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.
5504 Mid/Side to Right/Left.
5507 Left/Right to Left - Right.
5511 Set level of side signal. Default is 1.
5512 Allowed range is from 0.015625 to 64.
5515 Set balance of side signal. Default is 0.
5516 Allowed range is from -1 to 1.
5519 Set level of the middle signal. Default is 1.
5520 Allowed range is from 0.015625 to 64.
5523 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5526 Set stereo base between mono and inversed channels. Default is 0.
5527 Allowed range is from -1 to 1.
5530 Set delay in milliseconds how much to delay left from right channel and
5531 vice versa. Default is 0. Allowed range is from -20 to 20.
5534 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5537 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5539 @item bmode_in, bmode_out
5540 Set balance mode for balance_in/balance_out option.
5542 Can be one of the following:
5546 Classic balance mode. Attenuate one channel at time.
5547 Gain is raised up to 1.
5550 Similar as classic mode above but gain is raised up to 2.
5553 Equal power distribution, from -6dB to +6dB range.
5557 @subsection Commands
5559 This filter supports the all above options as @ref{commands}.
5561 @subsection Examples
5565 Apply karaoke like effect:
5567 stereotools=mlev=0.015625
5571 Convert M/S signal to L/R:
5573 "stereotools=mode=ms>lr"
5577 @section stereowiden
5579 This filter enhance the stereo effect by suppressing signal common to both
5580 channels and by delaying the signal of left into right and vice versa,
5581 thereby widening the stereo effect.
5583 The filter accepts the following options:
5587 Time in milliseconds of the delay of left signal into right and vice versa.
5588 Default is 20 milliseconds.
5591 Amount of gain in delayed signal into right and vice versa. Gives a delay
5592 effect of left signal in right output and vice versa which gives widening
5593 effect. Default is 0.3.
5596 Cross feed of left into right with inverted phase. This helps in suppressing
5597 the mono. If the value is 1 it will cancel all the signal common to both
5598 channels. Default is 0.3.
5601 Set level of input signal of original channel. Default is 0.8.
5604 @subsection Commands
5606 This filter supports the all above options except @code{delay} as @ref{commands}.
5608 @section superequalizer
5609 Apply 18 band equalizer.
5611 The filter accepts the following options:
5618 Set 131Hz band gain.
5620 Set 185Hz band gain.
5622 Set 262Hz band gain.
5624 Set 370Hz band gain.
5626 Set 523Hz band gain.
5628 Set 740Hz band gain.
5630 Set 1047Hz band gain.
5632 Set 1480Hz band gain.
5634 Set 2093Hz band gain.
5636 Set 2960Hz band gain.
5638 Set 4186Hz band gain.
5640 Set 5920Hz band gain.
5642 Set 8372Hz band gain.
5644 Set 11840Hz band gain.
5646 Set 16744Hz band gain.
5648 Set 20000Hz band gain.
5652 Apply audio surround upmix filter.
5654 This filter allows to produce multichannel output from audio stream.
5656 The filter accepts the following options:
5660 Set output channel layout. By default, this is @var{5.1}.
5662 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5663 for the required syntax.
5666 Set input channel layout. By default, this is @var{stereo}.
5668 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5669 for the required syntax.
5672 Set input volume level. By default, this is @var{1}.
5675 Set output volume level. By default, this is @var{1}.
5678 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5681 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5684 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5687 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5688 In @var{add} mode, LFE channel is created from input audio and added to output.
5689 In @var{sub} mode, LFE channel is created from input audio and added to output but
5690 also all non-LFE output channels are subtracted with output LFE channel.
5693 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5694 Default is @var{90}.
5697 Set front center input volume. By default, this is @var{1}.
5700 Set front center output volume. By default, this is @var{1}.
5703 Set front left input volume. By default, this is @var{1}.
5706 Set front left output volume. By default, this is @var{1}.
5709 Set front right input volume. By default, this is @var{1}.
5712 Set front right output volume. By default, this is @var{1}.
5715 Set side left input volume. By default, this is @var{1}.
5718 Set side left output volume. By default, this is @var{1}.
5721 Set side right input volume. By default, this is @var{1}.
5724 Set side right output volume. By default, this is @var{1}.
5727 Set back left input volume. By default, this is @var{1}.
5730 Set back left output volume. By default, this is @var{1}.
5733 Set back right input volume. By default, this is @var{1}.
5736 Set back right output volume. By default, this is @var{1}.
5739 Set back center input volume. By default, this is @var{1}.
5742 Set back center output volume. By default, this is @var{1}.
5745 Set LFE input volume. By default, this is @var{1}.
5748 Set LFE output volume. By default, this is @var{1}.
5751 Set spread usage of stereo image across X axis for all channels.
5754 Set spread usage of stereo image across Y axis for all channels.
5756 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5757 Set spread usage of stereo image across X axis for each channel.
5759 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5760 Set spread usage of stereo image across Y axis for each channel.
5763 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5766 Set window function.
5768 It accepts the following values:
5791 Default is @code{hann}.
5794 Set window overlap. If set to 1, the recommended overlap for selected
5795 window function will be picked. Default is @code{0.5}.
5798 @section treble, highshelf
5800 Boost or cut treble (upper) frequencies of the audio using a two-pole
5801 shelving filter with a response similar to that of a standard
5802 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5804 The filter accepts the following options:
5808 Give the gain at whichever is the lower of ~22 kHz and the
5809 Nyquist frequency. Its useful range is about -20 (for a large cut)
5810 to +20 (for a large boost). Beware of clipping when using a positive gain.
5813 Set the filter's central frequency and so can be used
5814 to extend or reduce the frequency range to be boosted or cut.
5815 The default value is @code{3000} Hz.
5818 Set method to specify band-width of filter.
5833 Determine how steep is the filter's shelf transition.
5836 How much to use filtered signal in output. Default is 1.
5837 Range is between 0 and 1.
5840 Specify which channels to filter, by default all available are filtered.
5843 Normalize biquad coefficients, by default is disabled.
5844 Enabling it will normalize magnitude response at DC to 0dB.
5847 Set transform type of IIR filter.
5856 @subsection Commands
5858 This filter supports the following commands:
5861 Change treble frequency.
5862 Syntax for the command is : "@var{frequency}"
5865 Change treble width_type.
5866 Syntax for the command is : "@var{width_type}"
5869 Change treble width.
5870 Syntax for the command is : "@var{width}"
5874 Syntax for the command is : "@var{gain}"
5878 Syntax for the command is : "@var{mix}"
5883 Sinusoidal amplitude modulation.
5885 The filter accepts the following options:
5889 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5890 (20 Hz or lower) will result in a tremolo effect.
5891 This filter may also be used as a ring modulator by specifying
5892 a modulation frequency higher than 20 Hz.
5893 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5896 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5897 Default value is 0.5.
5902 Sinusoidal phase modulation.
5904 The filter accepts the following options:
5908 Modulation frequency in Hertz.
5909 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5912 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5913 Default value is 0.5.
5918 Adjust the input audio volume.
5920 It accepts the following parameters:
5924 Set audio volume expression.
5926 Output values are clipped to the maximum value.
5928 The output audio volume is given by the relation:
5930 @var{output_volume} = @var{volume} * @var{input_volume}
5933 The default value for @var{volume} is "1.0".
5936 This parameter represents the mathematical precision.
5938 It determines which input sample formats will be allowed, which affects the
5939 precision of the volume scaling.
5943 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5945 32-bit floating-point; this limits input sample format to FLT. (default)
5947 64-bit floating-point; this limits input sample format to DBL.
5951 Choose the behaviour on encountering ReplayGain side data in input frames.
5955 Remove ReplayGain side data, ignoring its contents (the default).
5958 Ignore ReplayGain side data, but leave it in the frame.
5961 Prefer the track gain, if present.
5964 Prefer the album gain, if present.
5967 @item replaygain_preamp
5968 Pre-amplification gain in dB to apply to the selected replaygain gain.
5970 Default value for @var{replaygain_preamp} is 0.0.
5972 @item replaygain_noclip
5973 Prevent clipping by limiting the gain applied.
5975 Default value for @var{replaygain_noclip} is 1.
5978 Set when the volume expression is evaluated.
5980 It accepts the following values:
5983 only evaluate expression once during the filter initialization, or
5984 when the @samp{volume} command is sent
5987 evaluate expression for each incoming frame
5990 Default value is @samp{once}.
5993 The volume expression can contain the following parameters.
5997 frame number (starting at zero)
6000 @item nb_consumed_samples
6001 number of samples consumed by the filter
6003 number of samples in the current frame
6005 original frame position in the file
6011 PTS at start of stream
6013 time at start of stream
6019 last set volume value
6022 Note that when @option{eval} is set to @samp{once} only the
6023 @var{sample_rate} and @var{tb} variables are available, all other
6024 variables will evaluate to NAN.
6026 @subsection Commands
6028 This filter supports the following commands:
6031 Modify the volume expression.
6032 The command accepts the same syntax of the corresponding option.
6034 If the specified expression is not valid, it is kept at its current
6038 @subsection Examples
6042 Halve the input audio volume:
6046 volume=volume=-6.0206dB
6049 In all the above example the named key for @option{volume} can be
6050 omitted, for example like in:
6056 Increase input audio power by 6 decibels using fixed-point precision:
6058 volume=volume=6dB:precision=fixed
6062 Fade volume after time 10 with an annihilation period of 5 seconds:
6064 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6068 @section volumedetect
6070 Detect the volume of the input video.
6072 The filter has no parameters. The input is not modified. Statistics about
6073 the volume will be printed in the log when the input stream end is reached.
6075 In particular it will show the mean volume (root mean square), maximum
6076 volume (on a per-sample basis), and the beginning of a histogram of the
6077 registered volume values (from the maximum value to a cumulated 1/1000 of
6080 All volumes are in decibels relative to the maximum PCM value.
6082 @subsection Examples
6084 Here is an excerpt of the output:
6086 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6087 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6088 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6089 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6090 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6091 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6092 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6093 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6094 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6100 The mean square energy is approximately -27 dB, or 10^-2.7.
6102 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6104 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6107 In other words, raising the volume by +4 dB does not cause any clipping,
6108 raising it by +5 dB causes clipping for 6 samples, etc.
6110 @c man end AUDIO FILTERS
6112 @chapter Audio Sources
6113 @c man begin AUDIO SOURCES
6115 Below is a description of the currently available audio sources.
6119 Buffer audio frames, and make them available to the filter chain.
6121 This source is mainly intended for a programmatic use, in particular
6122 through the interface defined in @file{libavfilter/buffersrc.h}.
6124 It accepts the following parameters:
6128 The timebase which will be used for timestamps of submitted frames. It must be
6129 either a floating-point number or in @var{numerator}/@var{denominator} form.
6132 The sample rate of the incoming audio buffers.
6135 The sample format of the incoming audio buffers.
6136 Either a sample format name or its corresponding integer representation from
6137 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6139 @item channel_layout
6140 The channel layout of the incoming audio buffers.
6141 Either a channel layout name from channel_layout_map in
6142 @file{libavutil/channel_layout.c} or its corresponding integer representation
6143 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6146 The number of channels of the incoming audio buffers.
6147 If both @var{channels} and @var{channel_layout} are specified, then they
6152 @subsection Examples
6155 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6158 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6159 Since the sample format with name "s16p" corresponds to the number
6160 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6163 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6168 Generate an audio signal specified by an expression.
6170 This source accepts in input one or more expressions (one for each
6171 channel), which are evaluated and used to generate a corresponding
6174 This source accepts the following options:
6178 Set the '|'-separated expressions list for each separate channel. In case the
6179 @option{channel_layout} option is not specified, the selected channel layout
6180 depends on the number of provided expressions. Otherwise the last
6181 specified expression is applied to the remaining output channels.
6183 @item channel_layout, c
6184 Set the channel layout. The number of channels in the specified layout
6185 must be equal to the number of specified expressions.
6188 Set the minimum duration of the sourced audio. See
6189 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6190 for the accepted syntax.
6191 Note that the resulting duration may be greater than the specified
6192 duration, as the generated audio is always cut at the end of a
6195 If not specified, or the expressed duration is negative, the audio is
6196 supposed to be generated forever.
6199 Set the number of samples per channel per each output frame,
6202 @item sample_rate, s
6203 Specify the sample rate, default to 44100.
6206 Each expression in @var{exprs} can contain the following constants:
6210 number of the evaluated sample, starting from 0
6213 time of the evaluated sample expressed in seconds, starting from 0
6220 @subsection Examples
6230 Generate a sin signal with frequency of 440 Hz, set sample rate to
6233 aevalsrc="sin(440*2*PI*t):s=8000"
6237 Generate a two channels signal, specify the channel layout (Front
6238 Center + Back Center) explicitly:
6240 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6244 Generate white noise:
6246 aevalsrc="-2+random(0)"
6250 Generate an amplitude modulated signal:
6252 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6256 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6258 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6265 Generate a FIR coefficients using frequency sampling method.
6267 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6269 The filter accepts the following options:
6273 Set number of filter coefficents in output audio stream.
6274 Default value is 1025.
6277 Set frequency points from where magnitude and phase are set.
6278 This must be in non decreasing order, and first element must be 0, while last element
6279 must be 1. Elements are separated by white spaces.
6282 Set magnitude value for every frequency point set by @option{frequency}.
6283 Number of values must be same as number of frequency points.
6284 Values are separated by white spaces.
6287 Set phase value for every frequency point set by @option{frequency}.
6288 Number of values must be same as number of frequency points.
6289 Values are separated by white spaces.
6291 @item sample_rate, r
6292 Set sample rate, default is 44100.
6295 Set number of samples per each frame. Default is 1024.
6298 Set window function. Default is blackman.
6303 The null audio source, return unprocessed audio frames. It is mainly useful
6304 as a template and to be employed in analysis / debugging tools, or as
6305 the source for filters which ignore the input data (for example the sox
6308 This source accepts the following options:
6312 @item channel_layout, cl
6314 Specifies the channel layout, and can be either an integer or a string
6315 representing a channel layout. The default value of @var{channel_layout}
6318 Check the channel_layout_map definition in
6319 @file{libavutil/channel_layout.c} for the mapping between strings and
6320 channel layout values.
6322 @item sample_rate, r
6323 Specifies the sample rate, and defaults to 44100.
6326 Set the number of samples per requested frames.
6329 Set the duration of the sourced audio. See
6330 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6331 for the accepted syntax.
6333 If not specified, or the expressed duration is negative, the audio is
6334 supposed to be generated forever.
6337 @subsection Examples
6341 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6343 anullsrc=r=48000:cl=4
6347 Do the same operation with a more obvious syntax:
6349 anullsrc=r=48000:cl=mono
6353 All the parameters need to be explicitly defined.
6357 Synthesize a voice utterance using the libflite library.
6359 To enable compilation of this filter you need to configure FFmpeg with
6360 @code{--enable-libflite}.
6362 Note that versions of the flite library prior to 2.0 are not thread-safe.
6364 The filter accepts the following options:
6369 If set to 1, list the names of the available voices and exit
6370 immediately. Default value is 0.
6373 Set the maximum number of samples per frame. Default value is 512.
6376 Set the filename containing the text to speak.
6379 Set the text to speak.
6382 Set the voice to use for the speech synthesis. Default value is
6383 @code{kal}. See also the @var{list_voices} option.
6386 @subsection Examples
6390 Read from file @file{speech.txt}, and synthesize the text using the
6391 standard flite voice:
6393 flite=textfile=speech.txt
6397 Read the specified text selecting the @code{slt} voice:
6399 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6403 Input text to ffmpeg:
6405 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6409 Make @file{ffplay} speak the specified text, using @code{flite} and
6410 the @code{lavfi} device:
6412 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6416 For more information about libflite, check:
6417 @url{http://www.festvox.org/flite/}
6421 Generate a noise audio signal.
6423 The filter accepts the following options:
6426 @item sample_rate, r
6427 Specify the sample rate. Default value is 48000 Hz.
6430 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6434 Specify the duration of the generated audio stream. Not specifying this option
6435 results in noise with an infinite length.
6437 @item color, colour, c
6438 Specify the color of noise. Available noise colors are white, pink, brown,
6439 blue, violet and velvet. Default color is white.
6442 Specify a value used to seed the PRNG.
6445 Set the number of samples per each output frame, default is 1024.
6448 @subsection Examples
6453 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6455 anoisesrc=d=60:c=pink:r=44100:a=0.5
6461 Generate odd-tap Hilbert transform FIR coefficients.
6463 The resulting stream can be used with @ref{afir} filter for phase-shifting
6464 the signal by 90 degrees.
6466 This is used in many matrix coding schemes and for analytic signal generation.
6467 The process is often written as a multiplication by i (or j), the imaginary unit.
6469 The filter accepts the following options:
6473 @item sample_rate, s
6474 Set sample rate, default is 44100.
6477 Set length of FIR filter, default is 22051.
6480 Set number of samples per each frame.
6483 Set window function to be used when generating FIR coefficients.
6488 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6490 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6492 The filter accepts the following options:
6495 @item sample_rate, r
6496 Set sample rate, default is 44100.
6499 Set number of samples per each frame. Default is 1024.
6502 Set high-pass frequency. Default is 0.
6505 Set low-pass frequency. Default is 0.
6506 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6507 is higher than 0 then filter will create band-pass filter coefficients,
6508 otherwise band-reject filter coefficients.
6511 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6514 Set Kaiser window beta.
6517 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6520 Enable rounding, by default is disabled.
6523 Set number of taps for high-pass filter.
6526 Set number of taps for low-pass filter.
6531 Generate an audio signal made of a sine wave with amplitude 1/8.
6533 The audio signal is bit-exact.
6535 The filter accepts the following options:
6540 Set the carrier frequency. Default is 440 Hz.
6542 @item beep_factor, b
6543 Enable a periodic beep every second with frequency @var{beep_factor} times
6544 the carrier frequency. Default is 0, meaning the beep is disabled.
6546 @item sample_rate, r
6547 Specify the sample rate, default is 44100.
6550 Specify the duration of the generated audio stream.
6552 @item samples_per_frame
6553 Set the number of samples per output frame.
6555 The expression can contain the following constants:
6559 The (sequential) number of the output audio frame, starting from 0.
6562 The PTS (Presentation TimeStamp) of the output audio frame,
6563 expressed in @var{TB} units.
6566 The PTS of the output audio frame, expressed in seconds.
6569 The timebase of the output audio frames.
6572 Default is @code{1024}.
6575 @subsection Examples
6580 Generate a simple 440 Hz sine wave:
6586 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6590 sine=frequency=220:beep_factor=4:duration=5
6594 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6597 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6601 @c man end AUDIO SOURCES
6603 @chapter Audio Sinks
6604 @c man begin AUDIO SINKS
6606 Below is a description of the currently available audio sinks.
6608 @section abuffersink
6610 Buffer audio frames, and make them available to the end of filter chain.
6612 This sink is mainly intended for programmatic use, in particular
6613 through the interface defined in @file{libavfilter/buffersink.h}
6614 or the options system.
6616 It accepts a pointer to an AVABufferSinkContext structure, which
6617 defines the incoming buffers' formats, to be passed as the opaque
6618 parameter to @code{avfilter_init_filter} for initialization.
6621 Null audio sink; do absolutely nothing with the input audio. It is
6622 mainly useful as a template and for use in analysis / debugging
6625 @c man end AUDIO SINKS
6627 @chapter Video Filters
6628 @c man begin VIDEO FILTERS
6630 When you configure your FFmpeg build, you can disable any of the
6631 existing filters using @code{--disable-filters}.
6632 The configure output will show the video filters included in your
6635 Below is a description of the currently available video filters.
6639 Mark a region of interest in a video frame.
6641 The frame data is passed through unchanged, but metadata is attached
6642 to the frame indicating regions of interest which can affect the
6643 behaviour of later encoding. Multiple regions can be marked by
6644 applying the filter multiple times.
6648 Region distance in pixels from the left edge of the frame.
6650 Region distance in pixels from the top edge of the frame.
6652 Region width in pixels.
6654 Region height in pixels.
6656 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6657 and may contain the following variables:
6660 Width of the input frame.
6662 Height of the input frame.
6666 Quantisation offset to apply within the region.
6668 This must be a real value in the range -1 to +1. A value of zero
6669 indicates no quality change. A negative value asks for better quality
6670 (less quantisation), while a positive value asks for worse quality
6671 (greater quantisation).
6673 The range is calibrated so that the extreme values indicate the
6674 largest possible offset - if the rest of the frame is encoded with the
6675 worst possible quality, an offset of -1 indicates that this region
6676 should be encoded with the best possible quality anyway. Intermediate
6677 values are then interpolated in some codec-dependent way.
6679 For example, in 10-bit H.264 the quantisation parameter varies between
6680 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6681 this region should be encoded with a QP around one-tenth of the full
6682 range better than the rest of the frame. So, if most of the frame
6683 were to be encoded with a QP of around 30, this region would get a QP
6684 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6685 An extreme value of -1 would indicate that this region should be
6686 encoded with the best possible quality regardless of the treatment of
6687 the rest of the frame - that is, should be encoded at a QP of -12.
6689 If set to true, remove any existing regions of interest marked on the
6690 frame before adding the new one.
6693 @subsection Examples
6697 Mark the centre quarter of the frame as interesting.
6699 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6702 Mark the 100-pixel-wide region on the left edge of the frame as very
6703 uninteresting (to be encoded at much lower quality than the rest of
6706 addroi=0:0:100:ih:+1/5
6710 @section alphaextract
6712 Extract the alpha component from the input as a grayscale video. This
6713 is especially useful with the @var{alphamerge} filter.
6717 Add or replace the alpha component of the primary input with the
6718 grayscale value of a second input. This is intended for use with
6719 @var{alphaextract} to allow the transmission or storage of frame
6720 sequences that have alpha in a format that doesn't support an alpha
6723 For example, to reconstruct full frames from a normal YUV-encoded video
6724 and a separate video created with @var{alphaextract}, you might use:
6726 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6731 Amplify differences between current pixel and pixels of adjacent frames in
6732 same pixel location.
6734 This filter accepts the following options:
6738 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6739 For example radius of 3 will instruct filter to calculate average of 7 frames.
6742 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6745 Set threshold for difference amplification. Any difference greater or equal to
6746 this value will not alter source pixel. Default is 10.
6747 Allowed range is from 0 to 65535.
6750 Set tolerance for difference amplification. Any difference lower to
6751 this value will not alter source pixel. Default is 0.
6752 Allowed range is from 0 to 65535.
6755 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6756 This option controls maximum possible value that will decrease source pixel value.
6759 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6760 This option controls maximum possible value that will increase source pixel value.
6763 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6766 @subsection Commands
6768 This filter supports the following @ref{commands} that corresponds to option of same name:
6780 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6781 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6782 Substation Alpha) subtitles files.
6784 This filter accepts the following option in addition to the common options from
6785 the @ref{subtitles} filter:
6789 Set the shaping engine
6791 Available values are:
6794 The default libass shaping engine, which is the best available.
6796 Fast, font-agnostic shaper that can do only substitutions
6798 Slower shaper using OpenType for substitutions and positioning
6801 The default is @code{auto}.
6805 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6807 The filter accepts the following options:
6811 Set threshold A for 1st plane. Default is 0.02.
6812 Valid range is 0 to 0.3.
6815 Set threshold B for 1st plane. Default is 0.04.
6816 Valid range is 0 to 5.
6819 Set threshold A for 2nd plane. Default is 0.02.
6820 Valid range is 0 to 0.3.
6823 Set threshold B for 2nd plane. Default is 0.04.
6824 Valid range is 0 to 5.
6827 Set threshold A for 3rd plane. Default is 0.02.
6828 Valid range is 0 to 0.3.
6831 Set threshold B for 3rd plane. Default is 0.04.
6832 Valid range is 0 to 5.
6834 Threshold A is designed to react on abrupt changes in the input signal and
6835 threshold B is designed to react on continuous changes in the input signal.
6838 Set number of frames filter will use for averaging. Default is 9. Must be odd
6839 number in range [5, 129].
6842 Set what planes of frame filter will use for averaging. Default is all.
6845 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6846 Alternatively can be set to @code{s} serial.
6848 Parallel can be faster then serial, while other way around is never true.
6849 Parallel will abort early on first change being greater then thresholds, while serial
6850 will continue processing other side of frames if they are equal or below thresholds.
6853 @subsection Commands
6854 This filter supports same @ref{commands} as options except option @code{s}.
6855 The command accepts the same syntax of the corresponding option.
6859 Apply average blur filter.
6861 The filter accepts the following options:
6865 Set horizontal radius size.
6868 Set which planes to filter. By default all planes are filtered.
6871 Set vertical radius size, if zero it will be same as @code{sizeX}.
6872 Default is @code{0}.
6875 @subsection Commands
6876 This filter supports same commands as options.
6877 The command accepts the same syntax of the corresponding option.
6879 If the specified expression is not valid, it is kept at its current
6884 Compute the bounding box for the non-black pixels in the input frame
6887 This filter computes the bounding box containing all the pixels with a
6888 luminance value greater than the minimum allowed value.
6889 The parameters describing the bounding box are printed on the filter
6892 The filter accepts the following option:
6896 Set the minimal luminance value. Default is @code{16}.
6900 Apply bilateral filter, spatial smoothing while preserving edges.
6902 The filter accepts the following options:
6905 Set sigma of gaussian function to calculate spatial weight.
6906 Allowed range is 0 to 512. Default is 0.1.
6909 Set sigma of gaussian function to calculate range weight.
6910 Allowed range is 0 to 1. Default is 0.1.
6913 Set planes to filter. Default is first only.
6916 @section bitplanenoise
6918 Show and measure bit plane noise.
6920 The filter accepts the following options:
6924 Set which plane to analyze. Default is @code{1}.
6927 Filter out noisy pixels from @code{bitplane} set above.
6928 Default is disabled.
6931 @section blackdetect
6933 Detect video intervals that are (almost) completely black. Can be
6934 useful to detect chapter transitions, commercials, or invalid
6937 The filter outputs its detection analysis to both the log as well as
6938 frame metadata. If a black segment of at least the specified minimum
6939 duration is found, a line with the start and end timestamps as well
6940 as duration is printed to the log with level @code{info}. In addition,
6941 a log line with level @code{debug} is printed per frame showing the
6942 black amount detected for that frame.
6944 The filter also attaches metadata to the first frame of a black
6945 segment with key @code{lavfi.black_start} and to the first frame
6946 after the black segment ends with key @code{lavfi.black_end}. The
6947 value is the frame's timestamp. This metadata is added regardless
6948 of the minimum duration specified.
6950 The filter accepts the following options:
6953 @item black_min_duration, d
6954 Set the minimum detected black duration expressed in seconds. It must
6955 be a non-negative floating point number.
6957 Default value is 2.0.
6959 @item picture_black_ratio_th, pic_th
6960 Set the threshold for considering a picture "black".
6961 Express the minimum value for the ratio:
6963 @var{nb_black_pixels} / @var{nb_pixels}
6966 for which a picture is considered black.
6967 Default value is 0.98.
6969 @item pixel_black_th, pix_th
6970 Set the threshold for considering a pixel "black".
6972 The threshold expresses the maximum pixel luminance value for which a
6973 pixel is considered "black". The provided value is scaled according to
6974 the following equation:
6976 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6979 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6980 the input video format, the range is [0-255] for YUV full-range
6981 formats and [16-235] for YUV non full-range formats.
6983 Default value is 0.10.
6986 The following example sets the maximum pixel threshold to the minimum
6987 value, and detects only black intervals of 2 or more seconds:
6989 blackdetect=d=2:pix_th=0.00
6994 Detect frames that are (almost) completely black. Can be useful to
6995 detect chapter transitions or commercials. Output lines consist of
6996 the frame number of the detected frame, the percentage of blackness,
6997 the position in the file if known or -1 and the timestamp in seconds.
6999 In order to display the output lines, you need to set the loglevel at
7000 least to the AV_LOG_INFO value.
7002 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
7003 The value represents the percentage of pixels in the picture that
7004 are below the threshold value.
7006 It accepts the following parameters:
7011 The percentage of the pixels that have to be below the threshold; it defaults to
7014 @item threshold, thresh
7015 The threshold below which a pixel value is considered black; it defaults to
7023 Blend two video frames into each other.
7025 The @code{blend} filter takes two input streams and outputs one
7026 stream, the first input is the "top" layer and second input is
7027 "bottom" layer. By default, the output terminates when the longest input terminates.
7029 The @code{tblend} (time blend) filter takes two consecutive frames
7030 from one single stream, and outputs the result obtained by blending
7031 the new frame on top of the old frame.
7033 A description of the accepted options follows.
7041 Set blend mode for specific pixel component or all pixel components in case
7042 of @var{all_mode}. Default value is @code{normal}.
7044 Available values for component modes are:
7086 Set blend opacity for specific pixel component or all pixel components in case
7087 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7094 Set blend expression for specific pixel component or all pixel components in case
7095 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7097 The expressions can use the following variables:
7101 The sequential number of the filtered frame, starting from @code{0}.
7105 the coordinates of the current sample
7109 the width and height of currently filtered plane
7113 Width and height scale for the plane being filtered. It is the
7114 ratio between the dimensions of the current plane to the luma plane,
7115 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7116 the luma plane and @code{0.5,0.5} for the chroma planes.
7119 Time of the current frame, expressed in seconds.
7122 Value of pixel component at current location for first video frame (top layer).
7125 Value of pixel component at current location for second video frame (bottom layer).
7129 The @code{blend} filter also supports the @ref{framesync} options.
7131 @subsection Examples
7135 Apply transition from bottom layer to top layer in first 10 seconds:
7137 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7141 Apply linear horizontal transition from top layer to bottom layer:
7143 blend=all_expr='A*(X/W)+B*(1-X/W)'
7147 Apply 1x1 checkerboard effect:
7149 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7153 Apply uncover left effect:
7155 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7159 Apply uncover down effect:
7161 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7165 Apply uncover up-left effect:
7167 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7171 Split diagonally video and shows top and bottom layer on each side:
7173 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7177 Display differences between the current and the previous frame:
7179 tblend=all_mode=grainextract
7185 Denoise frames using Block-Matching 3D algorithm.
7187 The filter accepts the following options.
7191 Set denoising strength. Default value is 1.
7192 Allowed range is from 0 to 999.9.
7193 The denoising algorithm is very sensitive to sigma, so adjust it
7194 according to the source.
7197 Set local patch size. This sets dimensions in 2D.
7200 Set sliding step for processing blocks. Default value is 4.
7201 Allowed range is from 1 to 64.
7202 Smaller values allows processing more reference blocks and is slower.
7205 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7206 When set to 1, no block matching is done. Larger values allows more blocks
7208 Allowed range is from 1 to 256.
7211 Set radius for search block matching. Default is 9.
7212 Allowed range is from 1 to INT32_MAX.
7215 Set step between two search locations for block matching. Default is 1.
7216 Allowed range is from 1 to 64. Smaller is slower.
7219 Set threshold of mean square error for block matching. Valid range is 0 to
7223 Set thresholding parameter for hard thresholding in 3D transformed domain.
7224 Larger values results in stronger hard-thresholding filtering in frequency
7228 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7229 Default is @code{basic}.
7232 If enabled, filter will use 2nd stream for block matching.
7233 Default is disabled for @code{basic} value of @var{estim} option,
7234 and always enabled if value of @var{estim} is @code{final}.
7237 Set planes to filter. Default is all available except alpha.
7240 @subsection Examples
7244 Basic filtering with bm3d:
7246 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7250 Same as above, but filtering only luma:
7252 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7256 Same as above, but with both estimation modes:
7258 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
7262 Same as above, but prefilter with @ref{nlmeans} filter instead:
7264 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
7270 Apply a boxblur algorithm to the input video.
7272 It accepts the following parameters:
7276 @item luma_radius, lr
7277 @item luma_power, lp
7278 @item chroma_radius, cr
7279 @item chroma_power, cp
7280 @item alpha_radius, ar
7281 @item alpha_power, ap
7285 A description of the accepted options follows.
7288 @item luma_radius, lr
7289 @item chroma_radius, cr
7290 @item alpha_radius, ar
7291 Set an expression for the box radius in pixels used for blurring the
7292 corresponding input plane.
7294 The radius value must be a non-negative number, and must not be
7295 greater than the value of the expression @code{min(w,h)/2} for the
7296 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7299 Default value for @option{luma_radius} is "2". If not specified,
7300 @option{chroma_radius} and @option{alpha_radius} default to the
7301 corresponding value set for @option{luma_radius}.
7303 The expressions can contain the following constants:
7307 The input width and height in pixels.
7311 The input chroma image width and height in pixels.
7315 The horizontal and vertical chroma subsample values. For example, for the
7316 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7319 @item luma_power, lp
7320 @item chroma_power, cp
7321 @item alpha_power, ap
7322 Specify how many times the boxblur filter is applied to the
7323 corresponding plane.
7325 Default value for @option{luma_power} is 2. If not specified,
7326 @option{chroma_power} and @option{alpha_power} default to the
7327 corresponding value set for @option{luma_power}.
7329 A value of 0 will disable the effect.
7332 @subsection Examples
7336 Apply a boxblur filter with the luma, chroma, and alpha radii
7339 boxblur=luma_radius=2:luma_power=1
7344 Set the luma radius to 2, and alpha and chroma radius to 0:
7346 boxblur=2:1:cr=0:ar=0
7350 Set the luma and chroma radii to a fraction of the video dimension:
7352 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7358 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7359 Deinterlacing Filter").
7361 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7362 interpolation algorithms.
7363 It accepts the following parameters:
7367 The interlacing mode to adopt. It accepts one of the following values:
7371 Output one frame for each frame.
7373 Output one frame for each field.
7376 The default value is @code{send_field}.
7379 The picture field parity assumed for the input interlaced video. It accepts one
7380 of the following values:
7384 Assume the top field is first.
7386 Assume the bottom field is first.
7388 Enable automatic detection of field parity.
7391 The default value is @code{auto}.
7392 If the interlacing is unknown or the decoder does not export this information,
7393 top field first will be assumed.
7396 Specify which frames to deinterlace. Accepts one of the following
7401 Deinterlace all frames.
7403 Only deinterlace frames marked as interlaced.
7406 The default value is @code{all}.
7411 Apply Contrast Adaptive Sharpen filter to video stream.
7413 The filter accepts the following options:
7417 Set the sharpening strength. Default value is 0.
7420 Set planes to filter. Default value is to filter all
7421 planes except alpha plane.
7425 Remove all color information for all colors except for certain one.
7427 The filter accepts the following options:
7431 The color which will not be replaced with neutral chroma.
7434 Similarity percentage with the above color.
7435 0.01 matches only the exact key color, while 1.0 matches everything.
7439 0.0 makes pixels either fully gray, or not gray at all.
7440 Higher values result in more preserved color.
7443 Signals that the color passed is already in YUV instead of RGB.
7445 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7446 This can be used to pass exact YUV values as hexadecimal numbers.
7449 @subsection Commands
7450 This filter supports same @ref{commands} as options.
7451 The command accepts the same syntax of the corresponding option.
7453 If the specified expression is not valid, it is kept at its current
7457 YUV colorspace color/chroma keying.
7459 The filter accepts the following options:
7463 The color which will be replaced with transparency.
7466 Similarity percentage with the key color.
7468 0.01 matches only the exact key color, while 1.0 matches everything.
7473 0.0 makes pixels either fully transparent, or not transparent at all.
7475 Higher values result in semi-transparent pixels, with a higher transparency
7476 the more similar the pixels color is to the key color.
7479 Signals that the color passed is already in YUV instead of RGB.
7481 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7482 This can be used to pass exact YUV values as hexadecimal numbers.
7485 @subsection Commands
7486 This filter supports same @ref{commands} as options.
7487 The command accepts the same syntax of the corresponding option.
7489 If the specified expression is not valid, it is kept at its current
7492 @subsection Examples
7496 Make every green pixel in the input image transparent:
7498 ffmpeg -i input.png -vf chromakey=green out.png
7502 Overlay a greenscreen-video on top of a static black background.
7504 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
7509 Reduce chrominance noise.
7511 The filter accepts the following options:
7515 Set threshold for averaging chrominance values.
7516 Sum of absolute difference of U and V pixel components or current
7517 pixel and neighbour pixels lower than this threshold will be used in
7518 averaging. Luma component is left unchanged and is copied to output.
7519 Default value is 30. Allowed range is from 1 to 200.
7522 Set horizontal radius of rectangle used for averaging.
7523 Allowed range is from 1 to 100. Default value is 5.
7526 Set vertical radius of rectangle used for averaging.
7527 Allowed range is from 1 to 100. Default value is 5.
7530 Set horizontal step when averaging. Default value is 1.
7531 Allowed range is from 1 to 50.
7532 Mostly useful to speed-up filtering.
7535 Set vertical step when averaging. Default value is 1.
7536 Allowed range is from 1 to 50.
7537 Mostly useful to speed-up filtering.
7540 @subsection Commands
7541 This filter supports same @ref{commands} as options.
7542 The command accepts the same syntax of the corresponding option.
7544 @section chromashift
7545 Shift chroma pixels horizontally and/or vertically.
7547 The filter accepts the following options:
7550 Set amount to shift chroma-blue horizontally.
7552 Set amount to shift chroma-blue vertically.
7554 Set amount to shift chroma-red horizontally.
7556 Set amount to shift chroma-red vertically.
7558 Set edge mode, can be @var{smear}, default, or @var{warp}.
7561 @subsection Commands
7563 This filter supports the all above options as @ref{commands}.
7567 Display CIE color diagram with pixels overlaid onto it.
7569 The filter accepts the following options:
7584 @item uhdtv, rec2020
7598 Set what gamuts to draw.
7600 See @code{system} option for available values.
7603 Set ciescope size, by default set to 512.
7606 Set intensity used to map input pixel values to CIE diagram.
7609 Set contrast used to draw tongue colors that are out of active color system gamut.
7612 Correct gamma displayed on scope, by default enabled.
7615 Show white point on CIE diagram, by default disabled.
7618 Set input gamma. Used only with XYZ input color space.
7623 Visualize information exported by some codecs.
7625 Some codecs can export information through frames using side-data or other
7626 means. For example, some MPEG based codecs export motion vectors through the
7627 @var{export_mvs} flag in the codec @option{flags2} option.
7629 The filter accepts the following option:
7633 Set motion vectors to visualize.
7635 Available flags for @var{mv} are:
7639 forward predicted MVs of P-frames
7641 forward predicted MVs of B-frames
7643 backward predicted MVs of B-frames
7647 Display quantization parameters using the chroma planes.
7650 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7652 Available flags for @var{mv_type} are:
7656 forward predicted MVs
7658 backward predicted MVs
7661 @item frame_type, ft
7662 Set frame type to visualize motion vectors of.
7664 Available flags for @var{frame_type} are:
7668 intra-coded frames (I-frames)
7670 predicted frames (P-frames)
7672 bi-directionally predicted frames (B-frames)
7676 @subsection Examples
7680 Visualize forward predicted MVs of all frames using @command{ffplay}:
7682 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7686 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7688 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7692 @section colorbalance
7693 Modify intensity of primary colors (red, green and blue) of input frames.
7695 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7696 regions for the red-cyan, green-magenta or blue-yellow balance.
7698 A positive adjustment value shifts the balance towards the primary color, a negative
7699 value towards the complementary color.
7701 The filter accepts the following options:
7707 Adjust red, green and blue shadows (darkest pixels).
7712 Adjust red, green and blue midtones (medium pixels).
7717 Adjust red, green and blue highlights (brightest pixels).
7719 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7722 Preserve lightness when changing color balance. Default is disabled.
7725 @subsection Examples
7729 Add red color cast to shadows:
7735 @subsection Commands
7737 This filter supports the all above options as @ref{commands}.
7739 @section colorchannelmixer
7741 Adjust video input frames by re-mixing color channels.
7743 This filter modifies a color channel by adding the values associated to
7744 the other channels of the same pixels. For example if the value to
7745 modify is red, the output value will be:
7747 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7750 The filter accepts the following options:
7757 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7758 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7764 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7765 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7771 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7772 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7778 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7779 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7781 Allowed ranges for options are @code{[-2.0, 2.0]}.
