the minimum interval between dropped frames (if
14104 negative). If the value is 0, the frame is dropped disregarding the
14105 number of previous sequentially dropped frames.
14107 Default value is 0.
14112 Set the dropping threshold values.
14114 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14115 represent actual pixel value differences, so a threshold of 64
14116 corresponds to 1 unit of difference for each pixel, or the same spread
14117 out differently over the block.
14119 A frame is a candidate for dropping if no 8x8 blocks differ by more
14120 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14121 meaning the whole image) differ by more than a threshold of @option{lo}.
14123 Default value for @option{hi} is 64*12, default value for @option{lo} is
14124 64*5, and default value for @option{frac} is 0.33.
14130 Negate (invert) the input video.
14132 It accepts the following option:
14137 With value 1, it negates the alpha component, if present. Default value is 0.
14143 Denoise frames using Non-Local Means algorithm.
14145 Each pixel is adjusted by looking for other pixels with similar contexts. This
14146 context similarity is defined by comparing their surrounding patches of size
14147 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14150 Note that the research area defines centers for patches, which means some
14151 patches will be made of pixels outside that research area.
14153 The filter accepts the following options.
14157 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14160 Set patch size. Default is 7. Must be odd number in range [0, 99].
14163 Same as @option{p} but for chroma planes.
14165 The default value is @var{0} and means automatic.
14168 Set research size. Default is 15. Must be odd number in range [0, 99].
14171 Same as @option{r} but for chroma planes.
14173 The default value is @var{0} and means automatic.
14178 Deinterlace video using neural network edge directed interpolation.
14180 This filter accepts the following options:
14184 Mandatory option, without binary file filter can not work.
14185 Currently file can be found here:
14186 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14189 Set which frames to deinterlace, by default it is @code{all}.
14190 Can be @code{all} or @code{interlaced}.
14193 Set mode of operation.
14195 Can be one of the following:
14199 Use frame flags, both fields.
14201 Use frame flags, single field.
14203 Use top field only.
14205 Use bottom field only.
14207 Use both fields, top first.
14209 Use both fields, bottom first.
14213 Set which planes to process, by default filter process all frames.
14216 Set size of local neighborhood around each pixel, used by the predictor neural
14219 Can be one of the following:
14232 Set the number of neurons in predictor neural network.
14233 Can be one of the following:
14244 Controls the number of different neural network predictions that are blended
14245 together to compute the final output value. Can be @code{fast}, default or
14249 Set which set of weights to use in the predictor.
14250 Can be one of the following:
14254 weights trained to minimize absolute error
14256 weights trained to minimize squared error
14260 Controls whether or not the prescreener neural network is used to decide
14261 which pixels should be processed by the predictor neural network and which
14262 can be handled by simple cubic interpolation.
14263 The prescreener is trained to know whether cubic interpolation will be
14264 sufficient for a pixel or whether it should be predicted by the predictor nn.
14265 The computational complexity of the prescreener nn is much less than that of
14266 the predictor nn. Since most pixels can be handled by cubic interpolation,
14267 using the prescreener generally results in much faster processing.
14268 The prescreener is pretty accurate, so the difference between using it and not
14269 using it is almost always unnoticeable.
14271 Can be one of the following:
14279 Default is @code{new}.
14282 Set various debugging flags.
14287 Force libavfilter not to use any of the specified pixel formats for the
14288 input to the next filter.
14290 It accepts the following parameters:
14294 A '|'-separated list of pixel format names, such as
14295 pix_fmts=yuv420p|monow|rgb24".
14299 @subsection Examples
14303 Force libavfilter to use a format different from @var{yuv420p} for the
14304 input to the vflip filter:
14306 noformat=pix_fmts=yuv420p,vflip
14310 Convert the input video to any of the formats not contained in the list:
14312 noformat=yuv420p|yuv444p|yuv410p
14318 Add noise on video input frame.
14320 The filter accepts the following options:
14328 Set noise seed for specific pixel component or all pixel components in case
14329 of @var{all_seed}. Default value is @code{123457}.
14331 @item all_strength, alls
14332 @item c0_strength, c0s
14333 @item c1_strength, c1s
14334 @item c2_strength, c2s
14335 @item c3_strength, c3s
14336 Set noise strength for specific pixel component or all pixel components in case
14337 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14339 @item all_flags, allf
14340 @item c0_flags, c0f
14341 @item c1_flags, c1f
14342 @item c2_flags, c2f
14343 @item c3_flags, c3f
14344 Set pixel component flags or set flags for all components if @var{all_flags}.
14345 Available values for component flags are:
14348 averaged temporal noise (smoother)
14350 mix random noise with a (semi)regular pattern
14352 temporal noise (noise pattern changes between frames)
14354 uniform noise (gaussian otherwise)
14358 @subsection Examples
14360 Add temporal and uniform noise to input video:
14362 noise=alls=20:allf=t+u
14367 Normalize RGB video (aka histogram stretching, contrast stretching).
14368 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14370 For each channel of each frame, the filter computes the input range and maps
14371 it linearly to the user-specified output range. The output range defaults
14372 to the full dynamic range from pure black to pure white.
14374 Temporal smoothing can be used on the input range to reduce flickering (rapid
14375 changes in brightness) caused when small dark or bright objects enter or leave
14376 the scene. This is similar to the auto-exposure (automatic gain control) on a
14377 video camera, and, like a video camera, it may cause a period of over- or
14378 under-exposure of the video.
14380 The R,G,B channels can be normalized independently, which may cause some
14381 color shifting, or linked together as a single channel, which prevents
14382 color shifting. Linked normalization preserves hue. Independent normalization
14383 does not, so it can be used to remove some color casts. Independent and linked
14384 normalization can be combined in any ratio.
14386 The normalize filter accepts the following options:
14391 Colors which define the output range. The minimum input value is mapped to
14392 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14393 The defaults are black and white respectively. Specifying white for
14394 @var{blackpt} and black for @var{whitept} will give color-inverted,
14395 normalized video. Shades of grey can be used to reduce the dynamic range
14396 (contrast). Specifying saturated colors here can create some interesting
14400 The number of previous frames to use for temporal smoothing. The input range
14401 of each channel is smoothed using a rolling average over the current frame
14402 and the @var{smoothing} previous frames. The default is 0 (no temporal
14406 Controls the ratio of independent (color shifting) channel normalization to
14407 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14408 independent. Defaults to 1.0 (fully independent).
14411 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14412 expensive no-op. Defaults to 1.0 (full strength).
14416 @subsection Commands
14417 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14418 The command accepts the same syntax of the corresponding option.
14420 If the specified expression is not valid, it is kept at its current
14423 @subsection Examples
14425 Stretch video contrast to use the full dynamic range, with no temporal
14426 smoothing; may flicker depending on the source content:
14428 normalize=blackpt=black:whitept=white:smoothing=0
14431 As above, but with 50 frames of temporal smoothing; flicker should be
14432 reduced, depending on the source content:
14434 normalize=blackpt=black:whitept=white:smoothing=50
14437 As above, but with hue-preserving linked channel normalization:
14439 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14442 As above, but with half strength:
14444 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14447 Map the darkest input color to red, the brightest input color to cyan:
14449 normalize=blackpt=red:whitept=cyan
14454 Pass the video source unchanged to the output.
14457 Optical Character Recognition
14459 This filter uses Tesseract for optical character recognition. To enable
14460 compilation of this filter, you need to configure FFmpeg with
14461 @code{--enable-libtesseract}.
14463 It accepts the following options:
14467 Set datapath to tesseract data. Default is to use whatever was
14468 set at installation.
14471 Set language, default is "eng".
14474 Set character whitelist.
14477 Set character blacklist.
14480 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14481 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14485 Apply a video transform using libopencv.
14487 To enable this filter, install the libopencv library and headers and
14488 configure FFmpeg with @code{--enable-libopencv}.
14490 It accepts the following parameters:
14495 The name of the libopencv filter to apply.
14497 @item filter_params
14498 The parameters to pass to the libopencv filter. If not specified, the default
14499 values are assumed.
14503 Refer to the official libopencv documentation for more precise
14505 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14507 Several libopencv filters are supported; see the following subsections.
14512 Dilate an image by using a specific structuring element.
14513 It corresponds to the libopencv function @code{cvDilate}.
14515 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14517 @var{struct_el} represents a structuring element, and has the syntax:
14518 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14520 @var{cols} and @var{rows} represent the number of columns and rows of
14521 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14522 point, and @var{shape} the shape for the structuring element. @var{shape}
14523 must be "rect", "cross", "ellipse", or "custom".
14525 If the value for @var{shape} is "custom", it must be followed by a
14526 string of the form "=@var{filename}". The file with name
14527 @var{filename} is assumed to represent a binary image, with each
14528 printable character corresponding to a bright pixel. When a custom
14529 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14530 or columns and rows of the read file are assumed instead.
14532 The default value for @var{struct_el} is "3x3+0x0/rect".
14534 @var{nb_iterations} specifies the number of times the transform is
14535 applied to the image, and defaults to 1.
14539 # Use the default values
14542 # Dilate using a structuring element with a 5x5 cross, iterating two times
14543 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14545 # Read the shape from the file diamond.shape, iterating two times.
14546 # The file diamond.shape may contain a pattern of characters like this
14552 # The specified columns and rows are ignored
14553 # but the anchor point coordinates are not
14554 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14559 Erode an image by using a specific structuring element.
14560 It corresponds to the libopencv function @code{cvErode}.
14562 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14563 with the same syntax and semantics as the @ref{dilate} filter.
14567 Smooth the input video.
14569 The filter takes the following parameters:
14570 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14572 @var{type} is the type of smooth filter to apply, and must be one of
14573 the following values: "blur", "blur_no_scale", "median", "gaussian",
14574 or "bilateral". The default value is "gaussian".
14576 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14577 depends on the smooth type. @var{param1} and
14578 @var{param2} accept integer positive values or 0. @var{param3} and
14579 @var{param4} accept floating point values.
14581 The default value for @var{param1} is 3. The default value for the
14582 other parameters is 0.
14584 These parameters correspond to the parameters assigned to the
14585 libopencv function @code{cvSmooth}.
14587 @section oscilloscope
14589 2D Video Oscilloscope.
14591 Useful to measure spatial impulse, step responses, chroma delays, etc.
14593 It accepts the following parameters:
14597 Set scope center x position.
14600 Set scope center y position.
14603 Set scope size, relative to frame diagonal.
14606 Set scope tilt/rotation.
14612 Set trace center x position.
14615 Set trace center y position.
14618 Set trace width, relative to width of frame.
14621 Set trace height, relative to height of frame.
14624 Set which components to trace. By default it traces first three components.
14627 Draw trace grid. By default is enabled.
14630 Draw some statistics. By default is enabled.
14633 Draw scope. By default is enabled.
14636 @subsection Commands
14637 This filter supports same @ref{commands} as options.
14638 The command accepts the same syntax of the corresponding option.
14640 If the specified expression is not valid, it is kept at its current
14643 @subsection Examples
14647 Inspect full first row of video frame.
14649 oscilloscope=x=0.5:y=0:s=1
14653 Inspect full last row of video frame.
14655 oscilloscope=x=0.5:y=1:s=1
14659 Inspect full 5th line of video frame of height 1080.
14661 oscilloscope=x=0.5:y=5/1080:s=1
14665 Inspect full last column of video frame.
14667 oscilloscope=x=1:y=0.5:s=1:t=1
14675 Overlay one video on top of another.
14677 It takes two inputs and has one output. The first input is the "main"
14678 video on which the second input is overlaid.
14680 It accepts the following parameters:
14682 A description of the accepted options follows.
14687 Set the expression for the x and y coordinates of the overlaid video
14688 on the main video. Default value is "0" for both expressions. In case
14689 the expression is invalid, it is set to a huge value (meaning that the
14690 overlay will not be displayed within the output visible area).
14693 See @ref{framesync}.
14696 Set when the expressions for @option{x}, and @option{y} are evaluated.
14698 It accepts the following values:
14701 only evaluate expressions once during the filter initialization or
14702 when a command is processed
14705 evaluate expressions for each incoming frame
14708 Default value is @samp{frame}.
14711 See @ref{framesync}.
14714 Set the format for the output video.
14716 It accepts the following values:
14719 force YUV420 output
14722 force YUV420p10 output
14725 force YUV422 output
14728 force YUV422p10 output
14731 force YUV444 output
14734 force packed RGB output
14737 force planar RGB output
14740 automatically pick format
14743 Default value is @samp{yuv420}.
14746 See @ref{framesync}.
14749 Set format of alpha of the overlaid video, it can be @var{straight} or
14750 @var{premultiplied}. Default is @var{straight}.
14753 The @option{x}, and @option{y} expressions can contain the following
14759 The main input width and height.
14763 The overlay input width and height.
14767 The computed values for @var{x} and @var{y}. They are evaluated for
14772 horizontal and vertical chroma subsample values of the output
14773 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14777 the number of input frame, starting from 0
14780 the position in the file of the input frame, NAN if unknown
14783 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14787 This filter also supports the @ref{framesync} options.
14789 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14790 when evaluation is done @emph{per frame}, and will evaluate to NAN
14791 when @option{eval} is set to @samp{init}.
14793 Be aware that frames are taken from each input video in timestamp
14794 order, hence, if their initial timestamps differ, it is a good idea
14795 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14796 have them begin in the same zero timestamp, as the example for
14797 the @var{movie} filter does.
14799 You can chain together more overlays but you should test the
14800 efficiency of such approach.
14802 @subsection Commands
14804 This filter supports the following commands:
14808 Modify the x and y of the overlay input.
14809 The command accepts the same syntax of the corresponding option.
14811 If the specified expression is not valid, it is kept at its current
14815 @subsection Examples
14819 Draw the overlay at 10 pixels from the bottom right corner of the main
14822 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14825 Using named options the example above becomes:
14827 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14831 Insert a transparent PNG logo in the bottom left corner of the input,
14832 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14834 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14838 Insert 2 different transparent PNG logos (second logo on bottom
14839 right corner) using the @command{ffmpeg} tool:
14841 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
14845 Add a transparent color layer on top of the main video; @code{WxH}
14846 must specify the size of the main input to the overlay filter:
14848 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14852 Play an original video and a filtered version (here with the deshake
14853 filter) side by side using the @command{ffplay} tool:
14855 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14858 The above command is the same as:
14860 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14864 Make a sliding overlay appearing from the left to the right top part of the
14865 screen starting since time 2:
14867 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14871 Compose output by putting two input videos side to side:
14873 ffmpeg -i left.avi -i right.avi -filter_complex "
14874 nullsrc=size=200x100 [background];
14875 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14876 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14877 [background][left] overlay=shortest=1 [background+left];
14878 [background+left][right] overlay=shortest=1:x=100 [left+right]
14883 Mask 10-20 seconds of a video by applying the delogo filter to a section
14885 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14886 -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]'
14891 Chain several overlays in cascade:
14893 nullsrc=s=200x200 [bg];
14894 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14895 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14896 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14897 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14898 [in3] null, [mid2] overlay=100:100 [out0]
14903 @anchor{overlay_cuda}
14904 @section overlay_cuda
14906 Overlay one video on top of another.
14908 This is the CUDA variant of the @ref{overlay} filter.
14909 It only accepts CUDA frames. The underlying input pixel formats have to match.
14911 It takes two inputs and has one output. The first input is the "main"
14912 video on which the second input is overlaid.
14914 It accepts the following parameters:
14919 Set the x and y coordinates of the overlaid video on the main video.
14920 Default value is "0" for both expressions.
14923 See @ref{framesync}.
14926 See @ref{framesync}.
14929 See @ref{framesync}.
14933 This filter also supports the @ref{framesync} options.
14937 Apply Overcomplete Wavelet denoiser.
14939 The filter accepts the following options:
14945 Larger depth values will denoise lower frequency components more, but
14946 slow down filtering.
14948 Must be an int in the range 8-16, default is @code{8}.
14950 @item luma_strength, ls
14953 Must be a double value in the range 0-1000, default is @code{1.0}.
14955 @item chroma_strength, cs
14956 Set chroma strength.
14958 Must be a double value in the range 0-1000, default is @code{1.0}.
14964 Add paddings to the input image, and place the original input at the
14965 provided @var{x}, @var{y} coordinates.
14967 It accepts the following parameters:
14972 Specify an expression for the size of the output image with the
14973 paddings added. If the value for @var{width} or @var{height} is 0, the
14974 corresponding input size is used for the output.
14976 The @var{width} expression can reference the value set by the
14977 @var{height} expression, and vice versa.
14979 The default value of @var{width} and @var{height} is 0.
14983 Specify the offsets to place the input image at within the padded area,
14984 with respect to the top/left border of the output image.
14986 The @var{x} expression can reference the value set by the @var{y}
14987 expression, and vice versa.
14989 The default value of @var{x} and @var{y} is 0.
14991 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14992 so the input image is centered on the padded area.
14995 Specify the color of the padded area. For the syntax of this option,
14996 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14997 manual,ffmpeg-utils}.
14999 The default value of @var{color} is "black".
15002 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15004 It accepts the following values:
15008 Only evaluate expressions once during the filter initialization or when
15009 a command is processed.
15012 Evaluate expressions for each incoming frame.
15016 Default value is @samp{init}.
15019 Pad to aspect instead to a resolution.
15023 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15024 options are expressions containing the following constants:
15029 The input video width and height.
15033 These are the same as @var{in_w} and @var{in_h}.
15037 The output width and height (the size of the padded area), as
15038 specified by the @var{width} and @var{height} expressions.
15042 These are the same as @var{out_w} and @var{out_h}.
15046 The x and y offsets as specified by the @var{x} and @var{y}
15047 expressions, or NAN if not yet specified.
15050 same as @var{iw} / @var{ih}
15053 input sample aspect ratio
15056 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15060 The horizontal and vertical chroma subsample values. For example for the
15061 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15064 @subsection Examples
15068 Add paddings with the color "violet" to the input video. The output video
15069 size is 640x480, and the top-left corner of the input video is placed at
15072 pad=640:480:0:40:violet
15075 The example above is equivalent to the following command:
15077 pad=width=640:height=480:x=0:y=40:color=violet
15081 Pad the input to get an output with dimensions increased by 3/2,
15082 and put the input video at the center of the padded area:
15084 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15088 Pad the input to get a squared output with size equal to the maximum
15089 value between the input width and height, and put the input video at
15090 the center of the padded area:
15092 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15096 Pad the input to get a final w/h ratio of 16:9:
15098 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15102 In case of anamorphic video, in order to set the output display aspect
15103 correctly, it is necessary to use @var{sar} in the expression,
15104 according to the relation:
15106 (ih * X / ih) * sar = output_dar
15107 X = output_dar / sar
15110 Thus the previous example needs to be modified to:
15112 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15116 Double the output size and put the input video in the bottom-right
15117 corner of the output padded area:
15119 pad="2*iw:2*ih:ow-iw:oh-ih"
15123 @anchor{palettegen}
15124 @section palettegen
15126 Generate one palette for a whole video stream.
15128 It accepts the following options:
15132 Set the maximum number of colors to quantize in the palette.
15133 Note: the palette will still contain 256 colors; the unused palette entries
15136 @item reserve_transparent
15137 Create a palette of 255 colors maximum and reserve the last one for
15138 transparency. Reserving the transparency color is useful for GIF optimization.
15139 If not set, the maximum of colors in the palette will be 256. You probably want
15140 to disable this option for a standalone image.
15143 @item transparency_color
15144 Set the color that will be used as background for transparency.
15147 Set statistics mode.
15149 It accepts the following values:
15152 Compute full frame histograms.
15154 Compute histograms only for the part that differs from previous frame. This
15155 might be relevant to give more importance to the moving part of your input if
15156 the background is static.
15158 Compute new histogram for each frame.
15161 Default value is @var{full}.
15164 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15165 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15166 color quantization of the palette. This information is also visible at
15167 @var{info} logging level.
15169 @subsection Examples
15173 Generate a representative palette of a given video using @command{ffmpeg}:
15175 ffmpeg -i input.mkv -vf palettegen palette.png
15179 @section paletteuse
15181 Use a palette to downsample an input video stream.
15183 The filter takes two inputs: one video stream and a palette. The palette must
15184 be a 256 pixels image.
15186 It accepts the following options:
15190 Select dithering mode. Available algorithms are:
15193 Ordered 8x8 bayer dithering (deterministic)
15195 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15196 Note: this dithering is sometimes considered "wrong" and is included as a
15198 @item floyd_steinberg
15199 Floyd and Steingberg dithering (error diffusion)
15201 Frankie Sierra dithering v2 (error diffusion)
15203 Frankie Sierra dithering v2 "Lite" (error diffusion)
15206 Default is @var{sierra2_4a}.
15209 When @var{bayer} dithering is selected, this option defines the scale of the
15210 pattern (how much the crosshatch pattern is visible). A low value means more
15211 visible pattern for less banding, and higher value means less visible pattern
15212 at the cost of more banding.
15214 The option must be an integer value in the range [0,5]. Default is @var{2}.
15217 If set, define the zone to process
15221 Only the changing rectangle will be reprocessed. This is similar to GIF
15222 cropping/offsetting compression mechanism. This option can be useful for speed
15223 if only a part of the image is changing, and has use cases such as limiting the
15224 scope of the error diffusal @option{dither} to the rectangle that bounds the
15225 moving scene (it leads to more deterministic output if the scene doesn't change
15226 much, and as a result less moving noise and better GIF compression).
15229 Default is @var{none}.
15232 Take new palette for each output frame.
15234 @item alpha_threshold
15235 Sets the alpha threshold for transparency. Alpha values above this threshold
15236 will be treated as completely opaque, and values below this threshold will be
15237 treated as completely transparent.
15239 The option must be an integer value in the range [0,255]. Default is @var{128}.
15242 @subsection Examples
15246 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15247 using @command{ffmpeg}:
15249 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15253 @section perspective
15255 Correct perspective of video not recorded perpendicular to the screen.
15257 A description of the accepted parameters follows.
15268 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15269 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15270 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15271 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15272 then the corners of the source will be sent to the specified coordinates.
15274 The expressions can use the following variables:
15279 the width and height of video frame.
15283 Output frame count.
15286 @item interpolation
15287 Set interpolation for perspective correction.
15289 It accepts the following values:
15295 Default value is @samp{linear}.
15298 Set interpretation of coordinate options.
15300 It accepts the following values:
15304 Send point in the source specified by the given coordinates to
15305 the corners of the destination.
15307 @item 1, destination
15309 Send the corners of the source to the point in the destination specified
15310 by the given coordinates.
15312 Default value is @samp{source}.
15316 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15318 It accepts the following values:
15321 only evaluate expressions once during the filter initialization or
15322 when a command is processed
15325 evaluate expressions for each incoming frame
15328 Default value is @samp{init}.
15333 Delay interlaced video by one field time so that the field order changes.
15335 The intended use is to fix PAL movies that have been captured with the
15336 opposite field order to the film-to-video transfer.
15338 A description of the accepted parameters follows.
15344 It accepts the following values:
15347 Capture field order top-first, transfer bottom-first.
15348 Filter will delay the bottom field.
15351 Capture field order bottom-first, transfer top-first.
15352 Filter will delay the top field.
15355 Capture and transfer with the same field order. This mode only exists
15356 for the documentation of the other options to refer to, but if you
15357 actually select it, the filter will faithfully do nothing.
15360 Capture field order determined automatically by field flags, transfer
15362 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15363 basis using field flags. If no field information is available,
15364 then this works just like @samp{u}.
15367 Capture unknown or varying, transfer opposite.
15368 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15369 analyzing the images and selecting the alternative that produces best
15370 match between the fields.
15373 Capture top-first, transfer unknown or varying.
15374 Filter selects among @samp{t} and @samp{p} using image analysis.
15377 Capture bottom-first, transfer unknown or varying.
15378 Filter selects among @samp{b} and @samp{p} using image analysis.
15381 Capture determined by field flags, transfer unknown or varying.
15382 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15383 image analysis. If no field information is available, then this works just
15384 like @samp{U}. This is the default mode.
15387 Both capture and transfer unknown or varying.
15388 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15392 @section photosensitivity
15393 Reduce various flashes in video, so to help users with epilepsy.
15395 It accepts the following options:
15398 Set how many frames to use when filtering. Default is 30.
15401 Set detection threshold factor. Default is 1.
15405 Set how many pixels to skip when sampling frames. Default is 1.
15406 Allowed range is from 1 to 1024.
15409 Leave frames unchanged. Default is disabled.
15412 @section pixdesctest
15414 Pixel format descriptor test filter, mainly useful for internal
15415 testing. The output video should be equal to the input video.
15419 format=monow, pixdesctest
15422 can be used to test the monowhite pixel format descriptor definition.
15426 Display sample values of color channels. Mainly useful for checking color
15427 and levels. Minimum supported resolution is 640x480.
15429 The filters accept the following options:
15433 Set scope X position, relative offset on X axis.
15436 Set scope Y position, relative offset on Y axis.
15445 Set window opacity. This window also holds statistics about pixel area.
15448 Set window X position, relative offset on X axis.
15451 Set window Y position, relative offset on Y axis.
15456 Enable the specified chain of postprocessing subfilters using libpostproc. This
15457 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15458 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15459 Each subfilter and some options have a short and a long name that can be used
15460 interchangeably, i.e. dr/dering are the same.
15462 The filters accept the following options:
15466 Set postprocessing subfilters string.
15469 All subfilters share common options to determine their scope:
15473 Honor the quality commands for this subfilter.
15476 Do chrominance filtering, too (default).
15479 Do luminance filtering only (no chrominance).
15482 Do chrominance filtering only (no luminance).
15485 These options can be appended after the subfilter name, separated by a '|'.
15487 Available subfilters are:
15490 @item hb/hdeblock[|difference[|flatness]]
15491 Horizontal deblocking filter
15494 Difference factor where higher values mean more deblocking (default: @code{32}).
15496 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15499 @item vb/vdeblock[|difference[|flatness]]
15500 Vertical deblocking filter
15503 Difference factor where higher values mean more deblocking (default: @code{32}).
15505 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15508 @item ha/hadeblock[|difference[|flatness]]
15509 Accurate horizontal deblocking filter
15512 Difference factor where higher values mean more deblocking (default: @code{32}).
15514 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15517 @item va/vadeblock[|difference[|flatness]]
15518 Accurate vertical deblocking filter
15521 Difference factor where higher values mean more deblocking (default: @code{32}).
15523 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15527 The horizontal and vertical deblocking filters share the difference and
15528 flatness values so you cannot set different horizontal and vertical
15532 @item h1/x1hdeblock
15533 Experimental horizontal deblocking filter
15535 @item v1/x1vdeblock
15536 Experimental vertical deblocking filter
15541 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15544 larger -> stronger filtering
15546 larger -> stronger filtering
15548 larger -> stronger filtering
15551 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15554 Stretch luminance to @code{0-255}.