7784 @subsection Examples
7788 Convert source to grayscale:
7790 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7793 Simulate sepia tones:
7795 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7799 @subsection Commands
7801 This filter supports the all above options as @ref{commands}.
7804 RGB colorspace color keying.
7806 The filter accepts the following options:
7810 The color which will be replaced with transparency.
7813 Similarity percentage with the key color.
7815 0.01 matches only the exact key color, while 1.0 matches everything.
7820 0.0 makes pixels either fully transparent, or not transparent at all.
7822 Higher values result in semi-transparent pixels, with a higher transparency
7823 the more similar the pixels color is to the key color.
7826 @subsection Examples
7830 Make every green pixel in the input image transparent:
7832 ffmpeg -i input.png -vf colorkey=green out.png
7836 Overlay a greenscreen-video on top of a static background image.
7838 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
7842 @subsection Commands
7843 This filter supports same @ref{commands} as options.
7844 The command accepts the same syntax of the corresponding option.
7846 If the specified expression is not valid, it is kept at its current
7850 Remove all color information for all RGB colors except for certain one.
7852 The filter accepts the following options:
7856 The color which will not be replaced with neutral gray.
7859 Similarity percentage with the above color.
7860 0.01 matches only the exact key color, while 1.0 matches everything.
7863 Blend percentage. 0.0 makes pixels fully gray.
7864 Higher values result in more preserved color.
7867 @subsection Commands
7868 This filter supports same @ref{commands} as options.
7869 The command accepts the same syntax of the corresponding option.
7871 If the specified expression is not valid, it is kept at its current
7874 @section colorlevels
7876 Adjust video input frames using levels.
7878 The filter accepts the following options:
7885 Adjust red, green, blue and alpha input black point.
7886 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7892 Adjust red, green, blue and alpha input white point.
7893 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7895 Input levels are used to lighten highlights (bright tones), darken shadows
7896 (dark tones), change the balance of bright and dark tones.
7902 Adjust red, green, blue and alpha output black point.
7903 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7909 Adjust red, green, blue and alpha output white point.
7910 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7912 Output levels allows manual selection of a constrained output level range.
7915 @subsection Examples
7919 Make video output darker:
7921 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7927 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7931 Make video output lighter:
7933 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7937 Increase brightness:
7939 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7943 @subsection Commands
7945 This filter supports the all above options as @ref{commands}.
7947 @section colormatrix
7949 Convert color matrix.
7951 The filter accepts the following options:
7956 Specify the source and destination color matrix. Both values must be
7959 The accepted values are:
7987 For example to convert from BT.601 to SMPTE-240M, use the command:
7989 colormatrix=bt601:smpte240m
7994 Convert colorspace, transfer characteristics or color primaries.
7995 Input video needs to have an even size.
7997 The filter accepts the following options:
8002 Specify all color properties at once.
8004 The accepted values are:
8034 Specify output colorspace.
8036 The accepted values are:
8045 BT.470BG or BT.601-6 625
8048 SMPTE-170M or BT.601-6 525
8057 BT.2020 with non-constant luminance
8063 Specify output transfer characteristics.
8065 The accepted values are:
8077 Constant gamma of 2.2
8080 Constant gamma of 2.8
8083 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8101 BT.2020 for 10-bits content
8104 BT.2020 for 12-bits content
8110 Specify output color primaries.
8112 The accepted values are:
8121 BT.470BG or BT.601-6 625
8124 SMPTE-170M or BT.601-6 525
8148 Specify output color range.
8150 The accepted values are:
8153 TV (restricted) range
8156 MPEG (restricted) range
8167 Specify output color format.
8169 The accepted values are:
8172 YUV 4:2:0 planar 8-bits
8175 YUV 4:2:0 planar 10-bits
8178 YUV 4:2:0 planar 12-bits
8181 YUV 4:2:2 planar 8-bits
8184 YUV 4:2:2 planar 10-bits
8187 YUV 4:2:2 planar 12-bits
8190 YUV 4:4:4 planar 8-bits
8193 YUV 4:4:4 planar 10-bits
8196 YUV 4:4:4 planar 12-bits
8201 Do a fast conversion, which skips gamma/primary correction. This will take
8202 significantly less CPU, but will be mathematically incorrect. To get output
8203 compatible with that produced by the colormatrix filter, use fast=1.
8206 Specify dithering mode.
8208 The accepted values are:
8214 Floyd-Steinberg dithering
8218 Whitepoint adaptation mode.
8220 The accepted values are:
8223 Bradford whitepoint adaptation
8226 von Kries whitepoint adaptation
8229 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8233 Override all input properties at once. Same accepted values as @ref{all}.
8236 Override input colorspace. Same accepted values as @ref{space}.
8239 Override input color primaries. Same accepted values as @ref{primaries}.
8242 Override input transfer characteristics. Same accepted values as @ref{trc}.
8245 Override input color range. Same accepted values as @ref{range}.
8249 The filter converts the transfer characteristics, color space and color
8250 primaries to the specified user values. The output value, if not specified,
8251 is set to a default value based on the "all" property. If that property is
8252 also not specified, the filter will log an error. The output color range and
8253 format default to the same value as the input color range and format. The
8254 input transfer characteristics, color space, color primaries and color range
8255 should be set on the input data. If any of these are missing, the filter will
8256 log an error and no conversion will take place.
8258 For example to convert the input to SMPTE-240M, use the command:
8260 colorspace=smpte240m
8263 @section convolution
8265 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8267 The filter accepts the following options:
8274 Set matrix for each plane.
8275 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8276 and from 1 to 49 odd number of signed integers in @var{row} mode.
8282 Set multiplier for calculated value for each plane.
8283 If unset or 0, it will be sum of all matrix elements.
8289 Set bias for each plane. This value is added to the result of the multiplication.
8290 Useful for making the overall image brighter or darker. Default is 0.0.
8296 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8297 Default is @var{square}.
8300 @subsection Examples
8306 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"
8312 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"
8318 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"
8324 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"
8328 Apply laplacian edge detector which includes diagonals:
8330 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"
8336 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"
8342 Apply 2D convolution of video stream in frequency domain using second stream
8345 The filter accepts the following options:
8349 Set which planes to process.
8352 Set which impulse video frames will be processed, can be @var{first}
8353 or @var{all}. Default is @var{all}.
8356 The @code{convolve} filter also supports the @ref{framesync} options.
8360 Copy the input video source unchanged to the output. This is mainly useful for
8365 Video filtering on GPU using Apple's CoreImage API on OSX.
8367 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8368 processed by video hardware. However, software-based OpenGL implementations
8369 exist which means there is no guarantee for hardware processing. It depends on
8372 There are many filters and image generators provided by Apple that come with a
8373 large variety of options. The filter has to be referenced by its name along
8376 The coreimage filter accepts the following options:
8379 List all available filters and generators along with all their respective
8380 options as well as possible minimum and maximum values along with the default
8387 Specify all filters by their respective name and options.
8388 Use @var{list_filters} to determine all valid filter names and options.
8389 Numerical options are specified by a float value and are automatically clamped
8390 to their respective value range. Vector and color options have to be specified
8391 by a list of space separated float values. Character escaping has to be done.
8392 A special option name @code{default} is available to use default options for a
8395 It is required to specify either @code{default} or at least one of the filter options.
8396 All omitted options are used with their default values.
8397 The syntax of the filter string is as follows:
8399 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8403 Specify a rectangle where the output of the filter chain is copied into the
8404 input image. It is given by a list of space separated float values:
8406 output_rect=x\ y\ width\ height
8408 If not given, the output rectangle equals the dimensions of the input image.
8409 The output rectangle is automatically cropped at the borders of the input
8410 image. Negative values are valid for each component.
8412 output_rect=25\ 25\ 100\ 100
8416 Several filters can be chained for successive processing without GPU-HOST
8417 transfers allowing for fast processing of complex filter chains.
8418 Currently, only filters with zero (generators) or exactly one (filters) input
8419 image and one output image are supported. Also, transition filters are not yet
8422 Some filters generate output images with additional padding depending on the
8423 respective filter kernel. The padding is automatically removed to ensure the
8424 filter output has the same size as the input image.
8426 For image generators, the size of the output image is determined by the
8427 previous output image of the filter chain or the input image of the whole
8428 filterchain, respectively. The generators do not use the pixel information of
8429 this image to generate their output. However, the generated output is
8430 blended onto this image, resulting in partial or complete coverage of the
8433 The @ref{coreimagesrc} video source can be used for generating input images
8434 which are directly fed into the filter chain. By using it, providing input
8435 images by another video source or an input video is not required.
8437 @subsection Examples
8442 List all filters available:
8444 coreimage=list_filters=true
8448 Use the CIBoxBlur filter with default options to blur an image:
8450 coreimage=filter=CIBoxBlur@@default
8454 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8455 its center at 100x100 and a radius of 50 pixels:
8457 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8461 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8462 given as complete and escaped command-line for Apple's standard bash shell:
8464 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8470 Cover a rectangular object
8472 It accepts the following options:
8476 Filepath of the optional cover image, needs to be in yuv420.
8481 It accepts the following values:
8484 cover it by the supplied image
8486 cover it by interpolating the surrounding pixels
8489 Default value is @var{blur}.
8492 @subsection Examples
8496 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8498 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8504 Crop the input video to given dimensions.
8506 It accepts the following parameters:
8510 The width of the output video. It defaults to @code{iw}.
8511 This expression is evaluated only once during the filter
8512 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8515 The height of the output video. It defaults to @code{ih}.
8516 This expression is evaluated only once during the filter
8517 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8520 The horizontal position, in the input video, of the left edge of the output
8521 video. It defaults to @code{(in_w-out_w)/2}.
8522 This expression is evaluated per-frame.
8525 The vertical position, in the input video, of the top edge of the output video.
8526 It defaults to @code{(in_h-out_h)/2}.
8527 This expression is evaluated per-frame.
8530 If set to 1 will force the output display aspect ratio
8531 to be the same of the input, by changing the output sample aspect
8532 ratio. It defaults to 0.
8535 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8536 width/height/x/y as specified and will not be rounded to nearest smaller value.
8540 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8541 expressions containing the following constants:
8546 The computed values for @var{x} and @var{y}. They are evaluated for
8551 The input width and height.
8555 These are the same as @var{in_w} and @var{in_h}.
8559 The output (cropped) width and height.
8563 These are the same as @var{out_w} and @var{out_h}.
8566 same as @var{iw} / @var{ih}
8569 input sample aspect ratio
8572 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8576 horizontal and vertical chroma subsample values. For example for the
8577 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8580 The number of the input frame, starting from 0.
8583 the position in the file of the input frame, NAN if unknown
8586 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8590 The expression for @var{out_w} may depend on the value of @var{out_h},
8591 and the expression for @var{out_h} may depend on @var{out_w}, but they
8592 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8593 evaluated after @var{out_w} and @var{out_h}.
8595 The @var{x} and @var{y} parameters specify the expressions for the
8596 position of the top-left corner of the output (non-cropped) area. They
8597 are evaluated for each frame. If the evaluated value is not valid, it
8598 is approximated to the nearest valid value.
8600 The expression for @var{x} may depend on @var{y}, and the expression
8601 for @var{y} may depend on @var{x}.
8603 @subsection Examples
8607 Crop area with size 100x100 at position (12,34).
8612 Using named options, the example above becomes:
8614 crop=w=100:h=100:x=12:y=34
8618 Crop the central input area with size 100x100:
8624 Crop the central input area with size 2/3 of the input video:
8626 crop=2/3*in_w:2/3*in_h
8630 Crop the input video central square:
8637 Delimit the rectangle with the top-left corner placed at position
8638 100:100 and the right-bottom corner corresponding to the right-bottom
8639 corner of the input image.
8641 crop=in_w-100:in_h-100:100:100
8645 Crop 10 pixels from the left and right borders, and 20 pixels from
8646 the top and bottom borders
8648 crop=in_w-2*10:in_h-2*20
8652 Keep only the bottom right quarter of the input image:
8654 crop=in_w/2:in_h/2:in_w/2:in_h/2
8658 Crop height for getting Greek harmony:
8660 crop=in_w:1/PHI*in_w
8664 Apply trembling effect:
8666 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)
8670 Apply erratic camera effect depending on timestamp:
8672 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)"
8676 Set x depending on the value of y:
8678 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8682 @subsection Commands
8684 This filter supports the following commands:
8690 Set width/height of the output video and the horizontal/vertical position
8692 The command accepts the same syntax of the corresponding option.
8694 If the specified expression is not valid, it is kept at its current
8700 Auto-detect the crop size.
8702 It calculates the necessary cropping parameters and prints the
8703 recommended parameters via the logging system. The detected dimensions
8704 correspond to the non-black area of the input video.
8706 It accepts the following parameters:
8711 Set higher black value threshold, which can be optionally specified
8712 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8713 value greater to the set value is considered non-black. It defaults to 24.
8714 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8715 on the bitdepth of the pixel format.
8718 The value which the width/height should be divisible by. It defaults to
8719 16. The offset is automatically adjusted to center the video. Use 2 to
8720 get only even dimensions (needed for 4:2:2 video). 16 is best when
8721 encoding to most video codecs.
8723 @item reset_count, reset
8724 Set the counter that determines after how many frames cropdetect will
8725 reset the previously detected largest video area and start over to
8726 detect the current optimal crop area. Default value is 0.
8728 This can be useful when channel logos distort the video area. 0
8729 indicates 'never reset', and returns the largest area encountered during
8736 Delay video filtering until a given wallclock timestamp. The filter first
8737 passes on @option{preroll} amount of frames, then it buffers at most
8738 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8739 it forwards the buffered frames and also any subsequent frames coming in its
8742 The filter can be used synchronize the output of multiple ffmpeg processes for
8743 realtime output devices like decklink. By putting the delay in the filtering
8744 chain and pre-buffering frames the process can pass on data to output almost
8745 immediately after the target wallclock timestamp is reached.
8747 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8753 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8756 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8759 The maximum duration of content to buffer before waiting for the cue expressed
8760 in seconds. Default is 0.
8767 Apply color adjustments using curves.
8769 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8770 component (red, green and blue) has its values defined by @var{N} key points
8771 tied from each other using a smooth curve. The x-axis represents the pixel
8772 values from the input frame, and the y-axis the new pixel values to be set for
8775 By default, a component curve is defined by the two points @var{(0;0)} and
8776 @var{(1;1)}. This creates a straight line where each original pixel value is
8777 "adjusted" to its own value, which means no change to the image.
8779 The filter allows you to redefine these two points and add some more. A new
8780 curve (using a natural cubic spline interpolation) will be define to pass
8781 smoothly through all these new coordinates. The new defined points needs to be
8782 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8783 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8784 the vector spaces, the values will be clipped accordingly.
8786 The filter accepts the following options:
8790 Select one of the available color presets. This option can be used in addition
8791 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8792 options takes priority on the preset values.
8793 Available presets are:
8796 @item color_negative
8799 @item increase_contrast
8801 @item linear_contrast
8802 @item medium_contrast
8804 @item strong_contrast
8807 Default is @code{none}.
8809 Set the master key points. These points will define a second pass mapping. It
8810 is sometimes called a "luminance" or "value" mapping. It can be used with
8811 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8812 post-processing LUT.
8814 Set the key points for the red component.
8816 Set the key points for the green component.
8818 Set the key points for the blue component.
8820 Set the key points for all components (not including master).
8821 Can be used in addition to the other key points component
8822 options. In this case, the unset component(s) will fallback on this
8823 @option{all} setting.
8825 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8827 Save Gnuplot script of the curves in specified file.
8830 To avoid some filtergraph syntax conflicts, each key points list need to be
8831 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8833 @subsection Examples
8837 Increase slightly the middle level of blue:
8839 curves=blue='0/0 0.5/0.58 1/1'
8845 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'
8847 Here we obtain the following coordinates for each components:
8850 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8852 @code{(0;0) (0.50;0.48) (1;1)}
8854 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8858 The previous example can also be achieved with the associated built-in preset:
8860 curves=preset=vintage
8870 Use a Photoshop preset and redefine the points of the green component:
8872 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8876 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8877 and @command{gnuplot}:
8879 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8880 gnuplot -p /tmp/curves.plt
8886 Video data analysis filter.
8888 This filter shows hexadecimal pixel values of part of video.
8890 The filter accepts the following options:
8894 Set output video size.
8897 Set x offset from where to pick pixels.
8900 Set y offset from where to pick pixels.
8903 Set scope mode, can be one of the following:
8906 Draw hexadecimal pixel values with white color on black background.
8909 Draw hexadecimal pixel values with input video pixel color on black
8913 Draw hexadecimal pixel values on color background picked from input video,
8914 the text color is picked in such way so its always visible.
8918 Draw rows and columns numbers on left and top of video.
8921 Set background opacity.
8924 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
8928 Apply Directional blur filter.
8930 The filter accepts the following options:
8934 Set angle of directional blur. Default is @code{45}.
8937 Set radius of directional blur. Default is @code{5}.
8940 Set which planes to filter. By default all planes are filtered.
8943 @subsection Commands
8944 This filter supports same @ref{commands} as options.
8945 The command accepts the same syntax of the corresponding option.
8947 If the specified expression is not valid, it is kept at its current
8952 Denoise frames using 2D DCT (frequency domain filtering).
8954 This filter is not designed for real time.
8956 The filter accepts the following options:
8960 Set the noise sigma constant.
8962 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8963 coefficient (absolute value) below this threshold with be dropped.
8965 If you need a more advanced filtering, see @option{expr}.
8967 Default is @code{0}.
8970 Set number overlapping pixels for each block. Since the filter can be slow, you
8971 may want to reduce this value, at the cost of a less effective filter and the
8972 risk of various artefacts.
8974 If the overlapping value doesn't permit processing the whole input width or
8975 height, a warning will be displayed and according borders won't be denoised.
8977 Default value is @var{blocksize}-1, which is the best possible setting.
8980 Set the coefficient factor expression.
8982 For each coefficient of a DCT block, this expression will be evaluated as a
8983 multiplier value for the coefficient.
8985 If this is option is set, the @option{sigma} option will be ignored.
8987 The absolute value of the coefficient can be accessed through the @var{c}
8991 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8992 @var{blocksize}, which is the width and height of the processed blocks.
8994 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8995 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8996 on the speed processing. Also, a larger block size does not necessarily means a
9000 @subsection Examples
9002 Apply a denoise with a @option{sigma} of @code{4.5}:
9007 The same operation can be achieved using the expression system:
9009 dctdnoiz=e='gte(c, 4.5*3)'
9012 Violent denoise using a block size of @code{16x16}:
9019 Remove banding artifacts from input video.
9020 It works by replacing banded pixels with average value of referenced pixels.
9022 The filter accepts the following options:
9029 Set banding detection threshold for each plane. Default is 0.02.
9030 Valid range is 0.00003 to 0.5.
9031 If difference between current pixel and reference pixel is less than threshold,
9032 it will be considered as banded.
9035 Banding detection range in pixels. Default is 16. If positive, random number
9036 in range 0 to set value will be used. If negative, exact absolute value
9038 The range defines square of four pixels around current pixel.
9041 Set direction in radians from which four pixel will be compared. If positive,
9042 random direction from 0 to set direction will be picked. If negative, exact of
9043 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9044 will pick only pixels on same row and -PI/2 will pick only pixels on same
9048 If enabled, current pixel is compared with average value of all four
9049 surrounding pixels. The default is enabled. If disabled current pixel is
9050 compared with all four surrounding pixels. The pixel is considered banded
9051 if only all four differences with surrounding pixels are less than threshold.
9054 If enabled, current pixel is changed if and only if all pixel components are banded,
9055 e.g. banding detection threshold is triggered for all color components.
9056 The default is disabled.
9061 Remove blocking artifacts from input video.
9063 The filter accepts the following options:
9067 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9068 This controls what kind of deblocking is applied.
9071 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9077 Set blocking detection thresholds. Allowed range is 0 to 1.
9078 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9079 Using higher threshold gives more deblocking strength.
9080 Setting @var{alpha} controls threshold detection at exact edge of block.
9081 Remaining options controls threshold detection near the edge. Each one for
9082 below/above or left/right. Setting any of those to @var{0} disables
9086 Set planes to filter. Default is to filter all available planes.
9089 @subsection Examples
9093 Deblock using weak filter and block size of 4 pixels.
9095 deblock=filter=weak:block=4
9099 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9100 deblocking more edges.
9102 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9106 Similar as above, but filter only first plane.
9108 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9112 Similar as above, but filter only second and third plane.
9114 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9121 Drop duplicated frames at regular intervals.
9123 The filter accepts the following options:
9127 Set the number of frames from which one will be dropped. Setting this to
9128 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9129 Default is @code{5}.
9132 Set the threshold for duplicate detection. If the difference metric for a frame
9133 is less than or equal to this value, then it is declared as duplicate. Default
9137 Set scene change threshold. Default is @code{15}.
9141 Set the size of the x and y-axis blocks used during metric calculations.
9142 Larger blocks give better noise suppression, but also give worse detection of
9143 small movements. Must be a power of two. Default is @code{32}.
9146 Mark main input as a pre-processed input and activate clean source input
9147 stream. This allows the input to be pre-processed with various filters to help
9148 the metrics calculation while keeping the frame selection lossless. When set to
9149 @code{1}, the first stream is for the pre-processed input, and the second
9150 stream is the clean source from where the kept frames are chosen. Default is
9154 Set whether or not chroma is considered in the metric calculations. Default is
9160 Apply 2D deconvolution of video stream in frequency domain using second stream
9163 The filter accepts the following options:
9167 Set which planes to process.
9170 Set which impulse video frames will be processed, can be @var{first}
9171 or @var{all}. Default is @var{all}.
9174 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9175 and height are not same and not power of 2 or if stream prior to convolving
9179 The @code{deconvolve} filter also supports the @ref{framesync} options.
9183 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9185 It accepts the following options:
9189 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9190 @var{rainbows} for cross-color reduction.
9193 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9196 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9199 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9202 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9207 Apply deflate effect to the video.
9209 This filter replaces the pixel by the local(3x3) average by taking into account
9210 only values lower than the pixel.
9212 It accepts the following options:
9219 Limit the maximum change for each plane, default is 65535.
9220 If 0, plane will remain unchanged.
9223 @subsection Commands
9225 This filter supports the all above options as @ref{commands}.
9229 Remove temporal frame luminance variations.
9231 It accepts the following options:
9235 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9238 Set averaging mode to smooth temporal luminance variations.
9240 Available values are:
9265 Do not actually modify frame. Useful when one only wants metadata.
9270 Remove judder produced by partially interlaced telecined content.
9272 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9273 source was partially telecined content then the output of @code{pullup,dejudder}
9274 will have a variable frame rate. May change the recorded frame rate of the
9275 container. Aside from that change, this filter will not affect constant frame
9278 The option available in this filter is:
9282 Specify the length of the window over which the judder repeats.
9284 Accepts any integer greater than 1. Useful values are:
9288 If the original was telecined from 24 to 30 fps (Film to NTSC).
9291 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9294 If a mixture of the two.
9297 The default is @samp{4}.
9302 Suppress a TV station logo by a simple interpolation of the surrounding
9303 pixels. Just set a rectangle covering the logo and watch it disappear
9304 (and sometimes something even uglier appear - your mileage may vary).
9306 It accepts the following parameters:
9311 Specify the top left corner coordinates of the logo. They must be
9316 Specify the width and height of the logo to clear. They must be
9320 Specify the thickness of the fuzzy edge of the rectangle (added to
9321 @var{w} and @var{h}). The default value is 1. This option is
9322 deprecated, setting higher values should no longer be necessary and
9326 When set to 1, a green rectangle is drawn on the screen to simplify
9327 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9328 The default value is 0.
9330 The rectangle is drawn on the outermost pixels which will be (partly)
9331 replaced with interpolated values. The values of the next pixels
9332 immediately outside this rectangle in each direction will be used to
9333 compute the interpolated pixel values inside the rectangle.
9337 @subsection Examples
9341 Set a rectangle covering the area with top left corner coordinates 0,0
9342 and size 100x77, and a band of size 10:
9344 delogo=x=0:y=0:w=100:h=77:band=10
9352 Remove the rain in the input image/video by applying the derain methods based on
9353 convolutional neural networks. Supported models:
9357 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9358 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9361 Training as well as model generation scripts are provided in
9362 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9364 Native model files (.model) can be generated from TensorFlow model
9365 files (.pb) by using tools/python/convert.py
9367 The filter accepts the following options:
9371 Specify which filter to use. This option accepts the following values:
9375 Derain filter. To conduct derain filter, you need to use a derain model.
9378 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9380 Default value is @samp{derain}.
9383 Specify which DNN backend to use for model loading and execution. This option accepts
9384 the following values:
9388 Native implementation of DNN loading and execution.
9391 TensorFlow backend. To enable this backend you
9392 need to install the TensorFlow for C library (see
9393 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9394 @code{--enable-libtensorflow}
9396 Default value is @samp{native}.
9399 Set path to model file specifying network architecture and its parameters.
9400 Note that different backends use different file formats. TensorFlow and native
9401 backend can load files for only its format.
9404 It can also be finished with @ref{dnn_processing} filter.
9408 Attempt to fix small changes in horizontal and/or vertical shift. This
9409 filter helps remove camera shake from hand-holding a camera, bumping a
9410 tripod, moving on a vehicle, etc.
9412 The filter accepts the following options:
9420 Specify a rectangular area where to limit the search for motion
9422 If desired the search for motion vectors can be limited to a
9423 rectangular area of the frame defined by its top left corner, width
9424 and height. These parameters have the same meaning as the drawbox
9425 filter which can be used to visualise the position of the bounding
9428 This is useful when simultaneous movement of subjects within the frame
9429 might be confused for camera motion by the motion vector search.
9431 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9432 then the full frame is used. This allows later options to be set
9433 without specifying the bounding box for the motion vector search.
9435 Default - search the whole frame.
9439 Specify the maximum extent of movement in x and y directions in the
9440 range 0-64 pixels. Default 16.
9443 Specify how to generate pixels to fill blanks at the edge of the
9444 frame. Available values are:
9447 Fill zeroes at blank locations
9449 Original image at blank locations
9451 Extruded edge value at blank locations
9453 Mirrored edge at blank locations
9455 Default value is @samp{mirror}.
9458 Specify the blocksize to use for motion search. Range 4-128 pixels,
9462 Specify the contrast threshold for blocks. Only blocks with more than
9463 the specified contrast (difference between darkest and lightest
9464 pixels) will be considered. Range 1-255, default 125.
9467 Specify the search strategy. Available values are:
9470 Set exhaustive search
9472 Set less exhaustive search.
9474 Default value is @samp{exhaustive}.
9477 If set then a detailed log of the motion search is written to the
9484 Remove unwanted contamination of foreground colors, caused by reflected color of
9485 greenscreen or bluescreen.
9487 This filter accepts the following options:
9491 Set what type of despill to use.
9494 Set how spillmap will be generated.
9497 Set how much to get rid of still remaining spill.
9500 Controls amount of red in spill area.
9503 Controls amount of green in spill area.
9504 Should be -1 for greenscreen.
9507 Controls amount of blue in spill area.
9508 Should be -1 for bluescreen.
9511 Controls brightness of spill area, preserving colors.
9514 Modify alpha from generated spillmap.
9517 @subsection Commands
9519 This filter supports the all above options as @ref{commands}.
9523 Apply an exact inverse of the telecine operation. It requires a predefined
9524 pattern specified using the pattern option which must be the same as that passed
9525 to the telecine filter.
9527 This filter accepts the following options:
9536 The default value is @code{top}.
9540 A string of numbers representing the pulldown pattern you wish to apply.
9541 The default value is @code{23}.
9544 A number representing position of the first frame with respect to the telecine
9545 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9550 Apply dilation effect to the video.
9552 This filter replaces the pixel by the local(3x3) maximum.
9554 It accepts the following options:
9561 Limit the maximum change for each plane, default is 65535.
9562 If 0, plane will remain unchanged.
9565 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9568 Flags to local 3x3 coordinates maps like this:
9575 @subsection Commands
9577 This filter supports the all above options as @ref{commands}.
9581 Displace pixels as indicated by second and third input stream.
9583 It takes three input streams and outputs one stream, the first input is the
9584 source, and second and third input are displacement maps.
9586 The second input specifies how much to displace pixels along the
9587 x-axis, while the third input specifies how much to displace pixels
9589 If one of displacement map streams terminates, last frame from that
9590 displacement map will be used.
9592 Note that once generated, displacements maps can be reused over and over again.
9594 A description of the accepted options follows.
9598 Set displace behavior for pixels that are out of range.
9600 Available values are:
9603 Missing pixels are replaced by black pixels.
9606 Adjacent pixels will spread out to replace missing pixels.
9609 Out of range pixels are wrapped so they point to pixels of other side.
9612 Out of range pixels will be replaced with mirrored pixels.
9614 Default is @samp{smear}.
9618 @subsection Examples
9622 Add ripple effect to rgb input of video size hd720:
9624 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
9628 Add wave effect to rgb input of video size hd720:
9630 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
9634 @anchor{dnn_processing}
9635 @section dnn_processing
9637 Do image processing with deep neural networks. It works together with another filter
9638 which converts the pixel format of the Frame to what the dnn network requires.
9640 The filter accepts the following options:
9644 Specify which DNN backend to use for model loading and execution. This option accepts
9645 the following values:
9649 Native implementation of DNN loading and execution.
9652 TensorFlow backend. To enable this backend you
9653 need to install the TensorFlow for C library (see
9654 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9655 @code{--enable-libtensorflow}
9658 OpenVINO backend. To enable this backend you
9659 need to build and install the OpenVINO for C library (see
9660 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9661 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9662 be needed if the header files and libraries are not installed into system path)
9666 Default value is @samp{native}.
9669 Set path to model file specifying network architecture and its parameters.
9670 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9671 backend can load files for only its format.
9673 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9676 Set the input name of the dnn network.
9679 Set the output name of the dnn network.
9683 @subsection Examples
9687 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9689 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9693 Halve the pixel value of the frame with format gray32f:
9695 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
9699 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9701 ./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
9705 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9707 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9714 Draw a colored box on the input image.
9716 It accepts the following parameters:
9721 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9725 The expressions which specify the width and height of the box; if 0 they are interpreted as
9726 the input width and height. It defaults to 0.
9729 Specify the color of the box to write. For the general syntax of this option,
9730 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9731 value @code{invert} is used, the box edge color is the same as the
9732 video with inverted luma.
9735 The expression which sets the thickness of the box edge.
9736 A value of @code{fill} will create a filled box. Default value is @code{3}.
9738 See below for the list of accepted constants.
9741 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9742 will overwrite the video's color and alpha pixels.
9743 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9746 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9747 following constants:
9751 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9755 horizontal and vertical chroma subsample values. For example for the
9756 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9760 The input width and height.
9763 The input sample aspect ratio.
9767 The x and y offset coordinates where the box is drawn.
9771 The width and height of the drawn box.
9774 The thickness of the drawn box.
9776 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9777 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9781 @subsection Examples
9785 Draw a black box around the edge of the input image:
9791 Draw a box with color red and an opacity of 50%:
9793 drawbox=10:20:200:60:red@@0.5
9796 The previous example can be specified as:
9798 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9802 Fill the box with pink color:
9804 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9808 Draw a 2-pixel red 2.40:1 mask:
9810 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
9814 @subsection Commands
9815 This filter supports same commands as options.
9816 The command accepts the same syntax of the corresponding option.
9818 If the specified expression is not valid, it is kept at its current
9823 Draw a graph using input video metadata.
9825 It accepts the following parameters:
9829 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9832 Set 1st foreground color expression.
9835 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9838 Set 2nd foreground color expression.
9841 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9844 Set 3rd foreground color expression.
9847 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9850 Set 4th foreground color expression.
9853 Set minimal value of metadata value.
9856 Set maximal value of metadata value.
9859 Set graph background color. Default is white.
9864 Available values for mode is:
9871 Default is @code{line}.
9876 Available values for slide is:
9879 Draw new frame when right border is reached.
9882 Replace old columns with new ones.
9885 Scroll from right to left.
9888 Scroll from left to right.
9891 Draw single picture.
9894 Default is @code{frame}.
9897 Set size of graph video. For the syntax of this option, check the
9898 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9899 The default value is @code{900x256}.
9902 Set the output frame rate. Default value is @code{25}.
9904 The foreground color expressions can use the following variables:
9907 Minimal value of metadata value.
9910 Maximal value of metadata value.
9913 Current metadata key value.
9916 The color is defined as 0xAABBGGRR.
9919 Example using metadata from @ref{signalstats} filter:
9921 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9924 Example using metadata from @ref{ebur128} filter:
9926 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9931 Draw a grid on the input image.
9933 It accepts the following parameters:
9938 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9942 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9943 input width and height, respectively, minus @code{thickness}, so image gets
9944 framed. Default to 0.
9947 Specify the color of the grid. For the general syntax of this option,
9948 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9949 value @code{invert} is used, the grid color is the same as the
9950 video with inverted luma.
9953 The expression which sets the thickness of the grid line. Default value is @code{1}.
9955 See below for the list of accepted constants.
9958 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9959 will overwrite the video's color and alpha pixels.
9960 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9963 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9964 following constants:
9968 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9972 horizontal and vertical chroma subsample values. For example for the
9973 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9977 The input grid cell width and height.
9980 The input sample aspect ratio.
9984 The x and y coordinates of some point of grid intersection (meant to configure offset).
9988 The width and height of the drawn cell.
9991 The thickness of the drawn cell.
9993 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9994 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9998 @subsection Examples
10002 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
10004 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
10008 Draw a white 3x3 grid with an opacity of 50%:
10010 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10014 @subsection Commands
10015 This filter supports same commands as options.
10016 The command accepts the same syntax of the corresponding option.
10018 If the specified expression is not valid, it is kept at its current
10024 Draw a text string or text from a specified file on top of a video, using the
10025 libfreetype library.
10027 To enable compilation of this filter, you need to configure FFmpeg with
10028 @code{--enable-libfreetype}.
10029 To enable default font fallback and the @var{font} option you need to
10030 configure FFmpeg with @code{--enable-libfontconfig}.
10031 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10032 @code{--enable-libfribidi}.
10036 It accepts the following parameters:
10041 Used to draw a box around text using the background color.
10042 The value must be either 1 (enable) or 0 (disable).
10043 The default value of @var{box} is 0.
10046 Set the width of the border to be drawn around the box using @var{boxcolor}.
10047 The default value of @var{boxborderw} is 0.
10050 The color to be used for drawing box around text. For the syntax of this
10051 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10053 The default value of @var{boxcolor} is "white".
10056 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10057 The default value of @var{line_spacing} is 0.
10060 Set the width of the border to be drawn around the text using @var{bordercolor}.
10061 The default value of @var{borderw} is 0.
10064 Set the color to be used for drawing border around text. For the syntax of this
10065 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10067 The default value of @var{bordercolor} is "black".
10070 Select how the @var{text} is expanded. Can be either @code{none},
10071 @code{strftime} (deprecated) or
10072 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10076 Set a start time for the count. Value is in microseconds. Only applied
10077 in the deprecated strftime expansion mode. To emulate in normal expansion
10078 mode use the @code{pts} function, supplying the start time (in seconds)
10079 as the second argument.
10082 If true, check and fix text coords to avoid clipping.
10085 The color to be used for drawing fonts. For the syntax of this option, check
10086 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10088 The default value of @var{fontcolor} is "black".
10090 @item fontcolor_expr
10091 String which is expanded the same way as @var{text} to obtain dynamic
10092 @var{fontcolor} value. By default this option has empty value and is not
10093 processed. When this option is set, it overrides @var{fontcolor} option.
10096 The font family to be used for drawing text. By default Sans.
10099 The font file to be used for drawing text. The path must be included.
10100 This parameter is mandatory if the fontconfig support is disabled.
10103 Draw the text applying alpha blending. The value can
10104 be a number between 0.0 and 1.0.
10105 The expression accepts the same variables @var{x, y} as well.
10106 The default value is 1.
10107 Please see @var{fontcolor_expr}.
10110 The font size to be used for drawing text.
10111 The default value of @var{fontsize} is 16.
10114 If set to 1, attempt to shape the text (for example, reverse the order of
10115 right-to-left text and join Arabic characters) before drawing it.
10116 Otherwise, just draw the text exactly as given.
10117 By default 1 (if supported).
10119 @item ft_load_flags
10120 The flags to be used for loading the fonts.
10122 The flags map the corresponding flags supported by libfreetype, and are
10123 a combination of the following values:
10130 @item vertical_layout
10131 @item force_autohint
10134 @item ignore_global_advance_width
10136 @item ignore_transform
10138 @item linear_design
10142 Default value is "default".
10144 For more information consult the documentation for the FT_LOAD_*
10148 The color to be used for drawing a shadow behind the drawn text. For the
10149 syntax of this option, check the @ref{color syntax,,"Color" section in the
10150 ffmpeg-utils manual,ffmpeg-utils}.
10152 The default value of @var{shadowcolor} is "black".
10156 The x and y offsets for the text shadow position with respect to the
10157 position of the text. They can be either positive or negative
10158 values. The default value for both is "0".
10161 The starting frame number for the n/frame_num variable. The default value
10165 The size in number of spaces to use for rendering the tab.
10166 Default value is 4.
10169 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10170 format. It can be used with or without text parameter. @var{timecode_rate}
10171 option must be specified.
10173 @item timecode_rate, rate, r
10174 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10175 integer. Minimum value is "1".
10176 Drop-frame timecode is supported for frame rates 30 & 60.
10179 If set to 1, the output of the timecode option will wrap around at 24 hours.
10180 Default is 0 (disabled).
10183 The text string to be drawn. The text must be a sequence of UTF-8
10184 encoded characters.
10185 This parameter is mandatory if no file is specified with the parameter
10189 A text file containing text to be drawn. The text must be a sequence
10190 of UTF-8 encoded characters.
10192 This parameter is mandatory if no text string is specified with the
10193 parameter @var{text}.
10195 If both @var{text} and @var{textfile} are specified, an error is thrown.
10198 If set to 1, the @var{textfile} will be reloaded before each frame.
10199 Be sure to update it atomically, or it may be read partially, or even fail.
10203 The expressions which specify the offsets where text will be drawn
10204 within the video frame. They are relative to the top/left border of the
10207 The default value of @var{x} and @var{y} is "0".
10209 See below for the list of accepted constants and functions.
10212 The parameters for @var{x} and @var{y} are expressions containing the
10213 following constants and functions:
10217 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10221 horizontal and vertical chroma subsample values. For example for the
10222 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10225 the height of each text line
10233 @item max_glyph_a, ascent
10234 the maximum distance from the baseline to the highest/upper grid
10235 coordinate used to place a glyph outline point, for all the rendered
10237 It is a positive value, due to the grid's orientation with the Y axis
10240 @item max_glyph_d, descent
10241 the maximum distance from the baseline to the lowest grid coordinate
10242 used to place a glyph outline point, for all the rendered glyphs.
10243 This is a negative value, due to the grid's orientation, with the Y axis
10247 maximum glyph height, that is the maximum height for all the glyphs
10248 contained in the rendered text, it is equivalent to @var{ascent} -
10252 maximum glyph width, that is the maximum width for all the glyphs
10253 contained in the rendered text
10256 the number of input frame, starting from 0
10258 @item rand(min, max)
10259 return a random number included between @var{min} and @var{max}
10262 The input sample aspect ratio.
10265 timestamp expressed in seconds, NAN if the input timestamp is unknown
10268 the height of the rendered text
10271 the width of the rendered text
10275 the x and y offset coordinates where the text is drawn.
10277 These parameters allow the @var{x} and @var{y} expressions to refer
10278 to each other, so you can for example specify @code{y=x/dar}.
10281 A one character description of the current frame's picture type.
10284 The current packet's position in the input file or stream
10285 (in bytes, from the start of the input). A value of -1 indicates
10286 this info is not available.
10289 The current packet's duration, in seconds.
10292 The current packet's size (in bytes).
10295 @anchor{drawtext_expansion}
10296 @subsection Text expansion
10298 If @option{expansion} is set to @code{strftime},
10299 the filter recognizes strftime() sequences in the provided text and
10300 expands them accordingly. Check the documentation of strftime(). This
10301 feature is deprecated.
10303 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10305 If @option{expansion} is set to @code{normal} (which is the default),
10306 the following expansion mechanism is used.
10308 The backslash character @samp{\}, followed by any character, always expands to
10309 the second character.
10311 Sequences of the form @code{%@{...@}} are expanded. The text between the
10312 braces is a function name, possibly followed by arguments separated by ':'.
10313 If the arguments contain special characters or delimiters (':' or '@}'),
10314 they should be escaped.
10316 Note that they probably must also be escaped as the value for the
10317 @option{text} option in the filter argument string and as the filter
10318 argument in the filtergraph description, and possibly also for the shell,
10319 that makes up to four levels of escaping; using a text file avoids these
10322 The following functions are available:
10327 The expression evaluation result.
10329 It must take one argument specifying the expression to be evaluated,
10330 which accepts the same constants and functions as the @var{x} and
10331 @var{y} values. Note that not all constants should be used, for
10332 example the text size is not known when evaluating the expression, so
10333 the constants @var{text_w} and @var{text_h} will have an undefined
10336 @item expr_int_format, eif
10337 Evaluate the expression's value and output as formatted integer.
10339 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10340 The second argument specifies the output format. Allowed values are @samp{x},
10341 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10342 @code{printf} function.
10343 The third parameter is optional and sets the number of positions taken by the output.
10344 It can be used to add padding with zeros from the left.
10347 The time at which the filter is running, expressed in UTC.
10348 It can accept an argument: a strftime() format string.
10351 The time at which the filter is running, expressed in the local time zone.
10352 It can accept an argument: a strftime() format string.
10355 Frame metadata. Takes one or two arguments.
10357 The first argument is mandatory and specifies the metadata key.
10359 The second argument is optional and specifies a default value, used when the
10360 metadata key is not found or empty.
10362 Available metadata can be identified by inspecting entries
10363 starting with TAG included within each frame section
10364 printed by running @code{ffprobe -show_frames}.
10366 String metadata generated in filters leading to
10367 the drawtext filter are also available.
10370 The frame number, starting from 0.
10373 A one character description of the current picture type.
10376 The timestamp of the current frame.
10377 It can take up to three arguments.
10379 The first argument is the format of the timestamp; it defaults to @code{flt}
10380 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10381 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10382 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10383 @code{localtime} stands for the timestamp of the frame formatted as
10384 local time zone time.
10386 The second argument is an offset added to the timestamp.
10388 If the format is set to @code{hms}, a third argument @code{24HH} may be
10389 supplied to present the hour part of the formatted timestamp in 24h format
10392 If the format is set to @code{localtime} or @code{gmtime},
10393 a third argument may be supplied: a strftime() format string.
10394 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10397 @subsection Commands
10399 This filter supports altering parameters via commands:
10402 Alter existing filter parameters.
10404 Syntax for the argument is the same as for filter invocation, e.g.
10407 fontsize=56:fontcolor=green:text='Hello World'
10410 Full filter invocation with sendcmd would look like this:
10413 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10417 If the entire argument can't be parsed or applied as valid values then the filter will
10418 continue with its existing parameters.
10420 @subsection Examples
10424 Draw "Test Text" with font FreeSerif, using the default values for the
10425 optional parameters.
10428 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10432 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10433 and y=50 (counting from the top-left corner of the screen), text is
10434 yellow with a red box around it. Both the text and the box have an
10438 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10439 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10442 Note that the double quotes are not necessary if spaces are not used
10443 within the parameter list.
10446 Show the text at the center of the video frame:
10448 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10452 Show the text at a random position, switching to a new position every 30 seconds:
10454 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)"
10458 Show a text line sliding from right to left in the last row of the video
10459 frame. The file @file{LONG_LINE} is assumed to contain a single line
10462 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10466 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10468 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10472 Draw a single green letter "g", at the center of the input video.
10473 The glyph baseline is placed at half screen height.
10475 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10479 Show text for 1 second every 3 seconds:
10481 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10485 Use fontconfig to set the font. Note that the colons need to be escaped.
10487 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10491 Draw "Test Text" with font size dependent on height of the video.
10493 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10497 Print the date of a real-time encoding (see strftime(3)):
10499 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10503 Show text fading in and out (appearing/disappearing):
10506 DS=1.0 # display start
10507 DE=10.0 # display end
10508 FID=1.5 # fade in duration
10509 FOD=5 # fade out duration
10510 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 @}"
10514 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10515 and the @option{fontsize} value are included in the @option{y} offset.
10517 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10518 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10522 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10523 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10524 must have option @option{-export_path_metadata 1} for the special metadata fields
10525 to be available for filters.
10527 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10532 For more information about libfreetype, check:
10533 @url{http://www.freetype.org/}.
10535 For more information about fontconfig, check:
10536 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10538 For more information about libfribidi, check:
10539 @url{http://fribidi.org/}.
10541 @section edgedetect
10543 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10545 The filter accepts the following options:
10550 Set low and high threshold values used by the Canny thresholding
10553 The high threshold selects the "strong" edge pixels, which are then
10554 connected through 8-connectivity with the "weak" edge pixels selected
10555 by the low threshold.
10557 @var{low} and @var{high} threshold values must be chosen in the range
10558 [0,1], and @var{low} should be lesser or equal to @var{high}.
10560 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10564 Define the drawing mode.
10568 Draw white/gray wires on black background.
10571 Mix the colors to create a paint/cartoon effect.
10574 Apply Canny edge detector on all selected planes.
10576 Default value is @var{wires}.
10579 Select planes for filtering. By default all available planes are filtered.
10582 @subsection Examples
10586 Standard edge detection with custom values for the hysteresis thresholding:
10588 edgedetect=low=0.1:high=0.4
10592 Painting effect without thresholding:
10594 edgedetect=mode=colormix:high=0
10600 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10602 For each input image, the filter will compute the optimal mapping from
10603 the input to the output given the codebook length, that is the number
10604 of distinct output colors.
10606 This filter accepts the following options.
10609 @item codebook_length, l
10610 Set codebook length. The value must be a positive integer, and
10611 represents the number of distinct output colors. Default value is 256.
10614 Set the maximum number of iterations to apply for computing the optimal
10615 mapping. The higher the value the better the result and the higher the
10616 computation time. Default value is 1.
10619 Set a random seed, must be an integer included between 0 and
10620 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10621 will try to use a good random seed on a best effort basis.
10624 Set pal8 output pixel format. This option does not work with codebook
10625 length greater than 256.
10630 Measure graylevel entropy in histogram of color channels of video frames.
10632 It accepts the following parameters:
10636 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10638 @var{diff} mode measures entropy of histogram delta values, absolute differences
10639 between neighbour histogram values.
10643 Set brightness, contrast, saturation and approximate gamma adjustment.
10645 The filter accepts the following options:
10649 Set the contrast expression. The value must be a float value in range
10650 @code{-1000.0} to @code{1000.0}. The default value is "1".
10653 Set the brightness expression. The value must be a float value in
10654 range @code{-1.0} to @code{1.0}. The default value is "0".
10657 Set the saturation expression. The value must be a float in
10658 range @code{0.0} to @code{3.0}. The default value is "1".
10661 Set the gamma expression. The value must be a float in range
10662 @code{0.1} to @code{10.0}. The default value is "1".
10665 Set the gamma expression for red. The value must be a float in
10666 range @code{0.1} to @code{10.0}. The default value is "1".
10669 Set the gamma expression for green. The value must be a float in range
10670 @code{0.1} to @code{10.0}. The default value is "1".
10673 Set the gamma expression for blue. The value must be a float in range
10674 @code{0.1} to @code{10.0}. The default value is "1".
10677 Set the gamma weight expression. It can be used to reduce the effect
10678 of a high gamma value on bright image areas, e.g. keep them from
10679 getting overamplified and just plain white. The value must be a float
10680 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10681 gamma correction all the way down while @code{1.0} leaves it at its
10682 full strength. Default is "1".
10685 Set when the expressions for brightness, contrast, saturation and
10686 gamma expressions are evaluated.
10688 It accepts the following values:
10691 only evaluate expressions once during the filter initialization or
10692 when a command is processed
10695 evaluate expressions for each incoming frame
10698 Default value is @samp{init}.
10701 The expressions accept the following parameters:
10704 frame count of the input frame starting from 0
10707 byte position of the corresponding packet in the input file, NAN if
10711 frame rate of the input video, NAN if the input frame rate is unknown
10714 timestamp expressed in seconds, NAN if the input timestamp is unknown
10717 @subsection Commands
10718 The filter supports the following commands:
10722 Set the contrast expression.
10725 Set the brightness expression.
10728 Set the saturation expression.
10731 Set the gamma expression.
10734 Set the gamma_r expression.
10737 Set gamma_g expression.
10740 Set gamma_b expression.
10743 Set gamma_weight expression.
10745 The command accepts the same syntax of the corresponding option.
10747 If the specified expression is not valid, it is kept at its current
10754 Apply erosion effect to the video.
10756 This filter replaces the pixel by the local(3x3) minimum.
10758 It accepts the following options:
10765 Limit the maximum change for each plane, default is 65535.
10766 If 0, plane will remain unchanged.
10769 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10772 Flags to local 3x3 coordinates maps like this:
10779 @subsection Commands
10781 This filter supports the all above options as @ref{commands}.
10783 @section extractplanes
10785 Extract color channel components from input video stream into
10786 separate grayscale video streams.
10788 The filter accepts the following option:
10792 Set plane(s) to extract.
10794 Available values for planes are:
10805 Choosing planes not available in the input will result in an error.
10806 That means you cannot select @code{r}, @code{g}, @code{b} planes
10807 with @code{y}, @code{u}, @code{v} planes at same time.
10810 @subsection Examples
10814 Extract luma, u and v color channel component from input video frame
10815 into 3 grayscale outputs:
10817 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
10823 Apply a fade-in/out effect to the input video.
10825 It accepts the following parameters:
10829 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10831 Default is @code{in}.
10833 @item start_frame, s
10834 Specify the number of the frame to start applying the fade
10835 effect at. Default is 0.
10838 The number of frames that the fade effect lasts. At the end of the
10839 fade-in effect, the output video will have the same intensity as the input video.
10840 At the end of the fade-out transition, the output video will be filled with the
10841 selected @option{color}.
10845 If set to 1, fade only alpha channel, if one exists on the input.
10846 Default value is 0.
10848 @item start_time, st
10849 Specify the timestamp (in seconds) of the frame to start to apply the fade
10850 effect. If both start_frame and start_time are specified, the fade will start at
10851 whichever comes last. Default is 0.
10854 The number of seconds for which the fade effect has to last. At the end of the
10855 fade-in effect the output video will have the same intensity as the input video,
10856 at the end of the fade-out transition the output video will be filled with the
10857 selected @option{color}.
10858 If both duration and nb_frames are specified, duration is used. Default is 0
10859 (nb_frames is used by default).
10862 Specify the color of the fade. Default is "black".
10865 @subsection Examples
10869 Fade in the first 30 frames of video:
10874 The command above is equivalent to:
10880 Fade out the last 45 frames of a 200-frame video:
10883 fade=type=out:start_frame=155:nb_frames=45
10887 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10889 fade=in:0:25, fade=out:975:25
10893 Make the first 5 frames yellow, then fade in from frame 5-24:
10895 fade=in:5:20:color=yellow
10899 Fade in alpha over first 25 frames of video:
10901 fade=in:0:25:alpha=1
10905 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10907 fade=t=in:st=5.5:d=0.5
10913 Denoise frames using 3D FFT (frequency domain filtering).
10915 The filter accepts the following options:
10919 Set the noise sigma constant. This sets denoising strength.
10920 Default value is 1. Allowed range is from 0 to 30.
10921 Using very high sigma with low overlap may give blocking artifacts.
10924 Set amount of denoising. By default all detected noise is reduced.
10925 Default value is 1. Allowed range is from 0 to 1.
10928 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10929 Actual size of block in pixels is 2 to power of @var{block}, so by default
10930 block size in pixels is 2^4 which is 16.
10933 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10936 Set number of previous frames to use for denoising. By default is set to 0.
10939 Set number of next frames to to use for denoising. By default is set to 0.
10942 Set planes which will be filtered, by default are all available filtered
10947 Apply arbitrary expressions to samples in frequency domain
10951 Adjust the dc value (gain) of the luma plane of the image. The filter
10952 accepts an integer value in range @code{0} to @code{1000}. The default
10953 value is set to @code{0}.
10956 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10957 filter accepts an integer value in range @code{0} to @code{1000}. The
10958 default value is set to @code{0}.
10961 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10962 filter accepts an integer value in range @code{0} to @code{1000}. The
10963 default value is set to @code{0}.
10966 Set the frequency domain weight expression for the luma plane.
10969 Set the frequency domain weight expression for the 1st chroma plane.
10972 Set the frequency domain weight expression for the 2nd chroma plane.
10975 Set when the expressions are evaluated.
10977 It accepts the following values:
10980 Only evaluate expressions once during the filter initialization.
10983 Evaluate expressions for each incoming frame.
10986 Default value is @samp{init}.
10988 The filter accepts the following variables:
10991 The coordinates of the current sample.
10995 The width and height of the image.
10998 The number of input frame, starting from 0.
11001 @subsection Examples
11007 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11013 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11019 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11025 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11032 Extract a single field from an interlaced image using stride
11033 arithmetic to avoid wasting CPU time. The output frames are marked as
11036 The filter accepts the following options:
11040 Specify whether to extract the top (if the value is @code{0} or
11041 @code{top}) or the bottom field (if the value is @code{1} or
11047 Create new frames by copying the top and bottom fields from surrounding frames
11048 supplied as numbers by the hint file.
11052 Set file containing hints: absolute/relative frame numbers.
11054 There must be one line for each frame in a clip. Each line must contain two
11055 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11056 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11057 is current frame number for @code{absolute} mode or out of [-1, 1] range
11058 for @code{relative} mode. First number tells from which frame to pick up top
11059 field and second number tells from which frame to pick up bottom field.
11061 If optionally followed by @code{+} output frame will be marked as interlaced,
11062 else if followed by @code{-} output frame will be marked as progressive, else
11063 it will be marked same as input frame.
11064 If optionally followed by @code{t} output frame will use only top field, or in
11065 case of @code{b} it will use only bottom field.
11066 If line starts with @code{#} or @code{;} that line is skipped.
11069 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11072 Example of first several lines of @code{hint} file for @code{relative} mode:
11074 0,0 - # first frame
11075 1,0 - # second frame, use third's frame top field and second's frame bottom field
11076 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11091 @section fieldmatch
11093 Field matching filter for inverse telecine. It is meant to reconstruct the
11094 progressive frames from a telecined stream. The filter does not drop duplicated
11095 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11096 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11098 The separation of the field matching and the decimation is notably motivated by
11099 the possibility of inserting a de-interlacing filter fallback between the two.
11100 If the source has mixed telecined and real interlaced content,
11101 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11102 But these remaining combed frames will be marked as interlaced, and thus can be
11103 de-interlaced by a later filter such as @ref{yadif} before decimation.
11105 In addition to the various configuration options, @code{fieldmatch} can take an
11106 optional second stream, activated through the @option{ppsrc} option. If
11107 enabled, the frames reconstruction will be based on the fields and frames from
11108 this second stream. This allows the first input to be pre-processed in order to
11109 help the various algorithms of the filter, while keeping the output lossless
11110 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11111 or brightness/contrast adjustments can help.
11113 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11114 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11115 which @code{fieldmatch} is based on. While the semantic and usage are very
11116 close, some behaviour and options names can differ.
11118 The @ref{decimate} filter currently only works for constant frame rate input.
11119 If your input has mixed telecined (30fps) and progressive content with a lower
11120 framerate like 24fps use the following filterchain to produce the necessary cfr
11121 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11123 The filter accepts the following options:
11127 Specify the assumed field order of the input stream. Available values are:
11131 Auto detect parity (use FFmpeg's internal parity value).
11133 Assume bottom field first.
11135 Assume top field first.
11138 Note that it is sometimes recommended not to trust the parity announced by the
11141 Default value is @var{auto}.
11144 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11145 sense that it won't risk creating jerkiness due to duplicate frames when
11146 possible, but if there are bad edits or blended fields it will end up
11147 outputting combed frames when a good match might actually exist. On the other
11148 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11149 but will almost always find a good frame if there is one. The other values are
11150 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11151 jerkiness and creating duplicate frames versus finding good matches in sections
11152 with bad edits, orphaned fields, blended fields, etc.
11154 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11156 Available values are:
11160 2-way matching (p/c)
11162 2-way matching, and trying 3rd match if still combed (p/c + n)
11164 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11166 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11167 still combed (p/c + n + u/b)
11169 3-way matching (p/c/n)
11171 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11172 detected as combed (p/c/n + u/b)
11175 The parenthesis at the end indicate the matches that would be used for that
11176 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11179 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11182 Default value is @var{pc_n}.
11185 Mark the main input stream as a pre-processed input, and enable the secondary
11186 input stream as the clean source to pick the fields from. See the filter
11187 introduction for more details. It is similar to the @option{clip2} feature from
11190 Default value is @code{0} (disabled).
11193 Set the field to match from. It is recommended to set this to the same value as
11194 @option{order} unless you experience matching failures with that setting. In
11195 certain circumstances changing the field that is used to match from can have a
11196 large impact on matching performance. Available values are:
11200 Automatic (same value as @option{order}).
11202 Match from the bottom field.
11204 Match from the top field.
11207 Default value is @var{auto}.
11210 Set whether or not chroma is included during the match comparisons. In most
11211 cases it is recommended to leave this enabled. You should set this to @code{0}
11212 only if your clip has bad chroma problems such as heavy rainbowing or other
11213 artifacts. Setting this to @code{0} could also be used to speed things up at
11214 the cost of some accuracy.
11216 Default value is @code{1}.
11220 These define an exclusion band which excludes the lines between @option{y0} and
11221 @option{y1} from being included in the field matching decision. An exclusion
11222 band can be used to ignore subtitles, a logo, or other things that may
11223 interfere with the matching. @option{y0} sets the starting scan line and
11224 @option{y1} sets the ending line; all lines in between @option{y0} and
11225 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11226 @option{y0} and @option{y1} to the same value will disable the feature.
11227 @option{y0} and @option{y1} defaults to @code{0}.
11230 Set the scene change detection threshold as a percentage of maximum change on
11231 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11232 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11233 @option{scthresh} is @code{[0.0, 100.0]}.
11235 Default value is @code{12.0}.
11238 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11239 account the combed scores of matches when deciding what match to use as the
11240 final match. Available values are:
11244 No final matching based on combed scores.
11246 Combed scores are only used when a scene change is detected.
11248 Use combed scores all the time.
11251 Default is @var{sc}.
11254 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11255 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11256 Available values are:
11260 No forced calculation.
11262 Force p/c/n calculations.
11264 Force p/c/n/u/b calculations.
11267 Default value is @var{none}.
11270 This is the area combing threshold used for combed frame detection. This
11271 essentially controls how "strong" or "visible" combing must be to be detected.
11272 Larger values mean combing must be more visible and smaller values mean combing
11273 can be less visible or strong and still be detected. Valid settings are from
11274 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11275 be detected as combed). This is basically a pixel difference value. A good
11276 range is @code{[8, 12]}.
11278 Default value is @code{9}.
11281 Sets whether or not chroma is considered in the combed frame decision. Only
11282 disable this if your source has chroma problems (rainbowing, etc.) that are
11283 causing problems for the combed frame detection with chroma enabled. Actually,
11284 using @option{chroma}=@var{0} is usually more reliable, except for the case
11285 where there is chroma only combing in the source.
11287 Default value is @code{0}.
11291 Respectively set the x-axis and y-axis size of the window used during combed
11292 frame detection. This has to do with the size of the area in which
11293 @option{combpel} pixels are required to be detected as combed for a frame to be
11294 declared combed. See the @option{combpel} parameter description for more info.
11295 Possible values are any number that is a power of 2 starting at 4 and going up
11298 Default value is @code{16}.
11301 The number of combed pixels inside any of the @option{blocky} by
11302 @option{blockx} size blocks on the frame for the frame to be detected as
11303 combed. While @option{cthresh} controls how "visible" the combing must be, this
11304 setting controls "how much" combing there must be in any localized area (a
11305 window defined by the @option{blockx} and @option{blocky} settings) on the
11306 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11307 which point no frames will ever be detected as combed). This setting is known
11308 as @option{MI} in TFM/VFM vocabulary.
11310 Default value is @code{80}.
11313 @anchor{p/c/n/u/b meaning}
11314 @subsection p/c/n/u/b meaning
11316 @subsubsection p/c/n
11318 We assume the following telecined stream:
11321 Top fields: 1 2 2 3 4
11322 Bottom fields: 1 2 3 4 4
11325 The numbers correspond to the progressive frame the fields relate to. Here, the
11326 first two frames are progressive, the 3rd and 4th are combed, and so on.
11328 When @code{fieldmatch} is configured to run a matching from bottom
11329 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11334 B 1 2 3 4 4 <-- matching reference
11343 As a result of the field matching, we can see that some frames get duplicated.
11344 To perform a complete inverse telecine, you need to rely on a decimation filter
11345 after this operation. See for instance the @ref{decimate} filter.
11347 The same operation now matching from top fields (@option{field}=@var{top})
11352 T 1 2 2 3 4 <-- matching reference
11362 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11363 basically, they refer to the frame and field of the opposite parity:
11366 @item @var{p} matches the field of the opposite parity in the previous frame
11367 @item @var{c} matches the field of the opposite parity in the current frame
11368 @item @var{n} matches the field of the opposite parity in the next frame
11373 The @var{u} and @var{b} matching are a bit special in the sense that they match
11374 from the opposite parity flag. In the following examples, we assume that we are
11375 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11376 'x' is placed above and below each matched fields.
11378 With bottom matching (@option{field}=@var{bottom}):
11383 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11384 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11392 With top matching (@option{field}=@var{top}):
11397 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11398 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11406 @subsection Examples
11408 Simple IVTC of a top field first telecined stream:
11410 fieldmatch=order=tff:combmatch=none, decimate
11413 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11415 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11418 @section fieldorder
11420 Transform the field order of the input video.
11422 It accepts the following parameters:
11427 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11428 for bottom field first.
11431 The default value is @samp{tff}.
11433 The transformation is done by shifting the picture content up or down
11434 by one line, and filling the remaining line with appropriate picture content.
11435 This method is consistent with most broadcast field order converters.
11437 If the input video is not flagged as being interlaced, or it is already
11438 flagged as being of the required output field order, then this filter does
11439 not alter the incoming video.
11441 It is very useful when converting to or from PAL DV material,
11442 which is bottom field first.
11446 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11449 @section fifo, afifo
11451 Buffer input images and send them when they are requested.
11453 It is mainly useful when auto-inserted by the libavfilter
11456 It does not take parameters.
11458 @section fillborders
11460 Fill borders of the input video, without changing video stream dimensions.
11461 Sometimes video can have garbage at the four edges and you may not want to
11462 crop video input to keep size multiple of some number.
11464 This filter accepts the following options:
11468 Number of pixels to fill from left border.
11471 Number of pixels to fill from right border.
11474 Number of pixels to fill from top border.
11477 Number of pixels to fill from bottom border.
11482 It accepts the following values:
11485 fill pixels using outermost pixels
11488 fill pixels using mirroring
11491 fill pixels with constant value
11494 Default is @var{smear}.
11497 Set color for pixels in fixed mode. Default is @var{black}.
11500 @subsection Commands
11501 This filter supports same @ref{commands} as options.
11502 The command accepts the same syntax of the corresponding option.
11504 If the specified expression is not valid, it is kept at its current
11509 Find a rectangular object
11511 It accepts the following options:
11515 Filepath of the object image, needs to be in gray8.
11518 Detection threshold, default is 0.5.
11521 Number of mipmaps, default is 3.
11523 @item xmin, ymin, xmax, ymax
11524 Specifies the rectangle in which to search.
11527 @subsection Examples
11531 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11533 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11539 Flood area with values of same pixel components with another values.
11541 It accepts the following options:
11544 Set pixel x coordinate.
11547 Set pixel y coordinate.
11550 Set source #0 component value.
11553 Set source #1 component value.
11556 Set source #2 component value.
11559 Set source #3 component value.
11562 Set destination #0 component value.
11565 Set destination #1 component value.
11568 Set destination #2 component value.
11571 Set destination #3 component value.
11577 Convert the input video to one of the specified pixel formats.
11578 Libavfilter will try to pick one that is suitable as input to
11581 It accepts the following parameters:
11585 A '|'-separated list of pixel format names, such as
11586 "pix_fmts=yuv420p|monow|rgb24".
11590 @subsection Examples
11594 Convert the input video to the @var{yuv420p} format
11596 format=pix_fmts=yuv420p
11599 Convert the input video to any of the formats in the list
11601 format=pix_fmts=yuv420p|yuv444p|yuv410p
11608 Convert the video to specified constant frame rate by duplicating or dropping
11609 frames as necessary.
11611 It accepts the following parameters:
11615 The desired output frame rate. The default is @code{25}.
11618 Assume the first PTS should be the given value, in seconds. This allows for
11619 padding/trimming at the start of stream. By default, no assumption is made
11620 about the first frame's expected PTS, so no padding or trimming is done.
11621 For example, this could be set to 0 to pad the beginning with duplicates of
11622 the first frame if a video stream starts after the audio stream or to trim any
11623 frames with a negative PTS.
11626 Timestamp (PTS) rounding method.
11628 Possible values are:
11635 round towards -infinity
11637 round towards +infinity
11641 The default is @code{near}.
11644 Action performed when reading the last frame.
11646 Possible values are:
11649 Use same timestamp rounding method as used for other frames.
11651 Pass through last frame if input duration has not been reached yet.
11653 The default is @code{round}.
11657 Alternatively, the options can be specified as a flat string:
11658 @var{fps}[:@var{start_time}[:@var{round}]].
11660 See also the @ref{setpts} filter.
11662 @subsection Examples
11666 A typical usage in order to set the fps to 25:
11672 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11674 fps=fps=film:round=near
11680 Pack two different video streams into a stereoscopic video, setting proper
11681 metadata on supported codecs. The two views should have the same size and
11682 framerate and processing will stop when the shorter video ends. Please note
11683 that you may conveniently adjust view properties with the @ref{scale} and
11686 It accepts the following parameters:
11690 The desired packing format. Supported values are:
11695 The views are next to each other (default).
11698 The views are on top of each other.
11701 The views are packed by line.
11704 The views are packed by column.
11707 The views are temporally interleaved.
11716 # Convert left and right views into a frame-sequential video
11717 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11719 # Convert views into a side-by-side video with the same output resolution as the input
11720 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
11725 Change the frame rate by interpolating new video output frames from the source
11728 This filter is not designed to function correctly with interlaced media. If
11729 you wish to change the frame rate of interlaced media then you are required
11730 to deinterlace before this filter and re-interlace after this filter.
11732 A description of the accepted options follows.
11736 Specify the output frames per second. This option can also be specified
11737 as a value alone. The default is @code{50}.
11740 Specify the start of a range where the output frame will be created as a
11741 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11742 the default is @code{15}.
11745 Specify the end of a range where the output frame will be created as a
11746 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11747 the default is @code{240}.
11750 Specify the level at which a scene change is detected as a value between
11751 0 and 100 to indicate a new scene; a low value reflects a low
11752 probability for the current frame to introduce a new scene, while a higher
11753 value means the current frame is more likely to be one.
11754 The default is @code{8.2}.
11757 Specify flags influencing the filter process.
11759 Available value for @var{flags} is:
11762 @item scene_change_detect, scd
11763 Enable scene change detection using the value of the option @var{scene}.
11764 This flag is enabled by default.
11770 Select one frame every N-th frame.
11772 This filter accepts the following option:
11775 Select frame after every @code{step} frames.
11776 Allowed values are positive integers higher than 0. Default value is @code{1}.
11779 @section freezedetect
11781 Detect frozen video.
11783 This filter logs a message and sets frame metadata when it detects that the
11784 input video has no significant change in content during a specified duration.
11785 Video freeze detection calculates the mean average absolute difference of all
11786 the components of video frames and compares it to a noise floor.
11788 The printed times and duration are expressed in seconds. The
11789 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11790 whose timestamp equals or exceeds the detection duration and it contains the
11791 timestamp of the first frame of the freeze. The
11792 @code{lavfi.freezedetect.freeze_duration} and
11793 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11796 The filter accepts the following options:
11800 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11801 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11805 Set freeze duration until notification (default is 2 seconds).
11808 @section freezeframes
11810 Freeze video frames.
11812 This filter freezes video frames using frame from 2nd input.
11814 The filter accepts the following options:
11818 Set number of first frame from which to start freeze.
11821 Set number of last frame from which to end freeze.
11824 Set number of frame from 2nd input which will be used instead of replaced frames.
11830 Apply a frei0r effect to the input video.
11832 To enable the compilation of this filter, you need to install the frei0r
11833 header and configure FFmpeg with @code{--enable-frei0r}.
11835 It accepts the following parameters:
11840 The name of the frei0r effect to load. If the environment variable
11841 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11842 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11843 Otherwise, the standard frei0r paths are searched, in this order:
11844 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11845 @file{/usr/lib/frei0r-1/}.
11847 @item filter_params
11848 A '|'-separated list of parameters to pass to the frei0r effect.
11852 A frei0r effect parameter can be a boolean (its value is either
11853 "y" or "n"), a double, a color (specified as
11854 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11855 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11856 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11857 a position (specified as @var{X}/@var{Y}, where
11858 @var{X} and @var{Y} are floating point numbers) and/or a string.
11860 The number and types of parameters depend on the loaded effect. If an
11861 effect parameter is not specified, the default value is set.
11863 @subsection Examples
11867 Apply the distort0r effect, setting the first two double parameters:
11869 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11873 Apply the colordistance effect, taking a color as the first parameter:
11875 frei0r=colordistance:0.2/0.3/0.4
11876 frei0r=colordistance:violet
11877 frei0r=colordistance:0x112233
11881 Apply the perspective effect, specifying the top left and top right image
11884 frei0r=perspective:0.2/0.2|0.8/0.2
11888 For more information, see
11889 @url{http://frei0r.dyne.org}
11891 @subsection Commands
11893 This filter supports the @option{filter_params} option as @ref{commands}.
11897 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11899 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11900 processing filter, one of them is performed once per block, not per pixel.
11901 This allows for much higher speed.
11903 The filter accepts the following options:
11907 Set quality. This option defines the number of levels for averaging. It accepts
11908 an integer in the range 4-5. Default value is @code{4}.
11911 Force a constant quantization parameter. It accepts an integer in range 0-63.
11912 If not set, the filter will use the QP from the video stream (if available).
11915 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11916 more details but also more artifacts, while higher values make the image smoother
11917 but also blurrier. Default value is @code{0} − PSNR optimal.
11919 @item use_bframe_qp
11920 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11921 option may cause flicker since the B-Frames have often larger QP. Default is
11922 @code{0} (not enabled).
11928 Apply Gaussian blur filter.
11930 The filter accepts the following options:
11934 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11937 Set number of steps for Gaussian approximation. Default is @code{1}.
11940 Set which planes to filter. By default all planes are filtered.
11943 Set vertical sigma, if negative it will be same as @code{sigma}.
11944 Default is @code{-1}.
11947 @subsection Commands
11948 This filter supports same commands as options.
11949 The command accepts the same syntax of the corresponding option.
11951 If the specified expression is not valid, it is kept at its current
11956 Apply generic equation to each pixel.
11958 The filter accepts the following options:
11961 @item lum_expr, lum
11962 Set the luminance expression.
11964 Set the chrominance blue expression.
11966 Set the chrominance red expression.
11967 @item alpha_expr, a
11968 Set the alpha expression.
11970 Set the red expression.
11971 @item green_expr, g
11972 Set the green expression.
11974 Set the blue expression.
11977 The colorspace is selected according to the specified options. If one
11978 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11979 options is specified, the filter will automatically select a YCbCr
11980 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11981 @option{blue_expr} options is specified, it will select an RGB
11984 If one of the chrominance expression is not defined, it falls back on the other
11985 one. If no alpha expression is specified it will evaluate to opaque value.
11986 If none of chrominance expressions are specified, they will evaluate
11987 to the luminance expression.
11989 The expressions can use the following variables and functions:
11993 The sequential number of the filtered frame, starting from @code{0}.
11997 The coordinates of the current sample.
12001 The width and height of the image.
12005 Width and height scale depending on the currently filtered plane. It is the
12006 ratio between the corresponding luma plane number of pixels and the current
12007 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
12008 @code{0.5,0.5} for chroma planes.
12011 Time of the current frame, expressed in seconds.
12014 Return the value of the pixel at location (@var{x},@var{y}) of the current
12018 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12022 Return the value of the pixel at location (@var{x},@var{y}) of the
12023 blue-difference chroma plane. Return 0 if there is no such plane.
12026 Return the value of the pixel at location (@var{x},@var{y}) of the
12027 red-difference chroma plane. Return 0 if there is no such plane.
12032 Return the value of the pixel at location (@var{x},@var{y}) of the
12033 red/green/blue component. Return 0 if there is no such component.
12036 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12037 plane. Return 0 if there is no such plane.
12039 @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)
12040 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12041 sums of samples within a rectangle. See the functions without the sum postfix.
12043 @item interpolation
12044 Set one of interpolation methods:
12049 Default is bilinear.
12052 For functions, if @var{x} and @var{y} are outside the area, the value will be
12053 automatically clipped to the closer edge.
12055 Please note that this filter can use multiple threads in which case each slice
12056 will have its own expression state. If you want to use only a single expression
12057 state because your expressions depend on previous state then you should limit
12058 the number of filter threads to 1.
12060 @subsection Examples
12064 Flip the image horizontally:
12070 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12071 wavelength of 100 pixels:
12073 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12077 Generate a fancy enigmatic moving light:
12079 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
12083 Generate a quick emboss effect:
12085 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12089 Modify RGB components depending on pixel position:
12091 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12095 Create a radial gradient that is the same size as the input (also see
12096 the @ref{vignette} filter):
12098 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12104 Fix the banding artifacts that are sometimes introduced into nearly flat
12105 regions by truncation to 8-bit color depth.
12106 Interpolate the gradients that should go where the bands are, and
12109 It is designed for playback only. Do not use it prior to
12110 lossy compression, because compression tends to lose the dither and
12111 bring back the bands.
12113 It accepts the following parameters:
12118 The maximum amount by which the filter will change any one pixel. This is also
12119 the threshold for detecting nearly flat regions. Acceptable values range from
12120 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12124 The neighborhood to fit the gradient to. A larger radius makes for smoother
12125 gradients, but also prevents the filter from modifying the pixels near detailed
12126 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12127 values will be clipped to the valid range.
12131 Alternatively, the options can be specified as a flat string:
12132 @var{strength}[:@var{radius}]
12134 @subsection Examples
12138 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12144 Specify radius, omitting the strength (which will fall-back to the default
12152 @anchor{graphmonitor}
12153 @section graphmonitor
12154 Show various filtergraph stats.
12156 With this filter one can debug complete filtergraph.
12157 Especially issues with links filling with queued frames.
12159 The filter accepts the following options:
12163 Set video output size. Default is @var{hd720}.
12166 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12169 Set output mode, can be @var{fulll} or @var{compact}.
12170 In @var{compact} mode only filters with some queued frames have displayed stats.
12173 Set flags which enable which stats are shown in video.
12175 Available values for flags are:
12178 Display number of queued frames in each link.
12180 @item frame_count_in
12181 Display number of frames taken from filter.
12183 @item frame_count_out
12184 Display number of frames given out from filter.
12187 Display current filtered frame pts.
12190 Display current filtered frame time.
12193 Display time base for filter link.
12196 Display used format for filter link.
12199 Display video size or number of audio channels in case of audio used by filter link.
12202 Display video frame rate or sample rate in case of audio used by filter link.
12205 Display link output status.
12209 Set upper limit for video rate of output stream, Default value is @var{25}.
12210 This guarantee that output video frame rate will not be higher than this value.
12214 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12215 and corrects the scene colors accordingly.
12217 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12219 The filter accepts the following options:
12223 The order of differentiation to be applied on the scene. Must be chosen in the range
12224 [0,2] and default value is 1.
12227 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12228 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12229 max value instead of calculating Minkowski distance.
12232 The standard deviation of Gaussian blur to be applied on the scene. Must be
12233 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12234 can't be equal to 0 if @var{difford} is greater than 0.
12237 @subsection Examples
12243 greyedge=difford=1:minknorm=5:sigma=2
12249 greyedge=difford=1:minknorm=0:sigma=2
12257 Apply a Hald CLUT to a video stream.
12259 First input is the video stream to process, and second one is the Hald CLUT.
12260 The Hald CLUT input can be a simple picture or a complete video stream.
12262 The filter accepts the following options:
12266 Force termination when the shortest input terminates. Default is @code{0}.
12268 Continue applying the last CLUT after the end of the stream. A value of
12269 @code{0} disable the filter after the last frame of the CLUT is reached.
12270 Default is @code{1}.
12273 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12274 filters share the same internals).
12276 This filter also supports the @ref{framesync} options.
12278 More information about the Hald CLUT can be found on Eskil Steenberg's website
12279 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12281 @subsection Workflow examples
12283 @subsubsection Hald CLUT video stream
12285 Generate an identity Hald CLUT stream altered with various effects:
12287 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
12290 Note: make sure you use a lossless codec.
12292 Then use it with @code{haldclut} to apply it on some random stream:
12294 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12297 The Hald CLUT will be applied to the 10 first seconds (duration of
12298 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12299 to the remaining frames of the @code{mandelbrot} stream.
12301 @subsubsection Hald CLUT with preview
12303 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12304 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12305 biggest possible square starting at the top left of the picture. The remaining
12306 padding pixels (bottom or right) will be ignored. This area can be used to add
12307 a preview of the Hald CLUT.
12309 Typically, the following generated Hald CLUT will be supported by the
12310 @code{haldclut} filter:
12313 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12314 pad=iw+320 [padded_clut];
12315 smptebars=s=320x256, split [a][b];
12316 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12317 [main][b] overlay=W-320" -frames:v 1 clut.png
12320 It contains the original and a preview of the effect of the CLUT: SMPTE color
12321 bars are displayed on the right-top, and below the same color bars processed by
12324 Then, the effect of this Hald CLUT can be visualized with:
12326 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12331 Flip the input video horizontally.
12333 For example, to horizontally flip the input video with @command{ffmpeg}:
12335 ffmpeg -i in.avi -vf "hflip" out.avi
12339 This filter applies a global color histogram equalization on a
12342 It can be used to correct video that has a compressed range of pixel
12343 intensities. The filter redistributes the pixel intensities to
12344 equalize their distribution across the intensity range. It may be
12345 viewed as an "automatically adjusting contrast filter". This filter is
12346 useful only for correcting degraded or poorly captured source
12349 The filter accepts the following options:
12353 Determine the amount of equalization to be applied. As the strength
12354 is reduced, the distribution of pixel intensities more-and-more
12355 approaches that of the input frame. The value must be a float number
12356 in the range [0,1] and defaults to 0.200.
12359 Set the maximum intensity that can generated and scale the output
12360 values appropriately. The strength should be set as desired and then
12361 the intensity can be limited if needed to avoid washing-out. The value
12362 must be a float number in the range [0,1] and defaults to 0.210.
12365 Set the antibanding level. If enabled the filter will randomly vary
12366 the luminance of output pixels by a small amount to avoid banding of
12367 the histogram. Possible values are @code{none}, @code{weak} or
12368 @code{strong}. It defaults to @code{none}.
12374 Compute and draw a color distribution histogram for the input video.
12376 The computed histogram is a representation of the color component
12377 distribution in an image.
12379 Standard histogram displays the color components distribution in an image.
12380 Displays color graph for each color component. Shows distribution of
12381 the Y, U, V, A or R, G, B components, depending on input format, in the
12382 current frame. Below each graph a color component scale meter is shown.
12384 The filter accepts the following options:
12388 Set height of level. Default value is @code{200}.
12389 Allowed range is [50, 2048].
12392 Set height of color scale. Default value is @code{12}.
12393 Allowed range is [0, 40].
12397 It accepts the following values:
12400 Per color component graphs are placed below each other.
12403 Per color component graphs are placed side by side.
12406 Presents information identical to that in the @code{parade}, except
12407 that the graphs representing color components are superimposed directly
12410 Default is @code{stack}.
12413 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12414 Default is @code{linear}.