15557 @item lb/linblenddeint
15558 Linear blend deinterlacing filter that deinterlaces the given block by
15559 filtering all lines with a @code{(1 2 1)} filter.
15561 @item li/linipoldeint
15562 Linear interpolating deinterlacing filter that deinterlaces the given block by
15563 linearly interpolating every second line.
15565 @item ci/cubicipoldeint
15566 Cubic interpolating deinterlacing filter deinterlaces the given block by
15567 cubically interpolating every second line.
15569 @item md/mediandeint
15570 Median deinterlacing filter that deinterlaces the given block by applying a
15571 median filter to every second line.
15573 @item fd/ffmpegdeint
15574 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15575 second line with a @code{(-1 4 2 4 -1)} filter.
15578 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15579 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15581 @item fq/forceQuant[|quantizer]
15582 Overrides the quantizer table from the input with the constant quantizer you
15590 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15593 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15596 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15599 @subsection Examples
15603 Apply horizontal and vertical deblocking, deringing and automatic
15604 brightness/contrast:
15610 Apply default filters without brightness/contrast correction:
15616 Apply default filters and temporal denoiser:
15618 pp=default/tmpnoise|1|2|3
15622 Apply deblocking on luminance only, and switch vertical deblocking on or off
15623 automatically depending on available CPU time:
15630 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15631 similar to spp = 6 with 7 point DCT, where only the center sample is
15634 The filter accepts the following options:
15638 Force a constant quantization parameter. It accepts an integer in range
15639 0 to 63. If not set, the filter will use the QP from the video stream
15643 Set thresholding mode. Available modes are:
15647 Set hard thresholding.
15649 Set soft thresholding (better de-ringing effect, but likely blurrier).
15651 Set medium thresholding (good results, default).
15655 @section premultiply
15656 Apply alpha premultiply effect to input video stream using first plane
15657 of second stream as alpha.
15659 Both streams must have same dimensions and same pixel format.
15661 The filter accepts the following option:
15665 Set which planes will be processed, unprocessed planes will be copied.
15666 By default value 0xf, all planes will be processed.
15669 Do not require 2nd input for processing, instead use alpha plane from input stream.
15673 Apply prewitt operator to input video stream.
15675 The filter accepts the following option:
15679 Set which planes will be processed, unprocessed planes will be copied.
15680 By default value 0xf, all planes will be processed.
15683 Set value which will be multiplied with filtered result.
15686 Set value which will be added to filtered result.
15689 @section pseudocolor
15691 Alter frame colors in video with pseudocolors.
15693 This filter accepts the following options:
15697 set pixel first component expression
15700 set pixel second component expression
15703 set pixel third component expression
15706 set pixel fourth component expression, corresponds to the alpha component
15709 set component to use as base for altering colors
15712 Each of them specifies the expression to use for computing the lookup table for
15713 the corresponding pixel component values.
15715 The expressions can contain the following constants and functions:
15720 The input width and height.
15723 The input value for the pixel component.
15725 @item ymin, umin, vmin, amin
15726 The minimum allowed component value.
15728 @item ymax, umax, vmax, amax
15729 The maximum allowed component value.
15732 All expressions default to "val".
15734 @subsection Examples
15738 Change too high luma values to gradient:
15740 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'"
15746 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15747 Ratio) between two input videos.
15749 This filter takes in input two input videos, the first input is
15750 considered the "main" source and is passed unchanged to the
15751 output. The second input is used as a "reference" video for computing
15754 Both video inputs must have the same resolution and pixel format for
15755 this filter to work correctly. Also it assumes that both inputs
15756 have the same number of frames, which are compared one by one.
15758 The obtained average PSNR is printed through the logging system.
15760 The filter stores the accumulated MSE (mean squared error) of each
15761 frame, and at the end of the processing it is averaged across all frames
15762 equally, and the following formula is applied to obtain the PSNR:
15765 PSNR = 10*log10(MAX^2/MSE)
15768 Where MAX is the average of the maximum values of each component of the
15771 The description of the accepted parameters follows.
15774 @item stats_file, f
15775 If specified the filter will use the named file to save the PSNR of
15776 each individual frame. When filename equals "-" the data is sent to
15779 @item stats_version
15780 Specifies which version of the stats file format to use. Details of
15781 each format are written below.
15782 Default value is 1.
15784 @item stats_add_max
15785 Determines whether the max value is output to the stats log.
15786 Default value is 0.
15787 Requires stats_version >= 2. If this is set and stats_version < 2,
15788 the filter will return an error.
15791 This filter also supports the @ref{framesync} options.
15793 The file printed if @var{stats_file} is selected, contains a sequence of
15794 key/value pairs of the form @var{key}:@var{value} for each compared
15797 If a @var{stats_version} greater than 1 is specified, a header line precedes
15798 the list of per-frame-pair stats, with key value pairs following the frame
15799 format with the following parameters:
15802 @item psnr_log_version
15803 The version of the log file format. Will match @var{stats_version}.
15806 A comma separated list of the per-frame-pair parameters included in
15810 A description of each shown per-frame-pair parameter follows:
15814 sequential number of the input frame, starting from 1
15817 Mean Square Error pixel-by-pixel average difference of the compared
15818 frames, averaged over all the image components.
15820 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15821 Mean Square Error pixel-by-pixel average difference of the compared
15822 frames for the component specified by the suffix.
15824 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15825 Peak Signal to Noise ratio of the compared frames for the component
15826 specified by the suffix.
15828 @item max_avg, max_y, max_u, max_v
15829 Maximum allowed value for each channel, and average over all
15833 @subsection Examples
15838 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15839 [main][ref] psnr="stats_file=stats.log" [out]
15842 On this example the input file being processed is compared with the
15843 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15844 is stored in @file{stats.log}.
15847 Another example with different containers:
15849 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 -
15856 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15857 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15860 The pullup filter is designed to take advantage of future context in making
15861 its decisions. This filter is stateless in the sense that it does not lock
15862 onto a pattern to follow, but it instead looks forward to the following
15863 fields in order to identify matches and rebuild progressive frames.
15865 To produce content with an even framerate, insert the fps filter after
15866 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15867 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15869 The filter accepts the following options:
15876 These options set the amount of "junk" to ignore at the left, right, top, and
15877 bottom of the image, respectively. Left and right are in units of 8 pixels,
15878 while top and bottom are in units of 2 lines.
15879 The default is 8 pixels on each side.
15882 Set the strict breaks. Setting this option to 1 will reduce the chances of
15883 filter generating an occasional mismatched frame, but it may also cause an
15884 excessive number of frames to be dropped during high motion sequences.
15885 Conversely, setting it to -1 will make filter match fields more easily.
15886 This may help processing of video where there is slight blurring between
15887 the fields, but may also cause there to be interlaced frames in the output.
15888 Default value is @code{0}.
15891 Set the metric plane to use. It accepts the following values:
15897 Use chroma blue plane.
15900 Use chroma red plane.
15903 This option may be set to use chroma plane instead of the default luma plane
15904 for doing filter's computations. This may improve accuracy on very clean
15905 source material, but more likely will decrease accuracy, especially if there
15906 is chroma noise (rainbow effect) or any grayscale video.
15907 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15908 load and make pullup usable in realtime on slow machines.
15911 For best results (without duplicated frames in the output file) it is
15912 necessary to change the output frame rate. For example, to inverse
15913 telecine NTSC input:
15915 ffmpeg -i input -vf pullup -r 24000/1001 ...
15920 Change video quantization parameters (QP).
15922 The filter accepts the following option:
15926 Set expression for quantization parameter.
15929 The expression is evaluated through the eval API and can contain, among others,
15930 the following constants:
15934 1 if index is not 129, 0 otherwise.
15937 Sequential index starting from -129 to 128.
15940 @subsection Examples
15944 Some equation like:
15952 Flush video frames from internal cache of frames into a random order.
15953 No frame is discarded.
15954 Inspired by @ref{frei0r} nervous filter.
15958 Set size in number of frames of internal cache, in range from @code{2} to
15959 @code{512}. Default is @code{30}.
15962 Set seed for random number generator, must be an integer included between
15963 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15964 less than @code{0}, the filter will try to use a good random seed on a
15968 @section readeia608
15970 Read closed captioning (EIA-608) information from the top lines of a video frame.
15972 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15973 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15974 with EIA-608 data (starting from 0). A description of each metadata value follows:
15977 @item lavfi.readeia608.X.cc
15978 The two bytes stored as EIA-608 data (printed in hexadecimal).
15980 @item lavfi.readeia608.X.line
15981 The number of the line on which the EIA-608 data was identified and read.
15984 This filter accepts the following options:
15988 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15991 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15994 Set the ratio of width reserved for sync code detection.
15995 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
15998 Enable checking the parity bit. In the event of a parity error, the filter will output
15999 @code{0x00} for that character. Default is false.
16002 Lowpass lines prior to further processing. Default is enabled.
16005 @subsection Commands
16007 This filter supports the all above options as @ref{commands}.
16009 @subsection Examples
16013 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16015 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
16021 Read vertical interval timecode (VITC) information from the top lines of a
16024 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16025 timecode value, if a valid timecode has been detected. Further metadata key
16026 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16027 timecode data has been found or not.
16029 This filter accepts the following options:
16033 Set the maximum number of lines to scan for VITC data. If the value is set to
16034 @code{-1} the full video frame is scanned. Default is @code{45}.
16037 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16038 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16041 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16042 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16045 @subsection Examples
16049 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16050 draw @code{--:--:--:--} as a placeholder:
16052 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16058 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16060 Destination pixel at position (X, Y) will be picked from source (x, y) position
16061 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16062 value for pixel will be used for destination pixel.
16064 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16065 will have Xmap/Ymap video stream dimensions.
16066 Xmap and Ymap input video streams are 16bit depth, single channel.
16070 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16071 Default is @code{color}.
16074 Specify the color of the unmapped pixels. For the syntax of this option,
16075 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16076 manual,ffmpeg-utils}. Default color is @code{black}.
16079 @section removegrain
16081 The removegrain filter is a spatial denoiser for progressive video.
16085 Set mode for the first plane.
16088 Set mode for the second plane.
16091 Set mode for the third plane.
16094 Set mode for the fourth plane.
16097 Range of mode is from 0 to 24. Description of each mode follows:
16101 Leave input plane unchanged. Default.
16104 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16107 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16110 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16113 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16114 This is equivalent to a median filter.
16117 Line-sensitive clipping giving the minimal change.
16120 Line-sensitive clipping, intermediate.
16123 Line-sensitive clipping, intermediate.
16126 Line-sensitive clipping, intermediate.
16129 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16132 Replaces the target pixel with the closest neighbour.
16135 [1 2 1] horizontal and vertical kernel blur.
16141 Bob mode, interpolates top field from the line where the neighbours
16142 pixels are the closest.
16145 Bob mode, interpolates bottom field from the line where the neighbours
16146 pixels are the closest.
16149 Bob mode, interpolates top field. Same as 13 but with a more complicated
16150 interpolation formula.
16153 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16154 interpolation formula.
16157 Clips the pixel with the minimum and maximum of respectively the maximum and
16158 minimum of each pair of opposite neighbour pixels.
16161 Line-sensitive clipping using opposite neighbours whose greatest distance from
16162 the current pixel is minimal.
16165 Replaces the pixel with the average of its 8 neighbours.
16168 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16171 Clips pixels using the averages of opposite neighbour.
16174 Same as mode 21 but simpler and faster.
16177 Small edge and halo removal, but reputed useless.
16183 @section removelogo
16185 Suppress a TV station logo, using an image file to determine which
16186 pixels comprise the logo. It works by filling in the pixels that
16187 comprise the logo with neighboring pixels.
16189 The filter accepts the following options:
16193 Set the filter bitmap file, which can be any image format supported by
16194 libavformat. The width and height of the image file must match those of the
16195 video stream being processed.
16198 Pixels in the provided bitmap image with a value of zero are not
16199 considered part of the logo, non-zero pixels are considered part of
16200 the logo. If you use white (255) for the logo and black (0) for the
16201 rest, you will be safe. For making the filter bitmap, it is
16202 recommended to take a screen capture of a black frame with the logo
16203 visible, and then using a threshold filter followed by the erode
16204 filter once or twice.
16206 If needed, little splotches can be fixed manually. Remember that if
16207 logo pixels are not covered, the filter quality will be much
16208 reduced. Marking too many pixels as part of the logo does not hurt as
16209 much, but it will increase the amount of blurring needed to cover over
16210 the image and will destroy more information than necessary, and extra
16211 pixels will slow things down on a large logo.
16213 @section repeatfields
16215 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16216 fields based on its value.
16220 Reverse a video clip.
16222 Warning: This filter requires memory to buffer the entire clip, so trimming
16225 @subsection Examples
16229 Take the first 5 seconds of a clip, and reverse it.
16236 Shift R/G/B/A pixels horizontally and/or vertically.
16238 The filter accepts the following options:
16241 Set amount to shift red horizontally.
16243 Set amount to shift red vertically.
16245 Set amount to shift green horizontally.
16247 Set amount to shift green vertically.
16249 Set amount to shift blue horizontally.
16251 Set amount to shift blue vertically.
16253 Set amount to shift alpha horizontally.
16255 Set amount to shift alpha vertically.
16257 Set edge mode, can be @var{smear}, default, or @var{warp}.
16260 @subsection Commands
16262 This filter supports the all above options as @ref{commands}.
16265 Apply roberts cross operator to input video stream.
16267 The filter accepts the following option:
16271 Set which planes will be processed, unprocessed planes will be copied.
16272 By default value 0xf, all planes will be processed.
16275 Set value which will be multiplied with filtered result.
16278 Set value which will be added to filtered result.
16283 Rotate video by an arbitrary angle expressed in radians.
16285 The filter accepts the following options:
16287 A description of the optional parameters follows.
16290 Set an expression for the angle by which to rotate the input video
16291 clockwise, expressed as a number of radians. A negative value will
16292 result in a counter-clockwise rotation. By default it is set to "0".
16294 This expression is evaluated for each frame.
16297 Set the output width expression, default value is "iw".
16298 This expression is evaluated just once during configuration.
16301 Set the output height expression, default value is "ih".
16302 This expression is evaluated just once during configuration.
16305 Enable bilinear interpolation if set to 1, a value of 0 disables
16306 it. Default value is 1.
16309 Set the color used to fill the output area not covered by the rotated
16310 image. For the general syntax of this option, check the
16311 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16312 If the special value "none" is selected then no
16313 background is printed (useful for example if the background is never shown).
16315 Default value is "black".
16318 The expressions for the angle and the output size can contain the
16319 following constants and functions:
16323 sequential number of the input frame, starting from 0. It is always NAN
16324 before the first frame is filtered.
16327 time in seconds of the input frame, it is set to 0 when the filter is
16328 configured. It is always NAN before the first frame is filtered.
16332 horizontal and vertical chroma subsample values. For example for the
16333 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16337 the input video width and height
16341 the output width and height, that is the size of the padded area as
16342 specified by the @var{width} and @var{height} expressions
16346 the minimal width/height required for completely containing the input
16347 video rotated by @var{a} radians.
16349 These are only available when computing the @option{out_w} and
16350 @option{out_h} expressions.
16353 @subsection Examples
16357 Rotate the input by PI/6 radians clockwise:
16363 Rotate the input by PI/6 radians counter-clockwise:
16369 Rotate the input by 45 degrees clockwise:
16375 Apply a constant rotation with period T, starting from an angle of PI/3:
16377 rotate=PI/3+2*PI*t/T
16381 Make the input video rotation oscillating with a period of T
16382 seconds and an amplitude of A radians:
16384 rotate=A*sin(2*PI/T*t)
16388 Rotate the video, output size is chosen so that the whole rotating
16389 input video is always completely contained in the output:
16391 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16395 Rotate the video, reduce the output size so that no background is ever
16398 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16402 @subsection Commands
16404 The filter supports the following commands:
16408 Set the angle expression.
16409 The command accepts the same syntax of the corresponding option.
16411 If the specified expression is not valid, it is kept at its current
16417 Apply Shape Adaptive Blur.
16419 The filter accepts the following options:
16422 @item luma_radius, lr
16423 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16424 value is 1.0. A greater value will result in a more blurred image, and
16425 in slower processing.
16427 @item luma_pre_filter_radius, lpfr
16428 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16431 @item luma_strength, ls
16432 Set luma maximum difference between pixels to still be considered, must
16433 be a value in the 0.1-100.0 range, default value is 1.0.
16435 @item chroma_radius, cr
16436 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16437 greater value will result in a more blurred image, and in slower
16440 @item chroma_pre_filter_radius, cpfr
16441 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16443 @item chroma_strength, cs
16444 Set chroma maximum difference between pixels to still be considered,
16445 must be a value in the -0.9-100.0 range.
16448 Each chroma option value, if not explicitly specified, is set to the
16449 corresponding luma option value.
16454 Scale (resize) the input video, using the libswscale library.
16456 The scale filter forces the output display aspect ratio to be the same
16457 of the input, by changing the output sample aspect ratio.
16459 If the input image format is different from the format requested by
16460 the next filter, the scale filter will convert the input to the
16463 @subsection Options
16464 The filter accepts the following options, or any of the options
16465 supported by the libswscale scaler.
16467 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16468 the complete list of scaler options.
16473 Set the output video dimension expression. Default value is the input
16476 If the @var{width} or @var{w} value is 0, the input width is used for
16477 the output. If the @var{height} or @var{h} value is 0, the input height
16478 is used for the output.
16480 If one and only one of the values is -n with n >= 1, the scale filter
16481 will use a value that maintains the aspect ratio of the input image,
16482 calculated from the other specified dimension. After that it will,
16483 however, make sure that the calculated dimension is divisible by n and
16484 adjust the value if necessary.
16486 If both values are -n with n >= 1, the behavior will be identical to
16487 both values being set to 0 as previously detailed.
16489 See below for the list of accepted constants for use in the dimension
16493 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16497 Only evaluate expressions once during the filter initialization or when a command is processed.
16500 Evaluate expressions for each incoming frame.
16504 Default value is @samp{init}.
16508 Set the interlacing mode. It accepts the following values:
16512 Force interlaced aware scaling.
16515 Do not apply interlaced scaling.
16518 Select interlaced aware scaling depending on whether the source frames
16519 are flagged as interlaced or not.
16522 Default value is @samp{0}.
16525 Set libswscale scaling flags. See
16526 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16527 complete list of values. If not explicitly specified the filter applies
16531 @item param0, param1
16532 Set libswscale input parameters for scaling algorithms that need them. See
16533 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16534 complete documentation. If not explicitly specified the filter applies
16540 Set the video size. For the syntax of this option, check the
16541 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16543 @item in_color_matrix
16544 @item out_color_matrix
16545 Set in/output YCbCr color space type.
16547 This allows the autodetected value to be overridden as well as allows forcing
16548 a specific value used for the output and encoder.
16550 If not specified, the color space type depends on the pixel format.
16556 Choose automatically.
16559 Format conforming to International Telecommunication Union (ITU)
16560 Recommendation BT.709.
16563 Set color space conforming to the United States Federal Communications
16564 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16569 Set color space conforming to:
16573 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16576 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16579 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16584 Set color space conforming to SMPTE ST 240:1999.
16587 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16592 Set in/output YCbCr sample range.
16594 This allows the autodetected value to be overridden as well as allows forcing
16595 a specific value used for the output and encoder. If not specified, the
16596 range depends on the pixel format. Possible values:
16600 Choose automatically.
16603 Set full range (0-255 in case of 8-bit luma).
16605 @item mpeg/limited/tv
16606 Set "MPEG" range (16-235 in case of 8-bit luma).
16609 @item force_original_aspect_ratio
16610 Enable decreasing or increasing output video width or height if necessary to
16611 keep the original aspect ratio. Possible values:
16615 Scale the video as specified and disable this feature.
16618 The output video dimensions will automatically be decreased if needed.
16621 The output video dimensions will automatically be increased if needed.
16625 One useful instance of this option is that when you know a specific device's
16626 maximum allowed resolution, you can use this to limit the output video to
16627 that, while retaining the aspect ratio. For example, device A allows
16628 1280x720 playback, and your video is 1920x800. Using this option (set it to
16629 decrease) and specifying 1280x720 to the command line makes the output
16632 Please note that this is a different thing than specifying -1 for @option{w}
16633 or @option{h}, you still need to specify the output resolution for this option
16636 @item force_divisible_by
16637 Ensures that both the output dimensions, width and height, are divisible by the
16638 given integer when used together with @option{force_original_aspect_ratio}. This
16639 works similar to using @code{-n} in the @option{w} and @option{h} options.
16641 This option respects the value set for @option{force_original_aspect_ratio},
16642 increasing or decreasing the resolution accordingly. The video's aspect ratio
16643 may be slightly modified.
16645 This option can be handy if you need to have a video fit within or exceed
16646 a defined resolution using @option{force_original_aspect_ratio} but also have
16647 encoder restrictions on width or height divisibility.
16651 The values of the @option{w} and @option{h} options are expressions
16652 containing the following constants:
16657 The input width and height
16661 These are the same as @var{in_w} and @var{in_h}.
16665 The output (scaled) width and height
16669 These are the same as @var{out_w} and @var{out_h}
16672 The same as @var{iw} / @var{ih}
16675 input sample aspect ratio
16678 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16682 horizontal and vertical input chroma subsample values. For example for the
16683 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16687 horizontal and vertical output chroma subsample values. For example for the
16688 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16691 The (sequential) number of the input frame, starting from 0.
16692 Only available with @code{eval=frame}.
16695 The presentation timestamp of the input frame, expressed as a number of
16696 seconds. Only available with @code{eval=frame}.
16699 The position (byte offset) of the frame in the input stream, or NaN if
16700 this information is unavailable and/or meaningless (for example in case of synthetic video).
16701 Only available with @code{eval=frame}.
16704 @subsection Examples
16708 Scale the input video to a size of 200x100
16713 This is equivalent to:
16724 Specify a size abbreviation for the output size:
16729 which can also be written as:
16735 Scale the input to 2x:
16737 scale=w=2*iw:h=2*ih
16741 The above is the same as:
16743 scale=2*in_w:2*in_h
16747 Scale the input to 2x with forced interlaced scaling:
16749 scale=2*iw:2*ih:interl=1
16753 Scale the input to half size:
16755 scale=w=iw/2:h=ih/2
16759 Increase the width, and set the height to the same size:
16765 Seek Greek harmony:
16772 Increase the height, and set the width to 3/2 of the height:
16774 scale=w=3/2*oh:h=3/5*ih
16778 Increase the size, making the size a multiple of the chroma
16781 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16785 Increase the width to a maximum of 500 pixels,
16786 keeping the same aspect ratio as the input:
16788 scale=w='min(500\, iw*3/2):h=-1'
16792 Make pixels square by combining scale and setsar:
16794 scale='trunc(ih*dar):ih',setsar=1/1
16798 Make pixels square by combining scale and setsar,
16799 making sure the resulting resolution is even (required by some codecs):
16801 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16805 @subsection Commands
16807 This filter supports the following commands:
16811 Set the output video dimension expression.
16812 The command accepts the same syntax of the corresponding option.
16814 If the specified expression is not valid, it is kept at its current
16820 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16821 format conversion on CUDA video frames. Setting the output width and height
16822 works in the same way as for the @var{scale} filter.
16824 The following additional options are accepted:
16827 The pixel format of the output CUDA frames. If set to the string "same" (the
16828 default), the input format will be kept. Note that automatic format negotiation
16829 and conversion is not yet supported for hardware frames
16832 The interpolation algorithm used for resizing. One of the following:
16839 @item cubic2p_bspline
16840 2-parameter cubic (B=1, C=0)
16842 @item cubic2p_catmullrom
16843 2-parameter cubic (B=0, C=1/2)
16845 @item cubic2p_b05c03
16846 2-parameter cubic (B=1/2, C=3/10)
16854 @item force_original_aspect_ratio
16855 Enable decreasing or increasing output video width or height if necessary to
16856 keep the original aspect ratio. Possible values:
16860 Scale the video as specified and disable this feature.
16863 The output video dimensions will automatically be decreased if needed.
16866 The output video dimensions will automatically be increased if needed.
16870 One useful instance of this option is that when you know a specific device's
16871 maximum allowed resolution, you can use this to limit the output video to
16872 that, while retaining the aspect ratio. For example, device A allows
16873 1280x720 playback, and your video is 1920x800. Using this option (set it to
16874 decrease) and specifying 1280x720 to the command line makes the output
16877 Please note that this is a different thing than specifying -1 for @option{w}
16878 or @option{h}, you still need to specify the output resolution for this option
16881 @item force_divisible_by
16882 Ensures that both the output dimensions, width and height, are divisible by the
16883 given integer when used together with @option{force_original_aspect_ratio}. This
16884 works similar to using @code{-n} in the @option{w} and @option{h} options.
16886 This option respects the value set for @option{force_original_aspect_ratio},
16887 increasing or decreasing the resolution accordingly. The video's aspect ratio
16888 may be slightly modified.
16890 This option can be handy if you need to have a video fit within or exceed
16891 a defined resolution using @option{force_original_aspect_ratio} but also have
16892 encoder restrictions on width or height divisibility.
16898 Scale (resize) the input video, based on a reference video.
16900 See the scale filter for available options, scale2ref supports the same but
16901 uses the reference video instead of the main input as basis. scale2ref also
16902 supports the following additional constants for the @option{w} and
16903 @option{h} options:
16908 The main input video's width and height
16911 The same as @var{main_w} / @var{main_h}
16914 The main input video's sample aspect ratio
16916 @item main_dar, mdar
16917 The main input video's display aspect ratio. Calculated from
16918 @code{(main_w / main_h) * main_sar}.
16922 The main input video's horizontal and vertical chroma subsample values.
16923 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16927 The (sequential) number of the main input frame, starting from 0.
16928 Only available with @code{eval=frame}.
16931 The presentation timestamp of the main input frame, expressed as a number of
16932 seconds. Only available with @code{eval=frame}.
16935 The position (byte offset) of the frame in the main input stream, or NaN if
16936 this information is unavailable and/or meaningless (for example in case of synthetic video).
16937 Only available with @code{eval=frame}.
16940 @subsection Examples
16944 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16946 'scale2ref[b][a];[a][b]overlay'
16950 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16952 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16956 @subsection Commands
16958 This filter supports the following commands:
16962 Set the output video dimension expression.
16963 The command accepts the same syntax of the corresponding option.
16965 If the specified expression is not valid, it is kept at its current
16970 Scroll input video horizontally and/or vertically by constant speed.
16972 The filter accepts the following options:
16974 @item horizontal, h
16975 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16976 Negative values changes scrolling direction.
16979 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16980 Negative values changes scrolling direction.
16983 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16986 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16989 @subsection Commands
16991 This filter supports the following @ref{commands}:
16993 @item horizontal, h
16994 Set the horizontal scrolling speed.
16996 Set the vertical scrolling speed.
17002 Detect video scene change.
17004 This filter sets frame metadata with mafd between frame, the scene score, and
17005 forward the frame to the next filter, so they can use these metadata to detect
17006 scene change or others.
17008 In addition, this filter logs a message and sets frame metadata when it detects
17009 a scene change by @option{threshold}.
17011 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17013 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17014 to detect scene change.
17016 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17017 detect scene change with @option{threshold}.
17019 The filter accepts the following options:
17023 Set the scene change detection threshold as a percentage of maximum change. Good
17024 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17027 Default value is @code{10.}.
17030 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17031 You can enable it if you want to get snapshot of scene change frames only.
17034 @anchor{selectivecolor}
17035 @section selectivecolor
17037 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17038 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17039 by the "purity" of the color (that is, how saturated it already is).
17041 This filter is similar to the Adobe Photoshop Selective Color tool.
17043 The filter accepts the following options:
17046 @item correction_method
17047 Select color correction method.
17049 Available values are:
17052 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17055 Specified adjustments are relative to the original component value.
17057 Default is @code{absolute}.
17059 Adjustments for red pixels (pixels where the red component is the maximum)
17061 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17063 Adjustments for green pixels (pixels where the green component is the maximum)
17065 Adjustments for cyan pixels (pixels where the red component is the minimum)
17067 Adjustments for blue pixels (pixels where the blue component is the maximum)
17069 Adjustments for magenta pixels (pixels where the green component is the minimum)
17071 Adjustments for white pixels (pixels where all components are greater than 128)
17073 Adjustments for all pixels except pure black and pure white
17075 Adjustments for black pixels (pixels where all components are lesser than 128)
17077 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17080 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17081 4 space separated floating point adjustment values in the [-1,1] range,
17082 respectively to adjust the amount of cyan, magenta, yellow and black for the
17083 pixels of its range.
17085 @subsection Examples
17089 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17090 increase magenta by 27% in blue areas:
17092 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17096 Use a Photoshop selective color preset:
17098 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17102 @anchor{separatefields}
17103 @section separatefields
17105 The @code{separatefields} takes a frame-based video input and splits
17106 each frame into its components fields, producing a new half height clip
17107 with twice the frame rate and twice the frame count.
17109 This filter use field-dominance information in frame to decide which
17110 of each pair of fields to place first in the output.
17111 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17113 @section setdar, setsar
17115 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17118 This is done by changing the specified Sample (aka Pixel) Aspect
17119 Ratio, according to the following equation:
17121 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17124 Keep in mind that the @code{setdar} filter does not modify the pixel
17125 dimensions of the video frame. Also, the display aspect ratio set by
17126 this filter may be changed by later filters in the filterchain,
17127 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17130 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17131 the filter output video.
17133 Note that as a consequence of the application of this filter, the
17134 output display aspect ratio will change according to the equation
17137 Keep in mind that the sample aspect ratio set by the @code{setsar}
17138 filter may be changed by later filters in the filterchain, e.g. if
17139 another "setsar" or a "setdar" filter is applied.
17141 It accepts the following parameters:
17144 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17145 Set the aspect ratio used by the filter.
17147 The parameter can be a floating point number string, an expression, or
17148 a string of the form @var{num}:@var{den}, where @var{num} and
17149 @var{den} are the numerator and denominator of the aspect ratio. If
17150 the parameter is not specified, it is assumed the value "0".
17151 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17155 Set the maximum integer value to use for expressing numerator and
17156 denominator when reducing the expressed aspect ratio to a rational.
17157 Default value is @code{100}.
17161 The parameter @var{sar} is an expression containing
17162 the following constants:
17166 These are approximated values for the mathematical constants e
17167 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17170 The input width and height.
17173 These are the same as @var{w} / @var{h}.
17176 The input sample aspect ratio.
17179 The input display aspect ratio. It is the same as
17180 (@var{w} / @var{h}) * @var{sar}.
17183 Horizontal and vertical chroma subsample values. For example, for the
17184 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17187 @subsection Examples
17192 To change the display aspect ratio to 16:9, specify one of the following:
17199 To change the sample aspect ratio to 10:11, specify:
17205 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17206 1000 in the aspect ratio reduction, use the command:
17208 setdar=ratio=16/9:max=1000
17216 Force field for the output video frame.
17218 The @code{setfield} filter marks the interlace type field for the
17219 output frames. It does not change the input frame, but only sets the
17220 corresponding property, which affects how the frame is treated by
17221 following filters (e.g. @code{fieldorder} or @code{yadif}).
17223 The filter accepts the following options:
17228 Available values are:
17232 Keep the same field property.
17235 Mark the frame as bottom-field-first.
17238 Mark the frame as top-field-first.
17241 Mark the frame as progressive.
17248 Force frame parameter for the output video frame.
17250 The @code{setparams} filter marks interlace and color range for the
17251 output frames. It does not change the input frame, but only sets the
17252 corresponding property, which affects how the frame is treated by
17257 Available values are:
17261 Keep the same field property (default).
17264 Mark the frame as bottom-field-first.
17267 Mark the frame as top-field-first.
17270 Mark the frame as progressive.
17274 Available values are:
17278 Keep the same color range property (default).
17280 @item unspecified, unknown
17281 Mark the frame as unspecified color range.
17283 @item limited, tv, mpeg
17284 Mark the frame as limited range.
17286 @item full, pc, jpeg
17287 Mark the frame as full range.
17290 @item color_primaries
17291 Set the color primaries.
17292 Available values are:
17296 Keep the same color primaries property (default).
17313 Set the color transfer.
17314 Available values are:
17318 Keep the same color trc property (default).
17340 Set the colorspace.
17341 Available values are:
17345 Keep the same colorspace property (default).
17358 @item chroma-derived-nc
17359 @item chroma-derived-c
17366 Show a line containing various information for each input video frame.
17367 The input video is not modified.
17369 This filter supports the following options:
17373 Calculate checksums of each plane. By default enabled.
17376 The shown line contains a sequence of key/value pairs of the form
17377 @var{key}:@var{value}.
17379 The following values are shown in the output:
17383 The (sequential) number of the input frame, starting from 0.
17386 The Presentation TimeStamp of the input frame, expressed as a number of
17387 time base units. The time base unit depends on the filter input pad.
17390 The Presentation TimeStamp of the input frame, expressed as a number of
17394 The position of the frame in the input stream, or -1 if this information is
17395 unavailable and/or meaningless (for example in case of synthetic video).
17398 The pixel format name.
17401 The sample aspect ratio of the input frame, expressed in the form
17402 @var{num}/@var{den}.
17405 The size of the input frame. For the syntax of this option, check the
17406 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17409 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17410 for bottom field first).
17413 This is 1 if the frame is a key frame, 0 otherwise.
17416 The picture type of the input frame ("I" for an I-frame, "P" for a
17417 P-frame, "B" for a B-frame, or "?" for an unknown type).
17418 Also refer to the documentation of the @code{AVPictureType} enum and of
17419 the @code{av_get_picture_type_char} function defined in
17420 @file{libavutil/avutil.h}.
17423 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17425 @item plane_checksum
17426 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17427 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17430 The mean value of pixels in each plane of the input frame, expressed in the form
17431 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17434 The standard deviation of pixel values in each plane of the input frame, expressed
17435 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17439 @section showpalette
17441 Displays the 256 colors palette of each frame. This filter is only relevant for
17442 @var{pal8} pixel format frames.
17444 It accepts the following option:
17448 Set the size of the box used to represent one palette color entry. Default is
17449 @code{30} (for a @code{30x30} pixel box).
17452 @section shuffleframes
17454 Reorder and/or duplicate and/or drop video frames.
17456 It accepts the following parameters:
17460 Set the destination indexes of input frames.
17461 This is space or '|' separated list of indexes that maps input frames to output
17462 frames. Number of indexes also sets maximal value that each index may have.
17463 '-1' index have special meaning and that is to drop frame.
17466 The first frame has the index 0. The default is to keep the input unchanged.
17468 @subsection Examples
17472 Swap second and third frame of every three frames of the input:
17474 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17478 Swap 10th and 1st frame of every ten frames of the input:
17480 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17484 @section shuffleplanes
17486 Reorder and/or duplicate video planes.
17488 It accepts the following parameters:
17493 The index of the input plane to be used as the first output plane.
17496 The index of the input plane to be used as the second output plane.
17499 The index of the input plane to be used as the third output plane.
17502 The index of the input plane to be used as the fourth output plane.
17506 The first plane has the index 0. The default is to keep the input unchanged.
17508 @subsection Examples
17512 Swap the second and third planes of the input:
17514 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17518 @anchor{signalstats}
17519 @section signalstats
17520 Evaluate various visual metrics that assist in determining issues associated
17521 with the digitization of analog video media.
17523 By default the filter will log these metadata values:
17527 Display the minimal Y value contained within the input frame. Expressed in
17531 Display the Y value at the 10% percentile within the input frame. Expressed in
17535 Display the average Y value within the input frame. Expressed in range of
17539 Display the Y value at the 90% percentile within the input frame. Expressed in
17543 Display the maximum Y value contained within the input frame. Expressed in
17547 Display the minimal U value contained within the input frame. Expressed in
17551 Display the U value at the 10% percentile within the input frame. Expressed in
17555 Display the average U value within the input frame. Expressed in range of
17559 Display the U value at the 90% percentile within the input frame. Expressed in
17563 Display the maximum U value contained within the input frame. Expressed in
17567 Display the minimal V value contained within the input frame. Expressed in
17571 Display the V value at the 10% percentile within the input frame. Expressed in
17575 Display the average V value within the input frame. Expressed in range of
17579 Display the V value at the 90% percentile within the input frame. Expressed in
17583 Display the maximum V value contained within the input frame. Expressed in
17587 Display the minimal saturation value contained within the input frame.
17588 Expressed in range of [0-~181.02].
17591 Display the saturation value at the 10% percentile within the input frame.
17592 Expressed in range of [0-~181.02].
17595 Display the average saturation value within the input frame. Expressed in range
17599 Display the saturation value at the 90% percentile within the input frame.
17600 Expressed in range of [0-~181.02].
17603 Display the maximum saturation value contained within the input frame.
17604 Expressed in range of [0-~181.02].
17607 Display the median value for hue within the input frame. Expressed in range of
17611 Display the average value for hue within the input frame. Expressed in range of
17615 Display the average of sample value difference between all values of the Y
17616 plane in the current frame and corresponding values of the previous input frame.
17617 Expressed in range of [0-255].
17620 Display the average of sample value difference between all values of the U
17621 plane in the current frame and corresponding values of the previous input frame.
17622 Expressed in range of [0-255].
17625 Display the average of sample value difference between all values of the V
17626 plane in the current frame and corresponding values of the previous input frame.
17627 Expressed in range of [0-255].
17630 Display bit depth of Y plane in current frame.
17631 Expressed in range of [0-16].
17634 Display bit depth of U plane in current frame.
17635 Expressed in range of [0-16].
17638 Display bit depth of V plane in current frame.
17639 Expressed in range of [0-16].
17642 The filter accepts the following options:
17648 @option{stat} specify an additional form of image analysis.
17649 @option{out} output video with the specified type of pixel highlighted.
17651 Both options accept the following values:
17655 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17656 unlike the neighboring pixels of the same field. Examples of temporal outliers
17657 include the results of video dropouts, head clogs, or tape tracking issues.
17660 Identify @var{vertical line repetition}. Vertical line repetition includes
17661 similar rows of pixels within a frame. In born-digital video vertical line
17662 repetition is common, but this pattern is uncommon in video digitized from an
17663 analog source. When it occurs in video that results from the digitization of an
17664 analog source it can indicate concealment from a dropout compensator.
17667 Identify pixels that fall outside of legal broadcast range.
17671 Set the highlight color for the @option{out} option. The default color is
17675 @subsection Examples
17679 Output data of various video metrics:
17681 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17685 Output specific data about the minimum and maximum values of the Y plane per frame:
17687 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17691 Playback video while highlighting pixels that are outside of broadcast range in red.
17693 ffplay example.mov -vf signalstats="out=brng:color=red"
17697 Playback video with signalstats metadata drawn over the frame.
17699 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17702 The contents of signalstat_drawtext.txt used in the command are:
17705 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
17706 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
17707 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
17708 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
17716 Calculates the MPEG-7 Video Signature. The filter can handle more than one
17717 input. In this case the matching between the inputs can be calculated additionally.
17718 The filter always passes through the first input. The signature of each stream can
17719 be written into a file.
17721 It accepts the following options:
17725 Enable or disable the matching process.
17727 Available values are:
17731 Disable the calculation of a matching (default).
17733 Calculate the matching for the whole video and output whether the whole video
17734 matches or only parts.
17736 Calculate only until a matching is found or the video ends. Should be faster in
17741 Set the number of inputs. The option value must be a non negative integer.
17742 Default value is 1.
17745 Set the path to which the output is written. If there is more than one input,
17746 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
17747 integer), that will be replaced with the input number. If no filename is
17748 specified, no output will be written. This is the default.
17751 Choose the output format.
17753 Available values are:
17757 Use the specified binary representation (default).
17759 Use the specified xml representation.
17763 Set threshold to detect one word as similar. The option value must be an integer
17764 greater than zero. The default value is 9000.
17767 Set threshold to detect all words as similar. The option value must be an integer
17768 greater than zero. The default value is 60000.
17771 Set threshold to detect frames as similar. The option value must be an integer
17772 greater than zero. The default value is 116.
17775 Set the minimum length of a sequence in frames to recognize it as matching
17776 sequence. The option value must be a non negative integer value.
17777 The default value is 0.
17780 Set the minimum relation, that matching frames to all frames must have.
17781 The option value must be a double value between 0 and 1. The default value is 0.5.
17784 @subsection Examples
17788 To calculate the signature of an input video and store it in signature.bin:
17790 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17794 To detect whether two videos match and store the signatures in XML format in
17795 signature0.xml and signature1.xml:
17797 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 -
17805 Blur the input video without impacting the outlines.
17807 It accepts the following options:
17810 @item luma_radius, lr
17811 Set the luma radius. The option value must be a float number in
17812 the range [0.1,5.0] that specifies the variance of the gaussian filter
17813 used to blur the image (slower if larger). Default value is 1.0.
17815 @item luma_strength, ls
17816 Set the luma strength. The option value must be a float number
17817 in the range [-1.0,1.0] that configures the blurring. A value included
17818 in [0.0,1.0] will blur the image whereas a value included in
17819 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17821 @item luma_threshold, lt
17822 Set the luma threshold used as a coefficient to determine
17823 whether a pixel should be blurred or not. The option value must be an
17824 integer in the range [-30,30]. A value of 0 will filter all the image,
17825 a value included in [0,30] will filter flat areas and a value included
17826 in [-30,0] will filter edges. Default value is 0.
17828 @item chroma_radius, cr
17829 Set the chroma radius. The option value must be a float number in
17830 the range [0.1,5.0] that specifies the variance of the gaussian filter
17831 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17833 @item chroma_strength, cs
17834 Set the chroma strength. The option value must be a float number
17835 in the range [-1.0,1.0] that configures the blurring. A value included
17836 in [0.0,1.0] will blur the image whereas a value included in
17837 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17839 @item chroma_threshold, ct
17840 Set the chroma threshold used as a coefficient to determine
17841 whether a pixel should be blurred or not. The option value must be an
17842 integer in the range [-30,30]. A value of 0 will filter all the image,
17843 a value included in [0,30] will filter flat areas and a value included
17844 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17847 If a chroma option is not explicitly set, the corresponding luma value
17851 Apply sobel operator to input video stream.
17853 The filter accepts the following option:
17857 Set which planes will be processed, unprocessed planes will be copied.
17858 By default value 0xf, all planes will be processed.
17861 Set value which will be multiplied with filtered result.
17864 Set value which will be added to filtered result.
17870 Apply a simple postprocessing filter that compresses and decompresses the image
17871 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17872 and average the results.
17874 The filter accepts the following options:
17878 Set quality. This option defines the number of levels for averaging. It accepts
17879 an integer in the range 0-6. If set to @code{0}, the filter will have no
17880 effect. A value of @code{6} means the higher quality. For each increment of
17881 that value the speed drops by a factor of approximately 2. Default value is
17885 Force a constant quantization parameter. If not set, the filter will use the QP
17886 from the video stream (if available).
17889 Set thresholding mode. Available modes are:
17893 Set hard thresholding (default).
17895 Set soft thresholding (better de-ringing effect, but likely blurrier).
17898 @item use_bframe_qp
17899 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17900 option may cause flicker since the B-Frames have often larger QP. Default is
17901 @code{0} (not enabled).
17904 @subsection Commands
17906 This filter supports the following commands:
17908 @item quality, level
17909 Set quality level. The value @code{max} can be used to set the maximum level,
17910 currently @code{6}.
17916 Scale the input by applying one of the super-resolution methods based on
17917 convolutional neural networks. Supported models:
17921 Super-Resolution Convolutional Neural Network model (SRCNN).
17922 See @url{https://arxiv.org/abs/1501.00092}.
17925 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17926 See @url{https://arxiv.org/abs/1609.05158}.
17929 Training scripts as well as scripts for model file (.pb) saving can be found at
17930 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17931 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17933 Native model files (.model) can be generated from TensorFlow model
17934 files (.pb) by using tools/python/convert.py
17936 The filter accepts the following options:
17940 Specify which DNN backend to use for model loading and execution. This option accepts
17941 the following values:
17945 Native implementation of DNN loading and execution.
17948 TensorFlow backend. To enable this backend you
17949 need to install the TensorFlow for C library (see
17950 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17951 @code{--enable-libtensorflow}
17954 Default value is @samp{native}.
17957 Set path to model file specifying network architecture and its parameters.
17958 Note that different backends use different file formats. TensorFlow backend
17959 can load files for both formats, while native backend can load files for only
17963 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17964 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17965 input upscaled using bicubic upscaling with proper scale factor.
17968 This feature can also be finished with @ref{dnn_processing} filter.
17972 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17974 This filter takes in input two input videos, the first input is
17975 considered the "main" source and is passed unchanged to the
17976 output. The second input is used as a "reference" video for computing
17979 Both video inputs must have the same resolution and pixel format for
17980 this filter to work correctly. Also it assumes that both inputs
17981 have the same number of frames, which are compared one by one.
17983 The filter stores the calculated SSIM of each frame.
17985 The description of the accepted parameters follows.
17988 @item stats_file, f
17989 If specified the filter will use the named file to save the SSIM of
17990 each individual frame. When filename equals "-" the data is sent to
17994 The file printed if @var{stats_file} is selected, contains a sequence of
17995 key/value pairs of the form @var{key}:@var{value} for each compared
17998 A description of each shown parameter follows:
18002 sequential number of the input frame, starting from 1
18004 @item Y, U, V, R, G, B
18005 SSIM of the compared frames for the component specified by the suffix.
18008 SSIM of the compared frames for the whole frame.
18011 Same as above but in dB representation.
18014 This filter also supports the @ref{framesync} options.
18016 @subsection Examples
18021 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18022 [main][ref] ssim="stats_file=stats.log" [out]
18025 On this example the input file being processed is compared with the
18026 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18027 is stored in @file{stats.log}.
18030 Another example with both psnr and ssim at same time:
18032 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18036 Another example with different containers:
18038 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 -
18044 Convert between different stereoscopic image formats.
18046 The filters accept the following options:
18050 Set stereoscopic image format of input.
18052 Available values for input image formats are:
18055 side by side parallel (left eye left, right eye right)
18058 side by side crosseye (right eye left, left eye right)
18061 side by side parallel with half width resolution
18062 (left eye left, right eye right)
18065 side by side crosseye with half width resolution
18066 (right eye left, left eye right)
18070 above-below (left eye above, right eye below)
18074 above-below (right eye above, left eye below)
18078 above-below with half height resolution
18079 (left eye above, right eye below)
18083 above-below with half height resolution
18084 (right eye above, left eye below)
18087 alternating frames (left eye first, right eye second)
18090 alternating frames (right eye first, left eye second)
18093 interleaved rows (left eye has top row, right eye starts on next row)
18096 interleaved rows (right eye has top row, left eye starts on next row)
18099 interleaved columns, left eye first
18102 interleaved columns, right eye first
18104 Default value is @samp{sbsl}.
18108 Set stereoscopic image format of output.
18112 side by side parallel (left eye left, right eye right)
18115 side by side crosseye (right eye left, left eye right)
18118 side by side parallel with half width resolution
18119 (left eye left, right eye right)
18122 side by side crosseye with half width resolution
18123 (right eye left, left eye right)
18127 above-below (left eye above, right eye below)
18131 above-below (right eye above, left eye below)
18135 above-below with half height resolution
18136 (left eye above, right eye below)
18140 above-below with half height resolution
18141 (right eye above, left eye below)
18144 alternating frames (left eye first, right eye second)
18147 alternating frames (right eye first, left eye second)
18150 interleaved rows (left eye has top row, right eye starts on next row)
18153 interleaved rows (right eye has top row, left eye starts on next row)
18156 anaglyph red/blue gray
18157 (red filter on left eye, blue filter on right eye)
18160 anaglyph red/green gray
18161 (red filter on left eye, green filter on right eye)
18164 anaglyph red/cyan gray
18165 (red filter on left eye, cyan filter on right eye)
18168 anaglyph red/cyan half colored
18169 (red filter on left eye, cyan filter on right eye)
18172 anaglyph red/cyan color
18173 (red filter on left eye, cyan filter on right eye)
18176 anaglyph red/cyan color optimized with the least squares projection of dubois
18177 (red filter on left eye, cyan filter on right eye)
18180 anaglyph green/magenta gray
18181 (green filter on left eye, magenta filter on right eye)
18184 anaglyph green/magenta half colored
18185 (green filter on left eye, magenta filter on right eye)
18188 anaglyph green/magenta colored
18189 (green filter on left eye, magenta filter on right eye)
18192 anaglyph green/magenta color optimized with the least squares projection of dubois
18193 (green filter on left eye, magenta filter on right eye)
18196 anaglyph yellow/blue gray
18197 (yellow filter on left eye, blue filter on right eye)
18200 anaglyph yellow/blue half colored
18201 (yellow filter on left eye, blue filter on right eye)
18204 anaglyph yellow/blue colored
18205 (yellow filter on left eye, blue filter on right eye)
18208 anaglyph yellow/blue color optimized with the least squares projection of dubois
18209 (yellow filter on left eye, blue filter on right eye)
18212 mono output (left eye only)
18215 mono output (right eye only)
18218 checkerboard, left eye first
18221 checkerboard, right eye first
18224 interleaved columns, left eye first
18227 interleaved columns, right eye first
18233 Default value is @samp{arcd}.
18236 @subsection Examples
18240 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18246 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18252 @section streamselect, astreamselect
18253 Select video or audio streams.
18255 The filter accepts the following options:
18259 Set number of inputs. Default is 2.
18262 Set input indexes to remap to outputs.
18265 @subsection Commands
18267 The @code{streamselect} and @code{astreamselect} filter supports the following
18272 Set input indexes to remap to outputs.
18275 @subsection Examples
18279 Select first 5 seconds 1st stream and rest of time 2nd stream:
18281 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18285 Same as above, but for audio:
18287 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18294 Draw subtitles on top of input video using the libass library.
18296 To enable compilation of this filter you need to configure FFmpeg with
18297 @code{--enable-libass}. This filter also requires a build with libavcodec and
18298 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18299 Alpha) subtitles format.
18301 The filter accepts the following options:
18305 Set the filename of the subtitle file to read. It must be specified.
18307 @item original_size
18308 Specify the size of the original video, the video for which the ASS file
18309 was composed. For the syntax of this option, check the
18310 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18311 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18312 correctly scale the fonts if the aspect ratio has been changed.
18315 Set a directory path containing fonts that can be used by the filter.
18316 These fonts will be used in addition to whatever the font provider uses.
18319 Process alpha channel, by default alpha channel is untouched.
18322 Set subtitles input character encoding. @code{subtitles} filter only. Only
18323 useful if not UTF-8.
18325 @item stream_index, si
18326 Set subtitles stream index. @code{subtitles} filter only.
18329 Override default style or script info parameters of the subtitles. It accepts a
18330 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18333 If the first key is not specified, it is assumed that the first value
18334 specifies the @option{filename}.
18336 For example, to render the file @file{sub.srt} on top of the input
18337 video, use the command:
18342 which is equivalent to:
18344 subtitles=filename=sub.srt
18347 To render the default subtitles stream from file @file{video.mkv}, use:
18349 subtitles=video.mkv
18352 To render the second subtitles stream from that file, use:
18354 subtitles=video.mkv:si=1
18357 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18358 @code{DejaVu Serif}, use:
18360 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18363 @section super2xsai
18365 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18366 Interpolate) pixel art scaling algorithm.
18368 Useful for enlarging pixel art images without reducing sharpness.
18372 Swap two rectangular objects in video.
18374 This filter accepts the following options:
18384 Set 1st rect x coordinate.
18387 Set 1st rect y coordinate.
18390 Set 2nd rect x coordinate.
18393 Set 2nd rect y coordinate.
18395 All expressions are evaluated once for each frame.
18398 The all options are expressions containing the following constants:
18403 The input width and height.
18406 same as @var{w} / @var{h}
18409 input sample aspect ratio
18412 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18415 The number of the input frame, starting from 0.
18418 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18421 the position in the file of the input frame, NAN if unknown
18428 Blend successive video frames.
18434 Apply telecine process to the video.
18436 This filter accepts the following options:
18445 The default value is @code{top}.
18449 A string of numbers representing the pulldown pattern you wish to apply.
18450 The default value is @code{23}.
18454 Some typical patterns:
18459 24p: 2332 (preferred)
18466 24p: 222222222223 ("Euro pulldown")
18471 @section thistogram
18473 Compute and draw a color distribution histogram for the input video across time.
18475 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18476 at certain time, this filter shows also past histograms of number of frames defined
18477 by @code{width} option.
18479 The computed histogram is a representation of the color component
18480 distribution in an image.
18482 The filter accepts the following options:
18486 Set width of single color component output. Default value is @code{0}.
18487 Value of @code{0} means width will be picked from input video.
18488 This also set number of passed histograms to keep.
18489 Allowed range is [0, 8192].
18491 @item display_mode, d
18493 It accepts the following values:
18496 Per color component graphs are placed below each other.
18499 Per color component graphs are placed side by side.
18502 Presents information identical to that in the @code{parade}, except
18503 that the graphs representing color components are superimposed directly
18506 Default is @code{stack}.
18508 @item levels_mode, m
18509 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18510 Default is @code{linear}.