12417 Set what color components to display.
12418 Default is @code{7}.
12421 Set foreground opacity. Default is @code{0.7}.
12424 Set background opacity. Default is @code{0.5}.
12427 @subsection Examples
12432 Calculate and draw histogram:
12434 ffplay -i input -vf histogram
12442 This is a high precision/quality 3d denoise filter. It aims to reduce
12443 image noise, producing smooth images and making still images really
12444 still. It should enhance compressibility.
12446 It accepts the following optional parameters:
12450 A non-negative floating point number which specifies spatial luma strength.
12451 It defaults to 4.0.
12453 @item chroma_spatial
12454 A non-negative floating point number which specifies spatial chroma strength.
12455 It defaults to 3.0*@var{luma_spatial}/4.0.
12458 A floating point number which specifies luma temporal strength. It defaults to
12459 6.0*@var{luma_spatial}/4.0.
12462 A floating point number which specifies chroma temporal strength. It defaults to
12463 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12466 @subsection Commands
12467 This filter supports same @ref{commands} as options.
12468 The command accepts the same syntax of the corresponding option.
12470 If the specified expression is not valid, it is kept at its current
12473 @anchor{hwdownload}
12474 @section hwdownload
12476 Download hardware frames to system memory.
12478 The input must be in hardware frames, and the output a non-hardware format.
12479 Not all formats will be supported on the output - it may be necessary to insert
12480 an additional @option{format} filter immediately following in the graph to get
12481 the output in a supported format.
12485 Map hardware frames to system memory or to another device.
12487 This filter has several different modes of operation; which one is used depends
12488 on the input and output formats:
12491 Hardware frame input, normal frame output
12493 Map the input frames to system memory and pass them to the output. If the
12494 original hardware frame is later required (for example, after overlaying
12495 something else on part of it), the @option{hwmap} filter can be used again
12496 in the next mode to retrieve it.
12498 Normal frame input, hardware frame output
12500 If the input is actually a software-mapped hardware frame, then unmap it -
12501 that is, return the original hardware frame.
12503 Otherwise, a device must be provided. Create new hardware surfaces on that
12504 device for the output, then map them back to the software format at the input
12505 and give those frames to the preceding filter. This will then act like the
12506 @option{hwupload} filter, but may be able to avoid an additional copy when
12507 the input is already in a compatible format.
12509 Hardware frame input and output
12511 A device must be supplied for the output, either directly or with the
12512 @option{derive_device} option. The input and output devices must be of
12513 different types and compatible - the exact meaning of this is
12514 system-dependent, but typically it means that they must refer to the same
12515 underlying hardware context (for example, refer to the same graphics card).
12517 If the input frames were originally created on the output device, then unmap
12518 to retrieve the original frames.
12520 Otherwise, map the frames to the output device - create new hardware frames
12521 on the output corresponding to the frames on the input.
12524 The following additional parameters are accepted:
12528 Set the frame mapping mode. Some combination of:
12531 The mapped frame should be readable.
12533 The mapped frame should be writeable.
12535 The mapping will always overwrite the entire frame.
12537 This may improve performance in some cases, as the original contents of the
12538 frame need not be loaded.
12540 The mapping must not involve any copying.
12542 Indirect mappings to copies of frames are created in some cases where either
12543 direct mapping is not possible or it would have unexpected properties.
12544 Setting this flag ensures that the mapping is direct and will fail if that is
12547 Defaults to @var{read+write} if not specified.
12549 @item derive_device @var{type}
12550 Rather than using the device supplied at initialisation, instead derive a new
12551 device of type @var{type} from the device the input frames exist on.
12554 In a hardware to hardware mapping, map in reverse - create frames in the sink
12555 and map them back to the source. This may be necessary in some cases where
12556 a mapping in one direction is required but only the opposite direction is
12557 supported by the devices being used.
12559 This option is dangerous - it may break the preceding filter in undefined
12560 ways if there are any additional constraints on that filter's output.
12561 Do not use it without fully understanding the implications of its use.
12567 Upload system memory frames to hardware surfaces.
12569 The device to upload to must be supplied when the filter is initialised. If
12570 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12571 option or with the @option{derive_device} option. The input and output devices
12572 must be of different types and compatible - the exact meaning of this is
12573 system-dependent, but typically it means that they must refer to the same
12574 underlying hardware context (for example, refer to the same graphics card).
12576 The following additional parameters are accepted:
12579 @item derive_device @var{type}
12580 Rather than using the device supplied at initialisation, instead derive a new
12581 device of type @var{type} from the device the input frames exist on.
12584 @anchor{hwupload_cuda}
12585 @section hwupload_cuda
12587 Upload system memory frames to a CUDA device.
12589 It accepts the following optional parameters:
12593 The number of the CUDA device to use
12598 Apply a high-quality magnification filter designed for pixel art. This filter
12599 was originally created by Maxim Stepin.
12601 It accepts the following option:
12605 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12606 @code{hq3x} and @code{4} for @code{hq4x}.
12607 Default is @code{3}.
12611 Stack input videos horizontally.
12613 All streams must be of same pixel format and of same height.
12615 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12616 to create same output.
12618 The filter accepts the following option:
12622 Set number of input streams. Default is 2.
12625 If set to 1, force the output to terminate when the shortest input
12626 terminates. Default value is 0.
12631 Modify the hue and/or the saturation of the input.
12633 It accepts the following parameters:
12637 Specify the hue angle as a number of degrees. It accepts an expression,
12638 and defaults to "0".
12641 Specify the saturation in the [-10,10] range. It accepts an expression and
12645 Specify the hue angle as a number of radians. It accepts an
12646 expression, and defaults to "0".
12649 Specify the brightness in the [-10,10] range. It accepts an expression and
12653 @option{h} and @option{H} are mutually exclusive, and can't be
12654 specified at the same time.
12656 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12657 expressions containing the following constants:
12661 frame count of the input frame starting from 0
12664 presentation timestamp of the input frame expressed in time base units
12667 frame rate of the input video, NAN if the input frame rate is unknown
12670 timestamp expressed in seconds, NAN if the input timestamp is unknown
12673 time base of the input video
12676 @subsection Examples
12680 Set the hue to 90 degrees and the saturation to 1.0:
12686 Same command but expressing the hue in radians:
12692 Rotate hue and make the saturation swing between 0
12693 and 2 over a period of 1 second:
12695 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12699 Apply a 3 seconds saturation fade-in effect starting at 0:
12701 hue="s=min(t/3\,1)"
12704 The general fade-in expression can be written as:
12706 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12710 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12712 hue="s=max(0\, min(1\, (8-t)/3))"
12715 The general fade-out expression can be written as:
12717 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12722 @subsection Commands
12724 This filter supports the following commands:
12730 Modify the hue and/or the saturation and/or brightness of the input video.
12731 The command accepts the same syntax of the corresponding option.
12733 If the specified expression is not valid, it is kept at its current
12737 @section hysteresis
12739 Grow first stream into second stream by connecting components.
12740 This makes it possible to build more robust edge masks.
12742 This filter accepts the following options:
12746 Set which planes will be processed as bitmap, unprocessed planes will be
12747 copied from first stream.
12748 By default value 0xf, all planes will be processed.
12751 Set threshold which is used in filtering. If pixel component value is higher than
12752 this value filter algorithm for connecting components is activated.
12753 By default value is 0.
12756 The @code{hysteresis} filter also supports the @ref{framesync} options.
12760 Detect video interlacing type.
12762 This filter tries to detect if the input frames are interlaced, progressive,
12763 top or bottom field first. It will also try to detect fields that are
12764 repeated between adjacent frames (a sign of telecine).
12766 Single frame detection considers only immediately adjacent frames when classifying each frame.
12767 Multiple frame detection incorporates the classification history of previous frames.
12769 The filter will log these metadata values:
12772 @item single.current_frame
12773 Detected type of current frame using single-frame detection. One of:
12774 ``tff'' (top field first), ``bff'' (bottom field first),
12775 ``progressive'', or ``undetermined''
12778 Cumulative number of frames detected as top field first using single-frame detection.
12781 Cumulative number of frames detected as top field first using multiple-frame detection.
12784 Cumulative number of frames detected as bottom field first using single-frame detection.
12786 @item multiple.current_frame
12787 Detected type of current frame using multiple-frame detection. One of:
12788 ``tff'' (top field first), ``bff'' (bottom field first),
12789 ``progressive'', or ``undetermined''
12792 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12794 @item single.progressive
12795 Cumulative number of frames detected as progressive using single-frame detection.
12797 @item multiple.progressive
12798 Cumulative number of frames detected as progressive using multiple-frame detection.
12800 @item single.undetermined
12801 Cumulative number of frames that could not be classified using single-frame detection.
12803 @item multiple.undetermined
12804 Cumulative number of frames that could not be classified using multiple-frame detection.
12806 @item repeated.current_frame
12807 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12809 @item repeated.neither
12810 Cumulative number of frames with no repeated field.
12813 Cumulative number of frames with the top field repeated from the previous frame's top field.
12815 @item repeated.bottom
12816 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12819 The filter accepts the following options:
12823 Set interlacing threshold.
12825 Set progressive threshold.
12827 Threshold for repeated field detection.
12829 Number of frames after which a given frame's contribution to the
12830 statistics is halved (i.e., it contributes only 0.5 to its
12831 classification). The default of 0 means that all frames seen are given
12832 full weight of 1.0 forever.
12833 @item analyze_interlaced_flag
12834 When this is not 0 then idet will use the specified number of frames to determine
12835 if the interlaced flag is accurate, it will not count undetermined frames.
12836 If the flag is found to be accurate it will be used without any further
12837 computations, if it is found to be inaccurate it will be cleared without any
12838 further computations. This allows inserting the idet filter as a low computational
12839 method to clean up the interlaced flag
12844 Deinterleave or interleave fields.
12846 This filter allows one to process interlaced images fields without
12847 deinterlacing them. Deinterleaving splits the input frame into 2
12848 fields (so called half pictures). Odd lines are moved to the top
12849 half of the output image, even lines to the bottom half.
12850 You can process (filter) them independently and then re-interleave them.
12852 The filter accepts the following options:
12856 @item chroma_mode, c
12857 @item alpha_mode, a
12858 Available values for @var{luma_mode}, @var{chroma_mode} and
12859 @var{alpha_mode} are:
12865 @item deinterleave, d
12866 Deinterleave fields, placing one above the other.
12868 @item interleave, i
12869 Interleave fields. Reverse the effect of deinterleaving.
12871 Default value is @code{none}.
12873 @item luma_swap, ls
12874 @item chroma_swap, cs
12875 @item alpha_swap, as
12876 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12879 @subsection Commands
12881 This filter supports the all above options as @ref{commands}.
12885 Apply inflate effect to the video.
12887 This filter replaces the pixel by the local(3x3) average by taking into account
12888 only values higher than the pixel.
12890 It accepts the following options:
12897 Limit the maximum change for each plane, default is 65535.
12898 If 0, plane will remain unchanged.
12901 @subsection Commands
12903 This filter supports the all above options as @ref{commands}.
12907 Simple interlacing filter from progressive contents. This interleaves upper (or
12908 lower) lines from odd frames with lower (or upper) lines from even frames,
12909 halving the frame rate and preserving image height.
12912 Original Original New Frame
12913 Frame 'j' Frame 'j+1' (tff)
12914 ========== =========== ==================
12915 Line 0 --------------------> Frame 'j' Line 0
12916 Line 1 Line 1 ----> Frame 'j+1' Line 1
12917 Line 2 ---------------------> Frame 'j' Line 2
12918 Line 3 Line 3 ----> Frame 'j+1' Line 3
12920 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12923 It accepts the following optional parameters:
12927 This determines whether the interlaced frame is taken from the even
12928 (tff - default) or odd (bff) lines of the progressive frame.
12931 Vertical lowpass filter to avoid twitter interlacing and
12932 reduce moire patterns.
12936 Disable vertical lowpass filter
12939 Enable linear filter (default)
12942 Enable complex filter. This will slightly less reduce twitter and moire
12943 but better retain detail and subjective sharpness impression.
12950 Deinterlace input video by applying Donald Graft's adaptive kernel
12951 deinterling. Work on interlaced parts of a video to produce
12952 progressive frames.
12954 The description of the accepted parameters follows.
12958 Set the threshold which affects the filter's tolerance when
12959 determining if a pixel line must be processed. It must be an integer
12960 in the range [0,255] and defaults to 10. A value of 0 will result in
12961 applying the process on every pixels.
12964 Paint pixels exceeding the threshold value to white if set to 1.
12968 Set the fields order. Swap fields if set to 1, leave fields alone if
12972 Enable additional sharpening if set to 1. Default is 0.
12975 Enable twoway sharpening if set to 1. Default is 0.
12978 @subsection Examples
12982 Apply default values:
12984 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12988 Enable additional sharpening:
12994 Paint processed pixels in white:
13002 Slowly update darker pixels.
13004 This filter makes short flashes of light appear longer.
13005 This filter accepts the following options:
13009 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13012 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13015 @section lenscorrection
13017 Correct radial lens distortion
13019 This filter can be used to correct for radial distortion as can result from the use
13020 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13021 one can use tools available for example as part of opencv or simply trial-and-error.
13022 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13023 and extract the k1 and k2 coefficients from the resulting matrix.
13025 Note that effectively the same filter is available in the open-source tools Krita and
13026 Digikam from the KDE project.
13028 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13029 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13030 brightness distribution, so you may want to use both filters together in certain
13031 cases, though you will have to take care of ordering, i.e. whether vignetting should
13032 be applied before or after lens correction.
13034 @subsection Options
13036 The filter accepts the following options:
13040 Relative x-coordinate of the focal point of the image, and thereby the center of the
13041 distortion. This value has a range [0,1] and is expressed as fractions of the image
13042 width. Default is 0.5.
13044 Relative y-coordinate of the focal point of the image, and thereby the center of the
13045 distortion. This value has a range [0,1] and is expressed as fractions of the image
13046 height. Default is 0.5.
13048 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13049 no correction. Default is 0.
13051 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13052 0 means no correction. Default is 0.
13055 The formula that generates the correction is:
13057 @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)
13059 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13060 distances from the focal point in the source and target images, respectively.
13064 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13066 The @code{lensfun} filter requires the camera make, camera model, and lens model
13067 to apply the lens correction. The filter will load the lensfun database and
13068 query it to find the corresponding camera and lens entries in the database. As
13069 long as these entries can be found with the given options, the filter can
13070 perform corrections on frames. Note that incomplete strings will result in the
13071 filter choosing the best match with the given options, and the filter will
13072 output the chosen camera and lens models (logged with level "info"). You must
13073 provide the make, camera model, and lens model as they are required.
13075 The filter accepts the following options:
13079 The make of the camera (for example, "Canon"). This option is required.
13082 The model of the camera (for example, "Canon EOS 100D"). This option is
13086 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13087 option is required.
13090 The type of correction to apply. The following values are valid options:
13094 Enables fixing lens vignetting.
13097 Enables fixing lens geometry. This is the default.
13100 Enables fixing chromatic aberrations.
13103 Enables fixing lens vignetting and lens geometry.
13106 Enables fixing lens vignetting and chromatic aberrations.
13109 Enables fixing both lens geometry and chromatic aberrations.
13112 Enables all possible corrections.
13116 The focal length of the image/video (zoom; expected constant for video). For
13117 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13118 range should be chosen when using that lens. Default 18.
13121 The aperture of the image/video (expected constant for video). Note that
13122 aperture is only used for vignetting correction. Default 3.5.
13124 @item focus_distance
13125 The focus distance of the image/video (expected constant for video). Note that
13126 focus distance is only used for vignetting and only slightly affects the
13127 vignetting correction process. If unknown, leave it at the default value (which
13131 The scale factor which is applied after transformation. After correction the
13132 video is no longer necessarily rectangular. This parameter controls how much of
13133 the resulting image is visible. The value 0 means that a value will be chosen
13134 automatically such that there is little or no unmapped area in the output
13135 image. 1.0 means that no additional scaling is done. Lower values may result
13136 in more of the corrected image being visible, while higher values may avoid
13137 unmapped areas in the output.
13139 @item target_geometry
13140 The target geometry of the output image/video. The following values are valid
13144 @item rectilinear (default)
13147 @item equirectangular
13148 @item fisheye_orthographic
13149 @item fisheye_stereographic
13150 @item fisheye_equisolid
13151 @item fisheye_thoby
13154 Apply the reverse of image correction (instead of correcting distortion, apply
13157 @item interpolation
13158 The type of interpolation used when correcting distortion. The following values
13163 @item linear (default)
13168 @subsection Examples
13172 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13173 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13177 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
13181 Apply the same as before, but only for the first 5 seconds of video.
13184 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
13191 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13192 score between two input videos.
13194 The obtained VMAF score is printed through the logging system.
13196 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13197 After installing the library it can be enabled using:
13198 @code{./configure --enable-libvmaf}.
13199 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13201 The filter has following options:
13205 Set the model path which is to be used for SVM.
13206 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13209 Set the file path to be used to store logs.
13212 Set the format of the log file (csv, json or xml).
13214 @item enable_transform
13215 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13216 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13217 Default value: @code{false}
13220 Invokes the phone model which will generate VMAF scores higher than in the
13221 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13222 Default value: @code{false}
13225 Enables computing psnr along with vmaf.
13226 Default value: @code{false}
13229 Enables computing ssim along with vmaf.
13230 Default value: @code{false}
13233 Enables computing ms_ssim along with vmaf.
13234 Default value: @code{false}
13237 Set the pool method to be used for computing vmaf.
13238 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13241 Set number of threads to be used when computing vmaf.
13242 Default value: @code{0}, which makes use of all available logical processors.
13245 Set interval for frame subsampling used when computing vmaf.
13246 Default value: @code{1}
13248 @item enable_conf_interval
13249 Enables confidence interval.
13250 Default value: @code{false}
13253 This filter also supports the @ref{framesync} options.
13255 @subsection Examples
13258 On the below examples the input file @file{main.mpg} being processed is
13259 compared with the reference file @file{ref.mpg}.
13262 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13266 Example with options:
13268 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13272 Example with options and different containers:
13274 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 -
13280 Limits the pixel components values to the specified range [min, max].
13282 The filter accepts the following options:
13286 Lower bound. Defaults to the lowest allowed value for the input.
13289 Upper bound. Defaults to the highest allowed value for the input.
13292 Specify which planes will be processed. Defaults to all available.
13299 The filter accepts the following options:
13303 Set the number of loops. Setting this value to -1 will result in infinite loops.
13307 Set maximal size in number of frames. Default is 0.
13310 Set first frame of loop. Default is 0.
13313 @subsection Examples
13317 Loop single first frame infinitely:
13319 loop=loop=-1:size=1:start=0
13323 Loop single first frame 10 times:
13325 loop=loop=10:size=1:start=0
13329 Loop 10 first frames 5 times:
13331 loop=loop=5:size=10:start=0
13337 Apply a 1D LUT to an input video.
13339 The filter accepts the following options:
13343 Set the 1D LUT file name.
13345 Currently supported formats:
13354 Select interpolation mode.
13356 Available values are:
13360 Use values from the nearest defined point.
13362 Interpolate values using the linear interpolation.
13364 Interpolate values using the cosine interpolation.
13366 Interpolate values using the cubic interpolation.
13368 Interpolate values using the spline interpolation.
13375 Apply a 3D LUT to an input video.
13377 The filter accepts the following options:
13381 Set the 3D LUT file name.
13383 Currently supported formats:
13397 Select interpolation mode.
13399 Available values are:
13403 Use values from the nearest defined point.
13405 Interpolate values using the 8 points defining a cube.
13407 Interpolate values using a tetrahedron.
13413 Turn certain luma values into transparency.
13415 The filter accepts the following options:
13419 Set the luma which will be used as base for transparency.
13420 Default value is @code{0}.
13423 Set the range of luma values to be keyed out.
13424 Default value is @code{0.01}.
13427 Set the range of softness. Default value is @code{0}.
13428 Use this to control gradual transition from zero to full transparency.
13431 @subsection Commands
13432 This filter supports same @ref{commands} as options.
13433 The command accepts the same syntax of the corresponding option.
13435 If the specified expression is not valid, it is kept at its current
13438 @section lut, lutrgb, lutyuv
13440 Compute a look-up table for binding each pixel component input value
13441 to an output value, and apply it to the input video.
13443 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13444 to an RGB input video.
13446 These filters accept the following parameters:
13449 set first pixel component expression
13451 set second pixel component expression
13453 set third pixel component expression
13455 set fourth pixel component expression, corresponds to the alpha component
13458 set red component expression
13460 set green component expression
13462 set blue component expression
13464 alpha component expression
13467 set Y/luminance component expression
13469 set U/Cb component expression
13471 set V/Cr component expression
13474 Each of them specifies the expression to use for computing the lookup table for
13475 the corresponding pixel component values.
13477 The exact component associated to each of the @var{c*} options depends on the
13480 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13481 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13483 The expressions can contain the following constants and functions:
13488 The input width and height.
13491 The input value for the pixel component.
13494 The input value, clipped to the @var{minval}-@var{maxval} range.
13497 The maximum value for the pixel component.
13500 The minimum value for the pixel component.
13503 The negated value for the pixel component value, clipped to the
13504 @var{minval}-@var{maxval} range; it corresponds to the expression
13505 "maxval-clipval+minval".
13508 The computed value in @var{val}, clipped to the
13509 @var{minval}-@var{maxval} range.
13511 @item gammaval(gamma)
13512 The computed gamma correction value of the pixel component value,
13513 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13515 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13519 All expressions default to "val".
13521 @subsection Examples
13525 Negate input video:
13527 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13528 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13531 The above is the same as:
13533 lutrgb="r=negval:g=negval:b=negval"
13534 lutyuv="y=negval:u=negval:v=negval"
13544 Remove chroma components, turning the video into a graytone image:
13546 lutyuv="u=128:v=128"
13550 Apply a luma burning effect:
13556 Remove green and blue components:
13562 Set a constant alpha channel value on input:
13564 format=rgba,lutrgb=a="maxval-minval/2"
13568 Correct luminance gamma by a factor of 0.5:
13570 lutyuv=y=gammaval(0.5)
13574 Discard least significant bits of luma:
13576 lutyuv=y='bitand(val, 128+64+32)'
13580 Technicolor like effect:
13582 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13586 @section lut2, tlut2
13588 The @code{lut2} filter takes two input streams and outputs one
13591 The @code{tlut2} (time lut2) filter takes two consecutive frames
13592 from one single stream.
13594 This filter accepts the following parameters:
13597 set first pixel component expression
13599 set second pixel component expression
13601 set third pixel component expression
13603 set fourth pixel component expression, corresponds to the alpha component
13606 set output bit depth, only available for @code{lut2} filter. By default is 0,
13607 which means bit depth is automatically picked from first input format.
13610 The @code{lut2} filter also supports the @ref{framesync} options.
13612 Each of them specifies the expression to use for computing the lookup table for
13613 the corresponding pixel component values.
13615 The exact component associated to each of the @var{c*} options depends on the
13618 The expressions can contain the following constants:
13623 The input width and height.
13626 The first input value for the pixel component.
13629 The second input value for the pixel component.
13632 The first input video bit depth.
13635 The second input video bit depth.
13638 All expressions default to "x".
13640 @subsection Examples
13644 Highlight differences between two RGB video streams:
13646 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)'
13650 Highlight differences between two YUV video streams:
13652 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)'
13656 Show max difference between two video streams:
13658 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)))'
13662 @section maskedclamp
13664 Clamp the first input stream with the second input and third input stream.
13666 Returns the value of first stream to be between second input
13667 stream - @code{undershoot} and third input stream + @code{overshoot}.
13669 This filter accepts the following options:
13672 Default value is @code{0}.
13675 Default value is @code{0}.
13678 Set which planes will be processed as bitmap, unprocessed planes will be
13679 copied from first stream.
13680 By default value 0xf, all planes will be processed.
13685 Merge the second and third input stream into output stream using absolute differences
13686 between second input stream and first input stream and absolute difference between
13687 third input stream and first input stream. The picked value will be from second input
13688 stream if second absolute difference is greater than first one or from third input stream
13691 This filter accepts the following options:
13694 Set which planes will be processed as bitmap, unprocessed planes will be
13695 copied from first stream.
13696 By default value 0xf, all planes will be processed.
13699 @section maskedmerge
13701 Merge the first input stream with the second input stream using per pixel
13702 weights in the third input stream.
13704 A value of 0 in the third stream pixel component means that pixel component
13705 from first stream is returned unchanged, while maximum value (eg. 255 for
13706 8-bit videos) means that pixel component from second stream is returned
13707 unchanged. Intermediate values define the amount of merging between both
13708 input stream's pixel components.
13710 This filter accepts the following options:
13713 Set which planes will be processed as bitmap, unprocessed planes will be
13714 copied from first stream.
13715 By default value 0xf, all planes will be processed.
13720 Merge the second and third input stream into output stream using absolute differences
13721 between second input stream and first input stream and absolute difference between
13722 third input stream and first input stream. The picked value will be from second input
13723 stream if second absolute difference is less than first one or from third input stream
13726 This filter accepts the following options:
13729 Set which planes will be processed as bitmap, unprocessed planes will be
13730 copied from first stream.
13731 By default value 0xf, all planes will be processed.
13734 @section maskedthreshold
13735 Pick pixels comparing absolute difference of two video streams with fixed
13738 If absolute difference between pixel component of first and second video
13739 stream is equal or lower than user supplied threshold than pixel component
13740 from first video stream is picked, otherwise pixel component from second
13741 video stream is picked.
13743 This filter accepts the following options:
13746 Set threshold used when picking pixels from absolute difference from two input
13750 Set which planes will be processed as bitmap, unprocessed planes will be
13751 copied from second stream.
13752 By default value 0xf, all planes will be processed.
13756 Create mask from input video.
13758 For example it is useful to create motion masks after @code{tblend} filter.
13760 This filter accepts the following options:
13764 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13767 Set high threshold. Any pixel component higher than this value will be set to max value
13768 allowed for current pixel format.
13771 Set planes to filter, by default all available planes are filtered.
13774 Fill all frame pixels with this value.
13777 Set max average pixel value for frame. If sum of all pixel components is higher that this
13778 average, output frame will be completely filled with value set by @var{fill} option.
13779 Typically useful for scene changes when used in combination with @code{tblend} filter.
13784 Apply motion-compensation deinterlacing.
13786 It needs one field per frame as input and must thus be used together
13787 with yadif=1/3 or equivalent.
13789 This filter accepts the following options:
13792 Set the deinterlacing mode.
13794 It accepts one of the following values:
13799 use iterative motion estimation
13801 like @samp{slow}, but use multiple reference frames.
13803 Default value is @samp{fast}.
13806 Set the picture field parity assumed for the input video. It must be
13807 one of the following values:
13811 assume top field first
13813 assume bottom field first
13816 Default value is @samp{bff}.
13819 Set per-block quantization parameter (QP) used by the internal
13822 Higher values should result in a smoother motion vector field but less
13823 optimal individual vectors. Default value is 1.
13828 Pick median pixel from certain rectangle defined by radius.
13830 This filter accepts the following options:
13834 Set horizontal radius size. Default value is @code{1}.
13835 Allowed range is integer from 1 to 127.
13838 Set which planes to process. Default is @code{15}, which is all available planes.
13841 Set vertical radius size. Default value is @code{0}.
13842 Allowed range is integer from 0 to 127.
13843 If it is 0, value will be picked from horizontal @code{radius} option.
13846 Set median percentile. Default value is @code{0.5}.
13847 Default value of @code{0.5} will pick always median values, while @code{0} will pick
13848 minimum values, and @code{1} maximum values.
13851 @subsection Commands
13852 This filter supports same @ref{commands} as options.
13853 The command accepts the same syntax of the corresponding option.
13855 If the specified expression is not valid, it is kept at its current
13858 @section mergeplanes
13860 Merge color channel components from several video streams.
13862 The filter accepts up to 4 input streams, and merge selected input
13863 planes to the output video.
13865 This filter accepts the following options:
13868 Set input to output plane mapping. Default is @code{0}.
13870 The mappings is specified as a bitmap. It should be specified as a
13871 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13872 mapping for the first plane of the output stream. 'A' sets the number of
13873 the input stream to use (from 0 to 3), and 'a' the plane number of the
13874 corresponding input to use (from 0 to 3). The rest of the mappings is
13875 similar, 'Bb' describes the mapping for the output stream second
13876 plane, 'Cc' describes the mapping for the output stream third plane and
13877 'Dd' describes the mapping for the output stream fourth plane.
13880 Set output pixel format. Default is @code{yuva444p}.
13883 @subsection Examples
13887 Merge three gray video streams of same width and height into single video stream:
13889 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13893 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13895 [a0][a1]mergeplanes=0x00010210:yuva444p
13899 Swap Y and A plane in yuva444p stream:
13901 format=yuva444p,mergeplanes=0x03010200:yuva444p
13905 Swap U and V plane in yuv420p stream:
13907 format=yuv420p,mergeplanes=0x000201:yuv420p
13911 Cast a rgb24 clip to yuv444p:
13913 format=rgb24,mergeplanes=0x000102:yuv444p
13919 Estimate and export motion vectors using block matching algorithms.
13920 Motion vectors are stored in frame side data to be used by other filters.
13922 This filter accepts the following options:
13925 Specify the motion estimation method. Accepts one of the following values:
13929 Exhaustive search algorithm.
13931 Three step search algorithm.
13933 Two dimensional logarithmic search algorithm.
13935 New three step search algorithm.
13937 Four step search algorithm.
13939 Diamond search algorithm.
13941 Hexagon-based search algorithm.
13943 Enhanced predictive zonal search algorithm.
13945 Uneven multi-hexagon search algorithm.
13947 Default value is @samp{esa}.
13950 Macroblock size. Default @code{16}.
13953 Search parameter. Default @code{7}.
13956 @section midequalizer
13958 Apply Midway Image Equalization effect using two video streams.
13960 Midway Image Equalization adjusts a pair of images to have the same
13961 histogram, while maintaining their dynamics as much as possible. It's
13962 useful for e.g. matching exposures from a pair of stereo cameras.
13964 This filter has two inputs and one output, which must be of same pixel format, but
13965 may be of different sizes. The output of filter is first input adjusted with
13966 midway histogram of both inputs.
13968 This filter accepts the following option:
13972 Set which planes to process. Default is @code{15}, which is all available planes.
13975 @section minterpolate
13977 Convert the video to specified frame rate using motion interpolation.
13979 This filter accepts the following options:
13982 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}.
13985 Motion interpolation mode. Following values are accepted:
13988 Duplicate previous or next frame for interpolating new ones.
13990 Blend source frames. Interpolated frame is mean of previous and next frames.
13992 Motion compensated interpolation. Following options are effective when this mode is selected:
13996 Motion compensation mode. Following values are accepted:
13999 Overlapped block motion compensation.
14001 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
14003 Default mode is @samp{obmc}.
14006 Motion estimation mode. Following values are accepted:
14009 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14011 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14013 Default mode is @samp{bilat}.
14016 The algorithm to be used for motion estimation. Following values are accepted:
14019 Exhaustive search algorithm.
14021 Three step search algorithm.
14023 Two dimensional logarithmic search algorithm.
14025 New three step search algorithm.
14027 Four step search algorithm.
14029 Diamond search algorithm.
14031 Hexagon-based search algorithm.
14033 Enhanced predictive zonal search algorithm.
14035 Uneven multi-hexagon search algorithm.
14037 Default algorithm is @samp{epzs}.
14040 Macroblock size. Default @code{16}.
14043 Motion estimation search parameter. Default @code{32}.
14046 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).
14051 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:
14054 Disable scene change detection.
14056 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14058 Default method is @samp{fdiff}.
14060 @item scd_threshold
14061 Scene change detection threshold. Default is @code{10.}.
14066 Mix several video input streams into one video stream.
14068 A description of the accepted options follows.
14072 The number of inputs. If unspecified, it defaults to 2.
14075 Specify weight of each input video stream as sequence.
14076 Each weight is separated by space. If number of weights
14077 is smaller than number of @var{frames} last specified
14078 weight will be used for all remaining unset weights.
14081 Specify scale, if it is set it will be multiplied with sum
14082 of each weight multiplied with pixel values to give final destination
14083 pixel value. By default @var{scale} is auto scaled to sum of weights.
14086 Specify how end of stream is determined.
14089 The duration of the longest input. (default)
14092 The duration of the shortest input.
14095 The duration of the first input.
14099 @section mpdecimate
14101 Drop frames that do not differ greatly from the previous frame in
14102 order to reduce frame rate.
14104 The main use of this filter is for very-low-bitrate encoding
14105 (e.g. streaming over dialup modem), but it could in theory be used for
14106 fixing movies that were inverse-telecined incorrectly.
14108 A description of the accepted options follows.
14112 Set the maximum number of consecutive frames which can be dropped (if
14113 positive), or the minimum interval between dropped frames (if
14114 negative). If the value is 0, the frame is dropped disregarding the
14115 number of previous sequentially dropped frames.
14117 Default value is 0.
14122 Set the dropping threshold values.
14124 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14125 represent actual pixel value differences, so a threshold of 64
14126 corresponds to 1 unit of difference for each pixel, or the same spread
14127 out differently over the block.
14129 A frame is a candidate for dropping if no 8x8 blocks differ by more
14130 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14131 meaning the whole image) differ by more than a threshold of @option{lo}.
14133 Default value for @option{hi} is 64*12, default value for @option{lo} is
14134 64*5, and default value for @option{frac} is 0.33.
14140 Negate (invert) the input video.
14142 It accepts the following option:
14147 With value 1, it negates the alpha component, if present. Default value is 0.
14153 Denoise frames using Non-Local Means algorithm.
14155 Each pixel is adjusted by looking for other pixels with similar contexts. This
14156 context similarity is defined by comparing their surrounding patches of size
14157 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14160 Note that the research area defines centers for patches, which means some
14161 patches will be made of pixels outside that research area.
14163 The filter accepts the following options.
14167 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14170 Set patch size. Default is 7. Must be odd number in range [0, 99].
14173 Same as @option{p} but for chroma planes.
14175 The default value is @var{0} and means automatic.
14178 Set research size. Default is 15. Must be odd number in range [0, 99].
14181 Same as @option{r} but for chroma planes.
14183 The default value is @var{0} and means automatic.
14188 Deinterlace video using neural network edge directed interpolation.
14190 This filter accepts the following options:
14194 Mandatory option, without binary file filter can not work.
14195 Currently file can be found here:
14196 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14199 Set which frames to deinterlace, by default it is @code{all}.
14200 Can be @code{all} or @code{interlaced}.
14203 Set mode of operation.
14205 Can be one of the following:
14209 Use frame flags, both fields.
14211 Use frame flags, single field.
14213 Use top field only.
14215 Use bottom field only.
14217 Use both fields, top first.
14219 Use both fields, bottom first.
14223 Set which planes to process, by default filter process all frames.
14226 Set size of local neighborhood around each pixel, used by the predictor neural
14229 Can be one of the following:
14242 Set the number of neurons in predictor neural network.
14243 Can be one of the following:
14254 Controls the number of different neural network predictions that are blended
14255 together to compute the final output value. Can be @code{fast}, default or
14259 Set which set of weights to use in the predictor.
14260 Can be one of the following:
14264 weights trained to minimize absolute error
14266 weights trained to minimize squared error
14270 Controls whether or not the prescreener neural network is used to decide
14271 which pixels should be processed by the predictor neural network and which
14272 can be handled by simple cubic interpolation.
14273 The prescreener is trained to know whether cubic interpolation will be
14274 sufficient for a pixel or whether it should be predicted by the predictor nn.
14275 The computational complexity of the prescreener nn is much less than that of
14276 the predictor nn. Since most pixels can be handled by cubic interpolation,
14277 using the prescreener generally results in much faster processing.
14278 The prescreener is pretty accurate, so the difference between using it and not
14279 using it is almost always unnoticeable.
14281 Can be one of the following:
14289 Default is @code{new}.
14292 Set various debugging flags.
14297 Force libavfilter not to use any of the specified pixel formats for the
14298 input to the next filter.
14300 It accepts the following parameters:
14304 A '|'-separated list of pixel format names, such as
14305 pix_fmts=yuv420p|monow|rgb24".
14309 @subsection Examples
14313 Force libavfilter to use a format different from @var{yuv420p} for the
14314 input to the vflip filter:
14316 noformat=pix_fmts=yuv420p,vflip
14320 Convert the input video to any of the formats not contained in the list:
14322 noformat=yuv420p|yuv444p|yuv410p
14328 Add noise on video input frame.
14330 The filter accepts the following options:
14338 Set noise seed for specific pixel component or all pixel components in case
14339 of @var{all_seed}. Default value is @code{123457}.
14341 @item all_strength, alls
14342 @item c0_strength, c0s
14343 @item c1_strength, c1s
14344 @item c2_strength, c2s
14345 @item c3_strength, c3s
14346 Set noise strength for specific pixel component or all pixel components in case
14347 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14349 @item all_flags, allf
14350 @item c0_flags, c0f
14351 @item c1_flags, c1f
14352 @item c2_flags, c2f
14353 @item c3_flags, c3f
14354 Set pixel component flags or set flags for all components if @var{all_flags}.
14355 Available values for component flags are:
14358 averaged temporal noise (smoother)
14360 mix random noise with a (semi)regular pattern
14362 temporal noise (noise pattern changes between frames)
14364 uniform noise (gaussian otherwise)
14368 @subsection Examples
14370 Add temporal and uniform noise to input video:
14372 noise=alls=20:allf=t+u
14377 Normalize RGB video (aka histogram stretching, contrast stretching).
14378 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14380 For each channel of each frame, the filter computes the input range and maps
14381 it linearly to the user-specified output range. The output range defaults
14382 to the full dynamic range from pure black to pure white.
14384 Temporal smoothing can be used on the input range to reduce flickering (rapid
14385 changes in brightness) caused when small dark or bright objects enter or leave
14386 the scene. This is similar to the auto-exposure (automatic gain control) on a
14387 video camera, and, like a video camera, it may cause a period of over- or
14388 under-exposure of the video.
14390 The R,G,B channels can be normalized independently, which may cause some
14391 color shifting, or linked together as a single channel, which prevents
14392 color shifting. Linked normalization preserves hue. Independent normalization
14393 does not, so it can be used to remove some color casts. Independent and linked
14394 normalization can be combined in any ratio.
14396 The normalize filter accepts the following options:
14401 Colors which define the output range. The minimum input value is mapped to
14402 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14403 The defaults are black and white respectively. Specifying white for
14404 @var{blackpt} and black for @var{whitept} will give color-inverted,
14405 normalized video. Shades of grey can be used to reduce the dynamic range
14406 (contrast). Specifying saturated colors here can create some interesting
14410 The number of previous frames to use for temporal smoothing. The input range
14411 of each channel is smoothed using a rolling average over the current frame
14412 and the @var{smoothing} previous frames. The default is 0 (no temporal
14416 Controls the ratio of independent (color shifting) channel normalization to
14417 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14418 independent. Defaults to 1.0 (fully independent).
14421 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14422 expensive no-op. Defaults to 1.0 (full strength).
14426 @subsection Commands
14427 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14428 The command accepts the same syntax of the corresponding option.
14430 If the specified expression is not valid, it is kept at its current
14433 @subsection Examples
14435 Stretch video contrast to use the full dynamic range, with no temporal
14436 smoothing; may flicker depending on the source content:
14438 normalize=blackpt=black:whitept=white:smoothing=0
14441 As above, but with 50 frames of temporal smoothing; flicker should be
14442 reduced, depending on the source content:
14444 normalize=blackpt=black:whitept=white:smoothing=50
14447 As above, but with hue-preserving linked channel normalization:
14449 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14452 As above, but with half strength:
14454 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14457 Map the darkest input color to red, the brightest input color to cyan:
14459 normalize=blackpt=red:whitept=cyan
14464 Pass the video source unchanged to the output.
14467 Optical Character Recognition
14469 This filter uses Tesseract for optical character recognition. To enable
14470 compilation of this filter, you need to configure FFmpeg with
14471 @code{--enable-libtesseract}.
14473 It accepts the following options:
14477 Set datapath to tesseract data. Default is to use whatever was
14478 set at installation.
14481 Set language, default is "eng".
14484 Set character whitelist.
14487 Set character blacklist.
14490 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14491 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14495 Apply a video transform using libopencv.
14497 To enable this filter, install the libopencv library and headers and
14498 configure FFmpeg with @code{--enable-libopencv}.