18512 @item components, c
18513 Set what color components to display.
18514 Default is @code{7}.
18517 Set background opacity. Default is @code{0.9}.
18520 Show envelope. Default is disabled.
18523 Set envelope color. Default is @code{gold}.
18528 Available values for slide is:
18531 Draw new frame when right border is reached.
18534 Replace old columns with new ones.
18537 Scroll from right to left.
18540 Scroll from left to right.
18543 Draw single picture.
18546 Default is @code{replace}.
18551 Apply threshold effect to video stream.
18553 This filter needs four video streams to perform thresholding.
18554 First stream is stream we are filtering.
18555 Second stream is holding threshold values, third stream is holding min values,
18556 and last, fourth stream is holding max values.
18558 The filter accepts the following option:
18562 Set which planes will be processed, unprocessed planes will be copied.
18563 By default value 0xf, all planes will be processed.
18566 For example if first stream pixel's component value is less then threshold value
18567 of pixel component from 2nd threshold stream, third stream value will picked,
18568 otherwise fourth stream pixel component value will be picked.
18570 Using color source filter one can perform various types of thresholding:
18572 @subsection Examples
18576 Binary threshold, using gray color as threshold:
18578 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18582 Inverted binary threshold, using gray color as threshold:
18584 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18588 Truncate binary threshold, using gray color as threshold:
18590 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18594 Threshold to zero, using gray color as threshold:
18596 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18600 Inverted threshold to zero, using gray color as threshold:
18602 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18607 Select the most representative frame in a given sequence of consecutive frames.
18609 The filter accepts the following options:
18613 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18614 will pick one of them, and then handle the next batch of @var{n} frames until
18615 the end. Default is @code{100}.
18618 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18619 value will result in a higher memory usage, so a high value is not recommended.
18621 @subsection Examples
18625 Extract one picture each 50 frames:
18631 Complete example of a thumbnail creation with @command{ffmpeg}:
18633 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18640 Tile several successive frames together.
18642 The @ref{untile} filter can do the reverse.
18644 The filter accepts the following options:
18649 Set the grid size (i.e. the number of lines and columns). For the syntax of
18650 this option, check the
18651 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18654 Set the maximum number of frames to render in the given area. It must be less
18655 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18656 the area will be used.
18659 Set the outer border margin in pixels.
18662 Set the inner border thickness (i.e. the number of pixels between frames). For
18663 more advanced padding options (such as having different values for the edges),
18664 refer to the pad video filter.
18667 Specify the color of the unused area. For the syntax of this option, check the
18668 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18669 The default value of @var{color} is "black".
18672 Set the number of frames to overlap when tiling several successive frames together.
18673 The value must be between @code{0} and @var{nb_frames - 1}.
18676 Set the number of frames to initially be empty before displaying first output frame.
18677 This controls how soon will one get first output frame.
18678 The value must be between @code{0} and @var{nb_frames - 1}.
18681 @subsection Examples
18685 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18687 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18689 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18690 duplicating each output frame to accommodate the originally detected frame
18694 Display @code{5} pictures in an area of @code{3x2} frames,
18695 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18696 mixed flat and named options:
18698 tile=3x2:nb_frames=5:padding=7:margin=2
18702 @section tinterlace
18704 Perform various types of temporal field interlacing.
18706 Frames are counted starting from 1, so the first input frame is
18709 The filter accepts the following options:
18714 Specify the mode of the interlacing. This option can also be specified
18715 as a value alone. See below for a list of values for this option.
18717 Available values are:
18721 Move odd frames into the upper field, even into the lower field,
18722 generating a double height frame at half frame rate.
18726 Frame 1 Frame 2 Frame 3 Frame 4
18728 11111 22222 33333 44444
18729 11111 22222 33333 44444
18730 11111 22222 33333 44444
18731 11111 22222 33333 44444
18745 Only output odd frames, even frames are dropped, generating a frame with
18746 unchanged height at half frame rate.
18751 Frame 1 Frame 2 Frame 3 Frame 4
18753 11111 22222 33333 44444
18754 11111 22222 33333 44444
18755 11111 22222 33333 44444
18756 11111 22222 33333 44444
18766 Only output even frames, odd frames are dropped, generating a frame with
18767 unchanged height at half frame rate.
18772 Frame 1 Frame 2 Frame 3 Frame 4
18774 11111 22222 33333 44444
18775 11111 22222 33333 44444
18776 11111 22222 33333 44444
18777 11111 22222 33333 44444
18787 Expand each frame to full height, but pad alternate lines with black,
18788 generating a frame with double height at the same input frame rate.
18793 Frame 1 Frame 2 Frame 3 Frame 4
18795 11111 22222 33333 44444
18796 11111 22222 33333 44444
18797 11111 22222 33333 44444
18798 11111 22222 33333 44444
18801 11111 ..... 33333 .....
18802 ..... 22222 ..... 44444
18803 11111 ..... 33333 .....
18804 ..... 22222 ..... 44444
18805 11111 ..... 33333 .....
18806 ..... 22222 ..... 44444
18807 11111 ..... 33333 .....
18808 ..... 22222 ..... 44444
18812 @item interleave_top, 4
18813 Interleave the upper field from odd frames with the lower field from
18814 even frames, generating a frame with unchanged height at half frame rate.
18819 Frame 1 Frame 2 Frame 3 Frame 4
18821 11111<- 22222 33333<- 44444
18822 11111 22222<- 33333 44444<-
18823 11111<- 22222 33333<- 44444
18824 11111 22222<- 33333 44444<-
18834 @item interleave_bottom, 5
18835 Interleave the lower field from odd frames with the upper field from
18836 even frames, generating a frame with unchanged height at half frame rate.
18841 Frame 1 Frame 2 Frame 3 Frame 4
18843 11111 22222<- 33333 44444<-
18844 11111<- 22222 33333<- 44444
18845 11111 22222<- 33333 44444<-
18846 11111<- 22222 33333<- 44444
18856 @item interlacex2, 6
18857 Double frame rate with unchanged height. Frames are inserted each
18858 containing the second temporal field from the previous input frame and
18859 the first temporal field from the next input frame. This mode relies on
18860 the top_field_first flag. Useful for interlaced video displays with no
18861 field synchronisation.
18866 Frame 1 Frame 2 Frame 3 Frame 4
18868 11111 22222 33333 44444
18869 11111 22222 33333 44444
18870 11111 22222 33333 44444
18871 11111 22222 33333 44444
18874 11111 22222 22222 33333 33333 44444 44444
18875 11111 11111 22222 22222 33333 33333 44444
18876 11111 22222 22222 33333 33333 44444 44444
18877 11111 11111 22222 22222 33333 33333 44444
18882 Move odd frames into the upper field, even into the lower field,
18883 generating a double height frame at same frame rate.
18888 Frame 1 Frame 2 Frame 3 Frame 4
18890 11111 22222 33333 44444
18891 11111 22222 33333 44444
18892 11111 22222 33333 44444
18893 11111 22222 33333 44444
18896 11111 33333 33333 55555
18897 22222 22222 44444 44444
18898 11111 33333 33333 55555
18899 22222 22222 44444 44444
18900 11111 33333 33333 55555
18901 22222 22222 44444 44444
18902 11111 33333 33333 55555
18903 22222 22222 44444 44444
18908 Numeric values are deprecated but are accepted for backward
18909 compatibility reasons.
18911 Default mode is @code{merge}.
18914 Specify flags influencing the filter process.
18916 Available value for @var{flags} is:
18919 @item low_pass_filter, vlpf
18920 Enable linear vertical low-pass filtering in the filter.
18921 Vertical low-pass filtering is required when creating an interlaced
18922 destination from a progressive source which contains high-frequency
18923 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18926 @item complex_filter, cvlpf
18927 Enable complex vertical low-pass filtering.
18928 This will slightly less reduce interlace 'twitter' and Moire
18929 patterning but better retain detail and subjective sharpness impression.
18932 Bypass already interlaced frames, only adjust the frame rate.
18935 Vertical low-pass filtering and bypassing already interlaced frames can only be
18936 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
18941 Pick median pixels from several successive input video frames.
18943 The filter accepts the following options:
18947 Set radius of median filter.
18948 Default is 1. Allowed range is from 1 to 127.
18951 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
18954 Set median percentile. Default value is @code{0.5}.
18955 Default value of @code{0.5} will pick always median values, while @code{0} will pick
18956 minimum values, and @code{1} maximum values.
18961 Mix successive video frames.
18963 A description of the accepted options follows.
18967 The number of successive frames to mix. If unspecified, it defaults to 3.
18970 Specify weight of each input video frame.
18971 Each weight is separated by space. If number of weights is smaller than
18972 number of @var{frames} last specified weight will be used for all remaining
18976 Specify scale, if it is set it will be multiplied with sum
18977 of each weight multiplied with pixel values to give final destination
18978 pixel value. By default @var{scale} is auto scaled to sum of weights.
18981 @subsection Examples
18985 Average 7 successive frames:
18987 tmix=frames=7:weights="1 1 1 1 1 1 1"
18991 Apply simple temporal convolution:
18993 tmix=frames=3:weights="-1 3 -1"
18997 Similar as above but only showing temporal differences:
18999 tmix=frames=3:weights="-1 2 -1":scale=1
19005 Tone map colors from different dynamic ranges.
19007 This filter expects data in single precision floating point, as it needs to
19008 operate on (and can output) out-of-range values. Another filter, such as
19009 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19011 The tonemapping algorithms implemented only work on linear light, so input
19012 data should be linearized beforehand (and possibly correctly tagged).
19015 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19018 @subsection Options
19019 The filter accepts the following options.
19023 Set the tone map algorithm to use.
19025 Possible values are:
19028 Do not apply any tone map, only desaturate overbright pixels.
19031 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19032 in-range values, while distorting out-of-range values.
19035 Stretch the entire reference gamut to a linear multiple of the display.
19038 Fit a logarithmic transfer between the tone curves.
19041 Preserve overall image brightness with a simple curve, using nonlinear
19042 contrast, which results in flattening details and degrading color accuracy.
19045 Preserve both dark and bright details better than @var{reinhard}, at the cost
19046 of slightly darkening everything. Use it when detail preservation is more
19047 important than color and brightness accuracy.
19050 Smoothly map out-of-range values, while retaining contrast and colors for
19051 in-range material as much as possible. Use it when color accuracy is more
19052 important than detail preservation.
19058 Tune the tone mapping algorithm.
19060 This affects the following algorithms:
19066 Specifies the scale factor to use while stretching.
19070 Specifies the exponent of the function.
19074 Specify an extra linear coefficient to multiply into the signal before clipping.
19078 Specify the local contrast coefficient at the display peak.
19079 Default to 0.5, which means that in-gamut values will be about half as bright
19086 Specify the transition point from linear to mobius transform. Every value
19087 below this point is guaranteed to be mapped 1:1. The higher the value, the
19088 more accurate the result will be, at the cost of losing bright details.
19089 Default to 0.3, which due to the steep initial slope still preserves in-range
19090 colors fairly accurately.
19094 Apply desaturation for highlights that exceed this level of brightness. The
19095 higher the parameter, the more color information will be preserved. This
19096 setting helps prevent unnaturally blown-out colors for super-highlights, by
19097 (smoothly) turning into white instead. This makes images feel more natural,
19098 at the cost of reducing information about out-of-range colors.
19100 The default of 2.0 is somewhat conservative and will mostly just apply to
19101 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19103 This option works only if the input frame has a supported color tag.
19106 Override signal/nominal/reference peak with this value. Useful when the
19107 embedded peak information in display metadata is not reliable or when tone
19108 mapping from a lower range to a higher range.
19113 Temporarily pad video frames.
19115 The filter accepts the following options:
19119 Specify number of delay frames before input video stream. Default is 0.
19122 Specify number of padding frames after input video stream.
19123 Set to -1 to pad indefinitely. Default is 0.
19126 Set kind of frames added to beginning of stream.
19127 Can be either @var{add} or @var{clone}.
19128 With @var{add} frames of solid-color are added.
19129 With @var{clone} frames are clones of first frame.
19130 Default is @var{add}.
19133 Set kind of frames added to end of stream.
19134 Can be either @var{add} or @var{clone}.
19135 With @var{add} frames of solid-color are added.
19136 With @var{clone} frames are clones of last frame.
19137 Default is @var{add}.
19139 @item start_duration, stop_duration
19140 Specify the duration of the start/stop delay. See
19141 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19142 for the accepted syntax.
19143 These options override @var{start} and @var{stop}. Default is 0.
19146 Specify the color of the padded area. For the syntax of this option,
19147 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19148 manual,ffmpeg-utils}.
19150 The default value of @var{color} is "black".
19156 Transpose rows with columns in the input video and optionally flip it.
19158 It accepts the following parameters:
19163 Specify the transposition direction.
19165 Can assume the following values:
19167 @item 0, 4, cclock_flip
19168 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19176 Rotate by 90 degrees clockwise, that is:
19184 Rotate by 90 degrees counterclockwise, that is:
19191 @item 3, 7, clock_flip
19192 Rotate by 90 degrees clockwise and vertically flip, that is:
19200 For values between 4-7, the transposition is only done if the input
19201 video geometry is portrait and not landscape. These values are
19202 deprecated, the @code{passthrough} option should be used instead.
19204 Numerical values are deprecated, and should be dropped in favor of
19205 symbolic constants.
19208 Do not apply the transposition if the input geometry matches the one
19209 specified by the specified value. It accepts the following values:
19212 Always apply transposition.
19214 Preserve portrait geometry (when @var{height} >= @var{width}).
19216 Preserve landscape geometry (when @var{width} >= @var{height}).
19219 Default value is @code{none}.
19222 For example to rotate by 90 degrees clockwise and preserve portrait
19225 transpose=dir=1:passthrough=portrait
19228 The command above can also be specified as:
19230 transpose=1:portrait
19233 @section transpose_npp
19235 Transpose rows with columns in the input video and optionally flip it.
19236 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19238 It accepts the following parameters:
19243 Specify the transposition direction.
19245 Can assume the following values:
19248 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19251 Rotate by 90 degrees clockwise.
19254 Rotate by 90 degrees counterclockwise.
19257 Rotate by 90 degrees clockwise and vertically flip.
19261 Do not apply the transposition if the input geometry matches the one
19262 specified by the specified value. It accepts the following values:
19265 Always apply transposition. (default)
19267 Preserve portrait geometry (when @var{height} >= @var{width}).
19269 Preserve landscape geometry (when @var{width} >= @var{height}).
19275 Trim the input so that the output contains one continuous subpart of the input.
19277 It accepts the following parameters:
19280 Specify the time of the start of the kept section, i.e. the frame with the
19281 timestamp @var{start} will be the first frame in the output.
19284 Specify the time of the first frame that will be dropped, i.e. the frame
19285 immediately preceding the one with the timestamp @var{end} will be the last
19286 frame in the output.
19289 This is the same as @var{start}, except this option sets the start timestamp
19290 in timebase units instead of seconds.
19293 This is the same as @var{end}, except this option sets the end timestamp
19294 in timebase units instead of seconds.
19297 The maximum duration of the output in seconds.
19300 The number of the first frame that should be passed to the output.
19303 The number of the first frame that should be dropped.
19306 @option{start}, @option{end}, and @option{duration} are expressed as time
19307 duration specifications; see
19308 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19309 for the accepted syntax.
19311 Note that the first two sets of the start/end options and the @option{duration}
19312 option look at the frame timestamp, while the _frame variants simply count the
19313 frames that pass through the filter. Also note that this filter does not modify
19314 the timestamps. If you wish for the output timestamps to start at zero, insert a
19315 setpts filter after the trim filter.
19317 If multiple start or end options are set, this filter tries to be greedy and
19318 keep all the frames that match at least one of the specified constraints. To keep
19319 only the part that matches all the constraints at once, chain multiple trim
19322 The defaults are such that all the input is kept. So it is possible to set e.g.
19323 just the end values to keep everything before the specified time.
19328 Drop everything except the second minute of input:
19330 ffmpeg -i INPUT -vf trim=60:120
19334 Keep only the first second:
19336 ffmpeg -i INPUT -vf trim=duration=1
19341 @section unpremultiply
19342 Apply alpha unpremultiply effect to input video stream using first plane
19343 of second stream as alpha.
19345 Both streams must have same dimensions and same pixel format.
19347 The filter accepts the following option:
19351 Set which planes will be processed, unprocessed planes will be copied.
19352 By default value 0xf, all planes will be processed.
19354 If the format has 1 or 2 components, then luma is bit 0.
19355 If the format has 3 or 4 components:
19356 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19357 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19358 If present, the alpha channel is always the last bit.
19361 Do not require 2nd input for processing, instead use alpha plane from input stream.
19367 Sharpen or blur the input video.
19369 It accepts the following parameters:
19372 @item luma_msize_x, lx
19373 Set the luma matrix horizontal size. It must be an odd integer between
19374 3 and 23. The default value is 5.
19376 @item luma_msize_y, ly
19377 Set the luma matrix vertical size. It must be an odd integer between 3
19378 and 23. The default value is 5.
19380 @item luma_amount, la
19381 Set the luma effect strength. It must be a floating point number, reasonable
19382 values lay between -1.5 and 1.5.
19384 Negative values will blur the input video, while positive values will
19385 sharpen it, a value of zero will disable the effect.
19387 Default value is 1.0.
19389 @item chroma_msize_x, cx
19390 Set the chroma matrix horizontal size. It must be an odd integer
19391 between 3 and 23. The default value is 5.
19393 @item chroma_msize_y, cy
19394 Set the chroma matrix vertical size. It must be an odd integer
19395 between 3 and 23. The default value is 5.
19397 @item chroma_amount, ca
19398 Set the chroma effect strength. It must be a floating point number, reasonable
19399 values lay between -1.5 and 1.5.
19401 Negative values will blur the input video, while positive values will
19402 sharpen it, a value of zero will disable the effect.
19404 Default value is 0.0.
19408 All parameters are optional and default to the equivalent of the
19409 string '5:5:1.0:5:5:0.0'.
19411 @subsection Examples
19415 Apply strong luma sharpen effect:
19417 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19421 Apply a strong blur of both luma and chroma parameters:
19423 unsharp=7:7:-2:7:7:-2
19430 Decompose a video made of tiled images into the individual images.
19432 The frame rate of the output video is the frame rate of the input video
19433 multiplied by the number of tiles.
19435 This filter does the reverse of @ref{tile}.
19437 The filter accepts the following options:
19442 Set the grid size (i.e. the number of lines and columns). For the syntax of
19443 this option, check the
19444 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19447 @subsection Examples
19451 Produce a 1-second video from a still image file made of 25 frames stacked
19452 vertically, like an analogic film reel:
19454 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19460 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19461 the image at several (or - in the case of @option{quality} level @code{8} - all)
19462 shifts and average the results.
19464 The way this differs from the behavior of spp is that uspp actually encodes &
19465 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19466 DCT similar to MJPEG.
19468 The filter accepts the following options:
19472 Set quality. This option defines the number of levels for averaging. It accepts
19473 an integer in the range 0-8. If set to @code{0}, the filter will have no
19474 effect. A value of @code{8} means the higher quality. For each increment of
19475 that value the speed drops by a factor of approximately 2. Default value is
19479 Force a constant quantization parameter. If not set, the filter will use the QP
19480 from the video stream (if available).
19485 Convert 360 videos between various formats.
19487 The filter accepts the following options:
19493 Set format of the input/output video.
19501 Equirectangular projection.
19506 Cubemap with 3x2/6x1/1x6 layout.
19508 Format specific options:
19513 Set padding proportion for the input/output cubemap. Values in decimals.
19520 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)
19523 Default value is @b{@samp{0}}.
19524 Maximum value is @b{@samp{0.1}}.
19528 Set fixed padding for the input/output cubemap. Values in pixels.
19530 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19534 Set order of faces for the input/output cubemap. Choose one direction for each position.
19536 Designation of directions:
19552 Default value is @b{@samp{rludfb}}.
19556 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19558 Designation of angles:
19561 0 degrees clockwise
19563 90 degrees clockwise
19565 180 degrees clockwise
19567 270 degrees clockwise
19570 Default value is @b{@samp{000000}}.
19574 Equi-Angular Cubemap.
19581 Format specific options:
19586 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19588 If diagonal field of view is set it overrides horizontal and vertical field of view.
19593 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19595 If diagonal field of view is set it overrides horizontal and vertical field of view.
19601 Format specific options:
19606 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19608 If diagonal field of view is set it overrides horizontal and vertical field of view.
19613 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19615 If diagonal field of view is set it overrides horizontal and vertical field of view.
19621 Facebook's 360 formats.
19624 Stereographic format.
19626 Format specific options:
19631 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19633 If diagonal field of view is set it overrides horizontal and vertical field of view.
19638 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19640 If diagonal field of view is set it overrides horizontal and vertical field of view.
19647 Ball format, gives significant distortion toward the back.
19650 Hammer-Aitoff map projection format.
19653 Sinusoidal map projection format.
19656 Fisheye projection.
19658 Format specific options:
19663 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19665 If diagonal field of view is set it overrides horizontal and vertical field of view.
19670 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19672 If diagonal field of view is set it overrides horizontal and vertical field of view.
19676 Pannini projection.
19678 Format specific options:
19681 Set output pannini parameter.
19684 Set input pannini parameter.
19688 Cylindrical projection.
19690 Format specific options:
19695 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19697 If diagonal field of view is set it overrides horizontal and vertical field of view.
19702 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19704 If diagonal field of view is set it overrides horizontal and vertical field of view.
19708 Perspective projection. @i{(output only)}
19710 Format specific options:
19713 Set perspective parameter.
19717 Tetrahedron projection.
19720 Truncated square pyramid projection.
19724 Half equirectangular projection.
19729 Format specific options:
19734 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19736 If diagonal field of view is set it overrides horizontal and vertical field of view.
19741 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19743 If diagonal field of view is set it overrides horizontal and vertical field of view.
19747 Orthographic format.
19749 Format specific options:
19754 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19756 If diagonal field of view is set it overrides horizontal and vertical field of view.
19761 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19763 If diagonal field of view is set it overrides horizontal and vertical field of view.
19767 Octahedron projection.
19771 Set interpolation method.@*
19772 @i{Note: more complex interpolation methods require much more memory to run.}
19782 Bilinear interpolation.
19784 Lagrange9 interpolation.
19787 Bicubic interpolation.
19790 Lanczos interpolation.
19793 Spline16 interpolation.
19796 Gaussian interpolation.
19798 Mitchell interpolation.
19801 Default value is @b{@samp{line}}.
19805 Set the output video resolution.
19807 Default resolution depends on formats.
19811 Set the input/output stereo format.
19822 Default value is @b{@samp{2d}} for input and output format.
19827 Set rotation for the output video. Values in degrees.
19830 Set rotation order for the output video. Choose one item for each position.
19841 Default value is @b{@samp{ypr}}.
19846 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
19850 Set if input video is flipped horizontally/vertically. Boolean values.
19853 Set if input video is transposed. Boolean value, by default disabled.
19856 Set if output video needs to be transposed. Boolean value, by default disabled.
19859 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
19862 @subsection Examples
19866 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
19868 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19871 Extract back view of Equi-Angular Cubemap:
19873 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19876 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
19878 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19882 @subsection Commands
19884 This filter supports subset of above options as @ref{commands}.
19886 @section vaguedenoiser
19888 Apply a wavelet based denoiser.
19890 It transforms each frame from the video input into the wavelet domain,
19891 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19892 the obtained coefficients. It does an inverse wavelet transform after.
19893 Due to wavelet properties, it should give a nice smoothed result, and
19894 reduced noise, without blurring picture features.
19896 This filter accepts the following options:
19900 The filtering strength. The higher, the more filtered the video will be.
19901 Hard thresholding can use a higher threshold than soft thresholding
19902 before the video looks overfiltered. Default value is 2.
19905 The filtering method the filter will use.
19907 It accepts the following values:
19910 All values under the threshold will be zeroed.
19913 All values under the threshold will be zeroed. All values above will be
19914 reduced by the threshold.
19917 Scales or nullifies coefficients - intermediary between (more) soft and
19918 (less) hard thresholding.
19921 Default is garrote.
19924 Number of times, the wavelet will decompose the picture. Picture can't
19925 be decomposed beyond a particular point (typically, 8 for a 640x480
19926 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
19929 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
19932 A list of the planes to process. By default all planes are processed.
19935 The threshold type the filter will use.
19937 It accepts the following values:
19940 Threshold used is same for all decompositions.
19943 Threshold used depends also on each decomposition coefficients.
19946 Default is universal.
19949 @section vectorscope
19951 Display 2 color component values in the two dimensional graph (which is called
19954 This filter accepts the following options:
19958 Set vectorscope mode.
19960 It accepts the following values:
19964 Gray values are displayed on graph, higher brightness means more pixels have
19965 same component color value on location in graph. This is the default mode.
19968 Gray values are displayed on graph. Surrounding pixels values which are not
19969 present in video frame are drawn in gradient of 2 color components which are
19970 set by option @code{x} and @code{y}. The 3rd color component is static.
19973 Actual color components values present in video frame are displayed on graph.
19976 Similar as color2 but higher frequency of same values @code{x} and @code{y}
19977 on graph increases value of another color component, which is luminance by
19978 default values of @code{x} and @code{y}.
19981 Actual colors present in video frame are displayed on graph. If two different
19982 colors map to same position on graph then color with higher value of component
19983 not present in graph is picked.
19986 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
19987 component picked from radial gradient.
19991 Set which color component will be represented on X-axis. Default is @code{1}.
19994 Set which color component will be represented on Y-axis. Default is @code{2}.
19997 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
19998 of color component which represents frequency of (X, Y) location in graph.
20003 No envelope, this is default.
20006 Instant envelope, even darkest single pixel will be clearly highlighted.
20009 Hold maximum and minimum values presented in graph over time. This way you
20010 can still spot out of range values without constantly looking at vectorscope.
20013 Peak and instant envelope combined together.
20017 Set what kind of graticule to draw.
20026 Set graticule opacity.
20029 Set graticule flags.
20033 Draw graticule for white point.
20036 Draw graticule for black point.
20039 Draw color points short names.
20043 Set background opacity.
20045 @item lthreshold, l
20046 Set low threshold for color component not represented on X or Y axis.
20047 Values lower than this value will be ignored. Default is 0.
20048 Note this value is multiplied with actual max possible value one pixel component
20049 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20052 @item hthreshold, h
20053 Set high threshold for color component not represented on X or Y axis.
20054 Values higher than this value will be ignored. Default is 1.
20055 Note this value is multiplied with actual max possible value one pixel component
20056 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20057 is 0.9 * 255 = 230.
20059 @item colorspace, c
20060 Set what kind of colorspace to use when drawing graticule.
20070 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20071 This means no tint, and output will remain gray.
20074 @anchor{vidstabdetect}
20075 @section vidstabdetect
20077 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20078 @ref{vidstabtransform} for pass 2.
20080 This filter generates a file with relative translation and rotation
20081 transform information about subsequent frames, which is then used by
20082 the @ref{vidstabtransform} filter.
20084 To enable compilation of this filter you need to configure FFmpeg with
20085 @code{--enable-libvidstab}.
20087 This filter accepts the following options:
20091 Set the path to the file used to write the transforms information.
20092 Default value is @file{transforms.trf}.
20095 Set how shaky the video is and how quick the camera is. It accepts an
20096 integer in the range 1-10, a value of 1 means little shakiness, a
20097 value of 10 means strong shakiness. Default value is 5.
20100 Set the accuracy of the detection process. It must be a value in the
20101 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20102 accuracy. Default value is 15.
20105 Set stepsize of the search process. The region around minimum is
20106 scanned with 1 pixel resolution. Default value is 6.
20109 Set minimum contrast. Below this value a local measurement field is
20110 discarded. Must be a floating point value in the range 0-1. Default
20114 Set reference frame number for tripod mode.
20116 If enabled, the motion of the frames is compared to a reference frame
20117 in the filtered stream, identified by the specified number. The idea
20118 is to compensate all movements in a more-or-less static scene and keep
20119 the camera view absolutely still.
20121 If set to 0, it is disabled. The frames are counted starting from 1.
20124 Show fields and transforms in the resulting frames. It accepts an
20125 integer in the range 0-2. Default value is 0, which disables any
20129 @subsection Examples
20133 Use default values:
20139 Analyze strongly shaky movie and put the results in file
20140 @file{mytransforms.trf}:
20142 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20146 Visualize the result of internal transformations in the resulting
20149 vidstabdetect=show=1
20153 Analyze a video with medium shakiness using @command{ffmpeg}:
20155 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20159 @anchor{vidstabtransform}
20160 @section vidstabtransform
20162 Video stabilization/deshaking: pass 2 of 2,
20163 see @ref{vidstabdetect} for pass 1.
20165 Read a file with transform information for each frame and
20166 apply/compensate them. Together with the @ref{vidstabdetect}
20167 filter this can be used to deshake videos. See also
20168 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20169 the @ref{unsharp} filter, see below.
20171 To enable compilation of this filter you need to configure FFmpeg with
20172 @code{--enable-libvidstab}.
20174 @subsection Options
20178 Set path to the file used to read the transforms. Default value is
20179 @file{transforms.trf}.
20182 Set the number of frames (value*2 + 1) used for lowpass filtering the
20183 camera movements. Default value is 10.
20185 For example a number of 10 means that 21 frames are used (10 in the
20186 past and 10 in the future) to smoothen the motion in the video. A
20187 larger value leads to a smoother video, but limits the acceleration of
20188 the camera (pan/tilt movements). 0 is a special case where a static
20189 camera is simulated.
20192 Set the camera path optimization algorithm.
20194 Accepted values are:
20197 gaussian kernel low-pass filter on camera motion (default)
20199 averaging on transformations
20203 Set maximal number of pixels to translate frames. Default value is -1,
20207 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20208 value is -1, meaning no limit.
20211 Specify how to deal with borders that may be visible due to movement
20214 Available values are:
20217 keep image information from previous frame (default)
20219 fill the border black
20223 Invert transforms if set to 1. Default value is 0.
20226 Consider transforms as relative to previous frame if set to 1,
20227 absolute if set to 0. Default value is 0.
20230 Set percentage to zoom. A positive value will result in a zoom-in
20231 effect, a negative value in a zoom-out effect. Default value is 0 (no
20235 Set optimal zooming to avoid borders.
20237 Accepted values are:
20242 optimal static zoom value is determined (only very strong movements
20243 will lead to visible borders) (default)
20245 optimal adaptive zoom value is determined (no borders will be
20246 visible), see @option{zoomspeed}
20249 Note that the value given at zoom is added to the one calculated here.
20252 Set percent to zoom maximally each frame (enabled when
20253 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20257 Specify type of interpolation.
20259 Available values are:
20264 linear only horizontal
20266 linear in both directions (default)
20268 cubic in both directions (slow)
20272 Enable virtual tripod mode if set to 1, which is equivalent to
20273 @code{relative=0:smoothing=0}. Default value is 0.
20275 Use also @code{tripod} option of @ref{vidstabdetect}.
20278 Increase log verbosity if set to 1. Also the detected global motions
20279 are written to the temporary file @file{global_motions.trf}. Default
20283 @subsection Examples
20287 Use @command{ffmpeg} for a typical stabilization with default values:
20289 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20292 Note the use of the @ref{unsharp} filter which is always recommended.
20295 Zoom in a bit more and load transform data from a given file:
20297 vidstabtransform=zoom=5:input="mytransforms.trf"
20301 Smoothen the video even more:
20303 vidstabtransform=smoothing=30
20309 Flip the input video vertically.
20311 For example, to vertically flip a video with @command{ffmpeg}:
20313 ffmpeg -i in.avi -vf "vflip" out.avi
20318 Detect variable frame rate video.
20320 This filter tries to detect if the input is variable or constant frame rate.
20322 At end it will output number of frames detected as having variable delta pts,
20323 and ones with constant delta pts.
20324 If there was frames with variable delta, than it will also show min, max and
20325 average delta encountered.
20329 Boost or alter saturation.
20331 The filter accepts the following options:
20334 Set strength of boost if positive value or strength of alter if negative value.
20335 Default is 0. Allowed range is from -2 to 2.
20338 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20341 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20344 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20347 Set the red luma coefficient.
20350 Set the green luma coefficient.
20353 Set the blue luma coefficient.
20356 If @code{intensity} is negative and this is set to 1, colors will change,
20357 otherwise colors will be less saturated, more towards gray.
20360 @subsection Commands
20362 This filter supports the all above options as @ref{commands}.
20367 Make or reverse a natural vignetting effect.
20369 The filter accepts the following options:
20373 Set lens angle expression as a number of radians.
20375 The value is clipped in the @code{[0,PI/2]} range.
20377 Default value: @code{"PI/5"}
20381 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20385 Set forward/backward mode.
20387 Available modes are:
20390 The larger the distance from the central point, the darker the image becomes.
20393 The larger the distance from the central point, the brighter the image becomes.
20394 This can be used to reverse a vignette effect, though there is no automatic
20395 detection to extract the lens @option{angle} and other settings (yet). It can
20396 also be used to create a burning effect.
20399 Default value is @samp{forward}.
20402 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20404 It accepts the following values:
20407 Evaluate expressions only once during the filter initialization.
20410 Evaluate expressions for each incoming frame. This is way slower than the
20411 @samp{init} mode since it requires all the scalers to be re-computed, but it
20412 allows advanced dynamic expressions.
20415 Default value is @samp{init}.
20418 Set dithering to reduce the circular banding effects. Default is @code{1}
20422 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20423 Setting this value to the SAR of the input will make a rectangular vignetting
20424 following the dimensions of the video.
20426 Default is @code{1/1}.
20429 @subsection Expressions
20431 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20432 following parameters.
20437 input width and height
20440 the number of input frame, starting from 0
20443 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20444 @var{TB} units, NAN if undefined
20447 frame rate of the input video, NAN if the input frame rate is unknown
20450 the PTS (Presentation TimeStamp) of the filtered video frame,
20451 expressed in seconds, NAN if undefined
20454 time base of the input video
20458 @subsection Examples
20462 Apply simple strong vignetting effect:
20468 Make a flickering vignetting:
20470 vignette='PI/4+random(1)*PI/50':eval=frame
20475 @section vmafmotion
20477 Obtain the average VMAF motion score of a video.
20478 It is one of the component metrics of VMAF.
20480 The obtained average motion score is printed through the logging system.
20482 The filter accepts the following options:
20486 If specified, the filter will use the named file to save the motion score of
20487 each frame with respect to the previous frame.
20488 When filename equals "-" the data is sent to standard output.
20493 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20497 Stack input videos vertically.
20499 All streams must be of same pixel format and of same width.
20501 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20502 to create same output.
20504 The filter accepts the following options:
20508 Set number of input streams. Default is 2.
20511 If set to 1, force the output to terminate when the shortest input
20512 terminates. Default value is 0.
20517 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20518 Deinterlacing Filter").
20520 Based on the process described by Martin Weston for BBC R&D, and
20521 implemented based on the de-interlace algorithm written by Jim
20522 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20523 uses filter coefficients calculated by BBC R&D.
20525 This filter uses field-dominance information in frame to decide which
20526 of each pair of fields to place first in the output.
20527 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20529 There are two sets of filter coefficients, so called "simple"
20530 and "complex". Which set of filter coefficients is used can
20531 be set by passing an optional parameter:
20535 Set the interlacing filter coefficients. Accepts one of the following values:
20539 Simple filter coefficient set.
20541 More-complex filter coefficient set.
20543 Default value is @samp{complex}.
20546 Specify which frames to deinterlace. Accepts one of the following values:
20550 Deinterlace all frames,
20552 Only deinterlace frames marked as interlaced.
20555 Default value is @samp{all}.
20559 Video waveform monitor.
20561 The waveform monitor plots color component intensity. By default luminance
20562 only. Each column of the waveform corresponds to a column of pixels in the
20565 It accepts the following options:
20569 Can be either @code{row}, or @code{column}. Default is @code{column}.
20570 In row mode, the graph on the left side represents color component value 0 and
20571 the right side represents value = 255. In column mode, the top side represents
20572 color component value = 0 and bottom side represents value = 255.
20575 Set intensity. Smaller values are useful to find out how many values of the same
20576 luminance are distributed across input rows/columns.
20577 Default value is @code{0.04}. Allowed range is [0, 1].
20580 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20581 In mirrored mode, higher values will be represented on the left
20582 side for @code{row} mode and at the top for @code{column} mode. Default is
20583 @code{1} (mirrored).
20587 It accepts the following values:
20590 Presents information identical to that in the @code{parade}, except
20591 that the graphs representing color components are superimposed directly
20594 This display mode makes it easier to spot relative differences or similarities
20595 in overlapping areas of the color components that are supposed to be identical,
20596 such as neutral whites, grays, or blacks.
20599 Display separate graph for the color components side by side in
20600 @code{row} mode or one below the other in @code{column} mode.
20603 Display separate graph for the color components side by side in
20604 @code{column} mode or one below the other in @code{row} mode.
20606 Using this display mode makes it easy to spot color casts in the highlights
20607 and shadows of an image, by comparing the contours of the top and the bottom
20608 graphs of each waveform. Since whites, grays, and blacks are characterized
20609 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20610 should display three waveforms of roughly equal width/height. If not, the
20611 correction is easy to perform by making level adjustments the three waveforms.
20613 Default is @code{stack}.
20615 @item components, c
20616 Set which color components to display. Default is 1, which means only luminance
20617 or red color component if input is in RGB colorspace. If is set for example to
20618 7 it will display all 3 (if) available color components.
20623 No envelope, this is default.
20626 Instant envelope, minimum and maximum values presented in graph will be easily
20627 visible even with small @code{step} value.
20630 Hold minimum and maximum values presented in graph across time. This way you
20631 can still spot out of range values without constantly looking at waveforms.
20634 Peak and instant envelope combined together.
20640 No filtering, this is default.
20643 Luma and chroma combined together.
20646 Similar as above, but shows difference between blue and red chroma.
20649 Similar as above, but use different colors.
20652 Similar as above, but again with different colors.
20655 Displays only chroma.
20658 Displays actual color value on waveform.
20661 Similar as above, but with luma showing frequency of chroma values.
20665 Set which graticule to display.
20669 Do not display graticule.
20672 Display green graticule showing legal broadcast ranges.
20675 Display orange graticule showing legal broadcast ranges.
20678 Display invert graticule showing legal broadcast ranges.
20682 Set graticule opacity.
20685 Set graticule flags.
20689 Draw numbers above lines. By default enabled.
20692 Draw dots instead of lines.
20696 Set scale used for displaying graticule.
20703 Default is digital.
20706 Set background opacity.
20710 Set tint for output.
20711 Only used with lowpass filter and when display is not overlay and input
20712 pixel formats are not RGB.
20715 @section weave, doubleweave
20717 The @code{weave} takes a field-based video input and join
20718 each two sequential fields into single frame, producing a new double
20719 height clip with half the frame rate and half the frame count.
20721 The @code{doubleweave} works same as @code{weave} but without
20722 halving frame rate and frame count.
20724 It accepts the following option:
20728 Set first field. Available values are:
20732 Set the frame as top-field-first.
20735 Set the frame as bottom-field-first.
20739 @subsection Examples
20743 Interlace video using @ref{select} and @ref{separatefields} filter:
20745 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20750 Apply the xBR high-quality magnification filter which is designed for pixel
20751 art. It follows a set of edge-detection rules, see
20752 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
20754 It accepts the following option:
20758 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
20759 @code{3xBR} and @code{4} for @code{4xBR}.
20760 Default is @code{3}.
20765 Apply cross fade from one input video stream to another input video stream.
20766 The cross fade is applied for specified duration.
20768 The filter accepts the following options:
20772 Set one of available transition effects:
20820 Default transition effect is fade.
20823 Set cross fade duration in seconds.
20824 Default duration is 1 second.
20827 Set cross fade start relative to first input stream in seconds.
20828 Default offset is 0.
20831 Set expression for custom transition effect.
20833 The expressions can use the following variables and functions:
20838 The coordinates of the current sample.
20842 The width and height of the image.
20845 Progress of transition effect.
20848 Currently processed plane.
20851 Return value of first input at current location and plane.
20854 Return value of second input at current location and plane.
20860 Return the value of the pixel at location (@var{x},@var{y}) of the
20861 first/second/third/fourth component of first input.
20867 Return the value of the pixel at location (@var{x},@var{y}) of the
20868 first/second/third/fourth component of second input.
20872 @subsection Examples
20876 Cross fade from one input video to another input video, with fade transition and duration of transition
20877 of 2 seconds starting at offset of 5 seconds:
20879 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20884 Pick median pixels from several input videos.
20886 The filter accepts the following options:
20890 Set number of inputs.
20891 Default is 3. Allowed range is from 3 to 255.
20892 If number of inputs is even number, than result will be mean value between two median values.
20895 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
20898 Set median percentile. Default value is @code{0.5}.
20899 Default value of @code{0.5} will pick always median values, while @code{0} will pick
20900 minimum values, and @code{1} maximum values.
20904 Stack video inputs into custom layout.
20906 All streams must be of same pixel format.
20908 The filter accepts the following options:
20912 Set number of input streams. Default is 2.
20915 Specify layout of inputs.
20916 This option requires the desired layout configuration to be explicitly set by the user.
20917 This sets position of each video input in output. Each input
20918 is separated by '|'.
20919 The first number represents the column, and the second number represents the row.
20920 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
20921 where X is video input from which to take width or height.
20922 Multiple values can be used when separated by '+'. In such
20923 case values are summed together.
20925 Note that if inputs are of different sizes gaps may appear, as not all of
20926 the output video frame will be filled. Similarly, videos can overlap each
20927 other if their position doesn't leave enough space for the full frame of
20930 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
20931 a layout must be set by the user.
20934 If set to 1, force the output to terminate when the shortest input
20935 terminates. Default value is 0.
20938 If set to valid color, all unused pixels will be filled with that color.
20939 By default fill is set to none, so it is disabled.
20942 @subsection Examples
20946 Display 4 inputs into 2x2 grid.
20950 input1(0, 0) | input3(w0, 0)
20951 input2(0, h0) | input4(w0, h0)
20955 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20958 Note that if inputs are of different sizes, gaps or overlaps may occur.
20961 Display 4 inputs into 1x4 grid.
20968 input4(0, h0+h1+h2)
20972 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20975 Note that if inputs are of different widths, unused space will appear.
20978 Display 9 inputs into 3x3 grid.
20982 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20983 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20984 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20988 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
20991 Note that if inputs are of different sizes, gaps or overlaps may occur.
20994 Display 16 inputs into 4x4 grid.
20998 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
20999 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21000 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21001 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21005 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|
21006 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
21009 Note that if inputs are of different sizes, gaps or overlaps may occur.
21016 Deinterlace the input video ("yadif" means "yet another deinterlacing
21019 It accepts the following parameters:
21025 The interlacing mode to adopt. It accepts one of the following values:
21028 @item 0, send_frame
21029 Output one frame for each frame.
21030 @item 1, send_field
21031 Output one frame for each field.
21032 @item 2, send_frame_nospatial
21033 Like @code{send_frame}, but it skips the spatial interlacing check.
21034 @item 3, send_field_nospatial
21035 Like @code{send_field}, but it skips the spatial interlacing check.
21038 The default value is @code{send_frame}.
21041 The picture field parity assumed for the input interlaced video. It accepts one
21042 of the following values:
21046 Assume the top field is first.
21048 Assume the bottom field is first.
21050 Enable automatic detection of field parity.
21053 The default value is @code{auto}.
21054 If the interlacing is unknown or the decoder does not export this information,
21055 top field first will be assumed.
21058 Specify which frames to deinterlace. Accepts one of the following
21063 Deinterlace all frames.
21064 @item 1, interlaced
21065 Only deinterlace frames marked as interlaced.
21068 The default value is @code{all}.
21071 @section yadif_cuda
21073 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21074 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21077 It accepts the following parameters:
21083 The interlacing mode to adopt. It accepts one of the following values:
21086 @item 0, send_frame
21087 Output one frame for each frame.
21088 @item 1, send_field
21089 Output one frame for each field.
21090 @item 2, send_frame_nospatial
21091 Like @code{send_frame}, but it skips the spatial interlacing check.
21092 @item 3, send_field_nospatial
21093 Like @code{send_field}, but it skips the spatial interlacing check.
21096 The default value is @code{send_frame}.
21099 The picture field parity assumed for the input interlaced video. It accepts one
21100 of the following values:
21104 Assume the top field is first.
21106 Assume the bottom field is first.
21108 Enable automatic detection of field parity.
21111 The default value is @code{auto}.
21112 If the interlacing is unknown or the decoder does not export this information,
21113 top field first will be assumed.
21116 Specify which frames to deinterlace. Accepts one of the following
21121 Deinterlace all frames.
21122 @item 1, interlaced
21123 Only deinterlace frames marked as interlaced.
21126 The default value is @code{all}.
21131 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21132 The algorithm is described in
21133 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21135 It accepts the following parameters:
21139 Set the window radius. Default value is 3.
21142 Set which planes to filter. Default is only the first plane.
21145 Set blur strength. Default value is 128.
21148 @subsection Commands
21149 This filter supports same @ref{commands} as options.
21153 Apply Zoom & Pan effect.
21155 This filter accepts the following options:
21159 Set the zoom expression. Range is 1-10. Default is 1.
21163 Set the x and y expression. Default is 0.
21166 Set the duration expression in number of frames.
21167 This sets for how many number of frames effect will last for
21168 single input image.
21171 Set the output image size, default is 'hd720'.
21174 Set the output frame rate, default is '25'.
21177 Each expression can contain the following constants:
21196 Output frame count.
21199 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21201 @item out_time, time, ot
21202 The output timestamp expressed in seconds.
21206 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21207 for current input frame.
21211 'x' and 'y' of last output frame of previous input frame or 0 when there was
21212 not yet such frame (first input frame).
21215 Last calculated zoom from 'z' expression for current input frame.
21218 Last calculated zoom of last output frame of previous input frame.
21221 Number of output frames for current input frame. Calculated from 'd' expression
21222 for each input frame.
21225 number of output frames created for previous input frame
21228 Rational number: input width / input height
21231 sample aspect ratio
21234 display aspect ratio
21238 @subsection Examples
21242 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21244 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
21248 Zoom in up to 1.5x and pan always at center of picture:
21250 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21254 Same as above but without pausing:
21256 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21260 Zoom in 2x into center of picture only for the first second of the input video:
21262 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21269 Scale (resize) the input video, using the z.lib library:
21270 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21271 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21273 The zscale filter forces the output display aspect ratio to be the same
21274 as the input, by changing the output sample aspect ratio.
21276 If the input image format is different from the format requested by
21277 the next filter, the zscale filter will convert the input to the
21280 @subsection Options
21281 The filter accepts the following options.
21286 Set the output video dimension expression. Default value is the input
21289 If the @var{width} or @var{w} value is 0, the input width is used for
21290 the output. If the @var{height} or @var{h} value is 0, the input height
21291 is used for the output.
21293 If one and only one of the values is -n with n >= 1, the zscale filter
21294 will use a value that maintains the aspect ratio of the input image,
21295 calculated from the other specified dimension. After that it will,
21296 however, make sure that the calculated dimension is divisible by n and
21297 adjust the value if necessary.
21299 If both values are -n with n >= 1, the behavior will be identical to
21300 both values being set to 0 as previously detailed.
21302 See below for the list of accepted constants for use in the dimension
21306 Set the video size. For the syntax of this option, check the
21307 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21310 Set the dither type.
21312 Possible values are:
21317 @item error_diffusion
21323 Set the resize filter type.
21325 Possible values are:
21335 Default is bilinear.
21338 Set the color range.
21340 Possible values are:
21347 Default is same as input.
21350 Set the color primaries.
21352 Possible values are:
21362 Default is same as input.
21365 Set the transfer characteristics.
21367 Possible values are:
21381 Default is same as input.
21384 Set the colorspace matrix.
21386 Possible value are:
21397 Default is same as input.
21400 Set the input color range.
21402 Possible values are:
21409 Default is same as input.
21411 @item primariesin, pin
21412 Set the input color primaries.
21414 Possible values are:
21424 Default is same as input.
21426 @item transferin, tin
21427 Set the input transfer characteristics.
21429 Possible values are:
21440 Default is same as input.
21442 @item matrixin, min
21443 Set the input colorspace matrix.
21445 Possible value are:
21457 Set the output chroma location.
21459 Possible values are:
21470 @item chromalin, cin
21471 Set the input chroma location.
21473 Possible values are:
21485 Set the nominal peak luminance.
21488 The values of the @option{w} and @option{h} options are expressions
21489 containing the following constants:
21494 The input width and height
21498 These are the same as @var{in_w} and @var{in_h}.
21502 The output (scaled) width and height
21506 These are the same as @var{out_w} and @var{out_h}
21509 The same as @var{iw} / @var{ih}
21512 input sample aspect ratio
21515 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21519 horizontal and vertical input chroma subsample values. For example for the
21520 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21524 horizontal and vertical output chroma subsample values. For example for the
21525 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21528 @subsection Commands
21530 This filter supports the following commands:
21534 Set the output video dimension expression.
21535 The command accepts the same syntax of the corresponding option.
21537 If the specified expression is not valid, it is kept at its current
21541 @c man end VIDEO FILTERS
21543 @chapter OpenCL Video Filters
21544 @c man begin OPENCL VIDEO FILTERS
21546 Below is a description of the currently available OpenCL video filters.
21548 To enable compilation of these filters you need to configure FFmpeg with
21549 @code{--enable-opencl}.
21551 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21554 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21555 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21556 given device parameters.
21558 @item -filter_hw_device @var{name}
21559 Pass the hardware device called @var{name} to all filters in any filter graph.
21563 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21567 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21569 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21573 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.
21575 @section avgblur_opencl
21577 Apply average blur filter.
21579 The filter accepts the following options:
21583 Set horizontal radius size.
21584 Range is @code{[1, 1024]} and default value is @code{1}.
21587 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21590 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21593 @subsection Example
21597 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.
21599 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21603 @section boxblur_opencl
21605 Apply a boxblur algorithm to the input video.
21607 It accepts the following parameters:
21611 @item luma_radius, lr
21612 @item luma_power, lp
21613 @item chroma_radius, cr
21614 @item chroma_power, cp
21615 @item alpha_radius, ar
21616 @item alpha_power, ap
21620 A description of the accepted options follows.
21623 @item luma_radius, lr
21624 @item chroma_radius, cr
21625 @item alpha_radius, ar
21626 Set an expression for the box radius in pixels used for blurring the
21627 corresponding input plane.
21629 The radius value must be a non-negative number, and must not be
21630 greater than the value of the expression @code{min(w,h)/2} for the
21631 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21634 Default value for @option{luma_radius} is "2". If not specified,
21635 @option{chroma_radius} and @option{alpha_radius} default to the
21636 corresponding value set for @option{luma_radius}.
21638 The expressions can contain the following constants:
21642 The input width and height in pixels.
21646 The input chroma image width and height in pixels.
21650 The horizontal and vertical chroma subsample values. For example, for the
21651 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21654 @item luma_power, lp
21655 @item chroma_power, cp
21656 @item alpha_power, ap
21657 Specify how many times the boxblur filter is applied to the
21658 corresponding plane.
21660 Default value for @option{luma_power} is 2. If not specified,
21661 @option{chroma_power} and @option{alpha_power} default to the
21662 corresponding value set for @option{luma_power}.
21664 A value of 0 will disable the effect.
21667 @subsection Examples
21669 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.
21673 Apply a boxblur filter with the luma, chroma, and alpha radius
21674 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.
21676 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21677 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21681 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.
21683 For the luma plane, a 2x2 box radius will be run once.
21685 For the chroma plane, a 4x4 box radius will be run 5 times.
21687 For the alpha plane, a 3x3 box radius will be run 7 times.
21689 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21693 @section colorkey_opencl
21694 RGB colorspace color keying.
21696 The filter accepts the following options:
21700 The color which will be replaced with transparency.
21703 Similarity percentage with the key color.
21705 0.01 matches only the exact key color, while 1.0 matches everything.
21710 0.0 makes pixels either fully transparent, or not transparent at all.
21712 Higher values result in semi-transparent pixels, with a higher transparency
21713 the more similar the pixels color is to the key color.
21716 @subsection Examples
21720 Make every semi-green pixel in the input transparent with some slight blending:
21722 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
21726 @section convolution_opencl
21728 Apply convolution of 3x3, 5x5, 7x7 matrix.
21730 The filter accepts the following options:
21737 Set matrix for each plane.
21738 Matrix is sequence of 9, 25 or 49 signed numbers.
21739 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
21745 Set multiplier for calculated value for each plane.
21746 If unset or 0, it will be sum of all matrix elements.
21747 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
21753 Set bias for each plane. This value is added to the result of the multiplication.
21754 Useful for making the overall image brighter or darker.
21755 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
21759 @subsection Examples
21765 -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
21771 -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
21775 Apply edge enhance:
21777 -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
21783 -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
21787 Apply laplacian edge detector which includes diagonals:
21789 -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
21795 -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
21799 @section erosion_opencl
21801 Apply erosion effect to the video.
21803 This filter replaces the pixel by the local(3x3) minimum.
21805 It accepts the following options:
21812 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21813 If @code{0}, plane will remain unchanged.