14500 It accepts the following parameters:
14505 The name of the libopencv filter to apply.
14507 @item filter_params
14508 The parameters to pass to the libopencv filter. If not specified, the default
14509 values are assumed.
14513 Refer to the official libopencv documentation for more precise
14515 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14517 Several libopencv filters are supported; see the following subsections.
14522 Dilate an image by using a specific structuring element.
14523 It corresponds to the libopencv function @code{cvDilate}.
14525 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14527 @var{struct_el} represents a structuring element, and has the syntax:
14528 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14530 @var{cols} and @var{rows} represent the number of columns and rows of
14531 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14532 point, and @var{shape} the shape for the structuring element. @var{shape}
14533 must be "rect", "cross", "ellipse", or "custom".
14535 If the value for @var{shape} is "custom", it must be followed by a
14536 string of the form "=@var{filename}". The file with name
14537 @var{filename} is assumed to represent a binary image, with each
14538 printable character corresponding to a bright pixel. When a custom
14539 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14540 or columns and rows of the read file are assumed instead.
14542 The default value for @var{struct_el} is "3x3+0x0/rect".
14544 @var{nb_iterations} specifies the number of times the transform is
14545 applied to the image, and defaults to 1.
14549 # Use the default values
14552 # Dilate using a structuring element with a 5x5 cross, iterating two times
14553 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14555 # Read the shape from the file diamond.shape, iterating two times.
14556 # The file diamond.shape may contain a pattern of characters like this
14562 # The specified columns and rows are ignored
14563 # but the anchor point coordinates are not
14564 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14569 Erode an image by using a specific structuring element.
14570 It corresponds to the libopencv function @code{cvErode}.
14572 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14573 with the same syntax and semantics as the @ref{dilate} filter.
14577 Smooth the input video.
14579 The filter takes the following parameters:
14580 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14582 @var{type} is the type of smooth filter to apply, and must be one of
14583 the following values: "blur", "blur_no_scale", "median", "gaussian",
14584 or "bilateral". The default value is "gaussian".
14586 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14587 depends on the smooth type. @var{param1} and
14588 @var{param2} accept integer positive values or 0. @var{param3} and
14589 @var{param4} accept floating point values.
14591 The default value for @var{param1} is 3. The default value for the
14592 other parameters is 0.
14594 These parameters correspond to the parameters assigned to the
14595 libopencv function @code{cvSmooth}.
14597 @section oscilloscope
14599 2D Video Oscilloscope.
14601 Useful to measure spatial impulse, step responses, chroma delays, etc.
14603 It accepts the following parameters:
14607 Set scope center x position.
14610 Set scope center y position.
14613 Set scope size, relative to frame diagonal.
14616 Set scope tilt/rotation.
14622 Set trace center x position.
14625 Set trace center y position.
14628 Set trace width, relative to width of frame.
14631 Set trace height, relative to height of frame.
14634 Set which components to trace. By default it traces first three components.
14637 Draw trace grid. By default is enabled.
14640 Draw some statistics. By default is enabled.
14643 Draw scope. By default is enabled.
14646 @subsection Commands
14647 This filter supports same @ref{commands} as options.
14648 The command accepts the same syntax of the corresponding option.
14650 If the specified expression is not valid, it is kept at its current
14653 @subsection Examples
14657 Inspect full first row of video frame.
14659 oscilloscope=x=0.5:y=0:s=1
14663 Inspect full last row of video frame.
14665 oscilloscope=x=0.5:y=1:s=1
14669 Inspect full 5th line of video frame of height 1080.
14671 oscilloscope=x=0.5:y=5/1080:s=1
14675 Inspect full last column of video frame.
14677 oscilloscope=x=1:y=0.5:s=1:t=1
14685 Overlay one video on top of another.
14687 It takes two inputs and has one output. The first input is the "main"
14688 video on which the second input is overlaid.
14690 It accepts the following parameters:
14692 A description of the accepted options follows.
14697 Set the expression for the x and y coordinates of the overlaid video
14698 on the main video. Default value is "0" for both expressions. In case
14699 the expression is invalid, it is set to a huge value (meaning that the
14700 overlay will not be displayed within the output visible area).
14703 See @ref{framesync}.
14706 Set when the expressions for @option{x}, and @option{y} are evaluated.
14708 It accepts the following values:
14711 only evaluate expressions once during the filter initialization or
14712 when a command is processed
14715 evaluate expressions for each incoming frame
14718 Default value is @samp{frame}.
14721 See @ref{framesync}.
14724 Set the format for the output video.
14726 It accepts the following values:
14729 force YUV420 output
14732 force YUV420p10 output
14735 force YUV422 output
14738 force YUV422p10 output
14741 force YUV444 output
14744 force packed RGB output
14747 force planar RGB output
14750 automatically pick format
14753 Default value is @samp{yuv420}.
14756 See @ref{framesync}.
14759 Set format of alpha of the overlaid video, it can be @var{straight} or
14760 @var{premultiplied}. Default is @var{straight}.
14763 The @option{x}, and @option{y} expressions can contain the following
14769 The main input width and height.
14773 The overlay input width and height.
14777 The computed values for @var{x} and @var{y}. They are evaluated for
14782 horizontal and vertical chroma subsample values of the output
14783 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14787 the number of input frame, starting from 0
14790 the position in the file of the input frame, NAN if unknown
14793 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14797 This filter also supports the @ref{framesync} options.
14799 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14800 when evaluation is done @emph{per frame}, and will evaluate to NAN
14801 when @option{eval} is set to @samp{init}.
14803 Be aware that frames are taken from each input video in timestamp
14804 order, hence, if their initial timestamps differ, it is a good idea
14805 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14806 have them begin in the same zero timestamp, as the example for
14807 the @var{movie} filter does.
14809 You can chain together more overlays but you should test the
14810 efficiency of such approach.
14812 @subsection Commands
14814 This filter supports the following commands:
14818 Modify the x and y of the overlay input.
14819 The command accepts the same syntax of the corresponding option.
14821 If the specified expression is not valid, it is kept at its current
14825 @subsection Examples
14829 Draw the overlay at 10 pixels from the bottom right corner of the main
14832 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14835 Using named options the example above becomes:
14837 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14841 Insert a transparent PNG logo in the bottom left corner of the input,
14842 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14844 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14848 Insert 2 different transparent PNG logos (second logo on bottom
14849 right corner) using the @command{ffmpeg} tool:
14851 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
14855 Add a transparent color layer on top of the main video; @code{WxH}
14856 must specify the size of the main input to the overlay filter:
14858 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14862 Play an original video and a filtered version (here with the deshake
14863 filter) side by side using the @command{ffplay} tool:
14865 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14868 The above command is the same as:
14870 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14874 Make a sliding overlay appearing from the left to the right top part of the
14875 screen starting since time 2:
14877 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14881 Compose output by putting two input videos side to side:
14883 ffmpeg -i left.avi -i right.avi -filter_complex "
14884 nullsrc=size=200x100 [background];
14885 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14886 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14887 [background][left] overlay=shortest=1 [background+left];
14888 [background+left][right] overlay=shortest=1:x=100 [left+right]
14893 Mask 10-20 seconds of a video by applying the delogo filter to a section
14895 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14896 -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]'
14901 Chain several overlays in cascade:
14903 nullsrc=s=200x200 [bg];
14904 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14905 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14906 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14907 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14908 [in3] null, [mid2] overlay=100:100 [out0]
14913 @anchor{overlay_cuda}
14914 @section overlay_cuda
14916 Overlay one video on top of another.
14918 This is the CUDA variant of the @ref{overlay} filter.
14919 It only accepts CUDA frames. The underlying input pixel formats have to match.
14921 It takes two inputs and has one output. The first input is the "main"
14922 video on which the second input is overlaid.
14924 It accepts the following parameters:
14929 Set the x and y coordinates of the overlaid video on the main video.
14930 Default value is "0" for both expressions.
14933 See @ref{framesync}.
14936 See @ref{framesync}.
14939 See @ref{framesync}.
14943 This filter also supports the @ref{framesync} options.
14947 Apply Overcomplete Wavelet denoiser.
14949 The filter accepts the following options:
14955 Larger depth values will denoise lower frequency components more, but
14956 slow down filtering.
14958 Must be an int in the range 8-16, default is @code{8}.
14960 @item luma_strength, ls
14963 Must be a double value in the range 0-1000, default is @code{1.0}.
14965 @item chroma_strength, cs
14966 Set chroma strength.
14968 Must be a double value in the range 0-1000, default is @code{1.0}.
14974 Add paddings to the input image, and place the original input at the
14975 provided @var{x}, @var{y} coordinates.
14977 It accepts the following parameters:
14982 Specify an expression for the size of the output image with the
14983 paddings added. If the value for @var{width} or @var{height} is 0, the
14984 corresponding input size is used for the output.
14986 The @var{width} expression can reference the value set by the
14987 @var{height} expression, and vice versa.
14989 The default value of @var{width} and @var{height} is 0.
14993 Specify the offsets to place the input image at within the padded area,
14994 with respect to the top/left border of the output image.
14996 The @var{x} expression can reference the value set by the @var{y}
14997 expression, and vice versa.
14999 The default value of @var{x} and @var{y} is 0.
15001 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15002 so the input image is centered on the padded area.
15005 Specify the color of the padded area. For the syntax of this option,
15006 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15007 manual,ffmpeg-utils}.
15009 The default value of @var{color} is "black".
15012 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15014 It accepts the following values:
15018 Only evaluate expressions once during the filter initialization or when
15019 a command is processed.
15022 Evaluate expressions for each incoming frame.
15026 Default value is @samp{init}.
15029 Pad to aspect instead to a resolution.
15033 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15034 options are expressions containing the following constants:
15039 The input video width and height.
15043 These are the same as @var{in_w} and @var{in_h}.
15047 The output width and height (the size of the padded area), as
15048 specified by the @var{width} and @var{height} expressions.
15052 These are the same as @var{out_w} and @var{out_h}.
15056 The x and y offsets as specified by the @var{x} and @var{y}
15057 expressions, or NAN if not yet specified.
15060 same as @var{iw} / @var{ih}
15063 input sample aspect ratio
15066 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15070 The horizontal and vertical chroma subsample values. For example for the
15071 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15074 @subsection Examples
15078 Add paddings with the color "violet" to the input video. The output video
15079 size is 640x480, and the top-left corner of the input video is placed at
15082 pad=640:480:0:40:violet
15085 The example above is equivalent to the following command:
15087 pad=width=640:height=480:x=0:y=40:color=violet
15091 Pad the input to get an output with dimensions increased by 3/2,
15092 and put the input video at the center of the padded area:
15094 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15098 Pad the input to get a squared output with size equal to the maximum
15099 value between the input width and height, and put the input video at
15100 the center of the padded area:
15102 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15106 Pad the input to get a final w/h ratio of 16:9:
15108 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15112 In case of anamorphic video, in order to set the output display aspect
15113 correctly, it is necessary to use @var{sar} in the expression,
15114 according to the relation:
15116 (ih * X / ih) * sar = output_dar
15117 X = output_dar / sar
15120 Thus the previous example needs to be modified to:
15122 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15126 Double the output size and put the input video in the bottom-right
15127 corner of the output padded area:
15129 pad="2*iw:2*ih:ow-iw:oh-ih"
15133 @anchor{palettegen}
15134 @section palettegen
15136 Generate one palette for a whole video stream.
15138 It accepts the following options:
15142 Set the maximum number of colors to quantize in the palette.
15143 Note: the palette will still contain 256 colors; the unused palette entries
15146 @item reserve_transparent
15147 Create a palette of 255 colors maximum and reserve the last one for
15148 transparency. Reserving the transparency color is useful for GIF optimization.
15149 If not set, the maximum of colors in the palette will be 256. You probably want
15150 to disable this option for a standalone image.
15153 @item transparency_color
15154 Set the color that will be used as background for transparency.
15157 Set statistics mode.
15159 It accepts the following values:
15162 Compute full frame histograms.
15164 Compute histograms only for the part that differs from previous frame. This
15165 might be relevant to give more importance to the moving part of your input if
15166 the background is static.
15168 Compute new histogram for each frame.
15171 Default value is @var{full}.
15174 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15175 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15176 color quantization of the palette. This information is also visible at
15177 @var{info} logging level.
15179 @subsection Examples
15183 Generate a representative palette of a given video using @command{ffmpeg}:
15185 ffmpeg -i input.mkv -vf palettegen palette.png
15189 @section paletteuse
15191 Use a palette to downsample an input video stream.
15193 The filter takes two inputs: one video stream and a palette. The palette must
15194 be a 256 pixels image.
15196 It accepts the following options:
15200 Select dithering mode. Available algorithms are:
15203 Ordered 8x8 bayer dithering (deterministic)
15205 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15206 Note: this dithering is sometimes considered "wrong" and is included as a
15208 @item floyd_steinberg
15209 Floyd and Steingberg dithering (error diffusion)
15211 Frankie Sierra dithering v2 (error diffusion)
15213 Frankie Sierra dithering v2 "Lite" (error diffusion)
15216 Default is @var{sierra2_4a}.
15219 When @var{bayer} dithering is selected, this option defines the scale of the
15220 pattern (how much the crosshatch pattern is visible). A low value means more
15221 visible pattern for less banding, and higher value means less visible pattern
15222 at the cost of more banding.
15224 The option must be an integer value in the range [0,5]. Default is @var{2}.
15227 If set, define the zone to process
15231 Only the changing rectangle will be reprocessed. This is similar to GIF
15232 cropping/offsetting compression mechanism. This option can be useful for speed
15233 if only a part of the image is changing, and has use cases such as limiting the
15234 scope of the error diffusal @option{dither} to the rectangle that bounds the
15235 moving scene (it leads to more deterministic output if the scene doesn't change
15236 much, and as a result less moving noise and better GIF compression).
15239 Default is @var{none}.
15242 Take new palette for each output frame.
15244 @item alpha_threshold
15245 Sets the alpha threshold for transparency. Alpha values above this threshold
15246 will be treated as completely opaque, and values below this threshold will be
15247 treated as completely transparent.
15249 The option must be an integer value in the range [0,255]. Default is @var{128}.
15252 @subsection Examples
15256 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15257 using @command{ffmpeg}:
15259 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15263 @section perspective
15265 Correct perspective of video not recorded perpendicular to the screen.
15267 A description of the accepted parameters follows.
15278 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15279 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15280 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15281 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15282 then the corners of the source will be sent to the specified coordinates.
15284 The expressions can use the following variables:
15289 the width and height of video frame.
15293 Output frame count.
15296 @item interpolation
15297 Set interpolation for perspective correction.
15299 It accepts the following values:
15305 Default value is @samp{linear}.
15308 Set interpretation of coordinate options.
15310 It accepts the following values:
15314 Send point in the source specified by the given coordinates to
15315 the corners of the destination.
15317 @item 1, destination
15319 Send the corners of the source to the point in the destination specified
15320 by the given coordinates.
15322 Default value is @samp{source}.
15326 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15328 It accepts the following values:
15331 only evaluate expressions once during the filter initialization or
15332 when a command is processed
15335 evaluate expressions for each incoming frame
15338 Default value is @samp{init}.
15343 Delay interlaced video by one field time so that the field order changes.
15345 The intended use is to fix PAL movies that have been captured with the
15346 opposite field order to the film-to-video transfer.
15348 A description of the accepted parameters follows.
15354 It accepts the following values:
15357 Capture field order top-first, transfer bottom-first.
15358 Filter will delay the bottom field.
15361 Capture field order bottom-first, transfer top-first.
15362 Filter will delay the top field.
15365 Capture and transfer with the same field order. This mode only exists
15366 for the documentation of the other options to refer to, but if you
15367 actually select it, the filter will faithfully do nothing.
15370 Capture field order determined automatically by field flags, transfer
15372 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15373 basis using field flags. If no field information is available,
15374 then this works just like @samp{u}.
15377 Capture unknown or varying, transfer opposite.
15378 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15379 analyzing the images and selecting the alternative that produces best
15380 match between the fields.
15383 Capture top-first, transfer unknown or varying.
15384 Filter selects among @samp{t} and @samp{p} using image analysis.
15387 Capture bottom-first, transfer unknown or varying.
15388 Filter selects among @samp{b} and @samp{p} using image analysis.
15391 Capture determined by field flags, transfer unknown or varying.
15392 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15393 image analysis. If no field information is available, then this works just
15394 like @samp{U}. This is the default mode.
15397 Both capture and transfer unknown or varying.
15398 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15402 @section photosensitivity
15403 Reduce various flashes in video, so to help users with epilepsy.
15405 It accepts the following options:
15408 Set how many frames to use when filtering. Default is 30.
15411 Set detection threshold factor. Default is 1.
15415 Set how many pixels to skip when sampling frames. Default is 1.
15416 Allowed range is from 1 to 1024.
15419 Leave frames unchanged. Default is disabled.
15422 @section pixdesctest
15424 Pixel format descriptor test filter, mainly useful for internal
15425 testing. The output video should be equal to the input video.
15429 format=monow, pixdesctest
15432 can be used to test the monowhite pixel format descriptor definition.
15436 Display sample values of color channels. Mainly useful for checking color
15437 and levels. Minimum supported resolution is 640x480.
15439 The filters accept the following options:
15443 Set scope X position, relative offset on X axis.
15446 Set scope Y position, relative offset on Y axis.
15455 Set window opacity. This window also holds statistics about pixel area.
15458 Set window X position, relative offset on X axis.
15461 Set window Y position, relative offset on Y axis.
15466 Enable the specified chain of postprocessing subfilters using libpostproc. This
15467 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15468 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15469 Each subfilter and some options have a short and a long name that can be used
15470 interchangeably, i.e. dr/dering are the same.
15472 The filters accept the following options:
15476 Set postprocessing subfilters string.
15479 All subfilters share common options to determine their scope:
15483 Honor the quality commands for this subfilter.
15486 Do chrominance filtering, too (default).
15489 Do luminance filtering only (no chrominance).
15492 Do chrominance filtering only (no luminance).
15495 These options can be appended after the subfilter name, separated by a '|'.
15497 Available subfilters are:
15500 @item hb/hdeblock[|difference[|flatness]]
15501 Horizontal deblocking filter
15504 Difference factor where higher values mean more deblocking (default: @code{32}).
15506 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15509 @item vb/vdeblock[|difference[|flatness]]
15510 Vertical deblocking filter
15513 Difference factor where higher values mean more deblocking (default: @code{32}).
15515 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15518 @item ha/hadeblock[|difference[|flatness]]
15519 Accurate horizontal deblocking filter
15522 Difference factor where higher values mean more deblocking (default: @code{32}).
15524 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15527 @item va/vadeblock[|difference[|flatness]]
15528 Accurate vertical deblocking filter
15531 Difference factor where higher values mean more deblocking (default: @code{32}).
15533 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15537 The horizontal and vertical deblocking filters share the difference and
15538 flatness values so you cannot set different horizontal and vertical
15542 @item h1/x1hdeblock
15543 Experimental horizontal deblocking filter
15545 @item v1/x1vdeblock
15546 Experimental vertical deblocking filter
15551 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15554 larger -> stronger filtering
15556 larger -> stronger filtering
15558 larger -> stronger filtering
15561 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15564 Stretch luminance to @code{0-255}.
15567 @item lb/linblenddeint
15568 Linear blend deinterlacing filter that deinterlaces the given block by
15569 filtering all lines with a @code{(1 2 1)} filter.
15571 @item li/linipoldeint
15572 Linear interpolating deinterlacing filter that deinterlaces the given block by
15573 linearly interpolating every second line.
15575 @item ci/cubicipoldeint
15576 Cubic interpolating deinterlacing filter deinterlaces the given block by
15577 cubically interpolating every second line.
15579 @item md/mediandeint
15580 Median deinterlacing filter that deinterlaces the given block by applying a
15581 median filter to every second line.
15583 @item fd/ffmpegdeint
15584 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15585 second line with a @code{(-1 4 2 4 -1)} filter.
15588 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15589 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15591 @item fq/forceQuant[|quantizer]
15592 Overrides the quantizer table from the input with the constant quantizer you
15600 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15603 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15606 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15609 @subsection Examples
15613 Apply horizontal and vertical deblocking, deringing and automatic
15614 brightness/contrast:
15620 Apply default filters without brightness/contrast correction:
15626 Apply default filters and temporal denoiser:
15628 pp=default/tmpnoise|1|2|3
15632 Apply deblocking on luminance only, and switch vertical deblocking on or off
15633 automatically depending on available CPU time:
15640 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15641 similar to spp = 6 with 7 point DCT, where only the center sample is
15644 The filter accepts the following options:
15648 Force a constant quantization parameter. It accepts an integer in range
15649 0 to 63. If not set, the filter will use the QP from the video stream
15653 Set thresholding mode. Available modes are:
15657 Set hard thresholding.
15659 Set soft thresholding (better de-ringing effect, but likely blurrier).
15661 Set medium thresholding (good results, default).
15665 @section premultiply
15666 Apply alpha premultiply effect to input video stream using first plane
15667 of second stream as alpha.
15669 Both streams must have same dimensions and same pixel format.
15671 The filter accepts the following option:
15675 Set which planes will be processed, unprocessed planes will be copied.
15676 By default value 0xf, all planes will be processed.
15679 Do not require 2nd input for processing, instead use alpha plane from input stream.
15683 Apply prewitt operator to input video stream.
15685 The filter accepts the following option:
15689 Set which planes will be processed, unprocessed planes will be copied.
15690 By default value 0xf, all planes will be processed.
15693 Set value which will be multiplied with filtered result.
15696 Set value which will be added to filtered result.
15699 @section pseudocolor
15701 Alter frame colors in video with pseudocolors.
15703 This filter accepts the following options:
15707 set pixel first component expression
15710 set pixel second component expression
15713 set pixel third component expression
15716 set pixel fourth component expression, corresponds to the alpha component
15719 set component to use as base for altering colors
15722 Each of them specifies the expression to use for computing the lookup table for
15723 the corresponding pixel component values.
15725 The expressions can contain the following constants and functions:
15730 The input width and height.
15733 The input value for the pixel component.
15735 @item ymin, umin, vmin, amin
15736 The minimum allowed component value.
15738 @item ymax, umax, vmax, amax
15739 The maximum allowed component value.
15742 All expressions default to "val".
15744 @subsection Examples
15748 Change too high luma values to gradient:
15750 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'"
15756 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15757 Ratio) between two input videos.
15759 This filter takes in input two input videos, the first input is
15760 considered the "main" source and is passed unchanged to the
15761 output. The second input is used as a "reference" video for computing
15764 Both video inputs must have the same resolution and pixel format for
15765 this filter to work correctly. Also it assumes that both inputs
15766 have the same number of frames, which are compared one by one.
15768 The obtained average PSNR is printed through the logging system.
15770 The filter stores the accumulated MSE (mean squared error) of each
15771 frame, and at the end of the processing it is averaged across all frames
15772 equally, and the following formula is applied to obtain the PSNR:
15775 PSNR = 10*log10(MAX^2/MSE)
15778 Where MAX is the average of the maximum values of each component of the
15781 The description of the accepted parameters follows.
15784 @item stats_file, f
15785 If specified the filter will use the named file to save the PSNR of
15786 each individual frame. When filename equals "-" the data is sent to
15789 @item stats_version
15790 Specifies which version of the stats file format to use. Details of
15791 each format are written below.
15792 Default value is 1.
15794 @item stats_add_max
15795 Determines whether the max value is output to the stats log.
15796 Default value is 0.
15797 Requires stats_version >= 2. If this is set and stats_version < 2,
15798 the filter will return an error.
15801 This filter also supports the @ref{framesync} options.
15803 The file printed if @var{stats_file} is selected, contains a sequence of
15804 key/value pairs of the form @var{key}:@var{value} for each compared
15807 If a @var{stats_version} greater than 1 is specified, a header line precedes
15808 the list of per-frame-pair stats, with key value pairs following the frame
15809 format with the following parameters:
15812 @item psnr_log_version
15813 The version of the log file format. Will match @var{stats_version}.
15816 A comma separated list of the per-frame-pair parameters included in
15820 A description of each shown per-frame-pair parameter follows:
15824 sequential number of the input frame, starting from 1
15827 Mean Square Error pixel-by-pixel average difference of the compared
15828 frames, averaged over all the image components.
15830 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15831 Mean Square Error pixel-by-pixel average difference of the compared
15832 frames for the component specified by the suffix.
15834 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15835 Peak Signal to Noise ratio of the compared frames for the component
15836 specified by the suffix.
15838 @item max_avg, max_y, max_u, max_v
15839 Maximum allowed value for each channel, and average over all
15843 @subsection Examples
15848 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15849 [main][ref] psnr="stats_file=stats.log" [out]
15852 On this example the input file being processed is compared with the
15853 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15854 is stored in @file{stats.log}.
15857 Another example with different containers:
15859 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 -
15866 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15867 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15870 The pullup filter is designed to take advantage of future context in making
15871 its decisions. This filter is stateless in the sense that it does not lock
15872 onto a pattern to follow, but it instead looks forward to the following
15873 fields in order to identify matches and rebuild progressive frames.
15875 To produce content with an even framerate, insert the fps filter after
15876 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15877 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15879 The filter accepts the following options:
15886 These options set the amount of "junk" to ignore at the left, right, top, and
15887 bottom of the image, respectively. Left and right are in units of 8 pixels,
15888 while top and bottom are in units of 2 lines.
15889 The default is 8 pixels on each side.
15892 Set the strict breaks. Setting this option to 1 will reduce the chances of
15893 filter generating an occasional mismatched frame, but it may also cause an
15894 excessive number of frames to be dropped during high motion sequences.
15895 Conversely, setting it to -1 will make filter match fields more easily.
15896 This may help processing of video where there is slight blurring between
15897 the fields, but may also cause there to be interlaced frames in the output.
15898 Default value is @code{0}.
15901 Set the metric plane to use. It accepts the following values:
15907 Use chroma blue plane.
15910 Use chroma red plane.
15913 This option may be set to use chroma plane instead of the default luma plane
15914 for doing filter's computations. This may improve accuracy on very clean
15915 source material, but more likely will decrease accuracy, especially if there
15916 is chroma noise (rainbow effect) or any grayscale video.
15917 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15918 load and make pullup usable in realtime on slow machines.
15921 For best results (without duplicated frames in the output file) it is
15922 necessary to change the output frame rate. For example, to inverse
15923 telecine NTSC input:
15925 ffmpeg -i input -vf pullup -r 24000/1001 ...
15930 Change video quantization parameters (QP).
15932 The filter accepts the following option:
15936 Set expression for quantization parameter.
15939 The expression is evaluated through the eval API and can contain, among others,
15940 the following constants:
15944 1 if index is not 129, 0 otherwise.
15947 Sequential index starting from -129 to 128.
15950 @subsection Examples
15954 Some equation like:
15962 Flush video frames from internal cache of frames into a random order.
15963 No frame is discarded.
15964 Inspired by @ref{frei0r} nervous filter.
15968 Set size in number of frames of internal cache, in range from @code{2} to
15969 @code{512}. Default is @code{30}.
15972 Set seed for random number generator, must be an integer included between
15973 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15974 less than @code{0}, the filter will try to use a good random seed on a
15978 @section readeia608
15980 Read closed captioning (EIA-608) information from the top lines of a video frame.
15982 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15983 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15984 with EIA-608 data (starting from 0). A description of each metadata value follows:
15987 @item lavfi.readeia608.X.cc
15988 The two bytes stored as EIA-608 data (printed in hexadecimal).
15990 @item lavfi.readeia608.X.line
15991 The number of the line on which the EIA-608 data was identified and read.
15994 This filter accepts the following options:
15998 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16001 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16004 Set the ratio of width reserved for sync code detection.
16005 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16008 Enable checking the parity bit. In the event of a parity error, the filter will output
16009 @code{0x00} for that character. Default is false.
16012 Lowpass lines prior to further processing. Default is enabled.
16015 @subsection Commands
16017 This filter supports the all above options as @ref{commands}.
16019 @subsection Examples
16023 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16025 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
16031 Read vertical interval timecode (VITC) information from the top lines of a
16034 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16035 timecode value, if a valid timecode has been detected. Further metadata key
16036 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16037 timecode data has been found or not.
16039 This filter accepts the following options:
16043 Set the maximum number of lines to scan for VITC data. If the value is set to
16044 @code{-1} the full video frame is scanned. Default is @code{45}.
16047 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16048 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16051 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16052 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16055 @subsection Examples
16059 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16060 draw @code{--:--:--:--} as a placeholder:
16062 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16068 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16070 Destination pixel at position (X, Y) will be picked from source (x, y) position
16071 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16072 value for pixel will be used for destination pixel.
16074 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16075 will have Xmap/Ymap video stream dimensions.
16076 Xmap and Ymap input video streams are 16bit depth, single channel.
16080 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16081 Default is @code{color}.
16084 Specify the color of the unmapped pixels. For the syntax of this option,
16085 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16086 manual,ffmpeg-utils}. Default color is @code{black}.
16089 @section removegrain
16091 The removegrain filter is a spatial denoiser for progressive video.
16095 Set mode for the first plane.
16098 Set mode for the second plane.
16101 Set mode for the third plane.
16104 Set mode for the fourth plane.
16107 Range of mode is from 0 to 24. Description of each mode follows:
16111 Leave input plane unchanged. Default.
16114 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16117 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16120 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16123 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16124 This is equivalent to a median filter.
16127 Line-sensitive clipping giving the minimal change.
16130 Line-sensitive clipping, intermediate.
16133 Line-sensitive clipping, intermediate.
16136 Line-sensitive clipping, intermediate.
16139 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16142 Replaces the target pixel with the closest neighbour.
16145 [1 2 1] horizontal and vertical kernel blur.
16151 Bob mode, interpolates top field from the line where the neighbours
16152 pixels are the closest.
16155 Bob mode, interpolates bottom field from the line where the neighbours
16156 pixels are the closest.
16159 Bob mode, interpolates top field. Same as 13 but with a more complicated
16160 interpolation formula.
16163 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16164 interpolation formula.
16167 Clips the pixel with the minimum and maximum of respectively the maximum and
16168 minimum of each pair of opposite neighbour pixels.
16171 Line-sensitive clipping using opposite neighbours whose greatest distance from
16172 the current pixel is minimal.
16175 Replaces the pixel with the average of its 8 neighbours.
16178 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16181 Clips pixels using the averages of opposite neighbour.
16184 Same as mode 21 but simpler and faster.
16187 Small edge and halo removal, but reputed useless.
16193 @section removelogo
16195 Suppress a TV station logo, using an image file to determine which
16196 pixels comprise the logo. It works by filling in the pixels that
16197 comprise the logo with neighboring pixels.
16199 The filter accepts the following options:
16203 Set the filter bitmap file, which can be any image format supported by
16204 libavformat. The width and height of the image file must match those of the
16205 video stream being processed.
16208 Pixels in the provided bitmap image with a value of zero are not
16209 considered part of the logo, non-zero pixels are considered part of
16210 the logo. If you use white (255) for the logo and black (0) for the
16211 rest, you will be safe. For making the filter bitmap, it is
16212 recommended to take a screen capture of a black frame with the logo
16213 visible, and then using a threshold filter followed by the erode
16214 filter once or twice.
16216 If needed, little splotches can be fixed manually. Remember that if
16217 logo pixels are not covered, the filter quality will be much
16218 reduced. Marking too many pixels as part of the logo does not hurt as
16219 much, but it will increase the amount of blurring needed to cover over
16220 the image and will destroy more information than necessary, and extra
16221 pixels will slow things down on a large logo.
16223 @section repeatfields
16225 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16226 fields based on its value.
16230 Reverse a video clip.
16232 Warning: This filter requires memory to buffer the entire clip, so trimming
16235 @subsection Examples
16239 Take the first 5 seconds of a clip, and reverse it.
16246 Shift R/G/B/A pixels horizontally and/or vertically.
16248 The filter accepts the following options:
16251 Set amount to shift red horizontally.
16253 Set amount to shift red vertically.
16255 Set amount to shift green horizontally.
16257 Set amount to shift green vertically.
16259 Set amount to shift blue horizontally.
16261 Set amount to shift blue vertically.
16263 Set amount to shift alpha horizontally.
16265 Set amount to shift alpha vertically.
16267 Set edge mode, can be @var{smear}, default, or @var{warp}.
16270 @subsection Commands
16272 This filter supports the all above options as @ref{commands}.
16275 Apply roberts cross operator to input video stream.
16277 The filter accepts the following option:
16281 Set which planes will be processed, unprocessed planes will be copied.
16282 By default value 0xf, all planes will be processed.
16285 Set value which will be multiplied with filtered result.
16288 Set value which will be added to filtered result.
16293 Rotate video by an arbitrary angle expressed in radians.
16295 The filter accepts the following options:
16297 A description of the optional parameters follows.
16300 Set an expression for the angle by which to rotate the input video
16301 clockwise, expressed as a number of radians. A negative value will
16302 result in a counter-clockwise rotation. By default it is set to "0".
16304 This expression is evaluated for each frame.
16307 Set the output width expression, default value is "iw".
16308 This expression is evaluated just once during configuration.
16311 Set the output height expression, default value is "ih".
16312 This expression is evaluated just once during configuration.
16315 Enable bilinear interpolation if set to 1, a value of 0 disables
16316 it. Default value is 1.
16319 Set the color used to fill the output area not covered by the rotated
16320 image. For the general syntax of this option, check the
16321 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16322 If the special value "none" is selected then no
16323 background is printed (useful for example if the background is never shown).
16325 Default value is "black".
16328 The expressions for the angle and the output size can contain the
16329 following constants and functions:
16333 sequential number of the input frame, starting from 0. It is always NAN
16334 before the first frame is filtered.
16337 time in seconds of the input frame, it is set to 0 when the filter is
16338 configured. It is always NAN before the first frame is filtered.
16342 horizontal and vertical chroma subsample values. For example for the
16343 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16347 the input video width and height
16351 the output width and height, that is the size of the padded area as
16352 specified by the @var{width} and @var{height} expressions
16356 the minimal width/height required for completely containing the input
16357 video rotated by @var{a} radians.
16359 These are only available when computing the @option{out_w} and
16360 @option{out_h} expressions.
16363 @subsection Examples
16367 Rotate the input by PI/6 radians clockwise:
16373 Rotate the input by PI/6 radians counter-clockwise:
16379 Rotate the input by 45 degrees clockwise:
16385 Apply a constant rotation with period T, starting from an angle of PI/3:
16387 rotate=PI/3+2*PI*t/T
16391 Make the input video rotation oscillating with a period of T
16392 seconds and an amplitude of A radians:
16394 rotate=A*sin(2*PI/T*t)
16398 Rotate the video, output size is chosen so that the whole rotating
16399 input video is always completely contained in the output:
16401 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16405 Rotate the video, reduce the output size so that no background is ever
16408 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16412 @subsection Commands
16414 The filter supports the following commands:
16418 Set the angle expression.
16419 The command accepts the same syntax of the corresponding option.
16421 If the specified expression is not valid, it is kept at its current
16427 Apply Shape Adaptive Blur.
16429 The filter accepts the following options:
16432 @item luma_radius, lr
16433 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16434 value is 1.0. A greater value will result in a more blurred image, and
16435 in slower processing.
16437 @item luma_pre_filter_radius, lpfr
16438 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16441 @item luma_strength, ls
16442 Set luma maximum difference between pixels to still be considered, must
16443 be a value in the 0.1-100.0 range, default value is 1.0.
16445 @item chroma_radius, cr
16446 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16447 greater value will result in a more blurred image, and in slower
16450 @item chroma_pre_filter_radius, cpfr
16451 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16453 @item chroma_strength, cs
16454 Set chroma maximum difference between pixels to still be considered,
16455 must be a value in the -0.9-100.0 range.
16458 Each chroma option value, if not explicitly specified, is set to the
16459 corresponding luma option value.
16464 Scale (resize) the input video, using the libswscale library.
16466 The scale filter forces the output display aspect ratio to be the same
16467 of the input, by changing the output sample aspect ratio.
16469 If the input image format is different from the format requested by
16470 the next filter, the scale filter will convert the input to the
16473 @subsection Options
16474 The filter accepts the following options, or any of the options
16475 supported by the libswscale scaler.
16477 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16478 the complete list of scaler options.
16483 Set the output video dimension expression. Default value is the input
16486 If the @var{width} or @var{w} value is 0, the input width is used for
16487 the output. If the @var{height} or @var{h} value is 0, the input height
16488 is used for the output.
16490 If one and only one of the values is -n with n >= 1, the scale filter
16491 will use a value that maintains the aspect ratio of the input image,
16492 calculated from the other specified dimension. After that it will,
16493 however, make sure that the calculated dimension is divisible by n and
16494 adjust the value if necessary.
16496 If both values are -n with n >= 1, the behavior will be identical to
16497 both values being set to 0 as previously detailed.
16499 See below for the list of accepted constants for use in the dimension
16503 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16507 Only evaluate expressions once during the filter initialization or when a command is processed.
16510 Evaluate expressions for each incoming frame.
16514 Default value is @samp{init}.
16518 Set the interlacing mode. It accepts the following values:
16522 Force interlaced aware scaling.
16525 Do not apply interlaced scaling.
16528 Select interlaced aware scaling depending on whether the source frames
16529 are flagged as interlaced or not.
16532 Default value is @samp{0}.
16535 Set libswscale scaling flags. See
16536 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16537 complete list of values. If not explicitly specified the filter applies
16541 @item param0, param1
16542 Set libswscale input parameters for scaling algorithms that need them. See
16543 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16544 complete documentation. If not explicitly specified the filter applies
16550 Set the video size. For the syntax of this option, check the
16551 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16553 @item in_color_matrix
16554 @item out_color_matrix
16555 Set in/output YCbCr color space type.
16557 This allows the autodetected value to be overridden as well as allows forcing
16558 a specific value used for the output and encoder.
16560 If not specified, the color space type depends on the pixel format.
16566 Choose automatically.
16569 Format conforming to International Telecommunication Union (ITU)
16570 Recommendation BT.709.
16573 Set color space conforming to the United States Federal Communications
16574 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16579 Set color space conforming to:
16583 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16586 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16589 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16594 Set color space conforming to SMPTE ST 240:1999.
16597 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16602 Set in/output YCbCr sample range.
16604 This allows the autodetected value to be overridden as well as allows forcing
16605 a specific value used for the output and encoder. If not specified, the
16606 range depends on the pixel format. Possible values:
16610 Choose automatically.
16613 Set full range (0-255 in case of 8-bit luma).
16615 @item mpeg/limited/tv
16616 Set "MPEG" range (16-235 in case of 8-bit luma).
16619 @item force_original_aspect_ratio
16620 Enable decreasing or increasing output video width or height if necessary to
16621 keep the original aspect ratio. Possible values:
16625 Scale the video as specified and disable this feature.
16628 The output video dimensions will automatically be decreased if needed.
16631 The output video dimensions will automatically be increased if needed.
16635 One useful instance of this option is that when you know a specific device's
16636 maximum allowed resolution, you can use this to limit the output video to
16637 that, while retaining the aspect ratio. For example, device A allows
16638 1280x720 playback, and your video is 1920x800. Using this option (set it to
16639 decrease) and specifying 1280x720 to the command line makes the output
16642 Please note that this is a different thing than specifying -1 for @option{w}
16643 or @option{h}, you still need to specify the output resolution for this option
16646 @item force_divisible_by
16647 Ensures that both the output dimensions, width and height, are divisible by the
16648 given integer when used together with @option{force_original_aspect_ratio}. This
16649 works similar to using @code{-n} in the @option{w} and @option{h} options.
16651 This option respects the value set for @option{force_original_aspect_ratio},
16652 increasing or decreasing the resolution accordingly. The video's aspect ratio
16653 may be slightly modified.
16655 This option can be handy if you need to have a video fit within or exceed
16656 a defined resolution using @option{force_original_aspect_ratio} but also have
16657 encoder restrictions on width or height divisibility.
16661 The values of the @option{w} and @option{h} options are expressions
16662 containing the following constants:
16667 The input width and height
16671 These are the same as @var{in_w} and @var{in_h}.
16675 The output (scaled) width and height
16679 These are the same as @var{out_w} and @var{out_h}
16682 The same as @var{iw} / @var{ih}
16685 input sample aspect ratio
16688 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16692 horizontal and vertical input chroma subsample values. For example for the
16693 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16697 horizontal and vertical output chroma subsample values. For example for the
16698 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16701 The (sequential) number of the input frame, starting from 0.