21816 Flag which specifies the pixel to refer to.
21817 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21819 Flags to local 3x3 coordinates region centered on @code{x}:
21828 @subsection Example
21832 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.
21834 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21838 @section deshake_opencl
21839 Feature-point based video stabilization filter.
21841 The filter accepts the following options:
21845 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
21848 Whether or not additional debug info should be displayed, both in the processed output and in the console.
21850 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
21852 Viewing point matches in the output video is only supported for RGB input.
21854 Defaults to @code{0}.
21856 @item adaptive_crop
21857 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
21859 Defaults to @code{1}.
21861 @item refine_features
21862 Whether or not feature points should be refined at a sub-pixel level.
21864 This can be turned off for a slight performance gain at the cost of precision.
21866 Defaults to @code{1}.
21868 @item smooth_strength
21869 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
21871 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
21873 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
21875 Defaults to @code{0.0}.
21877 @item smooth_window_multiplier
21878 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
21880 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
21882 Acceptable values range from @code{0.1} to @code{10.0}.
21884 Larger values increase the amount of motion data available for determining how to smooth the camera path,
21885 potentially improving smoothness, but also increase latency and memory usage.
21887 Defaults to @code{2.0}.
21891 @subsection Examples
21895 Stabilize a video with a fixed, medium smoothing strength:
21897 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21901 Stabilize a video with debugging (both in console and in rendered video):
21903 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21907 @section dilation_opencl
21909 Apply dilation effect to the video.
21911 This filter replaces the pixel by the local(3x3) maximum.
21913 It accepts the following options:
21920 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21921 If @code{0}, plane will remain unchanged.
21924 Flag which specifies the pixel to refer to.
21925 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21927 Flags to local 3x3 coordinates region centered on @code{x}:
21936 @subsection Example
21940 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.
21942 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21946 @section nlmeans_opencl
21948 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
21950 @section overlay_opencl
21952 Overlay one video on top of another.
21954 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
21955 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
21957 The filter accepts the following options:
21962 Set the x coordinate of the overlaid video on the main video.
21963 Default value is @code{0}.
21966 Set the y coordinate of the overlaid video on the main video.
21967 Default value is @code{0}.
21971 @subsection Examples
21975 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
21977 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21980 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
21982 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21987 @section pad_opencl
21989 Add paddings to the input image, and place the original input at the
21990 provided @var{x}, @var{y} coordinates.
21992 It accepts the following options:
21997 Specify an expression for the size of the output image with the
21998 paddings added. If the value for @var{width} or @var{height} is 0, the
21999 corresponding input size is used for the output.
22001 The @var{width} expression can reference the value set by the
22002 @var{height} expression, and vice versa.
22004 The default value of @var{width} and @var{height} is 0.
22008 Specify the offsets to place the input image at within the padded area,
22009 with respect to the top/left border of the output image.
22011 The @var{x} expression can reference the value set by the @var{y}
22012 expression, and vice versa.
22014 The default value of @var{x} and @var{y} is 0.
22016 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22017 so the input image is centered on the padded area.
22020 Specify the color of the padded area. For the syntax of this option,
22021 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22022 manual,ffmpeg-utils}.
22025 Pad to an aspect instead to a resolution.
22028 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22029 options are expressions containing the following constants:
22034 The input video width and height.
22038 These are the same as @var{in_w} and @var{in_h}.
22042 The output width and height (the size of the padded area), as
22043 specified by the @var{width} and @var{height} expressions.
22047 These are the same as @var{out_w} and @var{out_h}.
22051 The x and y offsets as specified by the @var{x} and @var{y}
22052 expressions, or NAN if not yet specified.
22055 same as @var{iw} / @var{ih}
22058 input sample aspect ratio
22061 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22064 @section prewitt_opencl
22066 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22068 The filter accepts the following option:
22072 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22075 Set value which will be multiplied with filtered result.
22076 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22079 Set value which will be added to filtered result.
22080 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22083 @subsection Example
22087 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22089 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22093 @anchor{program_opencl}
22094 @section program_opencl
22096 Filter video using an OpenCL program.
22101 OpenCL program source file.
22104 Kernel name in program.
22107 Number of inputs to the filter. Defaults to 1.
22110 Size of output frames. Defaults to the same as the first input.
22114 The @code{program_opencl} filter also supports the @ref{framesync} options.
22116 The program source file must contain a kernel function with the given name,
22117 which will be run once for each plane of the output. Each run on a plane
22118 gets enqueued as a separate 2D global NDRange with one work-item for each
22119 pixel to be generated. The global ID offset for each work-item is therefore
22120 the coordinates of a pixel in the destination image.
22122 The kernel function needs to take the following arguments:
22125 Destination image, @var{__write_only image2d_t}.
22127 This image will become the output; the kernel should write all of it.
22129 Frame index, @var{unsigned int}.
22131 This is a counter starting from zero and increasing by one for each frame.
22133 Source images, @var{__read_only image2d_t}.
22135 These are the most recent images on each input. The kernel may read from
22136 them to generate the output, but they can't be written to.
22143 Copy the input to the output (output must be the same size as the input).
22145 __kernel void copy(__write_only image2d_t destination,
22146 unsigned int index,
22147 __read_only image2d_t source)
22149 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22151 int2 location = (int2)(get_global_id(0), get_global_id(1));
22153 float4 value = read_imagef(source, sampler, location);
22155 write_imagef(destination, location, value);
22160 Apply a simple transformation, rotating the input by an amount increasing
22161 with the index counter. Pixel values are linearly interpolated by the
22162 sampler, and the output need not have the same dimensions as the input.
22164 __kernel void rotate_image(__write_only image2d_t dst,
22165 unsigned int index,
22166 __read_only image2d_t src)
22168 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22169 CLK_FILTER_LINEAR);
22171 float angle = (float)index / 100.0f;
22173 float2 dst_dim = convert_float2(get_image_dim(dst));
22174 float2 src_dim = convert_float2(get_image_dim(src));
22176 float2 dst_cen = dst_dim / 2.0f;
22177 float2 src_cen = src_dim / 2.0f;
22179 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22181 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22183 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22184 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22186 src_pos = src_pos * src_dim / dst_dim;
22188 float2 src_loc = src_pos + src_cen;
22190 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22191 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22192 write_imagef(dst, dst_loc, 0.5f);
22194 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22199 Blend two inputs together, with the amount of each input used varying
22200 with the index counter.
22202 __kernel void blend_images(__write_only image2d_t dst,
22203 unsigned int index,
22204 __read_only image2d_t src1,
22205 __read_only image2d_t src2)
22207 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22208 CLK_FILTER_LINEAR);
22210 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22212 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22213 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22214 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22216 float4 val1 = read_imagef(src1, sampler, src1_loc);
22217 float4 val2 = read_imagef(src2, sampler, src2_loc);
22219 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22225 @section roberts_opencl
22226 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22228 The filter accepts the following option:
22232 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22235 Set value which will be multiplied with filtered result.
22236 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22239 Set value which will be added to filtered result.
22240 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22243 @subsection Example
22247 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22249 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22253 @section sobel_opencl
22255 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22257 The filter accepts the following option:
22261 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22264 Set value which will be multiplied with filtered result.
22265 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22268 Set value which will be added to filtered result.
22269 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22272 @subsection Example
22276 Apply sobel operator with scale set to 2 and delta set to 10
22278 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22282 @section tonemap_opencl
22284 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22286 It accepts the following parameters:
22290 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22293 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22296 Apply desaturation for highlights that exceed this level of brightness. The
22297 higher the parameter, the more color information will be preserved. This
22298 setting helps prevent unnaturally blown-out colors for super-highlights, by
22299 (smoothly) turning into white instead. This makes images feel more natural,
22300 at the cost of reducing information about out-of-range colors.
22302 The default value is 0.5, and the algorithm here is a little different from
22303 the cpu version tonemap currently. A setting of 0.0 disables this option.
22306 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22307 is used to detect whether the scene has changed or not. If the distance between
22308 the current frame average brightness and the current running average exceeds
22309 a threshold value, we would re-calculate scene average and peak brightness.
22310 The default value is 0.2.
22313 Specify the output pixel format.
22315 Currently supported formats are:
22322 Set the output color range.
22324 Possible values are:
22330 Default is same as input.
22333 Set the output color primaries.
22335 Possible values are:
22341 Default is same as input.
22344 Set the output transfer characteristics.
22346 Possible values are:
22355 Set the output colorspace matrix.
22357 Possible value are:
22363 Default is same as input.
22367 @subsection Example
22371 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22373 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22377 @section unsharp_opencl
22379 Sharpen or blur the input video.
22381 It accepts the following parameters:
22384 @item luma_msize_x, lx
22385 Set the luma matrix horizontal size.
22386 Range is @code{[1, 23]} and default value is @code{5}.
22388 @item luma_msize_y, ly
22389 Set the luma matrix vertical size.
22390 Range is @code{[1, 23]} and default value is @code{5}.
22392 @item luma_amount, la
22393 Set the luma effect strength.
22394 Range is @code{[-10, 10]} and default value is @code{1.0}.
22396 Negative values will blur the input video, while positive values will
22397 sharpen it, a value of zero will disable the effect.
22399 @item chroma_msize_x, cx
22400 Set the chroma matrix horizontal size.
22401 Range is @code{[1, 23]} and default value is @code{5}.
22403 @item chroma_msize_y, cy
22404 Set the chroma matrix vertical size.
22405 Range is @code{[1, 23]} and default value is @code{5}.
22407 @item chroma_amount, ca
22408 Set the chroma effect strength.
22409 Range is @code{[-10, 10]} and default value is @code{0.0}.
22411 Negative values will blur the input video, while positive values will
22412 sharpen it, a value of zero will disable the effect.
22416 All parameters are optional and default to the equivalent of the
22417 string '5:5:1.0:5:5:0.0'.
22419 @subsection Examples
22423 Apply strong luma sharpen effect:
22425 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22429 Apply a strong blur of both luma and chroma parameters:
22431 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22435 @section xfade_opencl
22437 Cross fade two videos with custom transition effect by using OpenCL.
22439 It accepts the following options:
22443 Set one of possible transition effects.
22447 Select custom transition effect, the actual transition description
22448 will be picked from source and kernel options.
22460 Default transition is fade.
22464 OpenCL program source file for custom transition.
22467 Set name of kernel to use for custom transition from program source file.
22470 Set duration of video transition.
22473 Set time of start of transition relative to first video.
22476 The program source file must contain a kernel function with the given name,
22477 which will be run once for each plane of the output. Each run on a plane
22478 gets enqueued as a separate 2D global NDRange with one work-item for each
22479 pixel to be generated. The global ID offset for each work-item is therefore
22480 the coordinates of a pixel in the destination image.
22482 The kernel function needs to take the following arguments:
22485 Destination image, @var{__write_only image2d_t}.
22487 This image will become the output; the kernel should write all of it.
22490 First Source image, @var{__read_only image2d_t}.
22491 Second Source image, @var{__read_only image2d_t}.
22493 These are the most recent images on each input. The kernel may read from
22494 them to generate the output, but they can't be written to.
22497 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22504 Apply dots curtain transition effect:
22506 __kernel void blend_images(__write_only image2d_t dst,
22507 __read_only image2d_t src1,
22508 __read_only image2d_t src2,
22511 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22512 CLK_FILTER_LINEAR);
22513 int2 p = (int2)(get_global_id(0), get_global_id(1));
22514 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22515 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22518 float2 dots = (float2)(20.0, 20.0);
22519 float2 center = (float2)(0,0);
22522 float4 val1 = read_imagef(src1, sampler, p);
22523 float4 val2 = read_imagef(src2, sampler, p);
22524 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22526 write_imagef(dst, p, next ? val1 : val2);
22532 @c man end OPENCL VIDEO FILTERS
22534 @chapter VAAPI Video Filters
22535 @c man begin VAAPI VIDEO FILTERS
22537 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22539 To enable compilation of these filters you need to configure FFmpeg with
22540 @code{--enable-vaapi}.
22542 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}
22544 @section tonemap_vaapi
22546 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22547 It maps the dynamic range of HDR10 content to the SDR content.
22548 It currently only accepts HDR10 as input.
22550 It accepts the following parameters:
22554 Specify the output pixel format.
22556 Currently supported formats are:
22565 Set the output color primaries.
22567 Default is same as input.
22570 Set the output transfer characteristics.
22575 Set the output colorspace matrix.
22577 Default is same as input.
22581 @subsection Example
22585 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22587 tonemap_vaapi=format=p010:t=bt2020-10
22591 @c man end VAAPI VIDEO FILTERS
22593 @chapter Video Sources
22594 @c man begin VIDEO SOURCES
22596 Below is a description of the currently available video sources.
22600 Buffer video frames, and make them available to the filter chain.
22602 This source is mainly intended for a programmatic use, in particular
22603 through the interface defined in @file{libavfilter/buffersrc.h}.
22605 It accepts the following parameters:
22610 Specify the size (width and height) of the buffered video frames. For the
22611 syntax of this option, check the
22612 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22615 The input video width.
22618 The input video height.
22621 A string representing the pixel format of the buffered video frames.
22622 It may be a number corresponding to a pixel format, or a pixel format
22626 Specify the timebase assumed by the timestamps of the buffered frames.
22629 Specify the frame rate expected for the video stream.
22631 @item pixel_aspect, sar
22632 The sample (pixel) aspect ratio of the input video.
22635 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22636 to the filtergraph description to specify swscale flags for automatically
22637 inserted scalers. See @ref{Filtergraph syntax}.
22639 @item hw_frames_ctx
22640 When using a hardware pixel format, this should be a reference to an
22641 AVHWFramesContext describing input frames.
22646 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22649 will instruct the source to accept video frames with size 320x240 and
22650 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22651 square pixels (1:1 sample aspect ratio).
22652 Since the pixel format with name "yuv410p" corresponds to the number 6
22653 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22654 this example corresponds to:
22656 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22659 Alternatively, the options can be specified as a flat string, but this
22660 syntax is deprecated:
22662 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22666 Create a pattern generated by an elementary cellular automaton.
22668 The initial state of the cellular automaton can be defined through the
22669 @option{filename} and @option{pattern} options. If such options are
22670 not specified an initial state is created randomly.
22672 At each new frame a new row in the video is filled with the result of
22673 the cellular automaton next generation. The behavior when the whole
22674 frame is filled is defined by the @option{scroll} option.
22676 This source accepts the following options:
22680 Read the initial cellular automaton state, i.e. the starting row, from
22681 the specified file.
22682 In the file, each non-whitespace character is considered an alive
22683 cell, a newline will terminate the row, and further characters in the
22684 file will be ignored.
22687 Read the initial cellular automaton state, i.e. the starting row, from
22688 the specified string.
22690 Each non-whitespace character in the string is considered an alive
22691 cell, a newline will terminate the row, and further characters in the
22692 string will be ignored.
22695 Set the video rate, that is the number of frames generated per second.
22698 @item random_fill_ratio, ratio
22699 Set the random fill ratio for the initial cellular automaton row. It
22700 is a floating point number value ranging from 0 to 1, defaults to
22703 This option is ignored when a file or a pattern is specified.
22705 @item random_seed, seed
22706 Set the seed for filling randomly the initial row, must be an integer
22707 included between 0 and UINT32_MAX. If not specified, or if explicitly
22708 set to -1, the filter will try to use a good random seed on a best
22712 Set the cellular automaton rule, it is a number ranging from 0 to 255.
22713 Default value is 110.
22716 Set the size of the output video. For the syntax of this option, check the
22717 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22719 If @option{filename} or @option{pattern} is specified, the size is set
22720 by default to the width of the specified initial state row, and the
22721 height is set to @var{width} * PHI.
22723 If @option{size} is set, it must contain the width of the specified
22724 pattern string, and the specified pattern will be centered in the
22727 If a filename or a pattern string is not specified, the size value
22728 defaults to "320x518" (used for a randomly generated initial state).
22731 If set to 1, scroll the output upward when all the rows in the output
22732 have been already filled. If set to 0, the new generated row will be
22733 written over the top row just after the bottom row is filled.
22736 @item start_full, full
22737 If set to 1, completely fill the output with generated rows before
22738 outputting the first frame.
22739 This is the default behavior, for disabling set the value to 0.
22742 If set to 1, stitch the left and right row edges together.
22743 This is the default behavior, for disabling set the value to 0.
22746 @subsection Examples
22750 Read the initial state from @file{pattern}, and specify an output of
22753 cellauto=f=pattern:s=200x400
22757 Generate a random initial row with a width of 200 cells, with a fill
22760 cellauto=ratio=2/3:s=200x200
22764 Create a pattern generated by rule 18 starting by a single alive cell
22765 centered on an initial row with width 100:
22767 cellauto=p=@@:s=100x400:full=0:rule=18
22771 Specify a more elaborated initial pattern:
22773 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
22778 @anchor{coreimagesrc}
22779 @section coreimagesrc
22780 Video source generated on GPU using Apple's CoreImage API on OSX.
22782 This video source is a specialized version of the @ref{coreimage} video filter.
22783 Use a core image generator at the beginning of the applied filterchain to
22784 generate the content.
22786 The coreimagesrc video source accepts the following options:
22788 @item list_generators
22789 List all available generators along with all their respective options as well as
22790 possible minimum and maximum values along with the default values.
22792 list_generators=true
22796 Specify the size of the sourced video. For the syntax of this option, check the
22797 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22798 The default value is @code{320x240}.
22801 Specify the frame rate of the sourced video, as the number of frames
22802 generated per second. It has to be a string in the format
22803 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22804 number or a valid video frame rate abbreviation. The default value is
22808 Set the sample aspect ratio of the sourced video.
22811 Set the duration of the sourced video. See
22812 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22813 for the accepted syntax.
22815 If not specified, or the expressed duration is negative, the video is
22816 supposed to be generated forever.
22819 Additionally, all options of the @ref{coreimage} video filter are accepted.
22820 A complete filterchain can be used for further processing of the
22821 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
22822 and examples for details.
22824 @subsection Examples
22829 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22830 given as complete and escaped command-line for Apple's standard bash shell:
22832 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
22834 This example is equivalent to the QRCode example of @ref{coreimage} without the
22835 need for a nullsrc video source.
22840 Generate several gradients.
22844 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22845 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22848 Set frame rate, expressed as number of frames per second. Default
22851 @item c0, c1, c2, c3, c4, c5, c6, c7
22852 Set 8 colors. Default values for colors is to pick random one.
22854 @item x0, y0, y0, y1
22855 Set gradient line source and destination points. If negative or out of range, random ones
22859 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
22862 Set seed for picking gradient line points.
22865 Set the duration of the sourced video. See
22866 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22867 for the accepted syntax.
22869 If not specified, or the expressed duration is negative, the video is
22870 supposed to be generated forever.
22873 Set speed of gradients rotation.
22877 @section mandelbrot
22879 Generate a Mandelbrot set fractal, and progressively zoom towards the
22880 point specified with @var{start_x} and @var{start_y}.
22882 This source accepts the following options:
22887 Set the terminal pts value. Default value is 400.
22890 Set the terminal scale value.
22891 Must be a floating point value. Default value is 0.3.
22894 Set the inner coloring mode, that is the algorithm used to draw the
22895 Mandelbrot fractal internal region.
22897 It shall assume one of the following values:
22902 Show time until convergence.
22904 Set color based on point closest to the origin of the iterations.
22909 Default value is @var{mincol}.
22912 Set the bailout value. Default value is 10.0.
22915 Set the maximum of iterations performed by the rendering
22916 algorithm. Default value is 7189.
22919 Set outer coloring mode.
22920 It shall assume one of following values:
22922 @item iteration_count
22923 Set iteration count mode.
22924 @item normalized_iteration_count
22925 set normalized iteration count mode.
22927 Default value is @var{normalized_iteration_count}.
22930 Set frame rate, expressed as number of frames per second. Default
22934 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22935 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22938 Set the initial scale value. Default value is 3.0.
22941 Set the initial x position. Must be a floating point value between
22942 -100 and 100. Default value is -0.743643887037158704752191506114774.
22945 Set the initial y position. Must be a floating point value between
22946 -100 and 100. Default value is -0.131825904205311970493132056385139.
22951 Generate various test patterns, as generated by the MPlayer test filter.
22953 The size of the generated video is fixed, and is 256x256.
22954 This source is useful in particular for testing encoding features.
22956 This source accepts the following options:
22961 Specify the frame rate of the sourced video, as the number of frames
22962 generated per second. It has to be a string in the format
22963 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22964 number or a valid video frame rate abbreviation. The default value is
22968 Set the duration of the sourced video. See
22969 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22970 for the accepted syntax.
22972 If not specified, or the expressed duration is negative, the video is
22973 supposed to be generated forever.
22977 Set the number or the name of the test to perform. Supported tests are:
22991 @item max_frames, m
22992 Set the maximum number of frames generated for each test, default value is 30.
22996 Default value is "all", which will cycle through the list of all tests.
23001 mptestsrc=t=dc_luma
23004 will generate a "dc_luma" test pattern.
23006 @section frei0r_src
23008 Provide a frei0r source.
23010 To enable compilation of this filter you need to install the frei0r
23011 header and configure FFmpeg with @code{--enable-frei0r}.
23013 This source accepts the following parameters:
23018 The size of the video to generate. For the syntax of this option, check the
23019 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23022 The framerate of the generated video. It may be a string of the form
23023 @var{num}/@var{den} or a frame rate abbreviation.
23026 The name to the frei0r source to load. For more information regarding frei0r and
23027 how to set the parameters, read the @ref{frei0r} section in the video filters
23030 @item filter_params
23031 A '|'-separated list of parameters to pass to the frei0r source.
23035 For example, to generate a frei0r partik0l source with size 200x200
23036 and frame rate 10 which is overlaid on the overlay filter main input:
23038 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23043 Generate a life pattern.
23045 This source is based on a generalization of John Conway's life game.
23047 The sourced input represents a life grid, each pixel represents a cell
23048 which can be in one of two possible states, alive or dead. Every cell
23049 interacts with its eight neighbours, which are the cells that are
23050 horizontally, vertically, or diagonally adjacent.
23052 At each interaction the grid evolves according to the adopted rule,
23053 which specifies the number of neighbor alive cells which will make a
23054 cell stay alive or born. The @option{rule} option allows one to specify
23057 This source accepts the following options:
23061 Set the file from which to read the initial grid state. In the file,
23062 each non-whitespace character is considered an alive cell, and newline
23063 is used to delimit the end of each row.
23065 If this option is not specified, the initial grid is generated
23069 Set the video rate, that is the number of frames generated per second.
23072 @item random_fill_ratio, ratio
23073 Set the random fill ratio for the initial random grid. It is a
23074 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23075 It is ignored when a file is specified.
23077 @item random_seed, seed
23078 Set the seed for filling the initial random grid, must be an integer
23079 included between 0 and UINT32_MAX. If not specified, or if explicitly
23080 set to -1, the filter will try to use a good random seed on a best
23086 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23087 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23088 @var{NS} specifies the number of alive neighbor cells which make a
23089 live cell stay alive, and @var{NB} the number of alive neighbor cells
23090 which make a dead cell to become alive (i.e. to "born").
23091 "s" and "b" can be used in place of "S" and "B", respectively.
23093 Alternatively a rule can be specified by an 18-bits integer. The 9
23094 high order bits are used to encode the next cell state if it is alive
23095 for each number of neighbor alive cells, the low order bits specify
23096 the rule for "borning" new cells. Higher order bits encode for an
23097 higher number of neighbor cells.
23098 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23099 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23101 Default value is "S23/B3", which is the original Conway's game of life
23102 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23103 cells, and will born a new cell if there are three alive cells around
23107 Set the size of the output video. For the syntax of this option, check the
23108 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23110 If @option{filename} is specified, the size is set by default to the
23111 same size of the input file. If @option{size} is set, it must contain
23112 the size specified in the input file, and the initial grid defined in
23113 that file is centered in the larger resulting area.
23115 If a filename is not specified, the size value defaults to "320x240"
23116 (used for a randomly generated initial grid).
23119 If set to 1, stitch the left and right grid edges together, and the
23120 top and bottom edges also. Defaults to 1.
23123 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23124 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23125 value from 0 to 255.
23128 Set the color of living (or new born) cells.
23131 Set the color of dead cells. If @option{mold} is set, this is the first color
23132 used to represent a dead cell.
23135 Set mold color, for definitely dead and moldy cells.
23137 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23138 ffmpeg-utils manual,ffmpeg-utils}.
23141 @subsection Examples
23145 Read a grid from @file{pattern}, and center it on a grid of size
23148 life=f=pattern:s=300x300
23152 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23154 life=ratio=2/3:s=200x200
23158 Specify a custom rule for evolving a randomly generated grid:
23164 Full example with slow death effect (mold) using @command{ffplay}:
23166 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23173 @anchor{haldclutsrc}
23176 @anchor{pal100bars}
23177 @anchor{rgbtestsrc}
23179 @anchor{smptehdbars}
23182 @anchor{yuvtestsrc}
23183 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23185 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23187 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23189 The @code{color} source provides an uniformly colored input.
23191 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23192 @ref{haldclut} filter.
23194 The @code{nullsrc} source returns unprocessed video frames. It is
23195 mainly useful to be employed in analysis / debugging tools, or as the
23196 source for filters which ignore the input data.
23198 The @code{pal75bars} source generates a color bars pattern, based on
23199 EBU PAL recommendations with 75% color levels.
23201 The @code{pal100bars} source generates a color bars pattern, based on
23202 EBU PAL recommendations with 100% color levels.
23204 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23205 detecting RGB vs BGR issues. You should see a red, green and blue
23206 stripe from top to bottom.
23208 The @code{smptebars} source generates a color bars pattern, based on
23209 the SMPTE Engineering Guideline EG 1-1990.
23211 The @code{smptehdbars} source generates a color bars pattern, based on
23212 the SMPTE RP 219-2002.
23214 The @code{testsrc} source generates a test video pattern, showing a
23215 color pattern, a scrolling gradient and a timestamp. This is mainly
23216 intended for testing purposes.
23218 The @code{testsrc2} source is similar to testsrc, but supports more
23219 pixel formats instead of just @code{rgb24}. This allows using it as an
23220 input for other tests without requiring a format conversion.
23222 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23223 see a y, cb and cr stripe from top to bottom.
23225 The sources accept the following parameters:
23230 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23231 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23232 pixels to be used as identity matrix for 3D lookup tables. Each component is
23233 coded on a @code{1/(N*N)} scale.
23236 Specify the color of the source, only available in the @code{color}
23237 source. For the syntax of this option, check the
23238 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23241 Specify the size of the sourced video. For the syntax of this option, check the
23242 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23243 The default value is @code{320x240}.
23245 This option is not available with the @code{allrgb}, @code{allyuv}, and
23246 @code{haldclutsrc} filters.
23249 Specify the frame rate of the sourced video, as the number of frames
23250 generated per second. It has to be a string in the format
23251 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23252 number or a valid video frame rate abbreviation. The default value is
23256 Set the duration of the sourced video. See
23257 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23258 for the accepted syntax.
23260 If not specified, or the expressed duration is negative, the video is
23261 supposed to be generated forever.
23263 Since the frame rate is used as time base, all frames including the last one
23264 will have their full duration. If the specified duration is not a multiple
23265 of the frame duration, it will be rounded up.
23268 Set the sample aspect ratio of the sourced video.
23271 Specify the alpha (opacity) of the background, only available in the
23272 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23273 255 (fully opaque, the default).
23276 Set the number of decimals to show in the timestamp, only available in the
23277 @code{testsrc} source.
23279 The displayed timestamp value will correspond to the original
23280 timestamp value multiplied by the power of 10 of the specified
23281 value. Default value is 0.
23284 @subsection Examples
23288 Generate a video with a duration of 5.3 seconds, with size
23289 176x144 and a frame rate of 10 frames per second:
23291 testsrc=duration=5.3:size=qcif:rate=10
23295 The following graph description will generate a red source
23296 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23299 color=c=red@@0.2:s=qcif:r=10
23303 If the input content is to be ignored, @code{nullsrc} can be used. The
23304 following command generates noise in the luminance plane by employing
23305 the @code{geq} filter:
23307 nullsrc=s=256x256, geq=random(1)*255:128:128
23311 @subsection Commands
23313 The @code{color} source supports the following commands:
23317 Set the color of the created image. Accepts the same syntax of the
23318 corresponding @option{color} option.
23323 Generate video using an OpenCL program.
23328 OpenCL program source file.
23331 Kernel name in program.
23334 Size of frames to generate. This must be set.
23337 Pixel format to use for the generated frames. This must be set.
23340 Number of frames generated every second. Default value is '25'.
23344 For details of how the program loading works, see the @ref{program_opencl}
23351 Generate a colour ramp by setting pixel values from the position of the pixel
23352 in the output image. (Note that this will work with all pixel formats, but
23353 the generated output will not be the same.)
23355 __kernel void ramp(__write_only image2d_t dst,
23356 unsigned int index)
23358 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23361 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23363 write_imagef(dst, loc, val);
23368 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23370 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23371 unsigned int index)
23373 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23375 float4 value = 0.0f;
23376 int x = loc.x + index;
23377 int y = loc.y + index;
23378 while (x > 0 || y > 0) {
23379 if (x % 3 == 1 && y % 3 == 1) {
23387 write_imagef(dst, loc, value);
23393 @section sierpinski
23395 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23397 This source accepts the following options:
23401 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23402 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23405 Set frame rate, expressed as number of frames per second. Default
23409 Set seed which is used for random panning.
23412 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23415 Set fractal type, can be default @code{carpet} or @code{triangle}.
23418 @c man end VIDEO SOURCES
23420 @chapter Video Sinks
23421 @c man begin VIDEO SINKS
23423 Below is a description of the currently available video sinks.
23425 @section buffersink
23427 Buffer video frames, and make them available to the end of the filter
23430 This sink is mainly intended for programmatic use, in particular
23431 through the interface defined in @file{libavfilter/buffersink.h}
23432 or the options system.
23434 It accepts a pointer to an AVBufferSinkContext structure, which
23435 defines the incoming buffers' formats, to be passed as the opaque
23436 parameter to @code{avfilter_init_filter} for initialization.
23440 Null video sink: do absolutely nothing with the input video. It is
23441 mainly useful as a template and for use in analysis / debugging
23444 @c man end VIDEO SINKS
23446 @chapter Multimedia Filters
23447 @c man begin MULTIMEDIA FILTERS
23449 Below is a description of the currently available multimedia filters.
23453 Convert input audio to a video output, displaying the audio bit scope.
23455 The filter accepts the following options:
23459 Set frame rate, expressed as number of frames per second. Default
23463 Specify the video size for the output. For the syntax of this option, check the
23464 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23465 Default value is @code{1024x256}.
23468 Specify list of colors separated by space or by '|' which will be used to
23469 draw channels. Unrecognized or missing colors will be replaced
23473 @section adrawgraph
23474 Draw a graph using input audio metadata.
23476 See @ref{drawgraph}
23478 @section agraphmonitor
23480 See @ref{graphmonitor}.
23482 @section ahistogram
23484 Convert input audio to a video output, displaying the volume histogram.
23486 The filter accepts the following options:
23490 Specify how histogram is calculated.
23492 It accepts the following values:
23495 Use single histogram for all channels.
23497 Use separate histogram for each channel.
23499 Default is @code{single}.
23502 Set frame rate, expressed as number of frames per second. Default
23506 Specify the video size for the output. For the syntax of this option, check the
23507 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23508 Default value is @code{hd720}.
23513 It accepts the following values:
23524 reverse logarithmic
23526 Default is @code{log}.
23529 Set amplitude scale.
23531 It accepts the following values:
23538 Default is @code{log}.
23541 Set how much frames to accumulate in histogram.
23542 Default is 1. Setting this to -1 accumulates all frames.
23545 Set histogram ratio of window height.
23548 Set sonogram sliding.
23550 It accepts the following values:
23553 replace old rows with new ones.
23555 scroll from top to bottom.
23557 Default is @code{replace}.
23560 @section aphasemeter
23562 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23563 representing mean phase of current audio frame. A video output can also be produced and is
23564 enabled by default. The audio is passed through as first output.
23566 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23567 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23568 and @code{1} means channels are in phase.
23570 The filter accepts the following options, all related to its video output:
23574 Set the output frame rate. Default value is @code{25}.
23577 Set the video size for the output. For the syntax of this option, check the
23578 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23579 Default value is @code{800x400}.
23584 Specify the red, green, blue contrast. Default values are @code{2},
23585 @code{7} and @code{1}.
23586 Allowed range is @code{[0, 255]}.
23589 Set color which will be used for drawing median phase. If color is
23590 @code{none} which is default, no median phase value will be drawn.
23593 Enable video output. Default is enabled.
23596 @subsection phasing detection
23598 The filter also detects out of phase and mono sequences in stereo streams.
23599 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23601 The filter accepts the following options for this detection:
23605 Enable mono and out of phase detection. Default is disabled.
23608 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23609 Allowed range is @code{[0, 1]}.
23612 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23613 Allowed range is @code{[90, 180]}.
23616 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23619 @subsection Examples
23623 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23625 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23629 @section avectorscope
23631 Convert input audio to a video output, representing the audio vector
23634 The filter is used to measure the difference between channels of stereo
23635 audio stream. A monaural signal, consisting of identical left and right
23636 signal, results in straight vertical line. Any stereo separation is visible
23637 as a deviation from this line, creating a Lissajous figure.
23638 If the straight (or deviation from it) but horizontal line appears this
23639 indicates that the left and right channels are out of phase.
23641 The filter accepts the following options:
23645 Set the vectorscope mode.
23647 Available values are:
23650 Lissajous rotated by 45 degrees.
23653 Same as above but not rotated.
23656 Shape resembling half of circle.
23659 Default value is @samp{lissajous}.
23662 Set the video size for the output. For the syntax of this option, check the
23663 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23664 Default value is @code{400x400}.
23667 Set the output frame rate. Default value is @code{25}.
23673 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23674 @code{160}, @code{80} and @code{255}.
23675 Allowed range is @code{[0, 255]}.
23681 Specify the red, green, blue and alpha fade. Default values are @code{15},
23682 @code{10}, @code{5} and @code{5}.
23683 Allowed range is @code{[0, 255]}.
23686 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23687 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23690 Set the vectorscope drawing mode.
23692 Available values are:
23695 Draw dot for each sample.
23698 Draw line between previous and current sample.
23701 Default value is @samp{dot}.
23704 Specify amplitude scale of audio samples.
23706 Available values are:
23722 Swap left channel axis with right channel axis.
23732 Mirror only x axis.
23735 Mirror only y axis.
23743 @subsection Examples
23747 Complete example using @command{ffplay}:
23749 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23750 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
23754 @section bench, abench
23756 Benchmark part of a filtergraph.
23758 The filter accepts the following options:
23762 Start or stop a timer.
23764 Available values are:
23767 Get the current time, set it as frame metadata (using the key
23768 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
23771 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
23772 the input frame metadata to get the time difference. Time difference, average,
23773 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
23774 @code{min}) are then printed. The timestamps are expressed in seconds.
23778 @subsection Examples
23782 Benchmark @ref{selectivecolor} filter:
23784 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
23790 Concatenate audio and video streams, joining them together one after the
23793 The filter works on segments of synchronized video and audio streams. All
23794 segments must have the same number of streams of each type, and that will
23795 also be the number of streams at output.
23797 The filter accepts the following options:
23802 Set the number of segments. Default is 2.
23805 Set the number of output video streams, that is also the number of video
23806 streams in each segment. Default is 1.
23809 Set the number of output audio streams, that is also the number of audio
23810 streams in each segment. Default is 0.
23813 Activate unsafe mode: do not fail if segments have a different format.
23817 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
23818 @var{a} audio outputs.
23820 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
23821 segment, in the same order as the outputs, then the inputs for the second
23824 Related streams do not always have exactly the same duration, for various
23825 reasons including codec frame size or sloppy authoring. For that reason,
23826 related synchronized streams (e.g. a video and its audio track) should be
23827 concatenated at once. The concat filter will use the duration of the longest
23828 stream in each segment (except the last one), and if necessary pad shorter
23829 audio streams with silence.
23831 For this filter to work correctly, all segments must start at timestamp 0.
23833 All corresponding streams must have the same parameters in all segments; the
23834 filtering system will automatically select a common pixel format for video
23835 streams, and a common sample format, sample rate and channel layout for
23836 audio streams, but other settings, such as resolution, must be converted
23837 explicitly by the user.
23839 Different frame rates are acceptable but will result in variable frame rate
23840 at output; be sure to configure the output file to handle it.
23842 @subsection Examples
23846 Concatenate an opening, an episode and an ending, all in bilingual version
23847 (video in stream 0, audio in streams 1 and 2):
23849 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23850 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23851 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23852 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23856 Concatenate two parts, handling audio and video separately, using the
23857 (a)movie sources, and adjusting the resolution:
23859 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23860 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23861 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23863 Note that a desync will happen at the stitch if the audio and video streams
23864 do not have exactly the same duration in the first file.
23868 @subsection Commands
23870 This filter supports the following commands:
23873 Close the current segment and step to the next one
23879 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
23880 level. By default, it logs a message at a frequency of 10Hz with the
23881 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
23882 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
23884 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
23885 sample format is double-precision floating point. The input stream will be converted to
23886 this specification, if needed. Users may need to insert aformat and/or aresample filters
23887 after this filter to obtain the original parameters.
23889 The filter also has a video output (see the @var{video} option) with a real
23890 time graph to observe the loudness evolution. The graphic contains the logged
23891 message mentioned above, so it is not printed anymore when this option is set,
23892 unless the verbose logging is set. The main graphing area contains the
23893 short-term loudness (3 seconds of analysis), and the gauge on the right is for
23894 the momentary loudness (400 milliseconds), but can optionally be configured
23895 to instead display short-term loudness (see @var{gauge}).
23897 The green area marks a +/- 1LU target range around the target loudness
23898 (-23LUFS by default, unless modified through @var{target}).
23900 More information about the Loudness Recommendation EBU R128 on
23901 @url{http://tech.ebu.ch/loudness}.
23903 The filter accepts the following options:
23908 Activate the video output. The audio stream is passed unchanged whether this
23909 option is set or no. The video stream will be the first output stream if
23910 activated. Default is @code{0}.
23913 Set the video size. This option is for video only. For the syntax of this
23915 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23916 Default and minimum resolution is @code{640x480}.
23919 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
23920 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
23921 other integer value between this range is allowed.
23924 Set metadata injection. If set to @code{1}, the audio input will be segmented
23925 into 100ms output frames, each of them containing various loudness information
23926 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
23928 Default is @code{0}.
23931 Force the frame logging level.
23933 Available values are:
23936 information logging level
23938 verbose logging level
23941 By default, the logging level is set to @var{info}. If the @option{video} or
23942 the @option{metadata} options are set, it switches to @var{verbose}.
23947 Available modes can be cumulated (the option is a @code{flag} type). Possible
23951 Disable any peak mode (default).
23953 Enable sample-peak mode.
23955 Simple peak mode looking for the higher sample value. It logs a message
23956 for sample-peak (identified by @code{SPK}).
23958 Enable true-peak mode.
23960 If enabled, the peak lookup is done on an over-sampled version of the input
23961 stream for better peak accuracy. It logs a message for true-peak.
23962 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
23963 This mode requires a build with @code{libswresample}.
23967 Treat mono input files as "dual mono". If a mono file is intended for playback
23968 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
23969 If set to @code{true}, this option will compensate for this effect.
23970 Multi-channel input files are not affected by this option.
23973 Set a specific pan law to be used for the measurement of dual mono files.
23974 This parameter is optional, and has a default value of -3.01dB.
23977 Set a specific target level (in LUFS) used as relative zero in the visualization.
23978 This parameter is optional and has a default value of -23LUFS as specified
23979 by EBU R128. However, material published online may prefer a level of -16LUFS
23980 (e.g. for use with podcasts or video platforms).
23983 Set the value displayed by the gauge. Valid values are @code{momentary} and s
23984 @code{shortterm}. By default the momentary value will be used, but in certain
23985 scenarios it may be more useful to observe the short term value instead (e.g.
23989 Sets the display scale for the loudness. Valid parameters are @code{absolute}
23990 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
23991 video output, not the summary or continuous log output.
23994 @subsection Examples
23998 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24000 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24004 Run an analysis with @command{ffmpeg}:
24006 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24010 @section interleave, ainterleave
24012 Temporally interleave frames from several inputs.
24014 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24016 These filters read frames from several inputs and send the oldest
24017 queued frame to the output.
24019 Input streams must have well defined, monotonically increasing frame
24022 In order to submit one frame to output, these filters need to enqueue
24023 at least one frame for each input, so they cannot work in case one
24024 input is not yet terminated and will not receive incoming frames.
24026 For example consider the case when one input is a @code{select} filter
24027 which always drops input frames. The @code{interleave} filter will keep
24028 reading from that input, but it will never be able to send new frames
24029 to output until the input sends an end-of-stream signal.
24031 Also, depending on inputs synchronization, the filters will drop
24032 frames in case one input receives more frames than the other ones, and
24033 the queue is already filled.
24035 These filters accept the following options:
24039 Set the number of different inputs, it is 2 by default.
24042 How to determine the end-of-stream.
24046 The duration of the longest input. (default)
24049 The duration of the shortest input.
24052 The duration of the first input.
24057 @subsection Examples
24061 Interleave frames belonging to different streams using @command{ffmpeg}:
24063 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24067 Add flickering blur effect:
24069 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24073 @section metadata, ametadata
24075 Manipulate frame metadata.
24077 This filter accepts the following options:
24081 Set mode of operation of the filter.
24083 Can be one of the following:
24087 If both @code{value} and @code{key} is set, select frames
24088 which have such metadata. If only @code{key} is set, select
24089 every frame that has such key in metadata.
24092 Add new metadata @code{key} and @code{value}. If key is already available
24096 Modify value of already present key.
24099 If @code{value} is set, delete only keys that have such value.
24100 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24104 Print key and its value if metadata was found. If @code{key} is not set print all
24105 metadata values available in frame.
24109 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24112 Set metadata value which will be used. This option is mandatory for
24113 @code{modify} and @code{add} mode.
24116 Which function to use when comparing metadata value and @code{value}.
24118 Can be one of following:
24122 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24125 Values are interpreted as strings, returns true if metadata value starts with
24126 the @code{value} option string.
24129 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24132 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24135 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24138 Values are interpreted as floats, returns true if expression from option @code{expr}
24142 Values are interpreted as strings, returns true if metadata value ends with
24143 the @code{value} option string.
24147 Set expression which is used when @code{function} is set to @code{expr}.
24148 The expression is evaluated through the eval API and can contain the following
24153 Float representation of @code{value} from metadata key.
24156 Float representation of @code{value} as supplied by user in @code{value} option.
24160 If specified in @code{print} mode, output is written to the named file. Instead of
24161 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24162 for standard output. If @code{file} option is not set, output is written to the log
24163 with AV_LOG_INFO loglevel.
24166 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24170 @subsection Examples
24174 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24177 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24180 Print silencedetect output to file @file{metadata.txt}.
24182 silencedetect,ametadata=mode=print:file=metadata.txt
24185 Direct all metadata to a pipe with file descriptor 4.
24187 metadata=mode=print:file='pipe\:4'
24191 @section perms, aperms
24193 Set read/write permissions for the output frames.
24195 These filters are mainly aimed at developers to test direct path in the
24196 following filter in the filtergraph.
24198 The filters accept the following options:
24202 Select the permissions mode.
24204 It accepts the following values:
24207 Do nothing. This is the default.
24209 Set all the output frames read-only.
24211 Set all the output frames directly writable.
24213 Make the frame read-only if writable, and writable if read-only.
24215 Set each output frame read-only or writable randomly.
24219 Set the seed for the @var{random} mode, must be an integer included between
24220 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24221 @code{-1}, the filter will try to use a good random seed on a best effort
24225 Note: in case of auto-inserted filter between the permission filter and the
24226 following one, the permission might not be received as expected in that
24227 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24228 perms/aperms filter can avoid this problem.
24230 @section realtime, arealtime
24232 Slow down filtering to match real time approximately.
24234 These filters will pause the filtering for a variable amount of time to
24235 match the output rate with the input timestamps.
24236 They are similar to the @option{re} option to @code{ffmpeg}.
24238 They accept the following options:
24242 Time limit for the pauses. Any pause longer than that will be considered
24243 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24245 Speed factor for processing. The value must be a float larger than zero.
24246 Values larger than 1.0 will result in faster than realtime processing,
24247 smaller will slow processing down. The @var{limit} is automatically adapted
24248 accordingly. Default is 1.0.
24250 A processing speed faster than what is possible without these filters cannot
24255 @section select, aselect
24257 Select frames to pass in output.
24259 This filter accepts the following options:
24264 Set expression, which is evaluated for each input frame.
24266 If the expression is evaluated to zero, the frame is discarded.
24268 If the evaluation result is negative or NaN, the frame is sent to the
24269 first output; otherwise it is sent to the output with index
24270 @code{ceil(val)-1}, assuming that the input index starts from 0.
24272 For example a value of @code{1.2} corresponds to the output with index
24273 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24276 Set the number of outputs. The output to which to send the selected
24277 frame is based on the result of the evaluation. Default value is 1.
24280 The expression can contain the following constants:
24284 The (sequential) number of the filtered frame, starting from 0.
24287 The (sequential) number of the selected frame, starting from 0.
24289 @item prev_selected_n
24290 The sequential number of the last selected frame. It's NAN if undefined.
24293 The timebase of the input timestamps.
24296 The PTS (Presentation TimeStamp) of the filtered video frame,
24297 expressed in @var{TB} units. It's NAN if undefined.
24300 The PTS of the filtered video frame,
24301 expressed in seconds. It's NAN if undefined.
24304 The PTS of the previously filtered video frame. It's NAN if undefined.
24306 @item prev_selected_pts
24307 The PTS of the last previously filtered video frame. It's NAN if undefined.
24309 @item prev_selected_t
24310 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24313 The PTS of the first video frame in the video. It's NAN if undefined.
24316 The time of the first video frame in the video. It's NAN if undefined.
24318 @item pict_type @emph{(video only)}
24319 The type of the filtered frame. It can assume one of the following
24331 @item interlace_type @emph{(video only)}
24332 The frame interlace type. It can assume one of the following values:
24335 The frame is progressive (not interlaced).
24337 The frame is top-field-first.
24339 The frame is bottom-field-first.
24342 @item consumed_sample_n @emph{(audio only)}
24343 the number of selected samples before the current frame
24345 @item samples_n @emph{(audio only)}
24346 the number of samples in the current frame
24348 @item sample_rate @emph{(audio only)}
24349 the input sample rate
24352 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24355 the position in the file of the filtered frame, -1 if the information
24356 is not available (e.g. for synthetic video)
24358 @item scene @emph{(video only)}
24359 value between 0 and 1 to indicate a new scene; a low value reflects a low
24360 probability for the current frame to introduce a new scene, while a higher
24361 value means the current frame is more likely to be one (see the example below)
24363 @item concatdec_select
24364 The concat demuxer can select only part of a concat input file by setting an
24365 inpoint and an outpoint, but the output packets may not be entirely contained
24366 in the selected interval. By using this variable, it is possible to skip frames
24367 generated by the concat demuxer which are not exactly contained in the selected
24370 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24371 and the @var{lavf.concat.duration} packet metadata values which are also
24372 present in the decoded frames.
24374 The @var{concatdec_select} variable is -1 if the frame pts is at least
24375 start_time and either the duration metadata is missing or the frame pts is less
24376 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24379 That basically means that an input frame is selected if its pts is within the
24380 interval set by the concat demuxer.
24384 The default value of the select expression is "1".
24386 @subsection Examples
24390 Select all frames in input:
24395 The example above is the same as:
24407 Select only I-frames:
24409 select='eq(pict_type\,I)'
24413 Select one frame every 100:
24415 select='not(mod(n\,100))'
24419 Select only frames contained in the 10-20 time interval:
24421 select=between(t\,10\,20)
24425 Select only I-frames contained in the 10-20 time interval:
24427 select=between(t\,10\,20)*eq(pict_type\,I)
24431 Select frames with a minimum distance of 10 seconds:
24433 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24437 Use aselect to select only audio frames with samples number > 100:
24439 aselect='gt(samples_n\,100)'
24443 Create a mosaic of the first scenes:
24445 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24448 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24452 Send even and odd frames to separate outputs, and compose them:
24454 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24458 Select useful frames from an ffconcat file which is using inpoints and
24459 outpoints but where the source files are not intra frame only.
24461 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24465 @section sendcmd, asendcmd
24467 Send commands to filters in the filtergraph.
24469 These filters read commands to be sent to other filters in the
24472 @code{sendcmd} must be inserted between two video filters,
24473 @code{asendcmd} must be inserted between two audio filters, but apart
24474 from that they act the same way.
24476 The specification of commands can be provided in the filter arguments
24477 with the @var{commands} option, or in a file specified by the
24478 @var{filename} option.
24480 These filters accept the following options:
24483 Set the commands to be read and sent to the other filters.
24485 Set the filename of the commands to be read and sent to the other
24489 @subsection Commands syntax
24491 A commands description consists of a sequence of interval
24492 specifications, comprising a list of commands to be executed when a
24493 particular event related to that interval occurs. The occurring event
24494 is typically the current frame time entering or leaving a given time
24497 An interval is specified by the following syntax:
24499 @var{START}[-@var{END}] @var{COMMANDS};
24502 The time interval is specified by the @var{START} and @var{END} times.
24503 @var{END} is optional and defaults to the maximum time.
24505 The current frame time is considered within the specified interval if
24506 it is included in the interval [@var{START}, @var{END}), that is when
24507 the time is greater or equal to @var{START} and is lesser than
24510 @var{COMMANDS} consists of a sequence of one or more command
24511 specifications, separated by ",", relating to that interval. The
24512 syntax of a command specification is given by:
24514 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24517 @var{FLAGS} is optional and specifies the type of events relating to
24518 the time interval which enable sending the specified command, and must
24519 be a non-null sequence of identifier flags separated by "+" or "|" and
24520 enclosed between "[" and "]".
24522 The following flags are recognized:
24525 The command is sent when the current frame timestamp enters the
24526 specified interval. In other words, the command is sent when the
24527 previous frame timestamp was not in the given interval, and the
24531 The command is sent when the current frame timestamp leaves the
24532 specified interval. In other words, the command is sent when the
24533 previous frame timestamp was in the given interval, and the
24537 The command @var{ARG} is interpreted as expression and result of
24538 expression is passed as @var{ARG}.
24540 The expression is evaluated through the eval API and can contain the following
24545 Original position in the file of the frame, or undefined if undefined
24546 for the current frame.
24549 The presentation timestamp in input.
24552 The count of the input frame for video or audio, starting from 0.
24555 The time in seconds of the current frame.
24558 The start time in seconds of the current command interval.
24561 The end time in seconds of the current command interval.
24564 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24569 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24572 @var{TARGET} specifies the target of the command, usually the name of
24573 the filter class or a specific filter instance name.
24575 @var{COMMAND} specifies the name of the command for the target filter.
24577 @var{ARG} is optional and specifies the optional list of argument for
24578 the given @var{COMMAND}.
24580 Between one interval specification and another, whitespaces, or
24581 sequences of characters starting with @code{#} until the end of line,
24582 are ignored and can be used to annotate comments.
24584 A simplified BNF description of the commands specification syntax
24587 @var{COMMAND_FLAG} ::= "enter" | "leave"
24588 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24589 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24590 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24591 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24592 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24595 @subsection Examples
24599 Specify audio tempo change at second 4:
24601 asendcmd=c='4.0 atempo tempo 1.5',atempo
24605 Target a specific filter instance:
24607 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24611 Specify a list of drawtext and hue commands in a file.
24613 # show text in the interval 5-10
24614 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24615 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24617 # desaturate the image in the interval 15-20
24618 15.0-20.0 [enter] hue s 0,
24619 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24621 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24623 # apply an exponential saturation fade-out effect, starting from time 25
24624 25 [enter] hue s exp(25-t)
24627 A filtergraph allowing to read and process the above command list
24628 stored in a file @file{test.cmd}, can be specified with:
24630 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24635 @section setpts, asetpts
24637 Change the PTS (presentation timestamp) of the input frames.
24639 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24641 This filter accepts the following options:
24646 The expression which is evaluated for each frame to construct its timestamp.
24650 The expression is evaluated through the eval API and can contain the following
24654 @item FRAME_RATE, FR
24655 frame rate, only defined for constant frame-rate video
24658 The presentation timestamp in input
24661 The count of the input frame for video or the number of consumed samples,
24662 not including the current frame for audio, starting from 0.