16702 Only available with @code{eval=frame}.
16705 The presentation timestamp of the input frame, expressed as a number of
16706 seconds. Only available with @code{eval=frame}.
16709 The position (byte offset) of the frame in the input stream, or NaN if
16710 this information is unavailable and/or meaningless (for example in case of synthetic video).
16711 Only available with @code{eval=frame}.
16714 @subsection Examples
16718 Scale the input video to a size of 200x100
16723 This is equivalent to:
16734 Specify a size abbreviation for the output size:
16739 which can also be written as:
16745 Scale the input to 2x:
16747 scale=w=2*iw:h=2*ih
16751 The above is the same as:
16753 scale=2*in_w:2*in_h
16757 Scale the input to 2x with forced interlaced scaling:
16759 scale=2*iw:2*ih:interl=1
16763 Scale the input to half size:
16765 scale=w=iw/2:h=ih/2
16769 Increase the width, and set the height to the same size:
16775 Seek Greek harmony:
16782 Increase the height, and set the width to 3/2 of the height:
16784 scale=w=3/2*oh:h=3/5*ih
16788 Increase the size, making the size a multiple of the chroma
16791 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16795 Increase the width to a maximum of 500 pixels,
16796 keeping the same aspect ratio as the input:
16798 scale=w='min(500\, iw*3/2):h=-1'
16802 Make pixels square by combining scale and setsar:
16804 scale='trunc(ih*dar):ih',setsar=1/1
16808 Make pixels square by combining scale and setsar,
16809 making sure the resulting resolution is even (required by some codecs):
16811 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16815 @subsection Commands
16817 This filter supports the following commands:
16821 Set the output video dimension expression.
16822 The command accepts the same syntax of the corresponding option.
16824 If the specified expression is not valid, it is kept at its current
16830 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16831 format conversion on CUDA video frames. Setting the output width and height
16832 works in the same way as for the @var{scale} filter.
16834 The following additional options are accepted:
16837 The pixel format of the output CUDA frames. If set to the string "same" (the
16838 default), the input format will be kept. Note that automatic format negotiation
16839 and conversion is not yet supported for hardware frames
16842 The interpolation algorithm used for resizing. One of the following:
16849 @item cubic2p_bspline
16850 2-parameter cubic (B=1, C=0)
16852 @item cubic2p_catmullrom
16853 2-parameter cubic (B=0, C=1/2)
16855 @item cubic2p_b05c03
16856 2-parameter cubic (B=1/2, C=3/10)
16864 @item force_original_aspect_ratio
16865 Enable decreasing or increasing output video width or height if necessary to
16866 keep the original aspect ratio. Possible values:
16870 Scale the video as specified and disable this feature.
16873 The output video dimensions will automatically be decreased if needed.
16876 The output video dimensions will automatically be increased if needed.
16880 One useful instance of this option is that when you know a specific device's
16881 maximum allowed resolution, you can use this to limit the output video to
16882 that, while retaining the aspect ratio. For example, device A allows
16883 1280x720 playback, and your video is 1920x800. Using this option (set it to
16884 decrease) and specifying 1280x720 to the command line makes the output
16887 Please note that this is a different thing than specifying -1 for @option{w}
16888 or @option{h}, you still need to specify the output resolution for this option
16891 @item force_divisible_by
16892 Ensures that both the output dimensions, width and height, are divisible by the
16893 given integer when used together with @option{force_original_aspect_ratio}. This
16894 works similar to using @code{-n} in the @option{w} and @option{h} options.
16896 This option respects the value set for @option{force_original_aspect_ratio},
16897 increasing or decreasing the resolution accordingly. The video's aspect ratio
16898 may be slightly modified.
16900 This option can be handy if you need to have a video fit within or exceed
16901 a defined resolution using @option{force_original_aspect_ratio} but also have
16902 encoder restrictions on width or height divisibility.
16908 Scale (resize) the input video, based on a reference video.
16910 See the scale filter for available options, scale2ref supports the same but
16911 uses the reference video instead of the main input as basis. scale2ref also
16912 supports the following additional constants for the @option{w} and
16913 @option{h} options:
16918 The main input video's width and height
16921 The same as @var{main_w} / @var{main_h}
16924 The main input video's sample aspect ratio
16926 @item main_dar, mdar
16927 The main input video's display aspect ratio. Calculated from
16928 @code{(main_w / main_h) * main_sar}.
16932 The main input video's horizontal and vertical chroma subsample values.
16933 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16937 The (sequential) number of the main input frame, starting from 0.
16938 Only available with @code{eval=frame}.
16941 The presentation timestamp of the main input frame, expressed as a number of
16942 seconds. Only available with @code{eval=frame}.
16945 The position (byte offset) of the frame in the main input stream, or NaN if
16946 this information is unavailable and/or meaningless (for example in case of synthetic video).
16947 Only available with @code{eval=frame}.
16950 @subsection Examples
16954 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16956 'scale2ref[b][a];[a][b]overlay'
16960 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16962 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16966 @subsection Commands
16968 This filter supports the following commands:
16972 Set the output video dimension expression.
16973 The command accepts the same syntax of the corresponding option.
16975 If the specified expression is not valid, it is kept at its current
16980 Scroll input video horizontally and/or vertically by constant speed.
16982 The filter accepts the following options:
16984 @item horizontal, h
16985 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16986 Negative values changes scrolling direction.
16989 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16990 Negative values changes scrolling direction.
16993 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16996 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16999 @subsection Commands
17001 This filter supports the following @ref{commands}:
17003 @item horizontal, h
17004 Set the horizontal scrolling speed.
17006 Set the vertical scrolling speed.
17012 Detect video scene change.
17014 This filter sets frame metadata with mafd between frame, the scene score, and
17015 forward the frame to the next filter, so they can use these metadata to detect
17016 scene change or others.
17018 In addition, this filter logs a message and sets frame metadata when it detects
17019 a scene change by @option{threshold}.
17021 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17023 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17024 to detect scene change.
17026 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17027 detect scene change with @option{threshold}.
17029 The filter accepts the following options:
17033 Set the scene change detection threshold as a percentage of maximum change. Good
17034 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17037 Default value is @code{10.}.
17040 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17041 You can enable it if you want to get snapshot of scene change frames only.
17044 @anchor{selectivecolor}
17045 @section selectivecolor
17047 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17048 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17049 by the "purity" of the color (that is, how saturated it already is).
17051 This filter is similar to the Adobe Photoshop Selective Color tool.
17053 The filter accepts the following options:
17056 @item correction_method
17057 Select color correction method.
17059 Available values are:
17062 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17065 Specified adjustments are relative to the original component value.
17067 Default is @code{absolute}.
17069 Adjustments for red pixels (pixels where the red component is the maximum)
17071 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17073 Adjustments for green pixels (pixels where the green component is the maximum)
17075 Adjustments for cyan pixels (pixels where the red component is the minimum)
17077 Adjustments for blue pixels (pixels where the blue component is the maximum)
17079 Adjustments for magenta pixels (pixels where the green component is the minimum)
17081 Adjustments for white pixels (pixels where all components are greater than 128)
17083 Adjustments for all pixels except pure black and pure white
17085 Adjustments for black pixels (pixels where all components are lesser than 128)
17087 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17090 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17091 4 space separated floating point adjustment values in the [-1,1] range,
17092 respectively to adjust the amount of cyan, magenta, yellow and black for the
17093 pixels of its range.
17095 @subsection Examples
17099 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17100 increase magenta by 27% in blue areas:
17102 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17106 Use a Photoshop selective color preset:
17108 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17112 @anchor{separatefields}
17113 @section separatefields
17115 The @code{separatefields} takes a frame-based video input and splits
17116 each frame into its components fields, producing a new half height clip
17117 with twice the frame rate and twice the frame count.
17119 This filter use field-dominance information in frame to decide which
17120 of each pair of fields to place first in the output.
17121 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17123 @section setdar, setsar
17125 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17128 This is done by changing the specified Sample (aka Pixel) Aspect
17129 Ratio, according to the following equation:
17131 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17134 Keep in mind that the @code{setdar} filter does not modify the pixel
17135 dimensions of the video frame. Also, the display aspect ratio set by
17136 this filter may be changed by later filters in the filterchain,
17137 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17140 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17141 the filter output video.
17143 Note that as a consequence of the application of this filter, the
17144 output display aspect ratio will change according to the equation
17147 Keep in mind that the sample aspect ratio set by the @code{setsar}
17148 filter may be changed by later filters in the filterchain, e.g. if
17149 another "setsar" or a "setdar" filter is applied.
17151 It accepts the following parameters:
17154 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17155 Set the aspect ratio used by the filter.
17157 The parameter can be a floating point number string, an expression, or
17158 a string of the form @var{num}:@var{den}, where @var{num} and
17159 @var{den} are the numerator and denominator of the aspect ratio. If
17160 the parameter is not specified, it is assumed the value "0".
17161 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17165 Set the maximum integer value to use for expressing numerator and
17166 denominator when reducing the expressed aspect ratio to a rational.
17167 Default value is @code{100}.
17171 The parameter @var{sar} is an expression containing
17172 the following constants:
17176 These are approximated values for the mathematical constants e
17177 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17180 The input width and height.
17183 These are the same as @var{w} / @var{h}.
17186 The input sample aspect ratio.
17189 The input display aspect ratio. It is the same as
17190 (@var{w} / @var{h}) * @var{sar}.
17193 Horizontal and vertical chroma subsample values. For example, for the
17194 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17197 @subsection Examples
17202 To change the display aspect ratio to 16:9, specify one of the following:
17209 To change the sample aspect ratio to 10:11, specify:
17215 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17216 1000 in the aspect ratio reduction, use the command:
17218 setdar=ratio=16/9:max=1000
17226 Force field for the output video frame.
17228 The @code{setfield} filter marks the interlace type field for the
17229 output frames. It does not change the input frame, but only sets the
17230 corresponding property, which affects how the frame is treated by
17231 following filters (e.g. @code{fieldorder} or @code{yadif}).
17233 The filter accepts the following options:
17238 Available values are:
17242 Keep the same field property.
17245 Mark the frame as bottom-field-first.
17248 Mark the frame as top-field-first.
17251 Mark the frame as progressive.
17258 Force frame parameter for the output video frame.
17260 The @code{setparams} filter marks interlace and color range for the
17261 output frames. It does not change the input frame, but only sets the
17262 corresponding property, which affects how the frame is treated by
17267 Available values are:
17271 Keep the same field property (default).
17274 Mark the frame as bottom-field-first.
17277 Mark the frame as top-field-first.
17280 Mark the frame as progressive.
17284 Available values are:
17288 Keep the same color range property (default).
17290 @item unspecified, unknown
17291 Mark the frame as unspecified color range.
17293 @item limited, tv, mpeg
17294 Mark the frame as limited range.
17296 @item full, pc, jpeg
17297 Mark the frame as full range.
17300 @item color_primaries
17301 Set the color primaries.
17302 Available values are:
17306 Keep the same color primaries property (default).
17323 Set the color transfer.
17324 Available values are:
17328 Keep the same color trc property (default).
17350 Set the colorspace.
17351 Available values are:
17355 Keep the same colorspace property (default).
17368 @item chroma-derived-nc
17369 @item chroma-derived-c
17376 Show a line containing various information for each input video frame.
17377 The input video is not modified.
17379 This filter supports the following options:
17383 Calculate checksums of each plane. By default enabled.
17386 The shown line contains a sequence of key/value pairs of the form
17387 @var{key}:@var{value}.
17389 The following values are shown in the output:
17393 The (sequential) number of the input frame, starting from 0.
17396 The Presentation TimeStamp of the input frame, expressed as a number of
17397 time base units. The time base unit depends on the filter input pad.
17400 The Presentation TimeStamp of the input frame, expressed as a number of
17404 The position of the frame in the input stream, or -1 if this information is
17405 unavailable and/or meaningless (for example in case of synthetic video).
17408 The pixel format name.
17411 The sample aspect ratio of the input frame, expressed in the form
17412 @var{num}/@var{den}.
17415 The size of the input frame. For the syntax of this option, check the
17416 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17419 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17420 for bottom field first).
17423 This is 1 if the frame is a key frame, 0 otherwise.
17426 The picture type of the input frame ("I" for an I-frame, "P" for a
17427 P-frame, "B" for a B-frame, or "?" for an unknown type).
17428 Also refer to the documentation of the @code{AVPictureType} enum and of
17429 the @code{av_get_picture_type_char} function defined in
17430 @file{libavutil/avutil.h}.
17433 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17435 @item plane_checksum
17436 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17437 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17440 The mean value of pixels in each plane of the input frame, expressed in the form
17441 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17444 The standard deviation of pixel values in each plane of the input frame, expressed
17445 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17449 @section showpalette
17451 Displays the 256 colors palette of each frame. This filter is only relevant for
17452 @var{pal8} pixel format frames.
17454 It accepts the following option:
17458 Set the size of the box used to represent one palette color entry. Default is
17459 @code{30} (for a @code{30x30} pixel box).
17462 @section shuffleframes
17464 Reorder and/or duplicate and/or drop video frames.
17466 It accepts the following parameters:
17470 Set the destination indexes of input frames.
17471 This is space or '|' separated list of indexes that maps input frames to output
17472 frames. Number of indexes also sets maximal value that each index may have.
17473 '-1' index have special meaning and that is to drop frame.
17476 The first frame has the index 0. The default is to keep the input unchanged.
17478 @subsection Examples
17482 Swap second and third frame of every three frames of the input:
17484 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17488 Swap 10th and 1st frame of every ten frames of the input:
17490 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17494 @section shuffleplanes
17496 Reorder and/or duplicate video planes.
17498 It accepts the following parameters:
17503 The index of the input plane to be used as the first output plane.
17506 The index of the input plane to be used as the second output plane.
17509 The index of the input plane to be used as the third output plane.
17512 The index of the input plane to be used as the fourth output plane.
17516 The first plane has the index 0. The default is to keep the input unchanged.
17518 @subsection Examples
17522 Swap the second and third planes of the input:
17524 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17528 @anchor{signalstats}
17529 @section signalstats
17530 Evaluate various visual metrics that assist in determining issues associated
17531 with the digitization of analog video media.
17533 By default the filter will log these metadata values:
17537 Display the minimal Y value contained within the input frame. Expressed in
17541 Display the Y value at the 10% percentile within the input frame. Expressed in
17545 Display the average Y value within the input frame. Expressed in range of
17549 Display the Y value at the 90% percentile within the input frame. Expressed in
17553 Display the maximum Y value contained within the input frame. Expressed in
17557 Display the minimal U value contained within the input frame. Expressed in
17561 Display the U value at the 10% percentile within the input frame. Expressed in
17565 Display the average U value within the input frame. Expressed in range of
17569 Display the U value at the 90% percentile within the input frame. Expressed in
17573 Display the maximum U value contained within the input frame. Expressed in
17577 Display the minimal V value contained within the input frame. Expressed in
17581 Display the V value at the 10% percentile within the input frame. Expressed in
17585 Display the average V value within the input frame. Expressed in range of
17589 Display the V value at the 90% percentile within the input frame. Expressed in
17593 Display the maximum V value contained within the input frame. Expressed in
17597 Display the minimal saturation value contained within the input frame.
17598 Expressed in range of [0-~181.02].
17601 Display the saturation value at the 10% percentile within the input frame.
17602 Expressed in range of [0-~181.02].
17605 Display the average saturation value within the input frame. Expressed in range
17609 Display the saturation value at the 90% percentile within the input frame.
17610 Expressed in range of [0-~181.02].
17613 Display the maximum saturation value contained within the input frame.
17614 Expressed in range of [0-~181.02].
17617 Display the median value for hue within the input frame. Expressed in range of
17621 Display the average value for hue within the input frame. Expressed in range of
17625 Display the average of sample value difference between all values of the Y
17626 plane in the current frame and corresponding values of the previous input frame.
17627 Expressed in range of [0-255].
17630 Display the average of sample value difference between all values of the U
17631 plane in the current frame and corresponding values of the previous input frame.
17632 Expressed in range of [0-255].
17635 Display the average of sample value difference between all values of the V
17636 plane in the current frame and corresponding values of the previous input frame.
17637 Expressed in range of [0-255].
17640 Display bit depth of Y plane in current frame.
17641 Expressed in range of [0-16].
17644 Display bit depth of U plane in current frame.
17645 Expressed in range of [0-16].
17648 Display bit depth of V plane in current frame.
17649 Expressed in range of [0-16].
17652 The filter accepts the following options:
17658 @option{stat} specify an additional form of image analysis.
17659 @option{out} output video with the specified type of pixel highlighted.
17661 Both options accept the following values:
17665 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17666 unlike the neighboring pixels of the same field. Examples of temporal outliers
17667 include the results of video dropouts, head clogs, or tape tracking issues.
17670 Identify @var{vertical line repetition}. Vertical line repetition includes
17671 similar rows of pixels within a frame. In born-digital video vertical line
17672 repetition is common, but this pattern is uncommon in video digitized from an
17673 analog source. When it occurs in video that results from the digitization of an
17674 analog source it can indicate concealment from a dropout compensator.
17677 Identify pixels that fall outside of legal broadcast range.
17681 Set the highlight color for the @option{out} option. The default color is
17685 @subsection Examples
17689 Output data of various video metrics:
17691 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17695 Output specific data about the minimum and maximum values of the Y plane per frame:
17697 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17701 Playback video while highlighting pixels that are outside of broadcast range in red.
17703 ffplay example.mov -vf signalstats="out=brng:color=red"
17707 Playback video with signalstats metadata drawn over the frame.
17709 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17712 The contents of signalstat_drawtext.txt used in the command are:
17715 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
17716 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
17717 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
17718 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
17726 Calculates the MPEG-7 Video Signature. The filter can handle more than one
17727 input. In this case the matching between the inputs can be calculated additionally.
17728 The filter always passes through the first input. The signature of each stream can
17729 be written into a file.
17731 It accepts the following options:
17735 Enable or disable the matching process.
17737 Available values are:
17741 Disable the calculation of a matching (default).
17743 Calculate the matching for the whole video and output whether the whole video
17744 matches or only parts.
17746 Calculate only until a matching is found or the video ends. Should be faster in
17751 Set the number of inputs. The option value must be a non negative integer.
17752 Default value is 1.
17755 Set the path to which the output is written. If there is more than one input,
17756 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
17757 integer), that will be replaced with the input number. If no filename is
17758 specified, no output will be written. This is the default.
17761 Choose the output format.
17763 Available values are:
17767 Use the specified binary representation (default).
17769 Use the specified xml representation.
17773 Set threshold to detect one word as similar. The option value must be an integer
17774 greater than zero. The default value is 9000.
17777 Set threshold to detect all words as similar. The option value must be an integer
17778 greater than zero. The default value is 60000.
17781 Set threshold to detect frames as similar. The option value must be an integer
17782 greater than zero. The default value is 116.
17785 Set the minimum length of a sequence in frames to recognize it as matching
17786 sequence. The option value must be a non negative integer value.
17787 The default value is 0.
17790 Set the minimum relation, that matching frames to all frames must have.
17791 The option value must be a double value between 0 and 1. The default value is 0.5.
17794 @subsection Examples
17798 To calculate the signature of an input video and store it in signature.bin:
17800 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17804 To detect whether two videos match and store the signatures in XML format in
17805 signature0.xml and signature1.xml:
17807 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 -
17815 Blur the input video without impacting the outlines.
17817 It accepts the following options:
17820 @item luma_radius, lr
17821 Set the luma radius. The option value must be a float number in
17822 the range [0.1,5.0] that specifies the variance of the gaussian filter
17823 used to blur the image (slower if larger). Default value is 1.0.
17825 @item luma_strength, ls
17826 Set the luma strength. The option value must be a float number
17827 in the range [-1.0,1.0] that configures the blurring. A value included
17828 in [0.0,1.0] will blur the image whereas a value included in
17829 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17831 @item luma_threshold, lt
17832 Set the luma threshold used as a coefficient to determine
17833 whether a pixel should be blurred or not. The option value must be an
17834 integer in the range [-30,30]. A value of 0 will filter all the image,
17835 a value included in [0,30] will filter flat areas and a value included
17836 in [-30,0] will filter edges. Default value is 0.
17838 @item chroma_radius, cr
17839 Set the chroma radius. The option value must be a float number in
17840 the range [0.1,5.0] that specifies the variance of the gaussian filter
17841 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17843 @item chroma_strength, cs
17844 Set the chroma strength. The option value must be a float number
17845 in the range [-1.0,1.0] that configures the blurring. A value included
17846 in [0.0,1.0] will blur the image whereas a value included in
17847 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17849 @item chroma_threshold, ct
17850 Set the chroma threshold used as a coefficient to determine
17851 whether a pixel should be blurred or not. The option value must be an
17852 integer in the range [-30,30]. A value of 0 will filter all the image,
17853 a value included in [0,30] will filter flat areas and a value included
17854 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17857 If a chroma option is not explicitly set, the corresponding luma value
17861 Apply sobel operator to input video stream.
17863 The filter accepts the following option:
17867 Set which planes will be processed, unprocessed planes will be copied.
17868 By default value 0xf, all planes will be processed.
17871 Set value which will be multiplied with filtered result.
17874 Set value which will be added to filtered result.
17880 Apply a simple postprocessing filter that compresses and decompresses the image
17881 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17882 and average the results.
17884 The filter accepts the following options:
17888 Set quality. This option defines the number of levels for averaging. It accepts
17889 an integer in the range 0-6. If set to @code{0}, the filter will have no
17890 effect. A value of @code{6} means the higher quality. For each increment of
17891 that value the speed drops by a factor of approximately 2. Default value is
17895 Force a constant quantization parameter. If not set, the filter will use the QP
17896 from the video stream (if available).
17899 Set thresholding mode. Available modes are:
17903 Set hard thresholding (default).
17905 Set soft thresholding (better de-ringing effect, but likely blurrier).
17908 @item use_bframe_qp
17909 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17910 option may cause flicker since the B-Frames have often larger QP. Default is
17911 @code{0} (not enabled).
17914 @subsection Commands
17916 This filter supports the following commands:
17918 @item quality, level
17919 Set quality level. The value @code{max} can be used to set the maximum level,
17920 currently @code{6}.
17926 Scale the input by applying one of the super-resolution methods based on
17927 convolutional neural networks. Supported models:
17931 Super-Resolution Convolutional Neural Network model (SRCNN).
17932 See @url{https://arxiv.org/abs/1501.00092}.
17935 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17936 See @url{https://arxiv.org/abs/1609.05158}.
17939 Training scripts as well as scripts for model file (.pb) saving can be found at
17940 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17941 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17943 Native model files (.model) can be generated from TensorFlow model
17944 files (.pb) by using tools/python/convert.py
17946 The filter accepts the following options:
17950 Specify which DNN backend to use for model loading and execution. This option accepts
17951 the following values:
17955 Native implementation of DNN loading and execution.
17958 TensorFlow backend. To enable this backend you
17959 need to install the TensorFlow for C library (see
17960 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17961 @code{--enable-libtensorflow}
17964 Default value is @samp{native}.
17967 Set path to model file specifying network architecture and its parameters.
17968 Note that different backends use different file formats. TensorFlow backend
17969 can load files for both formats, while native backend can load files for only
17973 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17974 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17975 input upscaled using bicubic upscaling with proper scale factor.
17978 This feature can also be finished with @ref{dnn_processing} filter.
17982 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17984 This filter takes in input two input videos, the first input is
17985 considered the "main" source and is passed unchanged to the
17986 output. The second input is used as a "reference" video for computing
17989 Both video inputs must have the same resolution and pixel format for
17990 this filter to work correctly. Also it assumes that both inputs
17991 have the same number of frames, which are compared one by one.
17993 The filter stores the calculated SSIM of each frame.
17995 The description of the accepted parameters follows.
17998 @item stats_file, f
17999 If specified the filter will use the named file to save the SSIM of
18000 each individual frame. When filename equals "-" the data is sent to
18004 The file printed if @var{stats_file} is selected, contains a sequence of
18005 key/value pairs of the form @var{key}:@var{value} for each compared
18008 A description of each shown parameter follows:
18012 sequential number of the input frame, starting from 1
18014 @item Y, U, V, R, G, B
18015 SSIM of the compared frames for the component specified by the suffix.
18018 SSIM of the compared frames for the whole frame.
18021 Same as above but in dB representation.
18024 This filter also supports the @ref{framesync} options.
18026 @subsection Examples
18031 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18032 [main][ref] ssim="stats_file=stats.log" [out]
18035 On this example the input file being processed is compared with the
18036 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18037 is stored in @file{stats.log}.
18040 Another example with both psnr and ssim at same time:
18042 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18046 Another example with different containers:
18048 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 -
18054 Convert between different stereoscopic image formats.
18056 The filters accept the following options:
18060 Set stereoscopic image format of input.
18062 Available values for input image formats are:
18065 side by side parallel (left eye left, right eye right)
18068 side by side crosseye (right eye left, left eye right)
18071 side by side parallel with half width resolution
18072 (left eye left, right eye right)
18075 side by side crosseye with half width resolution
18076 (right eye left, left eye right)
18080 above-below (left eye above, right eye below)
18084 above-below (right eye above, left eye below)
18088 above-below with half height resolution
18089 (left eye above, right eye below)
18093 above-below with half height resolution
18094 (right eye above, left eye below)
18097 alternating frames (left eye first, right eye second)
18100 alternating frames (right eye first, left eye second)
18103 interleaved rows (left eye has top row, right eye starts on next row)
18106 interleaved rows (right eye has top row, left eye starts on next row)
18109 interleaved columns, left eye first
18112 interleaved columns, right eye first
18114 Default value is @samp{sbsl}.
18118 Set stereoscopic image format of output.
18122 side by side parallel (left eye left, right eye right)
18125 side by side crosseye (right eye left, left eye right)
18128 side by side parallel with half width resolution
18129 (left eye left, right eye right)
18132 side by side crosseye with half width resolution
18133 (right eye left, left eye right)
18137 above-below (left eye above, right eye below)
18141 above-below (right eye above, left eye below)
18145 above-below with half height resolution
18146 (left eye above, right eye below)
18150 above-below with half height resolution
18151 (right eye above, left eye below)
18154 alternating frames (left eye first, right eye second)
18157 alternating frames (right eye first, left eye second)
18160 interleaved rows (left eye has top row, right eye starts on next row)
18163 interleaved rows (right eye has top row, left eye starts on next row)
18166 anaglyph red/blue gray
18167 (red filter on left eye, blue filter on right eye)
18170 anaglyph red/green gray
18171 (red filter on left eye, green filter on right eye)
18174 anaglyph red/cyan gray
18175 (red filter on left eye, cyan filter on right eye)
18178 anaglyph red/cyan half colored
18179 (red filter on left eye, cyan filter on right eye)
18182 anaglyph red/cyan color
18183 (red filter on left eye, cyan filter on right eye)
18186 anaglyph red/cyan color optimized with the least squares projection of dubois
18187 (red filter on left eye, cyan filter on right eye)
18190 anaglyph green/magenta gray
18191 (green filter on left eye, magenta filter on right eye)
18194 anaglyph green/magenta half colored
18195 (green filter on left eye, magenta filter on right eye)
18198 anaglyph green/magenta colored
18199 (green filter on left eye, magenta filter on right eye)
18202 anaglyph green/magenta color optimized with the least squares projection of dubois
18203 (green filter on left eye, magenta filter on right eye)
18206 anaglyph yellow/blue gray
18207 (yellow filter on left eye, blue filter on right eye)
18210 anaglyph yellow/blue half colored
18211 (yellow filter on left eye, blue filter on right eye)
18214 anaglyph yellow/blue colored
18215 (yellow filter on left eye, blue filter on right eye)
18218 anaglyph yellow/blue color optimized with the least squares projection of dubois
18219 (yellow filter on left eye, blue filter on right eye)
18222 mono output (left eye only)
18225 mono output (right eye only)
18228 checkerboard, left eye first
18231 checkerboard, right eye first
18234 interleaved columns, left eye first
18237 interleaved columns, right eye first
18243 Default value is @samp{arcd}.
18246 @subsection Examples
18250 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18256 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18262 @section streamselect, astreamselect
18263 Select video or audio streams.
18265 The filter accepts the following options:
18269 Set number of inputs. Default is 2.
18272 Set input indexes to remap to outputs.
18275 @subsection Commands
18277 The @code{streamselect} and @code{astreamselect} filter supports the following
18282 Set input indexes to remap to outputs.
18285 @subsection Examples
18289 Select first 5 seconds 1st stream and rest of time 2nd stream:
18291 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18295 Same as above, but for audio:
18297 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18304 Draw subtitles on top of input video using the libass library.
18306 To enable compilation of this filter you need to configure FFmpeg with
18307 @code{--enable-libass}. This filter also requires a build with libavcodec and
18308 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18309 Alpha) subtitles format.
18311 The filter accepts the following options:
18315 Set the filename of the subtitle file to read. It must be specified.
18317 @item original_size
18318 Specify the size of the original video, the video for which the ASS file
18319 was composed. For the syntax of this option, check the
18320 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18321 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18322 correctly scale the fonts if the aspect ratio has been changed.
18325 Set a directory path containing fonts that can be used by the filter.
18326 These fonts will be used in addition to whatever the font provider uses.
18329 Process alpha channel, by default alpha channel is untouched.
18332 Set subtitles input character encoding. @code{subtitles} filter only. Only
18333 useful if not UTF-8.
18335 @item stream_index, si
18336 Set subtitles stream index. @code{subtitles} filter only.
18339 Override default style or script info parameters of the subtitles. It accepts a
18340 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18343 If the first key is not specified, it is assumed that the first value
18344 specifies the @option{filename}.
18346 For example, to render the file @file{sub.srt} on top of the input
18347 video, use the command:
18352 which is equivalent to:
18354 subtitles=filename=sub.srt
18357 To render the default subtitles stream from file @file{video.mkv}, use:
18359 subtitles=video.mkv
18362 To render the second subtitles stream from that file, use:
18364 subtitles=video.mkv:si=1
18367 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18368 @code{DejaVu Serif}, use:
18370 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18373 @section super2xsai
18375 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18376 Interpolate) pixel art scaling algorithm.
18378 Useful for enlarging pixel art images without reducing sharpness.
18382 Swap two rectangular objects in video.
18384 This filter accepts the following options:
18394 Set 1st rect x coordinate.
18397 Set 1st rect y coordinate.
18400 Set 2nd rect x coordinate.
18403 Set 2nd rect y coordinate.
18405 All expressions are evaluated once for each frame.
18408 The all options are expressions containing the following constants:
18413 The input width and height.
18416 same as @var{w} / @var{h}
18419 input sample aspect ratio
18422 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18425 The number of the input frame, starting from 0.
18428 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18431 the position in the file of the input frame, NAN if unknown
18438 Blend successive video frames.
18444 Apply telecine process to the video.
18446 This filter accepts the following options:
18455 The default value is @code{top}.
18459 A string of numbers representing the pulldown pattern you wish to apply.
18460 The default value is @code{23}.
18464 Some typical patterns:
18469 24p: 2332 (preferred)
18476 24p: 222222222223 ("Euro pulldown")
18481 @section thistogram
18483 Compute and draw a color distribution histogram for the input video across time.
18485 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18486 at certain time, this filter shows also past histograms of number of frames defined
18487 by @code{width} option.
18489 The computed histogram is a representation of the color component
18490 distribution in an image.
18492 The filter accepts the following options:
18496 Set width of single color component output. Default value is @code{0}.
18497 Value of @code{0} means width will be picked from input video.
18498 This also set number of passed histograms to keep.
18499 Allowed range is [0, 8192].
18501 @item display_mode, d
18503 It accepts the following values:
18506 Per color component graphs are placed below each other.
18509 Per color component graphs are placed side by side.
18512 Presents information identical to that in the @code{parade}, except
18513 that the graphs representing color components are superimposed directly
18516 Default is @code{stack}.
18518 @item levels_mode, m
18519 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18520 Default is @code{linear}.
18522 @item components, c
18523 Set what color components to display.
18524 Default is @code{7}.
18527 Set background opacity. Default is @code{0.9}.
18530 Show envelope. Default is disabled.
18533 Set envelope color. Default is @code{gold}.
18538 Available values for slide is:
18541 Draw new frame when right border is reached.
18544 Replace old columns with new ones.
18547 Scroll from right to left.
18550 Scroll from left to right.
18553 Draw single picture.
18556 Default is @code{replace}.
18561 Apply threshold effect to video stream.
18563 This filter needs four video streams to perform thresholding.
18564 First stream is stream we are filtering.
18565 Second stream is holding threshold values, third stream is holding min values,
18566 and last, fourth stream is holding max values.
18568 The filter accepts the following option:
18572 Set which planes will be processed, unprocessed planes will be copied.
18573 By default value 0xf, all planes will be processed.
18576 For example if first stream pixel's component value is less then threshold value
18577 of pixel component from 2nd threshold stream, third stream value will picked,
18578 otherwise fourth stream pixel component value will be picked.
18580 Using color source filter one can perform various types of thresholding:
18582 @subsection Examples
18586 Binary threshold, using gray color as threshold:
18588 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18592 Inverted binary threshold, using gray color as threshold:
18594 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18598 Truncate binary threshold, using gray color as threshold:
18600 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18604 Threshold to zero, using gray color as threshold:
18606 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18610 Inverted threshold to zero, using gray color as threshold:
18612 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18617 Select the most representative frame in a given sequence of consecutive frames.
18619 The filter accepts the following options:
18623 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18624 will pick one of them, and then handle the next batch of @var{n} frames until
18625 the end. Default is @code{100}.
18628 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18629 value will result in a higher memory usage, so a high value is not recommended.
18631 @subsection Examples
18635 Extract one picture each 50 frames:
18641 Complete example of a thumbnail creation with @command{ffmpeg}:
18643 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18650 Tile several successive frames together.
18652 The @ref{untile} filter can do the reverse.
18654 The filter accepts the following options:
18659 Set the grid size (i.e. the number of lines and columns). For the syntax of
18660 this option, check the
18661 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18664 Set the maximum number of frames to render in the given area. It must be less
18665 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18666 the area will be used.
18669 Set the outer border margin in pixels.
18672 Set the inner border thickness (i.e. the number of pixels between frames). For
18673 more advanced padding options (such as having different values for the edges),
18674 refer to the pad video filter.
18677 Specify the color of the unused area. For the syntax of this option, check the
18678 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18679 The default value of @var{color} is "black".
18682 Set the number of frames to overlap when tiling several successive frames together.
18683 The value must be between @code{0} and @var{nb_frames - 1}.
18686 Set the number of frames to initially be empty before displaying first output frame.
18687 This controls how soon will one get first output frame.
18688 The value must be between @code{0} and @var{nb_frames - 1}.
18691 @subsection Examples
18695 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18697 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18699 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18700 duplicating each output frame to accommodate the originally detected frame
18704 Display @code{5} pictures in an area of @code{3x2} frames,
18705 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18706 mixed flat and named options:
18708 tile=3x2:nb_frames=5:padding=7:margin=2
18712 @section tinterlace
18714 Perform various types of temporal field interlacing.
18716 Frames are counted starting from 1, so the first input frame is
18719 The filter accepts the following options:
18724 Specify the mode of the interlacing. This option can also be specified
18725 as a value alone. See below for a list of values for this option.
18727 Available values are:
18731 Move odd frames into the upper field, even into the lower field,
18732 generating a double height frame at half frame rate.
18736 Frame 1 Frame 2 Frame 3 Frame 4
18738 11111 22222 33333 44444
18739 11111 22222 33333 44444
18740 11111 22222 33333 44444
18741 11111 22222 33333 44444
18755 Only output odd frames, even frames are dropped, generating a frame with
18756 unchanged height at half frame rate.
18761 Frame 1 Frame 2 Frame 3 Frame 4
18763 11111 22222 33333 44444
18764 11111 22222 33333 44444
18765 11111 22222 33333 44444
18766 11111 22222 33333 44444
18776 Only output even frames, odd frames are dropped, generating a frame with
18777 unchanged height at half frame rate.
18782 Frame 1 Frame 2 Frame 3 Frame 4
18784 11111 22222 33333 44444
18785 11111 22222 33333 44444
18786 11111 22222 33333 44444
18787 11111 22222 33333 44444
18797 Expand each frame to full height, but pad alternate lines with black,
18798 generating a frame with double height at the same input frame rate.
18803 Frame 1 Frame 2 Frame 3 Frame 4
18805 11111 22222 33333 44444
18806 11111 22222 33333 44444
18807 11111 22222 33333 44444
18808 11111 22222 33333 44444
18811 11111 ..... 33333 .....
18812 ..... 22222 ..... 44444
18813 11111 ..... 33333 .....
18814 ..... 22222 ..... 44444
18815 11111 ..... 33333 .....
18816 ..... 22222 ..... 44444
18817 11111 ..... 33333 .....
18818 ..... 22222 ..... 44444
18822 @item interleave_top, 4
18823 Interleave the upper field from odd frames with the lower field from
18824 even frames, generating a frame with unchanged height at half frame rate.
18829 Frame 1 Frame 2 Frame 3 Frame 4
18831 11111<- 22222 33333<- 44444
18832 11111 22222<- 33333 44444<-
18833 11111<- 22222 33333<- 44444
18834 11111 22222<- 33333 44444<-
18844 @item interleave_bottom, 5
18845 Interleave the lower field from odd frames with the upper field from
18846 even frames, generating a frame with unchanged height at half frame rate.
18851 Frame 1 Frame 2 Frame 3 Frame 4
18853 11111 22222<- 33333 44444<-
18854 11111<- 22222 33333<- 44444
18855 11111 22222<- 33333 44444<-
18856 11111<- 22222 33333<- 44444
18866 @item interlacex2, 6
18867 Double frame rate with unchanged height. Frames are inserted each
18868 containing the second temporal field from the previous input frame and
18869 the first temporal field from the next input frame. This mode relies on
18870 the top_field_first flag. Useful for interlaced video displays with no
18871 field synchronisation.
18876 Frame 1 Frame 2 Frame 3 Frame 4
18878 11111 22222 33333 44444
18879 11111 22222 33333 44444
18880 11111 22222 33333 44444
18881 11111 22222 33333 44444
18884 11111 22222 22222 33333 33333 44444 44444
18885 11111 11111 22222 22222 33333 33333 44444
18886 11111 22222 22222 33333 33333 44444 44444
18887 11111 11111 22222 22222 33333 33333 44444
18892 Move odd frames into the upper field, even into the lower field,
18893 generating a double height frame at same frame rate.
18898 Frame 1 Frame 2 Frame 3 Frame 4
18900 11111 22222 33333 44444
18901 11111 22222 33333 44444
18902 11111 22222 33333 44444
18903 11111 22222 33333 44444
18906 11111 33333 33333 55555
18907 22222 22222 44444 44444
18908 11111 33333 33333 55555
18909 22222 22222 44444 44444
18910 11111 33333 33333 55555
18911 22222 22222 44444 44444
18912 11111 33333 33333 55555
18913 22222 22222 44444 44444
18918 Numeric values are deprecated but are accepted for backward
18919 compatibility reasons.
18921 Default mode is @code{merge}.
18924 Specify flags influencing the filter process.
18926 Available value for @var{flags} is:
18929 @item low_pass_filter, vlpf
18930 Enable linear vertical low-pass filtering in the filter.
18931 Vertical low-pass filtering is required when creating an interlaced
18932 destination from a progressive source which contains high-frequency
18933 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18936 @item complex_filter, cvlpf
18937 Enable complex vertical low-pass filtering.
18938 This will slightly less reduce interlace 'twitter' and Moire
18939 patterning but better retain detail and subjective sharpness impression.
18942 Bypass already interlaced frames, only adjust the frame rate.
18945 Vertical low-pass filtering and bypassing already interlaced frames can only be
18946 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
18951 Pick median pixels from several successive input video frames.
18953 The filter accepts the following options:
18957 Set radius of median filter.
18958 Default is 1. Allowed range is from 1 to 127.
18961 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
18964 Set median percentile. Default value is @code{0.5}.
18965 Default value of @code{0.5} will pick always median values, while @code{0} will pick
18966 minimum values, and @code{1} maximum values.