24664 @item NB_CONSUMED_SAMPLES
24665 The number of consumed samples, not including the current frame (only
24668 @item NB_SAMPLES, S
24669 The number of samples in the current frame (only audio)
24671 @item SAMPLE_RATE, SR
24672 The audio sample rate.
24675 The PTS of the first frame.
24678 the time in seconds of the first frame
24681 State whether the current frame is interlaced.
24684 the time in seconds of the current frame
24687 original position in the file of the frame, or undefined if undefined
24688 for the current frame
24691 The previous input PTS.
24694 previous input time in seconds
24697 The previous output PTS.
24700 previous output time in seconds
24703 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
24707 The wallclock (RTC) time at the start of the movie in microseconds.
24710 The timebase of the input timestamps.
24714 @subsection Examples
24718 Start counting PTS from zero
24720 setpts=PTS-STARTPTS
24724 Apply fast motion effect:
24730 Apply slow motion effect:
24736 Set fixed rate of 25 frames per second:
24742 Set fixed rate 25 fps with some jitter:
24744 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
24748 Apply an offset of 10 seconds to the input PTS:
24754 Generate timestamps from a "live source" and rebase onto the current timebase:
24756 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
24760 Generate timestamps by counting samples:
24769 Force color range for the output video frame.
24771 The @code{setrange} filter marks the color range property for the
24772 output frames. It does not change the input frame, but only sets the
24773 corresponding property, which affects how the frame is treated by
24776 The filter accepts the following options:
24781 Available values are:
24785 Keep the same color range property.
24787 @item unspecified, unknown
24788 Set the color range as unspecified.
24790 @item limited, tv, mpeg
24791 Set the color range as limited.
24793 @item full, pc, jpeg
24794 Set the color range as full.
24798 @section settb, asettb
24800 Set the timebase to use for the output frames timestamps.
24801 It is mainly useful for testing timebase configuration.
24803 It accepts the following parameters:
24808 The expression which is evaluated into the output timebase.
24812 The value for @option{tb} is an arithmetic expression representing a
24813 rational. The expression can contain the constants "AVTB" (the default
24814 timebase), "intb" (the input timebase) and "sr" (the sample rate,
24815 audio only). Default value is "intb".
24817 @subsection Examples
24821 Set the timebase to 1/25:
24827 Set the timebase to 1/10:
24833 Set the timebase to 1001/1000:
24839 Set the timebase to 2*intb:
24845 Set the default timebase value:
24852 Convert input audio to a video output representing frequency spectrum
24853 logarithmically using Brown-Puckette constant Q transform algorithm with
24854 direct frequency domain coefficient calculation (but the transform itself
24855 is not really constant Q, instead the Q factor is actually variable/clamped),
24856 with musical tone scale, from E0 to D#10.
24858 The filter accepts the following options:
24862 Specify the video size for the output. It must be even. For the syntax of this option,
24863 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24864 Default value is @code{1920x1080}.
24867 Set the output frame rate. Default value is @code{25}.
24870 Set the bargraph height. It must be even. Default value is @code{-1} which
24871 computes the bargraph height automatically.
24874 Set the axis height. It must be even. Default value is @code{-1} which computes
24875 the axis height automatically.
24878 Set the sonogram height. It must be even. Default value is @code{-1} which
24879 computes the sonogram height automatically.
24882 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
24883 instead. Default value is @code{1}.
24885 @item sono_v, volume
24886 Specify the sonogram volume expression. It can contain variables:
24889 the @var{bar_v} evaluated expression
24890 @item frequency, freq, f
24891 the frequency where it is evaluated
24892 @item timeclamp, tc
24893 the value of @var{timeclamp} option
24897 @item a_weighting(f)
24898 A-weighting of equal loudness
24899 @item b_weighting(f)
24900 B-weighting of equal loudness
24901 @item c_weighting(f)
24902 C-weighting of equal loudness.
24904 Default value is @code{16}.
24906 @item bar_v, volume2
24907 Specify the bargraph volume expression. It can contain variables:
24910 the @var{sono_v} evaluated expression
24911 @item frequency, freq, f
24912 the frequency where it is evaluated
24913 @item timeclamp, tc
24914 the value of @var{timeclamp} option
24918 @item a_weighting(f)
24919 A-weighting of equal loudness
24920 @item b_weighting(f)
24921 B-weighting of equal loudness
24922 @item c_weighting(f)
24923 C-weighting of equal loudness.
24925 Default value is @code{sono_v}.
24927 @item sono_g, gamma
24928 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
24929 higher gamma makes the spectrum having more range. Default value is @code{3}.
24930 Acceptable range is @code{[1, 7]}.
24932 @item bar_g, gamma2
24933 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
24937 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
24938 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
24940 @item timeclamp, tc
24941 Specify the transform timeclamp. At low frequency, there is trade-off between
24942 accuracy in time domain and frequency domain. If timeclamp is lower,
24943 event in time domain is represented more accurately (such as fast bass drum),
24944 otherwise event in frequency domain is represented more accurately
24945 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
24948 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
24949 limits future samples by applying asymmetric windowing in time domain, useful
24950 when low latency is required. Accepted range is @code{[0, 1]}.
24953 Specify the transform base frequency. Default value is @code{20.01523126408007475},
24954 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
24957 Specify the transform end frequency. Default value is @code{20495.59681441799654},
24958 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
24961 This option is deprecated and ignored.
24964 Specify the transform length in time domain. Use this option to control accuracy
24965 trade-off between time domain and frequency domain at every frequency sample.
24966 It can contain variables:
24968 @item frequency, freq, f
24969 the frequency where it is evaluated
24970 @item timeclamp, tc
24971 the value of @var{timeclamp} option.
24973 Default value is @code{384*tc/(384+tc*f)}.
24976 Specify the transform count for every video frame. Default value is @code{6}.
24977 Acceptable range is @code{[1, 30]}.
24980 Specify the transform count for every single pixel. Default value is @code{0},
24981 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
24984 Specify font file for use with freetype to draw the axis. If not specified,
24985 use embedded font. Note that drawing with font file or embedded font is not
24986 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
24990 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
24991 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
24995 Specify font color expression. This is arithmetic expression that should return
24996 integer value 0xRRGGBB. It can contain variables:
24998 @item frequency, freq, f
24999 the frequency where it is evaluated
25000 @item timeclamp, tc
25001 the value of @var{timeclamp} option
25006 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25007 @item r(x), g(x), b(x)
25008 red, green, and blue value of intensity x.
25010 Default value is @code{st(0, (midi(f)-59.5)/12);
25011 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25012 r(1-ld(1)) + b(ld(1))}.
25015 Specify image file to draw the axis. This option override @var{fontfile} and
25016 @var{fontcolor} option.
25019 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25020 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25021 Default value is @code{1}.
25024 Set colorspace. The accepted values are:
25027 Unspecified (default)
25036 BT.470BG or BT.601-6 625
25039 SMPTE-170M or BT.601-6 525
25045 BT.2020 with non-constant luminance
25050 Set spectrogram color scheme. This is list of floating point values with format
25051 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25052 The default is @code{1|0.5|0|0|0.5|1}.
25056 @subsection Examples
25060 Playing audio while showing the spectrum:
25062 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25066 Same as above, but with frame rate 30 fps:
25068 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25072 Playing at 1280x720:
25074 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25078 Disable sonogram display:
25084 A1 and its harmonics: A1, A2, (near)E3, A3:
25086 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),
25087 asplit[a][out1]; [a] showcqt [out0]'
25091 Same as above, but with more accuracy in frequency domain:
25093 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),
25094 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25100 bar_v=10:sono_v=bar_v*a_weighting(f)
25104 Custom gamma, now spectrum is linear to the amplitude.
25110 Custom tlength equation:
25112 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)))'
25116 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25118 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25122 Custom font using fontconfig:
25124 font='Courier New,Monospace,mono|bold'
25128 Custom frequency range with custom axis using image file:
25130 axisfile=myaxis.png:basefreq=40:endfreq=10000
25136 Convert input audio to video output representing the audio power spectrum.
25137 Audio amplitude is on Y-axis while frequency is on X-axis.
25139 The filter accepts the following options:
25143 Specify size of video. For the syntax of this option, check the
25144 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25145 Default is @code{1024x512}.
25149 This set how each frequency bin will be represented.
25151 It accepts the following values:
25157 Default is @code{bar}.
25160 Set amplitude scale.
25162 It accepts the following values:
25176 Default is @code{log}.
25179 Set frequency scale.
25181 It accepts the following values:
25190 Reverse logarithmic scale.
25192 Default is @code{lin}.
25195 Set window size. Allowed range is from 16 to 65536.
25197 Default is @code{2048}
25200 Set windowing function.
25202 It accepts the following values:
25225 Default is @code{hanning}.
25228 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25229 which means optimal overlap for selected window function will be picked.
25232 Set time averaging. Setting this to 0 will display current maximal peaks.
25233 Default is @code{1}, which means time averaging is disabled.
25236 Specify list of colors separated by space or by '|' which will be used to
25237 draw channel frequencies. Unrecognized or missing colors will be replaced
25241 Set channel display mode.
25243 It accepts the following values:
25248 Default is @code{combined}.
25251 Set minimum amplitude used in @code{log} amplitude scaler.
25254 Set data display mode.
25256 It accepts the following values:
25262 Default is @code{magnitude}.
25265 @section showspatial
25267 Convert stereo input audio to a video output, representing the spatial relationship
25268 between two channels.
25270 The filter accepts the following options:
25274 Specify the video size for the output. For the syntax of this option, check the
25275 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25276 Default value is @code{512x512}.
25279 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25282 Set window function.
25284 It accepts the following values:
25309 Default value is @code{hann}.
25312 Set ratio of overlap window. Default value is @code{0.5}.
25313 When value is @code{1} overlap is set to recommended size for specific
25314 window function currently used.
25317 @anchor{showspectrum}
25318 @section showspectrum
25320 Convert input audio to a video output, representing the audio frequency
25323 The filter accepts the following options:
25327 Specify the video size for the output. For the syntax of this option, check the
25328 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25329 Default value is @code{640x512}.
25332 Specify how the spectrum should slide along the window.
25334 It accepts the following values:
25337 the samples start again on the left when they reach the right
25339 the samples scroll from right to left
25341 frames are only produced when the samples reach the right
25343 the samples scroll from left to right
25346 Default value is @code{replace}.
25349 Specify display mode.
25351 It accepts the following values:
25354 all channels are displayed in the same row
25356 all channels are displayed in separate rows
25359 Default value is @samp{combined}.
25362 Specify display color mode.
25364 It accepts the following values:
25367 each channel is displayed in a separate color
25369 each channel is displayed using the same color scheme
25371 each channel is displayed using the rainbow color scheme
25373 each channel is displayed using the moreland color scheme
25375 each channel is displayed using the nebulae color scheme
25377 each channel is displayed using the fire color scheme
25379 each channel is displayed using the fiery color scheme
25381 each channel is displayed using the fruit color scheme
25383 each channel is displayed using the cool color scheme
25385 each channel is displayed using the magma color scheme
25387 each channel is displayed using the green color scheme
25389 each channel is displayed using the viridis color scheme
25391 each channel is displayed using the plasma color scheme
25393 each channel is displayed using the cividis color scheme
25395 each channel is displayed using the terrain color scheme
25398 Default value is @samp{channel}.
25401 Specify scale used for calculating intensity color values.
25403 It accepts the following values:
25408 square root, default
25419 Default value is @samp{sqrt}.
25422 Specify frequency scale.
25424 It accepts the following values:
25432 Default value is @samp{lin}.
25435 Set saturation modifier for displayed colors. Negative values provide
25436 alternative color scheme. @code{0} is no saturation at all.
25437 Saturation must be in [-10.0, 10.0] range.
25438 Default value is @code{1}.
25441 Set window function.
25443 It accepts the following values:
25468 Default value is @code{hann}.
25471 Set orientation of time vs frequency axis. Can be @code{vertical} or
25472 @code{horizontal}. Default is @code{vertical}.
25475 Set ratio of overlap window. Default value is @code{0}.
25476 When value is @code{1} overlap is set to recommended size for specific
25477 window function currently used.
25480 Set scale gain for calculating intensity color values.
25481 Default value is @code{1}.
25484 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25487 Set color rotation, must be in [-1.0, 1.0] range.
25488 Default value is @code{0}.
25491 Set start frequency from which to display spectrogram. Default is @code{0}.
25494 Set stop frequency to which to display spectrogram. Default is @code{0}.
25497 Set upper frame rate limit. Default is @code{auto}, unlimited.
25500 Draw time and frequency axes and legends. Default is disabled.
25503 The usage is very similar to the showwaves filter; see the examples in that
25506 @subsection Examples
25510 Large window with logarithmic color scaling:
25512 showspectrum=s=1280x480:scale=log
25516 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25518 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25519 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25523 @section showspectrumpic
25525 Convert input audio to a single video frame, representing the audio frequency
25528 The filter accepts the following options:
25532 Specify the video size for the output. For the syntax of this option, check the
25533 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25534 Default value is @code{4096x2048}.
25537 Specify display mode.
25539 It accepts the following values:
25542 all channels are displayed in the same row
25544 all channels are displayed in separate rows
25546 Default value is @samp{combined}.
25549 Specify display color mode.
25551 It accepts the following values:
25554 each channel is displayed in a separate color
25556 each channel is displayed using the same color scheme
25558 each channel is displayed using the rainbow color scheme
25560 each channel is displayed using the moreland color scheme
25562 each channel is displayed using the nebulae color scheme
25564 each channel is displayed using the fire color scheme
25566 each channel is displayed using the fiery color scheme
25568 each channel is displayed using the fruit color scheme
25570 each channel is displayed using the cool color scheme
25572 each channel is displayed using the magma color scheme
25574 each channel is displayed using the green color scheme
25576 each channel is displayed using the viridis color scheme
25578 each channel is displayed using the plasma color scheme
25580 each channel is displayed using the cividis color scheme
25582 each channel is displayed using the terrain color scheme
25584 Default value is @samp{intensity}.
25587 Specify scale used for calculating intensity color values.
25589 It accepts the following values:
25594 square root, default
25604 Default value is @samp{log}.
25607 Specify frequency scale.
25609 It accepts the following values:
25617 Default value is @samp{lin}.
25620 Set saturation modifier for displayed colors. Negative values provide
25621 alternative color scheme. @code{0} is no saturation at all.
25622 Saturation must be in [-10.0, 10.0] range.
25623 Default value is @code{1}.
25626 Set window function.
25628 It accepts the following values:
25652 Default value is @code{hann}.
25655 Set orientation of time vs frequency axis. Can be @code{vertical} or
25656 @code{horizontal}. Default is @code{vertical}.
25659 Set scale gain for calculating intensity color values.
25660 Default value is @code{1}.
25663 Draw time and frequency axes and legends. Default is enabled.
25666 Set color rotation, must be in [-1.0, 1.0] range.
25667 Default value is @code{0}.
25670 Set start frequency from which to display spectrogram. Default is @code{0}.
25673 Set stop frequency to which to display spectrogram. Default is @code{0}.
25676 @subsection Examples
25680 Extract an audio spectrogram of a whole audio track
25681 in a 1024x1024 picture using @command{ffmpeg}:
25683 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25687 @section showvolume
25689 Convert input audio volume to a video output.
25691 The filter accepts the following options:
25698 Set border width, allowed range is [0, 5]. Default is 1.
25701 Set channel width, allowed range is [80, 8192]. Default is 400.
25704 Set channel height, allowed range is [1, 900]. Default is 20.
25707 Set fade, allowed range is [0, 1]. Default is 0.95.
25710 Set volume color expression.
25712 The expression can use the following variables:
25716 Current max volume of channel in dB.
25722 Current channel number, starting from 0.
25726 If set, displays channel names. Default is enabled.
25729 If set, displays volume values. Default is enabled.
25732 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
25733 default is @code{h}.
25736 Set step size, allowed range is [0, 5]. Default is 0, which means
25740 Set background opacity, allowed range is [0, 1]. Default is 0.
25743 Set metering mode, can be peak: @code{p} or rms: @code{r},
25744 default is @code{p}.
25747 Set display scale, can be linear: @code{lin} or log: @code{log},
25748 default is @code{lin}.
25752 If set to > 0., display a line for the max level
25753 in the previous seconds.
25754 default is disabled: @code{0.}
25757 The color of the max line. Use when @code{dm} option is set to > 0.
25758 default is: @code{orange}
25763 Convert input audio to a video output, representing the samples waves.
25765 The filter accepts the following options:
25769 Specify the video size for the output. For the syntax of this option, check the
25770 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25771 Default value is @code{600x240}.
25776 Available values are:
25779 Draw a point for each sample.
25782 Draw a vertical line for each sample.
25785 Draw a point for each sample and a line between them.
25788 Draw a centered vertical line for each sample.
25791 Default value is @code{point}.
25794 Set the number of samples which are printed on the same column. A
25795 larger value will decrease the frame rate. Must be a positive
25796 integer. This option can be set only if the value for @var{rate}
25797 is not explicitly specified.
25800 Set the (approximate) output frame rate. This is done by setting the
25801 option @var{n}. Default value is "25".
25803 @item split_channels
25804 Set if channels should be drawn separately or overlap. Default value is 0.
25807 Set colors separated by '|' which are going to be used for drawing of each channel.
25810 Set amplitude scale.
25812 Available values are:
25830 Set the draw mode. This is mostly useful to set for high @var{n}.
25832 Available values are:
25835 Scale pixel values for each drawn sample.
25838 Draw every sample directly.
25841 Default value is @code{scale}.
25844 @subsection Examples
25848 Output the input file audio and the corresponding video representation
25851 amovie=a.mp3,asplit[out0],showwaves[out1]
25855 Create a synthetic signal and show it with showwaves, forcing a
25856 frame rate of 30 frames per second:
25858 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
25862 @section showwavespic
25864 Convert input audio to a single video frame, representing the samples waves.
25866 The filter accepts the following options:
25870 Specify the video size for the output. For the syntax of this option, check the
25871 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25872 Default value is @code{600x240}.
25874 @item split_channels
25875 Set if channels should be drawn separately or overlap. Default value is 0.
25878 Set colors separated by '|' which are going to be used for drawing of each channel.
25881 Set amplitude scale.
25883 Available values are:
25903 Available values are:
25906 Scale pixel values for each drawn sample.
25909 Draw every sample directly.
25912 Default value is @code{scale}.
25915 Set the filter mode.
25917 Available values are:
25920 Use average samples values for each drawn sample.
25923 Use peak samples values for each drawn sample.
25926 Default value is @code{average}.
25929 @subsection Examples
25933 Extract a channel split representation of the wave form of a whole audio track
25934 in a 1024x800 picture using @command{ffmpeg}:
25936 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25940 @section sidedata, asidedata
25942 Delete frame side data, or select frames based on it.
25944 This filter accepts the following options:
25948 Set mode of operation of the filter.
25950 Can be one of the following:
25954 Select every frame with side data of @code{type}.
25957 Delete side data of @code{type}. If @code{type} is not set, delete all side
25963 Set side data type used with all modes. Must be set for @code{select} mode. For
25964 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
25965 in @file{libavutil/frame.h}. For example, to choose
25966 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
25970 @section spectrumsynth
25972 Synthesize audio from 2 input video spectrums, first input stream represents
25973 magnitude across time and second represents phase across time.
25974 The filter will transform from frequency domain as displayed in videos back
25975 to time domain as presented in audio output.
25977 This filter is primarily created for reversing processed @ref{showspectrum}
25978 filter outputs, but can synthesize sound from other spectrograms too.
25979 But in such case results are going to be poor if the phase data is not
25980 available, because in such cases phase data need to be recreated, usually
25981 it's just recreated from random noise.
25982 For best results use gray only output (@code{channel} color mode in
25983 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
25984 @code{lin} scale for phase video. To produce phase, for 2nd video, use
25985 @code{data} option. Inputs videos should generally use @code{fullframe}
25986 slide mode as that saves resources needed for decoding video.
25988 The filter accepts the following options:
25992 Specify sample rate of output audio, the sample rate of audio from which
25993 spectrum was generated may differ.
25996 Set number of channels represented in input video spectrums.
25999 Set scale which was used when generating magnitude input spectrum.
26000 Can be @code{lin} or @code{log}. Default is @code{log}.
26003 Set slide which was used when generating inputs spectrums.
26004 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26005 Default is @code{fullframe}.
26008 Set window function used for resynthesis.
26011 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26012 which means optimal overlap for selected window function will be picked.
26015 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26016 Default is @code{vertical}.
26019 @subsection Examples
26023 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26024 then resynthesize videos back to audio with spectrumsynth:
26026 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
26027 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
26028 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26032 @section split, asplit
26034 Split input into several identical outputs.
26036 @code{asplit} works with audio input, @code{split} with video.
26038 The filter accepts a single parameter which specifies the number of outputs. If
26039 unspecified, it defaults to 2.
26041 @subsection Examples
26045 Create two separate outputs from the same input:
26047 [in] split [out0][out1]
26051 To create 3 or more outputs, you need to specify the number of
26054 [in] asplit=3 [out0][out1][out2]
26058 Create two separate outputs from the same input, one cropped and
26061 [in] split [splitout1][splitout2];
26062 [splitout1] crop=100:100:0:0 [cropout];
26063 [splitout2] pad=200:200:100:100 [padout];
26067 Create 5 copies of the input audio with @command{ffmpeg}:
26069 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26075 Receive commands sent through a libzmq client, and forward them to
26076 filters in the filtergraph.
26078 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26079 must be inserted between two video filters, @code{azmq} between two
26080 audio filters. Both are capable to send messages to any filter type.
26082 To enable these filters you need to install the libzmq library and
26083 headers and configure FFmpeg with @code{--enable-libzmq}.
26085 For more information about libzmq see:
26086 @url{http://www.zeromq.org/}
26088 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26089 receives messages sent through a network interface defined by the
26090 @option{bind_address} (or the abbreviation "@option{b}") option.
26091 Default value of this option is @file{tcp://localhost:5555}. You may
26092 want to alter this value to your needs, but do not forget to escape any
26093 ':' signs (see @ref{filtergraph escaping}).
26095 The received message must be in the form:
26097 @var{TARGET} @var{COMMAND} [@var{ARG}]
26100 @var{TARGET} specifies the target of the command, usually the name of
26101 the filter class or a specific filter instance name. The default
26102 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26103 but you can override this by using the @samp{filter_name@@id} syntax
26104 (see @ref{Filtergraph syntax}).
26106 @var{COMMAND} specifies the name of the command for the target filter.
26108 @var{ARG} is optional and specifies the optional argument list for the
26109 given @var{COMMAND}.
26111 Upon reception, the message is processed and the corresponding command
26112 is injected into the filtergraph. Depending on the result, the filter
26113 will send a reply to the client, adopting the format:
26115 @var{ERROR_CODE} @var{ERROR_REASON}
26119 @var{MESSAGE} is optional.
26121 @subsection Examples
26123 Look at @file{tools/zmqsend} for an example of a zmq client which can
26124 be used to send commands processed by these filters.
26126 Consider the following filtergraph generated by @command{ffplay}.
26127 In this example the last overlay filter has an instance name. All other
26128 filters will have default instance names.
26131 ffplay -dumpgraph 1 -f lavfi "
26132 color=s=100x100:c=red [l];
26133 color=s=100x100:c=blue [r];
26134 nullsrc=s=200x100, zmq [bg];
26135 [bg][l] overlay [bg+l];
26136 [bg+l][r] overlay@@my=x=100 "
26139 To change the color of the left side of the video, the following
26140 command can be used:
26142 echo Parsed_color_0 c yellow | tools/zmqsend
26145 To change the right side:
26147 echo Parsed_color_1 c pink | tools/zmqsend
26150 To change the position of the right side:
26152 echo overlay@@my x 150 | tools/zmqsend
26156 @c man end MULTIMEDIA FILTERS
26158 @chapter Multimedia Sources
26159 @c man begin MULTIMEDIA SOURCES
26161 Below is a description of the currently available multimedia sources.
26165 This is the same as @ref{movie} source, except it selects an audio
26171 Read audio and/or video stream(s) from a movie container.
26173 It accepts the following parameters:
26177 The name of the resource to read (not necessarily a file; it can also be a
26178 device or a stream accessed through some protocol).
26180 @item format_name, f
26181 Specifies the format assumed for the movie to read, and can be either
26182 the name of a container or an input device. If not specified, the
26183 format is guessed from @var{movie_name} or by probing.
26185 @item seek_point, sp
26186 Specifies the seek point in seconds. The frames will be output
26187 starting from this seek point. The parameter is evaluated with
26188 @code{av_strtod}, so the numerical value may be suffixed by an IS
26189 postfix. The default value is "0".
26192 Specifies the streams to read. Several streams can be specified,
26193 separated by "+". The source will then have as many outputs, in the
26194 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26195 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26196 respectively the default (best suited) video and audio stream. Default
26197 is "dv", or "da" if the filter is called as "amovie".
26199 @item stream_index, si
26200 Specifies the index of the video stream to read. If the value is -1,
26201 the most suitable video stream will be automatically selected. The default
26202 value is "-1". Deprecated. If the filter is called "amovie", it will select
26203 audio instead of video.
26206 Specifies how many times to read the stream in sequence.
26207 If the value is 0, the stream will be looped infinitely.
26208 Default value is "1".
26210 Note that when the movie is looped the source timestamps are not
26211 changed, so it will generate non monotonically increasing timestamps.
26213 @item discontinuity
26214 Specifies the time difference between frames above which the point is
26215 considered a timestamp discontinuity which is removed by adjusting the later
26219 It allows overlaying a second video on top of the main input of
26220 a filtergraph, as shown in this graph:
26222 input -----------> deltapts0 --> overlay --> output
26225 movie --> scale--> deltapts1 -------+
26227 @subsection Examples
26231 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26232 on top of the input labelled "in":
26234 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26235 [in] setpts=PTS-STARTPTS [main];
26236 [main][over] overlay=16:16 [out]
26240 Read from a video4linux2 device, and overlay it on top of the input
26243 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26244 [in] setpts=PTS-STARTPTS [main];
26245 [main][over] overlay=16:16 [out]
26249 Read the first video stream and the audio stream with id 0x81 from
26250 dvd.vob; the video is connected to the pad named "video" and the audio is
26251 connected to the pad named "audio":
26253 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26257 @subsection Commands
26259 Both movie and amovie support the following commands:
26262 Perform seek using "av_seek_frame".
26263 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26266 @var{stream_index}: If stream_index is -1, a default
26267 stream is selected, and @var{timestamp} is automatically converted
26268 from AV_TIME_BASE units to the stream specific time_base.
26270 @var{timestamp}: Timestamp in AVStream.time_base units
26271 or, if no stream is specified, in AV_TIME_BASE units.
26273 @var{flags}: Flags which select direction and seeking mode.
26277 Get movie duration in AV_TIME_BASE units.
26281 @c man end MULTIMEDIA SOURCES