18971 Mix successive video frames.
18973 A description of the accepted options follows.
18977 The number of successive frames to mix. If unspecified, it defaults to 3.
18980 Specify weight of each input video frame.
18981 Each weight is separated by space. If number of weights is smaller than
18982 number of @var{frames} last specified weight will be used for all remaining
18986 Specify scale, if it is set it will be multiplied with sum
18987 of each weight multiplied with pixel values to give final destination
18988 pixel value. By default @var{scale} is auto scaled to sum of weights.
18991 @subsection Examples
18995 Average 7 successive frames:
18997 tmix=frames=7:weights="1 1 1 1 1 1 1"
19001 Apply simple temporal convolution:
19003 tmix=frames=3:weights="-1 3 -1"
19007 Similar as above but only showing temporal differences:
19009 tmix=frames=3:weights="-1 2 -1":scale=1
19015 Tone map colors from different dynamic ranges.
19017 This filter expects data in single precision floating point, as it needs to
19018 operate on (and can output) out-of-range values. Another filter, such as
19019 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19021 The tonemapping algorithms implemented only work on linear light, so input
19022 data should be linearized beforehand (and possibly correctly tagged).
19025 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19028 @subsection Options
19029 The filter accepts the following options.
19033 Set the tone map algorithm to use.
19035 Possible values are:
19038 Do not apply any tone map, only desaturate overbright pixels.
19041 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19042 in-range values, while distorting out-of-range values.
19045 Stretch the entire reference gamut to a linear multiple of the display.
19048 Fit a logarithmic transfer between the tone curves.
19051 Preserve overall image brightness with a simple curve, using nonlinear
19052 contrast, which results in flattening details and degrading color accuracy.
19055 Preserve both dark and bright details better than @var{reinhard}, at the cost
19056 of slightly darkening everything. Use it when detail preservation is more
19057 important than color and brightness accuracy.
19060 Smoothly map out-of-range values, while retaining contrast and colors for
19061 in-range material as much as possible. Use it when color accuracy is more
19062 important than detail preservation.
19068 Tune the tone mapping algorithm.
19070 This affects the following algorithms:
19076 Specifies the scale factor to use while stretching.
19080 Specifies the exponent of the function.
19084 Specify an extra linear coefficient to multiply into the signal before clipping.
19088 Specify the local contrast coefficient at the display peak.
19089 Default to 0.5, which means that in-gamut values will be about half as bright
19096 Specify the transition point from linear to mobius transform. Every value
19097 below this point is guaranteed to be mapped 1:1. The higher the value, the
19098 more accurate the result will be, at the cost of losing bright details.
19099 Default to 0.3, which due to the steep initial slope still preserves in-range
19100 colors fairly accurately.
19104 Apply desaturation for highlights that exceed this level of brightness. The
19105 higher the parameter, the more color information will be preserved. This
19106 setting helps prevent unnaturally blown-out colors for super-highlights, by
19107 (smoothly) turning into white instead. This makes images feel more natural,
19108 at the cost of reducing information about out-of-range colors.
19110 The default of 2.0 is somewhat conservative and will mostly just apply to
19111 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19113 This option works only if the input frame has a supported color tag.
19116 Override signal/nominal/reference peak with this value. Useful when the
19117 embedded peak information in display metadata is not reliable or when tone
19118 mapping from a lower range to a higher range.
19123 Temporarily pad video frames.
19125 The filter accepts the following options:
19129 Specify number of delay frames before input video stream. Default is 0.
19132 Specify number of padding frames after input video stream.
19133 Set to -1 to pad indefinitely. Default is 0.
19136 Set kind of frames added to beginning of stream.
19137 Can be either @var{add} or @var{clone}.
19138 With @var{add} frames of solid-color are added.
19139 With @var{clone} frames are clones of first frame.
19140 Default is @var{add}.
19143 Set kind of frames added to end of stream.
19144 Can be either @var{add} or @var{clone}.
19145 With @var{add} frames of solid-color are added.
19146 With @var{clone} frames are clones of last frame.
19147 Default is @var{add}.
19149 @item start_duration, stop_duration
19150 Specify the duration of the start/stop delay. See
19151 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19152 for the accepted syntax.
19153 These options override @var{start} and @var{stop}. Default is 0.
19156 Specify the color of the padded area. For the syntax of this option,
19157 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19158 manual,ffmpeg-utils}.
19160 The default value of @var{color} is "black".
19166 Transpose rows with columns in the input video and optionally flip it.
19168 It accepts the following parameters:
19173 Specify the transposition direction.
19175 Can assume the following values:
19177 @item 0, 4, cclock_flip
19178 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19186 Rotate by 90 degrees clockwise, that is:
19194 Rotate by 90 degrees counterclockwise, that is:
19201 @item 3, 7, clock_flip
19202 Rotate by 90 degrees clockwise and vertically flip, that is:
19210 For values between 4-7, the transposition is only done if the input
19211 video geometry is portrait and not landscape. These values are
19212 deprecated, the @code{passthrough} option should be used instead.
19214 Numerical values are deprecated, and should be dropped in favor of
19215 symbolic constants.
19218 Do not apply the transposition if the input geometry matches the one
19219 specified by the specified value. It accepts the following values:
19222 Always apply transposition.
19224 Preserve portrait geometry (when @var{height} >= @var{width}).
19226 Preserve landscape geometry (when @var{width} >= @var{height}).
19229 Default value is @code{none}.
19232 For example to rotate by 90 degrees clockwise and preserve portrait
19235 transpose=dir=1:passthrough=portrait
19238 The command above can also be specified as:
19240 transpose=1:portrait
19243 @section transpose_npp
19245 Transpose rows with columns in the input video and optionally flip it.
19246 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19248 It accepts the following parameters:
19253 Specify the transposition direction.
19255 Can assume the following values:
19258 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19261 Rotate by 90 degrees clockwise.
19264 Rotate by 90 degrees counterclockwise.
19267 Rotate by 90 degrees clockwise and vertically flip.
19271 Do not apply the transposition if the input geometry matches the one
19272 specified by the specified value. It accepts the following values:
19275 Always apply transposition. (default)
19277 Preserve portrait geometry (when @var{height} >= @var{width}).
19279 Preserve landscape geometry (when @var{width} >= @var{height}).
19285 Trim the input so that the output contains one continuous subpart of the input.
19287 It accepts the following parameters:
19290 Specify the time of the start of the kept section, i.e. the frame with the
19291 timestamp @var{start} will be the first frame in the output.
19294 Specify the time of the first frame that will be dropped, i.e. the frame
19295 immediately preceding the one with the timestamp @var{end} will be the last
19296 frame in the output.
19299 This is the same as @var{start}, except this option sets the start timestamp
19300 in timebase units instead of seconds.
19303 This is the same as @var{end}, except this option sets the end timestamp
19304 in timebase units instead of seconds.
19307 The maximum duration of the output in seconds.
19310 The number of the first frame that should be passed to the output.
19313 The number of the first frame that should be dropped.
19316 @option{start}, @option{end}, and @option{duration} are expressed as time
19317 duration specifications; see
19318 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19319 for the accepted syntax.
19321 Note that the first two sets of the start/end options and the @option{duration}
19322 option look at the frame timestamp, while the _frame variants simply count the
19323 frames that pass through the filter. Also note that this filter does not modify
19324 the timestamps. If you wish for the output timestamps to start at zero, insert a
19325 setpts filter after the trim filter.
19327 If multiple start or end options are set, this filter tries to be greedy and
19328 keep all the frames that match at least one of the specified constraints. To keep
19329 only the part that matches all the constraints at once, chain multiple trim
19332 The defaults are such that all the input is kept. So it is possible to set e.g.
19333 just the end values to keep everything before the specified time.
19338 Drop everything except the second minute of input:
19340 ffmpeg -i INPUT -vf trim=60:120
19344 Keep only the first second:
19346 ffmpeg -i INPUT -vf trim=duration=1
19351 @section unpremultiply
19352 Apply alpha unpremultiply effect to input video stream using first plane
19353 of second stream as alpha.
19355 Both streams must have same dimensions and same pixel format.
19357 The filter accepts the following option:
19361 Set which planes will be processed, unprocessed planes will be copied.
19362 By default value 0xf, all planes will be processed.
19364 If the format has 1 or 2 components, then luma is bit 0.
19365 If the format has 3 or 4 components:
19366 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19367 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19368 If present, the alpha channel is always the last bit.
19371 Do not require 2nd input for processing, instead use alpha plane from input stream.
19377 Sharpen or blur the input video.
19379 It accepts the following parameters:
19382 @item luma_msize_x, lx
19383 Set the luma matrix horizontal size. It must be an odd integer between
19384 3 and 23. The default value is 5.
19386 @item luma_msize_y, ly
19387 Set the luma matrix vertical size. It must be an odd integer between 3
19388 and 23. The default value is 5.
19390 @item luma_amount, la
19391 Set the luma effect strength. It must be a floating point number, reasonable
19392 values lay between -1.5 and 1.5.
19394 Negative values will blur the input video, while positive values will
19395 sharpen it, a value of zero will disable the effect.
19397 Default value is 1.0.
19399 @item chroma_msize_x, cx
19400 Set the chroma matrix horizontal size. It must be an odd integer
19401 between 3 and 23. The default value is 5.
19403 @item chroma_msize_y, cy
19404 Set the chroma matrix vertical size. It must be an odd integer
19405 between 3 and 23. The default value is 5.
19407 @item chroma_amount, ca
19408 Set the chroma effect strength. It must be a floating point number, reasonable
19409 values lay between -1.5 and 1.5.
19411 Negative values will blur the input video, while positive values will
19412 sharpen it, a value of zero will disable the effect.
19414 Default value is 0.0.
19418 All parameters are optional and default to the equivalent of the
19419 string '5:5:1.0:5:5:0.0'.
19421 @subsection Examples
19425 Apply strong luma sharpen effect:
19427 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19431 Apply a strong blur of both luma and chroma parameters:
19433 unsharp=7:7:-2:7:7:-2
19440 Decompose a video made of tiled images into the individual images.
19442 The frame rate of the output video is the frame rate of the input video
19443 multiplied by the number of tiles.
19445 This filter does the reverse of @ref{tile}.
19447 The filter accepts the following options:
19452 Set the grid size (i.e. the number of lines and columns). For the syntax of
19453 this option, check the
19454 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19457 @subsection Examples
19461 Produce a 1-second video from a still image file made of 25 frames stacked
19462 vertically, like an analogic film reel:
19464 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19470 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19471 the image at several (or - in the case of @option{quality} level @code{8} - all)
19472 shifts and average the results.
19474 The way this differs from the behavior of spp is that uspp actually encodes &
19475 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19476 DCT similar to MJPEG.
19478 The filter accepts the following options:
19482 Set quality. This option defines the number of levels for averaging. It accepts
19483 an integer in the range 0-8. If set to @code{0}, the filter will have no
19484 effect. A value of @code{8} means the higher quality. For each increment of
19485 that value the speed drops by a factor of approximately 2. Default value is
19489 Force a constant quantization parameter. If not set, the filter will use the QP
19490 from the video stream (if available).
19495 Convert 360 videos between various formats.
19497 The filter accepts the following options:
19503 Set format of the input/output video.
19511 Equirectangular projection.
19516 Cubemap with 3x2/6x1/1x6 layout.
19518 Format specific options:
19523 Set padding proportion for the input/output cubemap. Values in decimals.
19530 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)
19533 Default value is @b{@samp{0}}.
19534 Maximum value is @b{@samp{0.1}}.
19538 Set fixed padding for the input/output cubemap. Values in pixels.
19540 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19544 Set order of faces for the input/output cubemap. Choose one direction for each position.
19546 Designation of directions:
19562 Default value is @b{@samp{rludfb}}.
19566 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19568 Designation of angles:
19571 0 degrees clockwise
19573 90 degrees clockwise
19575 180 degrees clockwise
19577 270 degrees clockwise
19580 Default value is @b{@samp{000000}}.
19584 Equi-Angular Cubemap.
19591 Format specific options:
19596 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19598 If diagonal field of view is set it overrides horizontal and vertical field of view.
19603 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19605 If diagonal field of view is set it overrides horizontal and vertical field of view.
19611 Format specific options:
19616 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19618 If diagonal field of view is set it overrides horizontal and vertical field of view.
19623 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19625 If diagonal field of view is set it overrides horizontal and vertical field of view.
19631 Facebook's 360 formats.
19634 Stereographic format.
19636 Format specific options:
19641 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19643 If diagonal field of view is set it overrides horizontal and vertical field of view.
19648 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19650 If diagonal field of view is set it overrides horizontal and vertical field of view.
19657 Ball format, gives significant distortion toward the back.
19660 Hammer-Aitoff map projection format.
19663 Sinusoidal map projection format.
19666 Fisheye projection.
19668 Format specific options:
19673 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19675 If diagonal field of view is set it overrides horizontal and vertical field of view.
19680 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19682 If diagonal field of view is set it overrides horizontal and vertical field of view.
19686 Pannini projection.
19688 Format specific options:
19691 Set output pannini parameter.
19694 Set input pannini parameter.
19698 Cylindrical projection.
19700 Format specific options:
19705 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19707 If diagonal field of view is set it overrides horizontal and vertical field of view.
19712 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19714 If diagonal field of view is set it overrides horizontal and vertical field of view.
19718 Perspective projection. @i{(output only)}
19720 Format specific options:
19723 Set perspective parameter.
19727 Tetrahedron projection.
19730 Truncated square pyramid projection.
19734 Half equirectangular projection.
19739 Format specific options:
19744 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19746 If diagonal field of view is set it overrides horizontal and vertical field of view.
19751 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19753 If diagonal field of view is set it overrides horizontal and vertical field of view.
19757 Orthographic format.
19759 Format specific options:
19764 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19766 If diagonal field of view is set it overrides horizontal and vertical field of view.
19771 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19773 If diagonal field of view is set it overrides horizontal and vertical field of view.
19777 Octahedron projection.
19781 Set interpolation method.@*
19782 @i{Note: more complex interpolation methods require much more memory to run.}
19792 Bilinear interpolation.
19794 Lagrange9 interpolation.
19797 Bicubic interpolation.
19800 Lanczos interpolation.
19803 Spline16 interpolation.
19806 Gaussian interpolation.
19808 Mitchell interpolation.
19811 Default value is @b{@samp{line}}.
19815 Set the output video resolution.
19817 Default resolution depends on formats.
19821 Set the input/output stereo format.
19832 Default value is @b{@samp{2d}} for input and output format.
19837 Set rotation for the output video. Values in degrees.
19840 Set rotation order for the output video. Choose one item for each position.
19851 Default value is @b{@samp{ypr}}.
19856 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
19860 Set if input video is flipped horizontally/vertically. Boolean values.
19863 Set if input video is transposed. Boolean value, by default disabled.
19866 Set if output video needs to be transposed. Boolean value, by default disabled.
19869 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
19872 @subsection Examples
19876 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
19878 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19881 Extract back view of Equi-Angular Cubemap:
19883 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19886 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
19888 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19892 @subsection Commands
19894 This filter supports subset of above options as @ref{commands}.
19896 @section vaguedenoiser
19898 Apply a wavelet based denoiser.
19900 It transforms each frame from the video input into the wavelet domain,
19901 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19902 the obtained coefficients. It does an inverse wavelet transform after.
19903 Due to wavelet properties, it should give a nice smoothed result, and
19904 reduced noise, without blurring picture features.
19906 This filter accepts the following options:
19910 The filtering strength. The higher, the more filtered the video will be.
19911 Hard thresholding can use a higher threshold than soft thresholding
19912 before the video looks overfiltered. Default value is 2.
19915 The filtering method the filter will use.
19917 It accepts the following values:
19920 All values under the threshold will be zeroed.
19923 All values under the threshold will be zeroed. All values above will be
19924 reduced by the threshold.
19927 Scales or nullifies coefficients - intermediary between (more) soft and
19928 (less) hard thresholding.
19931 Default is garrote.
19934 Number of times, the wavelet will decompose the picture. Picture can't
19935 be decomposed beyond a particular point (typically, 8 for a 640x480
19936 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
19939 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
19942 A list of the planes to process. By default all planes are processed.
19945 The threshold type the filter will use.
19947 It accepts the following values:
19950 Threshold used is same for all decompositions.
19953 Threshold used depends also on each decomposition coefficients.
19956 Default is universal.
19959 @section vectorscope
19961 Display 2 color component values in the two dimensional graph (which is called
19964 This filter accepts the following options:
19968 Set vectorscope mode.
19970 It accepts the following values:
19974 Gray values are displayed on graph, higher brightness means more pixels have
19975 same component color value on location in graph. This is the default mode.
19978 Gray values are displayed on graph. Surrounding pixels values which are not
19979 present in video frame are drawn in gradient of 2 color components which are
19980 set by option @code{x} and @code{y}. The 3rd color component is static.
19983 Actual color components values present in video frame are displayed on graph.
19986 Similar as color2 but higher frequency of same values @code{x} and @code{y}
19987 on graph increases value of another color component, which is luminance by
19988 default values of @code{x} and @code{y}.
19991 Actual colors present in video frame are displayed on graph. If two different
19992 colors map to same position on graph then color with higher value of component
19993 not present in graph is picked.
19996 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
19997 component picked from radial gradient.
20001 Set which color component will be represented on X-axis. Default is @code{1}.
20004 Set which color component will be represented on Y-axis. Default is @code{2}.
20007 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20008 of color component which represents frequency of (X, Y) location in graph.
20013 No envelope, this is default.
20016 Instant envelope, even darkest single pixel will be clearly highlighted.
20019 Hold maximum and minimum values presented in graph over time. This way you
20020 can still spot out of range values without constantly looking at vectorscope.
20023 Peak and instant envelope combined together.
20027 Set what kind of graticule to draw.
20036 Set graticule opacity.
20039 Set graticule flags.
20043 Draw graticule for white point.
20046 Draw graticule for black point.
20049 Draw color points short names.
20053 Set background opacity.
20055 @item lthreshold, l
20056 Set low threshold for color component not represented on X or Y axis.
20057 Values lower than this value will be ignored. Default is 0.
20058 Note this value is multiplied with actual max possible value one pixel component
20059 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20062 @item hthreshold, h
20063 Set high threshold for color component not represented on X or Y axis.
20064 Values higher than this value will be ignored. Default is 1.
20065 Note this value is multiplied with actual max possible value one pixel component
20066 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20067 is 0.9 * 255 = 230.
20069 @item colorspace, c
20070 Set what kind of colorspace to use when drawing graticule.
20080 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20081 This means no tint, and output will remain gray.
20084 @anchor{vidstabdetect}
20085 @section vidstabdetect
20087 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20088 @ref{vidstabtransform} for pass 2.
20090 This filter generates a file with relative translation and rotation
20091 transform information about subsequent frames, which is then used by
20092 the @ref{vidstabtransform} filter.
20094 To enable compilation of this filter you need to configure FFmpeg with
20095 @code{--enable-libvidstab}.
20097 This filter accepts the following options:
20101 Set the path to the file used to write the transforms information.
20102 Default value is @file{transforms.trf}.
20105 Set how shaky the video is and how quick the camera is. It accepts an
20106 integer in the range 1-10, a value of 1 means little shakiness, a
20107 value of 10 means strong shakiness. Default value is 5.
20110 Set the accuracy of the detection process. It must be a value in the
20111 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20112 accuracy. Default value is 15.
20115 Set stepsize of the search process. The region around minimum is
20116 scanned with 1 pixel resolution. Default value is 6.
20119 Set minimum contrast. Below this value a local measurement field is
20120 discarded. Must be a floating point value in the range 0-1. Default
20124 Set reference frame number for tripod mode.
20126 If enabled, the motion of the frames is compared to a reference frame
20127 in the filtered stream, identified by the specified number. The idea
20128 is to compensate all movements in a more-or-less static scene and keep
20129 the camera view absolutely still.
20131 If set to 0, it is disabled. The frames are counted starting from 1.
20134 Show fields and transforms in the resulting frames. It accepts an
20135 integer in the range 0-2. Default value is 0, which disables any
20139 @subsection Examples
20143 Use default values:
20149 Analyze strongly shaky movie and put the results in file
20150 @file{mytransforms.trf}:
20152 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20156 Visualize the result of internal transformations in the resulting
20159 vidstabdetect=show=1
20163 Analyze a video with medium shakiness using @command{ffmpeg}:
20165 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20169 @anchor{vidstabtransform}
20170 @section vidstabtransform
20172 Video stabilization/deshaking: pass 2 of 2,
20173 see @ref{vidstabdetect} for pass 1.
20175 Read a file with transform information for each frame and
20176 apply/compensate them. Together with the @ref{vidstabdetect}
20177 filter this can be used to deshake videos. See also
20178 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20179 the @ref{unsharp} filter, see below.
20181 To enable compilation of this filter you need to configure FFmpeg with
20182 @code{--enable-libvidstab}.
20184 @subsection Options
20188 Set path to the file used to read the transforms. Default value is
20189 @file{transforms.trf}.
20192 Set the number of frames (value*2 + 1) used for lowpass filtering the
20193 camera movements. Default value is 10.
20195 For example a number of 10 means that 21 frames are used (10 in the
20196 past and 10 in the future) to smoothen the motion in the video. A
20197 larger value leads to a smoother video, but limits the acceleration of
20198 the camera (pan/tilt movements). 0 is a special case where a static
20199 camera is simulated.
20202 Set the camera path optimization algorithm.
20204 Accepted values are:
20207 gaussian kernel low-pass filter on camera motion (default)
20209 averaging on transformations
20213 Set maximal number of pixels to translate frames. Default value is -1,
20217 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20218 value is -1, meaning no limit.
20221 Specify how to deal with borders that may be visible due to movement
20224 Available values are:
20227 keep image information from previous frame (default)
20229 fill the border black
20233 Invert transforms if set to 1. Default value is 0.
20236 Consider transforms as relative to previous frame if set to 1,
20237 absolute if set to 0. Default value is 0.
20240 Set percentage to zoom. A positive value will result in a zoom-in
20241 effect, a negative value in a zoom-out effect. Default value is 0 (no
20245 Set optimal zooming to avoid borders.
20247 Accepted values are:
20252 optimal static zoom value is determined (only very strong movements
20253 will lead to visible borders) (default)
20255 optimal adaptive zoom value is determined (no borders will be
20256 visible), see @option{zoomspeed}
20259 Note that the value given at zoom is added to the one calculated here.
20262 Set percent to zoom maximally each frame (enabled when
20263 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20267 Specify type of interpolation.
20269 Available values are:
20274 linear only horizontal
20276 linear in both directions (default)
20278 cubic in both directions (slow)
20282 Enable virtual tripod mode if set to 1, which is equivalent to
20283 @code{relative=0:smoothing=0}. Default value is 0.
20285 Use also @code{tripod} option of @ref{vidstabdetect}.
20288 Increase log verbosity if set to 1. Also the detected global motions
20289 are written to the temporary file @file{global_motions.trf}. Default
20293 @subsection Examples
20297 Use @command{ffmpeg} for a typical stabilization with default values:
20299 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20302 Note the use of the @ref{unsharp} filter which is always recommended.
20305 Zoom in a bit more and load transform data from a given file:
20307 vidstabtransform=zoom=5:input="mytransforms.trf"
20311 Smoothen the video even more:
20313 vidstabtransform=smoothing=30
20319 Flip the input video vertically.
20321 For example, to vertically flip a video with @command{ffmpeg}:
20323 ffmpeg -i in.avi -vf "vflip" out.avi
20328 Detect variable frame rate video.
20330 This filter tries to detect if the input is variable or constant frame rate.
20332 At end it will output number of frames detected as having variable delta pts,
20333 and ones with constant delta pts.
20334 If there was frames with variable delta, than it will also show min, max and
20335 average delta encountered.
20339 Boost or alter saturation.
20341 The filter accepts the following options:
20344 Set strength of boost if positive value or strength of alter if negative value.
20345 Default is 0. Allowed range is from -2 to 2.
20348 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20351 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20354 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20357 Set the red luma coefficient.
20360 Set the green luma coefficient.
20363 Set the blue luma coefficient.
20366 If @code{intensity} is negative and this is set to 1, colors will change,
20367 otherwise colors will be less saturated, more towards gray.
20370 @subsection Commands
20372 This filter supports the all above options as @ref{commands}.
20377 Make or reverse a natural vignetting effect.
20379 The filter accepts the following options:
20383 Set lens angle expression as a number of radians.
20385 The value is clipped in the @code{[0,PI/2]} range.
20387 Default value: @code{"PI/5"}
20391 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20395 Set forward/backward mode.
20397 Available modes are:
20400 The larger the distance from the central point, the darker the image becomes.
20403 The larger the distance from the central point, the brighter the image becomes.
20404 This can be used to reverse a vignette effect, though there is no automatic
20405 detection to extract the lens @option{angle} and other settings (yet). It can
20406 also be used to create a burning effect.
20409 Default value is @samp{forward}.
20412 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20414 It accepts the following values:
20417 Evaluate expressions only once during the filter initialization.
20420 Evaluate expressions for each incoming frame. This is way slower than the
20421 @samp{init} mode since it requires all the scalers to be re-computed, but it
20422 allows advanced dynamic expressions.
20425 Default value is @samp{init}.
20428 Set dithering to reduce the circular banding effects. Default is @code{1}
20432 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20433 Setting this value to the SAR of the input will make a rectangular vignetting
20434 following the dimensions of the video.
20436 Default is @code{1/1}.
20439 @subsection Expressions
20441 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20442 following parameters.
20447 input width and height
20450 the number of input frame, starting from 0
20453 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20454 @var{TB} units, NAN if undefined
20457 frame rate of the input video, NAN if the input frame rate is unknown
20460 the PTS (Presentation TimeStamp) of the filtered video frame,
20461 expressed in seconds, NAN if undefined
20464 time base of the input video
20468 @subsection Examples
20472 Apply simple strong vignetting effect:
20478 Make a flickering vignetting:
20480 vignette='PI/4+random(1)*PI/50':eval=frame
20485 @section vmafmotion
20487 Obtain the average VMAF motion score of a video.
20488 It is one of the component metrics of VMAF.
20490 The obtained average motion score is printed through the logging system.
20492 The filter accepts the following options:
20496 If specified, the filter will use the named file to save the motion score of
20497 each frame with respect to the previous frame.
20498 When filename equals "-" the data is sent to standard output.
20503 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20507 Stack input videos vertically.
20509 All streams must be of same pixel format and of same width.
20511 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20512 to create same output.
20514 The filter accepts the following options:
20518 Set number of input streams. Default is 2.
20521 If set to 1, force the output to terminate when the shortest input
20522 terminates. Default value is 0.
20527 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20528 Deinterlacing Filter").
20530 Based on the process described by Martin Weston for BBC R&D, and
20531 implemented based on the de-interlace algorithm written by Jim
20532 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20533 uses filter coefficients calculated by BBC R&D.
20535 This filter uses field-dominance information in frame to decide which
20536 of each pair of fields to place first in the output.
20537 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20539 There are two sets of filter coefficients, so called "simple"
20540 and "complex". Which set of filter coefficients is used can
20541 be set by passing an optional parameter:
20545 Set the interlacing filter coefficients. Accepts one of the following values:
20549 Simple filter coefficient set.
20551 More-complex filter coefficient set.
20553 Default value is @samp{complex}.
20556 Specify which frames to deinterlace. Accepts one of the following values:
20560 Deinterlace all frames,
20562 Only deinterlace frames marked as interlaced.
20565 Default value is @samp{all}.
20569 Video waveform monitor.
20571 The waveform monitor plots color component intensity. By default luminance
20572 only. Each column of the waveform corresponds to a column of pixels in the
20575 It accepts the following options:
20579 Can be either @code{row}, or @code{column}. Default is @code{column}.
20580 In row mode, the graph on the left side represents color component value 0 and
20581 the right side represents value = 255. In column mode, the top side represents
20582 color component value = 0 and bottom side represents value = 255.
20585 Set intensity. Smaller values are useful to find out how many values of the same
20586 luminance are distributed across input rows/columns.
20587 Default value is @code{0.04}. Allowed range is [0, 1].
20590 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20591 In mirrored mode, higher values will be represented on the left
20592 side for @code{row} mode and at the top for @code{column} mode. Default is
20593 @code{1} (mirrored).
20597 It accepts the following values:
20600 Presents information identical to that in the @code{parade}, except
20601 that the graphs representing color components are superimposed directly
20604 This display mode makes it easier to spot relative differences or similarities
20605 in overlapping areas of the color components that are supposed to be identical,
20606 such as neutral whites, grays, or blacks.
20609 Display separate graph for the color components side by side in
20610 @code{row} mode or one below the other in @code{column} mode.
20613 Display separate graph for the color components side by side in
20614 @code{column} mode or one below the other in @code{row} mode.
20616 Using this display mode makes it easy to spot color casts in the highlights
20617 and shadows of an image, by comparing the contours of the top and the bottom
20618 graphs of each waveform. Since whites, grays, and blacks are characterized
20619 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20620 should display three waveforms of roughly equal width/height. If not, the
20621 correction is easy to perform by making level adjustments the three waveforms.
20623 Default is @code{stack}.
20625 @item components, c
20626 Set which color components to display. Default is 1, which means only luminance
20627 or red color component if input is in RGB colorspace. If is set for example to
20628 7 it will display all 3 (if) available color components.
20633 No envelope, this is default.
20636 Instant envelope, minimum and maximum values presented in graph will be easily
20637 visible even with small @code{step} value.
20640 Hold minimum and maximum values presented in graph across time. This way you
20641 can still spot out of range values without constantly looking at waveforms.
20644 Peak and instant envelope combined together.
20650 No filtering, this is default.
20653 Luma and chroma combined together.
20656 Similar as above, but shows difference between blue and red chroma.
20659 Similar as above, but use different colors.
20662 Similar as above, but again with different colors.
20665 Displays only chroma.
20668 Displays actual color value on waveform.
20671 Similar as above, but with luma showing frequency of chroma values.
20675 Set which graticule to display.
20679 Do not display graticule.
20682 Display green graticule showing legal broadcast ranges.
20685 Display orange graticule showing legal broadcast ranges.
20688 Display invert graticule showing legal broadcast ranges.
20692 Set graticule opacity.
20695 Set graticule flags.
20699 Draw numbers above lines. By default enabled.
20702 Draw dots instead of lines.
20706 Set scale used for displaying graticule.
20713 Default is digital.
20716 Set background opacity.
20720 Set tint for output.
20721 Only used with lowpass filter and when display is not overlay and input
20722 pixel formats are not RGB.
20725 @section weave, doubleweave
20727 The @code{weave} takes a field-based video input and join
20728 each two sequential fields into single frame, producing a new double
20729 height clip with half the frame rate and half the frame count.
20731 The @code{doubleweave} works same as @code{weave} but without
20732 halving frame rate and frame count.
20734 It accepts the following option:
20738 Set first field. Available values are:
20742 Set the frame as top-field-first.
20745 Set the frame as bottom-field-first.
20749 @subsection Examples
20753 Interlace video using @ref{select} and @ref{separatefields} filter:
20755 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20760 Apply the xBR high-quality magnification filter which is designed for pixel
20761 art. It follows a set of edge-detection rules, see
20762 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
20764 It accepts the following option:
20768 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
20769 @code{3xBR} and @code{4} for @code{4xBR}.
20770 Default is @code{3}.
20775 Apply cross fade from one input video stream to another input video stream.
20776 The cross fade is applied for specified duration.
20778 The filter accepts the following options:
20782 Set one of available transition effects:
20830 Default transition effect is fade.
20833 Set cross fade duration in seconds.
20834 Default duration is 1 second.
20837 Set cross fade start relative to first input stream in seconds.
20838 Default offset is 0.
20841 Set expression for custom transition effect.
20843 The expressions can use the following variables and functions:
20848 The coordinates of the current sample.
20852 The width and height of the image.
20855 Progress of transition effect.
20858 Currently processed plane.
20861 Return value of first input at current location and plane.
20864 Return value of second input at current location and plane.
20870 Return the value of the pixel at location (@var{x},@var{y}) of the
20871 first/second/third/fourth component of first input.
20877 Return the value of the pixel at location (@var{x},@var{y}) of the
20878 first/second/third/fourth component of second input.
20882 @subsection Examples
20886 Cross fade from one input video to another input video, with fade transition and duration of transition
20887 of 2 seconds starting at offset of 5 seconds:
20889 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20894 Pick median pixels from several input videos.
20896 The filter accepts the following options:
20900 Set number of inputs.
20901 Default is 3. Allowed range is from 3 to 255.
20902 If number of inputs is even number, than result will be mean value between two median values.
20905 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
20908 Set median percentile. Default value is @code{0.5}.
20909 Default value of @code{0.5} will pick always median values, while @code{0} will pick
20910 minimum values, and @code{1} maximum values.
20914 Stack video inputs into custom layout.
20916 All streams must be of same pixel format.
20918 The filter accepts the following options:
20922 Set number of input streams. Default is 2.
20925 Specify layout of inputs.
20926 This option requires the desired layout configuration to be explicitly set by the user.
20927 This sets position of each video input in output. Each input
20928 is separated by '|'.
20929 The first number represents the column, and the second number represents the row.
20930 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
20931 where X is video input from which to take width or height.
20932 Multiple values can be used when separated by '+'. In such
20933 case values are summed together.
20935 Note that if inputs are of different sizes gaps may appear, as not all of
20936 the output video frame will be filled. Similarly, videos can overlap each
20937 other if their position doesn't leave enough space for the full frame of
20940 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
20941 a layout must be set by the user.
20944 If set to 1, force the output to terminate when the shortest input
20945 terminates. Default value is 0.
20948 If set to valid color, all unused pixels will be filled with that color.
20949 By default fill is set to none, so it is disabled.
20952 @subsection Examples
20956 Display 4 inputs into 2x2 grid.
20960 input1(0, 0) | input3(w0, 0)
20961 input2(0, h0) | input4(w0, h0)
20965 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20968 Note that if inputs are of different sizes, gaps or overlaps may occur.
20971 Display 4 inputs into 1x4 grid.
20978 input4(0, h0+h1+h2)
20982 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20985 Note that if inputs are of different widths, unused space will appear.
20988 Display 9 inputs into 3x3 grid.
20992 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20993 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20994 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20998 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
21001 Note that if inputs are of different sizes, gaps or overlaps may occur.
21004 Display 16 inputs into 4x4 grid.
21008 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21009 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21010 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21011 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21015 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|
21016 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
21019 Note that if inputs are of different sizes, gaps or overlaps may occur.
21026 Deinterlace the input video ("yadif" means "yet another deinterlacing
21029 It accepts the following parameters:
21035 The interlacing mode to adopt. It accepts one of the following values:
21038 @item 0, send_frame
21039 Output one frame for each frame.
21040 @item 1, send_field
21041 Output one frame for each field.
21042 @item 2, send_frame_nospatial
21043 Like @code{send_frame}, but it skips the spatial interlacing check.
21044 @item 3, send_field_nospatial
21045 Like @code{send_field}, but it skips the spatial interlacing check.
21048 The default value is @code{send_frame}.
21051 The picture field parity assumed for the input interlaced video. It accepts one
21052 of the following values:
21056 Assume the top field is first.
21058 Assume the bottom field is first.
21060 Enable automatic detection of field parity.
21063 The default value is @code{auto}.
21064 If the interlacing is unknown or the decoder does not export this information,
21065 top field first will be assumed.
21068 Specify which frames to deinterlace. Accepts one of the following
21073 Deinterlace all frames.
21074 @item 1, interlaced
21075 Only deinterlace frames marked as interlaced.
21078 The default value is @code{all}.
21081 @section yadif_cuda
21083 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21084 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21087 It accepts the following parameters:
21093 The interlacing mode to adopt. It accepts one of the following values:
21096 @item 0, send_frame
21097 Output one frame for each frame.
21098 @item 1, send_field
21099 Output one frame for each field.
21100 @item 2, send_frame_nospatial
21101 Like @code{send_frame}, but it skips the spatial interlacing check.
21102 @item 3, send_field_nospatial
21103 Like @code{send_field}, but it skips the spatial interlacing check.
21106 The default value is @code{send_frame}.
21109 The picture field parity assumed for the input interlaced video. It accepts one
21110 of the following values:
21114 Assume the top field is first.
21116 Assume the bottom field is first.
21118 Enable automatic detection of field parity.
21121 The default value is @code{auto}.
21122 If the interlacing is unknown or the decoder does not export this information,
21123 top field first will be assumed.
21126 Specify which frames to deinterlace. Accepts one of the following
21131 Deinterlace all frames.
21132 @item 1, interlaced
21133 Only deinterlace frames marked as interlaced.
21136 The default value is @code{all}.
21141 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21142 The algorithm is described in
21143 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21145 It accepts the following parameters:
21149 Set the window radius. Default value is 3.
21152 Set which planes to filter. Default is only the first plane.
21155 Set blur strength. Default value is 128.
21158 @subsection Commands
21159 This filter supports same @ref{commands} as options.
21163 Apply Zoom & Pan effect.
21165 This filter accepts the following options:
21169 Set the zoom expression. Range is 1-10. Default is 1.
21173 Set the x and y expression. Default is 0.
21176 Set the duration expression in number of frames.
21177 This sets for how many number of frames effect will last for
21178 single input image.
21181 Set the output image size, default is 'hd720'.
21184 Set the output frame rate, default is '25'.
21187 Each expression can contain the following constants:
21206 Output frame count.
21209 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21211 @item out_time, time, ot
21212 The output timestamp expressed in seconds.
21216 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21217 for current input frame.
21221 'x' and 'y' of last output frame of previous input frame or 0 when there was
21222 not yet such frame (first input frame).
21225 Last calculated zoom from 'z' expression for current input frame.
21228 Last calculated zoom of last output frame of previous input frame.
21231 Number of output frames for current input frame. Calculated from 'd' expression
21232 for each input frame.
21235 number of output frames created for previous input frame
21238 Rational number: input width / input height
21241 sample aspect ratio
21244 display aspect ratio
21248 @subsection Examples
21252 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21254 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
21258 Zoom in up to 1.5x and pan always at center of picture:
21260 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21264 Same as above but without pausing:
21266 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21270 Zoom in 2x into center of picture only for the first second of the input video:
21272 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21279 Scale (resize) the input video, using the z.lib library:
21280 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21281 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21283 The zscale filter forces the output display aspect ratio to be the same
21284 as the input, by changing the output sample aspect ratio.
21286 If the input image format is different from the format requested by
21287 the next filter, the zscale filter will convert the input to the
21290 @subsection Options
21291 The filter accepts the following options.
21296 Set the output video dimension expression. Default value is the input
21299 If the @var{width} or @var{w} value is 0, the input width is used for
21300 the output. If the @var{height} or @var{h} value is 0, the input height
21301 is used for the output.
21303 If one and only one of the values is -n with n >= 1, the zscale filter
21304 will use a value that maintains the aspect ratio of the input image,
21305 calculated from the other specified dimension. After that it will,
21306 however, make sure that the calculated dimension is divisible by n and
21307 adjust the value if necessary.
21309 If both values are -n with n >= 1, the behavior will be identical to
21310 both values being set to 0 as previously detailed.
21312 See below for the list of accepted constants for use in the dimension
21316 Set the video size. For the syntax of this option, check the
21317 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21320 Set the dither type.
21322 Possible values are:
21327 @item error_diffusion
21333 Set the resize filter type.
21335 Possible values are:
21345 Default is bilinear.
21348 Set the color range.
21350 Possible values are:
21357 Default is same as input.
21360 Set the color primaries.
21362 Possible values are:
21372 Default is same as input.
21375 Set the transfer characteristics.
21377 Possible values are:
21391 Default is same as input.
21394 Set the colorspace matrix.
21396 Possible value are:
21407 Default is same as input.
21410 Set the input color range.
21412 Possible values are:
21419 Default is same as input.
21421 @item primariesin, pin
21422 Set the input color primaries.
21424 Possible values are:
21434 Default is same as input.
21436 @item transferin, tin
21437 Set the input transfer characteristics.
21439 Possible values are:
21450 Default is same as input.
21452 @item matrixin, min
21453 Set the input colorspace matrix.
21455 Possible value are:
21467 Set the output chroma location.
21469 Possible values are:
21480 @item chromalin, cin
21481 Set the input chroma location.
21483 Possible values are:
21495 Set the nominal peak luminance.
21498 The values of the @option{w} and @option{h} options are expressions
21499 containing the following constants:
21504 The input width and height
21508 These are the same as @var{in_w} and @var{in_h}.
21512 The output (scaled) width and height
21516 These are the same as @var{out_w} and @var{out_h}
21519 The same as @var{iw} / @var{ih}
21522 input sample aspect ratio
21525 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21529 horizontal and vertical input chroma subsample values. For example for the
21530 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21534 horizontal and vertical output chroma subsample values. For example for the
21535 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21538 @subsection Commands
21540 This filter supports the following commands:
21544 Set the output video dimension expression.
21545 The command accepts the same syntax of the corresponding option.
21547 If the specified expression is not valid, it is kept at its current
21551 @c man end VIDEO FILTERS
21553 @chapter OpenCL Video Filters
21554 @c man begin OPENCL VIDEO FILTERS
21556 Below is a description of the currently available OpenCL video filters.
21558 To enable compilation of these filters you need to configure FFmpeg with
21559 @code{--enable-opencl}.
21561 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21564 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21565 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21566 given device parameters.
21568 @item -filter_hw_device @var{name}
21569 Pass the hardware device called @var{name} to all filters in any filter graph.
21573 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21577 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21579 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21583 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.
21585 @section avgblur_opencl
21587 Apply average blur filter.
21589 The filter accepts the following options:
21593 Set horizontal radius size.
21594 Range is @code{[1, 1024]} and default value is @code{1}.
21597 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21600 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21603 @subsection Example
21607 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.
21609 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21613 @section boxblur_opencl
21615 Apply a boxblur algorithm to the input video.
21617 It accepts the following parameters:
21621 @item luma_radius, lr
21622 @item luma_power, lp
21623 @item chroma_radius, cr
21624 @item chroma_power, cp
21625 @item alpha_radius, ar
21626 @item alpha_power, ap
21630 A description of the accepted options follows.
21633 @item luma_radius, lr
21634 @item chroma_radius, cr
21635 @item alpha_radius, ar
21636 Set an expression for the box radius in pixels used for blurring the
21637 corresponding input plane.
21639 The radius value must be a non-negative number, and must not be
21640 greater than the value of the expression @code{min(w,h)/2} for the
21641 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21644 Default value for @option{luma_radius} is "2". If not specified,
21645 @option{chroma_radius} and @option{alpha_radius} default to the
21646 corresponding value set for @option{luma_radius}.
21648 The expressions can contain the following constants:
21652 The input width and height in pixels.
21656 The input chroma image width and height in pixels.
21660 The horizontal and vertical chroma subsample values. For example, for the
21661 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21664 @item luma_power, lp
21665 @item chroma_power, cp
21666 @item alpha_power, ap
21667 Specify how many times the boxblur filter is applied to the
21668 corresponding plane.
21670 Default value for @option{luma_power} is 2. If not specified,
21671 @option{chroma_power} and @option{alpha_power} default to the
21672 corresponding value set for @option{luma_power}.
21674 A value of 0 will disable the effect.
21677 @subsection Examples
21679 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.
21683 Apply a boxblur filter with the luma, chroma, and alpha radius
21684 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.
21686 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21687 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21691 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.
21693 For the luma plane, a 2x2 box radius will be run once.
21695 For the chroma plane, a 4x4 box radius will be run 5 times.
21697 For the alpha plane, a 3x3 box radius will be run 7 times.
21699 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
21703 @section colorkey_opencl
21704 RGB colorspace color keying.
21706 The filter accepts the following options:
21710 The color which will be replaced with transparency.
21713 Similarity percentage with the key color.
21715 0.01 matches only the exact key color, while 1.0 matches everything.
21720 0.0 makes pixels either fully transparent, or not transparent at all.
21722 Higher values result in semi-transparent pixels, with a higher transparency
21723 the more similar the pixels color is to the key color.
21726 @subsection Examples
21730 Make every semi-green pixel in the input transparent with some slight blending:
21732 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
21736 @section convolution_opencl
21738 Apply convolution of 3x3, 5x5, 7x7 matrix.
21740 The filter accepts the following options:
21747 Set matrix for each plane.
21748 Matrix is sequence of 9, 25 or 49 signed numbers.
21749 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
21755 Set multiplier for calculated value for each plane.
21756 If unset or 0, it will be sum of all matrix elements.
21757 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
21763 Set bias for each plane. This value is added to the result of the multiplication.
21764 Useful for making the overall image brighter or darker.
21765 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
21769 @subsection Examples
21775 -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
21781 -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
21785 Apply edge enhance:
21787 -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
21793 -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
21797 Apply laplacian edge detector which includes diagonals:
21799 -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
21805 -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
21809 @section erosion_opencl
21811 Apply erosion effect to the video.
21813 This filter replaces the pixel by the local(3x3) minimum.
21815 It accepts the following options:
21822 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21823 If @code{0}, plane will remain unchanged.
21826 Flag which specifies the pixel to refer to.
21827 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21829 Flags to local 3x3 coordinates region centered on @code{x}:
21838 @subsection Example
21842 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.
21844 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21848 @section deshake_opencl
21849 Feature-point based video stabilization filter.
21851 The filter accepts the following options:
21855 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
21858 Whether or not additional debug info should be displayed, both in the processed output and in the console.
21860 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
21862 Viewing point matches in the output video is only supported for RGB input.
21864 Defaults to @code{0}.
21866 @item adaptive_crop
21867 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
21869 Defaults to @code{1}.
21871 @item refine_features
21872 Whether or not feature points should be refined at a sub-pixel level.
21874 This can be turned off for a slight performance gain at the cost of precision.
21876 Defaults to @code{1}.
21878 @item smooth_strength
21879 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
21881 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
21883 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
21885 Defaults to @code{0.0}.
21887 @item smooth_window_multiplier
21888 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
21890 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
21892 Acceptable values range from @code{0.1} to @code{10.0}.
21894 Larger values increase the amount of motion data available for determining how to smooth the camera path,
21895 potentially improving smoothness, but also increase latency and memory usage.
21897 Defaults to @code{2.0}.
21901 @subsection Examples
21905 Stabilize a video with a fixed, medium smoothing strength:
21907 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21911 Stabilize a video with debugging (both in console and in rendered video):
21913 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21917 @section dilation_opencl
21919 Apply dilation effect to the video.
21921 This filter replaces the pixel by the local(3x3) maximum.
21923 It accepts the following options:
21930 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21931 If @code{0}, plane will remain unchanged.
21934 Flag which specifies the pixel to refer to.
21935 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21937 Flags to local 3x3 coordinates region centered on @code{x}:
21946 @subsection Example
21950 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.
21952 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21956 @section nlmeans_opencl
21958 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
21960 @section overlay_opencl
21962 Overlay one video on top of another.
21964 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
21965 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
21967 The filter accepts the following options:
21972 Set the x coordinate of the overlaid video on the main video.
21973 Default value is @code{0}.
21976 Set the y coordinate of the overlaid video on the main video.
21977 Default value is @code{0}.
21981 @subsection Examples
21985 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
21987 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21990 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
21992 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21997 @section pad_opencl
21999 Add paddings to the input image, and place the original input at the
22000 provided @var{x}, @var{y} coordinates.
22002 It accepts the following options:
22007 Specify an expression for the size of the output image with the
22008 paddings added. If the value for @var{width} or @var{height} is 0, the
22009 corresponding input size is used for the output.
22011 The @var{width} expression can reference the value set by the
22012 @var{height} expression, and vice versa.
22014 The default value of @var{width} and @var{height} is 0.
22018 Specify the offsets to place the input image at within the padded area,
22019 with respect to the top/left border of the output image.
22021 The @var{x} expression can reference the value set by the @var{y}
22022 expression, and vice versa.
22024 The default value of @var{x} and @var{y} is 0.
22026 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22027 so the input image is centered on the padded area.
22030 Specify the color of the padded area. For the syntax of this option,
22031 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22032 manual,ffmpeg-utils}.
22035 Pad to an aspect instead to a resolution.
22038 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22039 options are expressions containing the following constants:
22044 The input video width and height.
22048 These are the same as @var{in_w} and @var{in_h}.
22052 The output width and height (the size of the padded area), as
22053 specified by the @var{width} and @var{height} expressions.
22057 These are the same as @var{out_w} and @var{out_h}.
22061 The x and y offsets as specified by the @var{x} and @var{y}
22062 expressions, or NAN if not yet specified.
22065 same as @var{iw} / @var{ih}
22068 input sample aspect ratio
22071 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22074 @section prewitt_opencl
22076 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22078 The filter accepts the following option:
22082 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22085 Set value which will be multiplied with filtered result.
22086 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22089 Set value which will be added to filtered result.
22090 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22093 @subsection Example
22097 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22099 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22103 @anchor{program_opencl}
22104 @section program_opencl
22106 Filter video using an OpenCL program.
22111 OpenCL program source file.
22114 Kernel name in program.
22117 Number of inputs to the filter. Defaults to 1.
22120 Size of output frames. Defaults to the same as the first input.
22124 The @code{program_opencl} filter also supports the @ref{framesync} options.
22126 The program source file must contain a kernel function with the given name,
22127 which will be run once for each plane of the output. Each run on a plane
22128 gets enqueued as a separate 2D global NDRange with one work-item for each
22129 pixel to be generated. The global ID offset for each work-item is therefore
22130 the coordinates of a pixel in the destination image.
22132 The kernel function needs to take the following arguments:
22135 Destination image, @var{__write_only image2d_t}.
22137 This image will become the output; the kernel should write all of it.
22139 Frame index, @var{unsigned int}.
22141 This is a counter starting from zero and increasing by one for each frame.
22143 Source images, @var{__read_only image2d_t}.
22145 These are the most recent images on each input. The kernel may read from
22146 them to generate the output, but they can't be written to.
22153 Copy the input to the output (output must be the same size as the input).
22155 __kernel void copy(__write_only image2d_t destination,
22156 unsigned int index,
22157 __read_only image2d_t source)
22159 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22161 int2 location = (int2)(get_global_id(0), get_global_id(1));
22163 float4 value = read_imagef(source, sampler, location);
22165 write_imagef(destination, location, value);
22170 Apply a simple transformation, rotating the input by an amount increasing
22171 with the index counter. Pixel values are linearly interpolated by the
22172 sampler, and the output need not have the same dimensions as the input.
22174 __kernel void rotate_image(__write_only image2d_t dst,
22175 unsigned int index,
22176 __read_only image2d_t src)
22178 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22179 CLK_FILTER_LINEAR);
22181 float angle = (float)index / 100.0f;
22183 float2 dst_dim = convert_float2(get_image_dim(dst));
22184 float2 src_dim = convert_float2(get_image_dim(src));
22186 float2 dst_cen = dst_dim / 2.0f;
22187 float2 src_cen = src_dim / 2.0f;
22189 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22191 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22193 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22194 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22196 src_pos = src_pos * src_dim / dst_dim;
22198 float2 src_loc = src_pos + src_cen;
22200 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22201 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22202 write_imagef(dst, dst_loc, 0.5f);
22204 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22209 Blend two inputs together, with the amount of each input used varying
22210 with the index counter.
22212 __kernel void blend_images(__write_only image2d_t dst,
22213 unsigned int index,
22214 __read_only image2d_t src1,
22215 __read_only image2d_t src2)
22217 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22218 CLK_FILTER_LINEAR);
22220 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22222 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22223 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22224 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22226 float4 val1 = read_imagef(src1, sampler, src1_loc);
22227 float4 val2 = read_imagef(src2, sampler, src2_loc);
22229 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22235 @section roberts_opencl
22236 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22238 The filter accepts the following option:
22242 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22245 Set value which will be multiplied with filtered result.
22246 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22249 Set value which will be added to filtered result.
22250 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22253 @subsection Example
22257 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22259 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22263 @section sobel_opencl
22265 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22267 The filter accepts the following option:
22271 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22274 Set value which will be multiplied with filtered result.
22275 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22278 Set value which will be added to filtered result.
22279 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22282 @subsection Example
22286 Apply sobel operator with scale set to 2 and delta set to 10
22288 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22292 @section tonemap_opencl
22294 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22296 It accepts the following parameters:
22300 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22303 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22306 Apply desaturation for highlights that exceed this level of brightness. The
22307 higher the parameter, the more color information will be preserved. This
22308 setting helps prevent unnaturally blown-out colors for super-highlights, by
22309 (smoothly) turning into white instead. This makes images feel more natural,
22310 at the cost of reducing information about out-of-range colors.
22312 The default value is 0.5, and the algorithm here is a little different from
22313 the cpu version tonemap currently. A setting of 0.0 disables this option.
22316 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22317 is used to detect whether the scene has changed or not. If the distance between
22318 the current frame average brightness and the current running average exceeds
22319 a threshold value, we would re-calculate scene average and peak brightness.
22320 The default value is 0.2.
22323 Specify the output pixel format.
22325 Currently supported formats are:
22332 Set the output color range.
22334 Possible values are:
22340 Default is same as input.
22343 Set the output color primaries.
22345 Possible values are:
22351 Default is same as input.
22354 Set the output transfer characteristics.
22356 Possible values are:
22365 Set the output colorspace matrix.
22367 Possible value are:
22373 Default is same as input.
22377 @subsection Example
22381 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22383 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22387 @section unsharp_opencl
22389 Sharpen or blur the input video.
22391 It accepts the following parameters:
22394 @item luma_msize_x, lx
22395 Set the luma matrix horizontal size.
22396 Range is @code{[1, 23]} and default value is @code{5}.
22398 @item luma_msize_y, ly
22399 Set the luma matrix vertical size.
22400 Range is @code{[1, 23]} and default value is @code{5}.
22402 @item luma_amount, la
22403 Set the luma effect strength.
22404 Range is @code{[-10, 10]} and default value is @code{1.0}.
22406 Negative values will blur the input video, while positive values will
22407 sharpen it, a value of zero will disable the effect.
22409 @item chroma_msize_x, cx
22410 Set the chroma matrix horizontal size.
22411 Range is @code{[1, 23]} and default value is @code{5}.
22413 @item chroma_msize_y, cy
22414 Set the chroma matrix vertical size.
22415 Range is @code{[1, 23]} and default value is @code{5}.
22417 @item chroma_amount, ca
22418 Set the chroma effect strength.
22419 Range is @code{[-10, 10]} and default value is @code{0.0}.
22421 Negative values will blur the input video, while positive values will
22422 sharpen it, a value of zero will disable the effect.
22426 All parameters are optional and default to the equivalent of the
22427 string '5:5:1.0:5:5:0.0'.
22429 @subsection Examples
22433 Apply strong luma sharpen effect:
22435 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22439 Apply a strong blur of both luma and chroma parameters:
22441 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22445 @section xfade_opencl
22447 Cross fade two videos with custom transition effect by using OpenCL.
22449 It accepts the following options:
22453 Set one of possible transition effects.
22457 Select custom transition effect, the actual transition description
22458 will be picked from source and kernel options.
22470 Default transition is fade.
22474 OpenCL program source file for custom transition.
22477 Set name of kernel to use for custom transition from program source file.
22480 Set duration of video transition.
22483 Set time of start of transition relative to first video.
22486 The program source file must contain a kernel function with the given name,
22487 which will be run once for each plane of the output. Each run on a plane
22488 gets enqueued as a separate 2D global NDRange with one work-item for each
22489 pixel to be generated. The global ID offset for each work-item is therefore
22490 the coordinates of a pixel in the destination image.
22492 The kernel function needs to take the following arguments:
22495 Destination image, @var{__write_only image2d_t}.
22497 This image will become the output; the kernel should write all of it.
22500 First Source image, @var{__read_only image2d_t}.
22501 Second Source image, @var{__read_only image2d_t}.
22503 These are the most recent images on each input. The kernel may read from
22504 them to generate the output, but they can't be written to.
22507 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22514 Apply dots curtain transition effect:
22516 __kernel void blend_images(__write_only image2d_t dst,
22517 __read_only image2d_t src1,
22518 __read_only image2d_t src2,
22521 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22522 CLK_FILTER_LINEAR);
22523 int2 p = (int2)(get_global_id(0), get_global_id(1));
22524 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22525 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22528 float2 dots = (float2)(20.0, 20.0);
22529 float2 center = (float2)(0,0);
22532 float4 val1 = read_imagef(src1, sampler, p);
22533 float4 val2 = read_imagef(src2, sampler, p);
22534 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22536 write_imagef(dst, p, next ? val1 : val2);
22542 @c man end OPENCL VIDEO FILTERS
22544 @chapter VAAPI Video Filters
22545 @c man begin VAAPI VIDEO FILTERS
22547 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22549 To enable compilation of these filters you need to configure FFmpeg with
22550 @code{--enable-vaapi}.
22552 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}
22554 @section tonemap_vaapi
22556 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22557 It maps the dynamic range of HDR10 content to the SDR content.
22558 It currently only accepts HDR10 as input.
22560 It accepts the following parameters:
22564 Specify the output pixel format.
22566 Currently supported formats are:
22575 Set the output color primaries.
22577 Default is same as input.
22580 Set the output transfer characteristics.
22585 Set the output colorspace matrix.
22587 Default is same as input.
22591 @subsection Example
22595 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22597 tonemap_vaapi=format=p010:t=bt2020-10
22601 @c man end VAAPI VIDEO FILTERS
22603 @chapter Video Sources
22604 @c man begin VIDEO SOURCES
22606 Below is a description of the currently available video sources.
22610 Buffer video frames, and make them available to the filter chain.
22612 This source is mainly intended for a programmatic use, in particular
22613 through the interface defined in @file{libavfilter/buffersrc.h}.
22615 It accepts the following parameters:
22620 Specify the size (width and height) of the buffered video frames. For the
22621 syntax of this option, check the
22622 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22625 The input video width.
22628 The input video height.
22631 A string representing the pixel format of the buffered video frames.
22632 It may be a number corresponding to a pixel format, or a pixel format
22636 Specify the timebase assumed by the timestamps of the buffered frames.
22639 Specify the frame rate expected for the video stream.
22641 @item pixel_aspect, sar
22642 The sample (pixel) aspect ratio of the input video.
22645 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22646 to the filtergraph description to specify swscale flags for automatically
22647 inserted scalers. See @ref{Filtergraph syntax}.
22649 @item hw_frames_ctx
22650 When using a hardware pixel format, this should be a reference to an
22651 AVHWFramesContext describing input frames.
22656 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22659 will instruct the source to accept video frames with size 320x240 and
22660 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22661 square pixels (1:1 sample aspect ratio).
22662 Since the pixel format with name "yuv410p" corresponds to the number 6
22663 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22664 this example corresponds to:
22666 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22669 Alternatively, the options can be specified as a flat string, but this
22670 syntax is deprecated:
22672 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22676 Create a pattern generated by an elementary cellular automaton.
22678 The initial state of the cellular automaton can be defined through the
22679 @option{filename} and @option{pattern} options. If such options are
22680 not specified an initial state is created randomly.
22682 At each new frame a new row in the video is filled with the result of
22683 the cellular automaton next generation. The behavior when the whole
22684 frame is filled is defined by the @option{scroll} option.
22686 This source accepts the following options:
22690 Read the initial cellular automaton state, i.e. the starting row, from
22691 the specified file.
22692 In the file, each non-whitespace character is considered an alive
22693 cell, a newline will terminate the row, and further characters in the
22694 file will be ignored.
22697 Read the initial cellular automaton state, i.e. the starting row, from
22698 the specified string.
22700 Each non-whitespace character in the string is considered an alive
22701 cell, a newline will terminate the row, and further characters in the
22702 string will be ignored.
22705 Set the video rate, that is the number of frames generated per second.
22708 @item random_fill_ratio, ratio
22709 Set the random fill ratio for the initial cellular automaton row. It
22710 is a floating point number value ranging from 0 to 1, defaults to
22713 This option is ignored when a file or a pattern is specified.
22715 @item random_seed, seed
22716 Set the seed for filling randomly the initial row, must be an integer
22717 included between 0 and UINT32_MAX. If not specified, or if explicitly
22718 set to -1, the filter will try to use a good random seed on a best
22722 Set the cellular automaton rule, it is a number ranging from 0 to 255.
22723 Default value is 110.
22726 Set the size of the output video. For the syntax of this option, check the
22727 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22729 If @option{filename} or @option{pattern} is specified, the size is set
22730 by default to the width of the specified initial state row, and the
22731 height is set to @var{width} * PHI.
22733 If @option{size} is set, it must contain the width of the specified
22734 pattern string, and the specified pattern will be centered in the
22737 If a filename or a pattern string is not specified, the size value
22738 defaults to "320x518" (used for a randomly generated initial state).
22741 If set to 1, scroll the output upward when all the rows in the output
22742 have been already filled. If set to 0, the new generated row will be
22743 written over the top row just after the bottom row is filled.
22746 @item start_full, full
22747 If set to 1, completely fill the output with generated rows before
22748 outputting the first frame.
22749 This is the default behavior, for disabling set the value to 0.
22752 If set to 1, stitch the left and right row edges together.
22753 This is the default behavior, for disabling set the value to 0.
22756 @subsection Examples
22760 Read the initial state from @file{pattern}, and specify an output of
22763 cellauto=f=pattern:s=200x400
22767 Generate a random initial row with a width of 200 cells, with a fill
22770 cellauto=ratio=2/3:s=200x200
22774 Create a pattern generated by rule 18 starting by a single alive cell
22775 centered on an initial row with width 100:
22777 cellauto=p=@@:s=100x400:full=0:rule=18
22781 Specify a more elaborated initial pattern:
22783 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
22788 @anchor{coreimagesrc}
22789 @section coreimagesrc
22790 Video source generated on GPU using Apple's CoreImage API on OSX.
22792 This video source is a specialized version of the @ref{coreimage} video filter.
22793 Use a core image generator at the beginning of the applied filterchain to
22794 generate the content.
22796 The coreimagesrc video source accepts the following options:
22798 @item list_generators
22799 List all available generators along with all their respective options as well as
22800 possible minimum and maximum values along with the default values.
22802 list_generators=true
22806 Specify the size of the sourced video. For the syntax of this option, check the
22807 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22808 The default value is @code{320x240}.
22811 Specify the frame rate of the sourced video, as the number of frames
22812 generated per second. It has to be a string in the format
22813 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22814 number or a valid video frame rate abbreviation. The default value is
22818 Set the sample aspect ratio of the sourced video.
22821 Set the duration of the sourced video. See
22822 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22823 for the accepted syntax.
22825 If not specified, or the expressed duration is negative, the video is
22826 supposed to be generated forever.
22829 Additionally, all options of the @ref{coreimage} video filter are accepted.
22830 A complete filterchain can be used for further processing of the
22831 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
22832 and examples for details.
22834 @subsection Examples
22839 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22840 given as complete and escaped command-line for Apple's standard bash shell:
22842 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
22844 This example is equivalent to the QRCode example of @ref{coreimage} without the
22845 need for a nullsrc video source.
22850 Generate several gradients.
22854 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22855 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22858 Set frame rate, expressed as number of frames per second. Default
22861 @item c0, c1, c2, c3, c4, c5, c6, c7
22862 Set 8 colors. Default values for colors is to pick random one.
22864 @item x0, y0, y0, y1
22865 Set gradient line source and destination points. If negative or out of range, random ones
22869 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
22872 Set seed for picking gradient line points.
22875 Set the duration of the sourced video. See
22876 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22877 for the accepted syntax.
22879 If not specified, or the expressed duration is negative, the video is
22880 supposed to be generated forever.
22883 Set speed of gradients rotation.
22887 @section mandelbrot
22889 Generate a Mandelbrot set fractal, and progressively zoom towards the
22890 point specified with @var{start_x} and @var{start_y}.
22892 This source accepts the following options:
22897 Set the terminal pts value. Default value is 400.
22900 Set the terminal scale value.
22901 Must be a floating point value. Default value is 0.3.
22904 Set the inner coloring mode, that is the algorithm used to draw the
22905 Mandelbrot fractal internal region.
22907 It shall assume one of the following values:
22912 Show time until convergence.
22914 Set color based on point closest to the origin of the iterations.
22919 Default value is @var{mincol}.
22922 Set the bailout value. Default value is 10.0.
22925 Set the maximum of iterations performed by the rendering
22926 algorithm. Default value is 7189.
22929 Set outer coloring mode.
22930 It shall assume one of following values:
22932 @item iteration_count
22933 Set iteration count mode.
22934 @item normalized_iteration_count
22935 set normalized iteration count mode.
22937 Default value is @var{normalized_iteration_count}.
22940 Set frame rate, expressed as number of frames per second. Default
22944 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22945 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22948 Set the initial scale value. Default value is 3.0.
22951 Set the initial x position. Must be a floating point value between
22952 -100 and 100. Default value is -0.743643887037158704752191506114774.
22955 Set the initial y position. Must be a floating point value between
22956 -100 and 100. Default value is -0.131825904205311970493132056385139.
22961 Generate various test patterns, as generated by the MPlayer test filter.
22963 The size of the generated video is fixed, and is 256x256.
22964 This source is useful in particular for testing encoding features.
22966 This source accepts the following options:
22971 Specify the frame rate of the sourced video, as the number of frames
22972 generated per second. It has to be a string in the format
22973 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22974 number or a valid video frame rate abbreviation. The default value is
22978 Set the duration of the sourced video. See
22979 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22980 for the accepted syntax.
22982 If not specified, or the expressed duration is negative, the video is
22983 supposed to be generated forever.
22987 Set the number or the name of the test to perform. Supported tests are:
23001 @item max_frames, m
23002 Set the maximum number of frames generated for each test, default value is 30.
23006 Default value is "all", which will cycle through the list of all tests.
23011 mptestsrc=t=dc_luma
23014 will generate a "dc_luma" test pattern.
23016 @section frei0r_src
23018 Provide a frei0r source.
23020 To enable compilation of this filter you need to install the frei0r
23021 header and configure FFmpeg with @code{--enable-frei0r}.
23023 This source accepts the following parameters:
23028 The size of the video to generate. For the syntax of this option, check the
23029 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23032 The framerate of the generated video. It may be a string of the form
23033 @var{num}/@var{den} or a frame rate abbreviation.
23036 The name to the frei0r source to load. For more information regarding frei0r and
23037 how to set the parameters, read the @ref{frei0r} section in the video filters
23040 @item filter_params
23041 A '|'-separated list of parameters to pass to the frei0r source.
23045 For example, to generate a frei0r partik0l source with size 200x200
23046 and frame rate 10 which is overlaid on the overlay filter main input:
23048 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23053 Generate a life pattern.
23055 This source is based on a generalization of John Conway's life game.
23057 The sourced input represents a life grid, each pixel represents a cell
23058 which can be in one of two possible states, alive or dead. Every cell
23059 interacts with its eight neighbours, which are the cells that are
23060 horizontally, vertically, or diagonally adjacent.
23062 At each interaction the grid evolves according to the adopted rule,
23063 which specifies the number of neighbor alive cells which will make a
23064 cell stay alive or born. The @option{rule} option allows one to specify
23067 This source accepts the following options:
23071 Set the file from which to read the initial grid state. In the file,
23072 each non-whitespace character is considered an alive cell, and newline
23073 is used to delimit the end of each row.
23075 If this option is not specified, the initial grid is generated
23079 Set the video rate, that is the number of frames generated per second.
23082 @item random_fill_ratio, ratio
23083 Set the random fill ratio for the initial random grid. It is a
23084 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23085 It is ignored when a file is specified.
23087 @item random_seed, seed
23088 Set the seed for filling the initial random grid, must be an integer
23089 included between 0 and UINT32_MAX. If not specified, or if explicitly
23090 set to -1, the filter will try to use a good random seed on a best
23096 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23097 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23098 @var{NS} specifies the number of alive neighbor cells which make a
23099 live cell stay alive, and @var{NB} the number of alive neighbor cells
23100 which make a dead cell to become alive (i.e. to "born").
23101 "s" and "b" can be used in place of "S" and "B", respectively.
23103 Alternatively a rule can be specified by an 18-bits integer. The 9
23104 high order bits are used to encode the next cell state if it is alive
23105 for each number of neighbor alive cells, the low order bits specify
23106 the rule for "borning" new cells. Higher order bits encode for an
23107 higher number of neighbor cells.
23108 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23109 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23111 Default value is "S23/B3", which is the original Conway's game of life
23112 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23113 cells, and will born a new cell if there are three alive cells around
23117 Set the size of the output video. For the syntax of this option, check the
23118 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23120 If @option{filename} is specified, the size is set by default to the
23121 same size of the input file. If @option{size} is set, it must contain
23122 the size specified in the input file, and the initial grid defined in
23123 that file is centered in the larger resulting area.
23125 If a filename is not specified, the size value defaults to "320x240"
23126 (used for a randomly generated initial grid).
23129 If set to 1, stitch the left and right grid edges together, and the
23130 top and bottom edges also. Defaults to 1.
23133 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23134 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23135 value from 0 to 255.
23138 Set the color of living (or new born) cells.
23141 Set the color of dead cells. If @option{mold} is set, this is the first color
23142 used to represent a dead cell.
23145 Set mold color, for definitely dead and moldy cells.
23147 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23148 ffmpeg-utils manual,ffmpeg-utils}.
23151 @subsection Examples
23155 Read a grid from @file{pattern}, and center it on a grid of size
23158 life=f=pattern:s=300x300
23162 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23164 life=ratio=2/3:s=200x200
23168 Specify a custom rule for evolving a randomly generated grid:
23174 Full example with slow death effect (mold) using @command{ffplay}:
23176 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23183 @anchor{haldclutsrc}
23186 @anchor{pal100bars}
23187 @anchor{rgbtestsrc}
23189 @anchor{smptehdbars}
23192 @anchor{yuvtestsrc}
23193 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23195 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23197 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23199 The @code{color} source provides an uniformly colored input.
23201 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23202 @ref{haldclut} filter.
23204 The @code{nullsrc} source returns unprocessed video frames. It is
23205 mainly useful to be employed in analysis / debugging tools, or as the
23206 source for filters which ignore the input data.
23208 The @code{pal75bars} source generates a color bars pattern, based on
23209 EBU PAL recommendations with 75% color levels.
23211 The @code{pal100bars} source generates a color bars pattern, based on
23212 EBU PAL recommendations with 100% color levels.
23214 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23215 detecting RGB vs BGR issues. You should see a red, green and blue
23216 stripe from top to bottom.
23218 The @code{smptebars} source generates a color bars pattern, based on
23219 the SMPTE Engineering Guideline EG 1-1990.
23221 The @code{smptehdbars} source generates a color bars pattern, based on
23222 the SMPTE RP 219-2002.
23224 The @code{testsrc} source generates a test video pattern, showing a
23225 color pattern, a scrolling gradient and a timestamp. This is mainly
23226 intended for testing purposes.
23228 The @code{testsrc2} source is similar to testsrc, but supports more
23229 pixel formats instead of just @code{rgb24}. This allows using it as an
23230 input for other tests without requiring a format conversion.
23232 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23233 see a y, cb and cr stripe from top to bottom.
23235 The sources accept the following parameters:
23240 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23241 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23242 pixels to be used as identity matrix for 3D lookup tables. Each component is
23243 coded on a @code{1/(N*N)} scale.
23246 Specify the color of the source, only available in the @code{color}
23247 source. For the syntax of this option, check the
23248 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23251 Specify the size of the sourced video. For the syntax of this option, check the
23252 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23253 The default value is @code{320x240}.
23255 This option is not available with the @code{allrgb}, @code{allyuv}, and
23256 @code{haldclutsrc} filters.
23259 Specify the frame rate of the sourced video, as the number of frames
23260 generated per second. It has to be a string in the format
23261 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23262 number or a valid video frame rate abbreviation. The default value is
23266 Set the duration of the sourced video. See
23267 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23268 for the accepted syntax.
23270 If not specified, or the expressed duration is negative, the video is
23271 supposed to be generated forever.
23273 Since the frame rate is used as time base, all frames including the last one
23274 will have their full duration. If the specified duration is not a multiple
23275 of the frame duration, it will be rounded up.
23278 Set the sample aspect ratio of the sourced video.
23281 Specify the alpha (opacity) of the background, only available in the
23282 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23283 255 (fully opaque, the default).
23286 Set the number of decimals to show in the timestamp, only available in the
23287 @code{testsrc} source.
23289 The displayed timestamp value will correspond to the original
23290 timestamp value multiplied by the power of 10 of the specified
23291 value. Default value is 0.
23294 @subsection Examples
23298 Generate a video with a duration of 5.3 seconds, with size
23299 176x144 and a frame rate of 10 frames per second:
23301 testsrc=duration=5.3:size=qcif:rate=10
23305 The following graph description will generate a red source
23306 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23309 color=c=red@@0.2:s=qcif:r=10
23313 If the input content is to be ignored, @code{nullsrc} can be used. The
23314 following command generates noise in the luminance plane by employing
23315 the @code{geq} filter:
23317 nullsrc=s=256x256, geq=random(1)*255:128:128
23321 @subsection Commands
23323 The @code{color} source supports the following commands:
23327 Set the color of the created image. Accepts the same syntax of the
23328 corresponding @option{color} option.
23333 Generate video using an OpenCL program.
23338 OpenCL program source file.
23341 Kernel name in program.
23344 Size of frames to generate. This must be set.
23347 Pixel format to use for the generated frames. This must be set.
23350 Number of frames generated every second. Default value is '25'.
23354 For details of how the program loading works, see the @ref{program_opencl}
23361 Generate a colour ramp by setting pixel values from the position of the pixel
23362 in the output image. (Note that this will work with all pixel formats, but
23363 the generated output will not be the same.)
23365 __kernel void ramp(__write_only image2d_t dst,
23366 unsigned int index)
23368 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23371 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23373 write_imagef(dst, loc, val);
23378 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23380 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23381 unsigned int index)
23383 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23385 float4 value = 0.0f;
23386 int x = loc.x + index;
23387 int y = loc.y + index;
23388 while (x > 0 || y > 0) {
23389 if (x % 3 == 1 && y % 3 == 1) {
23397 write_imagef(dst, loc, value);
23403 @section sierpinski
23405 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23407 This source accepts the following options:
23411 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23412 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23415 Set frame rate, expressed as number of frames per second. Default
23419 Set seed which is used for random panning.
23422 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23425 Set fractal type, can be default @code{carpet} or @code{triangle}.
23428 @c man end VIDEO SOURCES
23430 @chapter Video Sinks
23431 @c man begin VIDEO SINKS
23433 Below is a description of the currently available video sinks.
23435 @section buffersink
23437 Buffer video frames, and make them available to the end of the filter
23440 This sink is mainly intended for programmatic use, in particular
23441 through the interface defined in @file{libavfilter/buffersink.h}
23442 or the options system.
23444 It accepts a pointer to an AVBufferSinkContext structure, which
23445 defines the incoming buffers' formats, to be passed as the opaque
23446 parameter to @code{avfilter_init_filter} for initialization.
23450 Null video sink: do absolutely nothing with the input video. It is
23451 mainly useful as a template and for use in analysis / debugging
23454 @c man end VIDEO SINKS
23456 @chapter Multimedia Filters
23457 @c man begin MULTIMEDIA FILTERS
23459 Below is a description of the currently available multimedia filters.
23463 Convert input audio to a video output, displaying the audio bit scope.
23465 The filter accepts the following options:
23469 Set frame rate, expressed as number of frames per second. Default
23473 Specify the video size for the output. For the syntax of this option, check the
23474 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23475 Default value is @code{1024x256}.
23478 Specify list of colors separated by space or by '|' which will be used to
23479 draw channels. Unrecognized or missing colors will be replaced
23483 @section adrawgraph
23484 Draw a graph using input audio metadata.
23486 See @ref{drawgraph}
23488 @section agraphmonitor
23490 See @ref{graphmonitor}.
23492 @section ahistogram
23494 Convert input audio to a video output, displaying the volume histogram.
23496 The filter accepts the following options:
23500 Specify how histogram is calculated.
23502 It accepts the following values:
23505 Use single histogram for all channels.
23507 Use separate histogram for each channel.
23509 Default is @code{single}.
23512 Set frame rate, expressed as number of frames per second. Default
23516 Specify the video size for the output. For the syntax of this option, check the
23517 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23518 Default value is @code{hd720}.
23523 It accepts the following values:
23534 reverse logarithmic
23536 Default is @code{log}.
23539 Set amplitude scale.
23541 It accepts the following values:
23548 Default is @code{log}.
23551 Set how much frames to accumulate in histogram.
23552 Default is 1. Setting this to -1 accumulates all frames.
23555 Set histogram ratio of window height.
23558 Set sonogram sliding.
23560 It accepts the following values:
23563 replace old rows with new ones.
23565 scroll from top to bottom.
23567 Default is @code{replace}.
23570 @section aphasemeter
23572 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23573 representing mean phase of current audio frame. A video output can also be produced and is
23574 enabled by default. The audio is passed through as first output.
23576 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23577 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23578 and @code{1} means channels are in phase.
23580 The filter accepts the following options, all related to its video output:
23584 Set the output frame rate. Default value is @code{25}.
23587 Set the video size for the output. For the syntax of this option, check the
23588 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23589 Default value is @code{800x400}.
23594 Specify the red, green, blue contrast. Default values are @code{2},
23595 @code{7} and @code{1}.
23596 Allowed range is @code{[0, 255]}.
23599 Set color which will be used for drawing median phase. If color is
23600 @code{none} which is default, no median phase value will be drawn.
23603 Enable video output. Default is enabled.
23606 @subsection phasing detection
23608 The filter also detects out of phase and mono sequences in stereo streams.
23609 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23611 The filter accepts the following options for this detection:
23615 Enable mono and out of phase detection. Default is disabled.
23618 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23619 Allowed range is @code{[0, 1]}.
23622 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23623 Allowed range is @code{[90, 180]}.
23626 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23629 @subsection Examples
23633 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23635 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23639 @section avectorscope
23641 Convert input audio to a video output, representing the audio vector
23644 The filter is used to measure the difference between channels of stereo
23645 audio stream. A monaural signal, consisting of identical left and right
23646 signal, results in straight vertical line. Any stereo separation is visible
23647 as a deviation from this line, creating a Lissajous figure.
23648 If the straight (or deviation from it) but horizontal line appears this
23649 indicates that the left and right channels are out of phase.
23651 The filter accepts the following options:
23655 Set the vectorscope mode.
23657 Available values are:
23660 Lissajous rotated by 45 degrees.
23663 Same as above but not rotated.
23666 Shape resembling half of circle.
23669 Default value is @samp{lissajous}.
23672 Set the video size for the output. For the syntax of this option, check the
23673 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23674 Default value is @code{400x400}.
23677 Set the output frame rate. Default value is @code{25}.
23683 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23684 @code{160}, @code{80} and @code{255}.
23685 Allowed range is @code{[0, 255]}.
23691 Specify the red, green, blue and alpha fade. Default values are @code{15},
23692 @code{10}, @code{5} and @code{5}.
23693 Allowed range is @code{[0, 255]}.
23696 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23697 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23700 Set the vectorscope drawing mode.
23702 Available values are:
23705 Draw dot for each sample.
23708 Draw line between previous and current sample.
23711 Default value is @samp{dot}.
23714 Specify amplitude scale of audio samples.
23716 Available values are:
23732 Swap left channel axis with right channel axis.
23742 Mirror only x axis.
23745 Mirror only y axis.
23753 @subsection Examples
23757 Complete example using @command{ffplay}:
23759 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23760 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
23764 @section bench, abench
23766 Benchmark part of a filtergraph.
23768 The filter accepts the following options:
23772 Start or stop a timer.
23774 Available values are:
23777 Get the current time, set it as frame metadata (using the key
23778 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
23781 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
23782 the input frame metadata to get the time difference. Time difference, average,
23783 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
23784 @code{min}) are then printed. The timestamps are expressed in seconds.
23788 @subsection Examples
23792 Benchmark @ref{selectivecolor} filter:
23794 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
23800 Concatenate audio and video streams, joining them together one after the
23803 The filter works on segments of synchronized video and audio streams. All
23804 segments must have the same number of streams of each type, and that will
23805 also be the number of streams at output.
23807 The filter accepts the following options:
23812 Set the number of segments. Default is 2.
23815 Set the number of output video streams, that is also the number of video
23816 streams in each segment. Default is 1.
23819 Set the number of output audio streams, that is also the number of audio
23820 streams in each segment. Default is 0.
23823 Activate unsafe mode: do not fail if segments have a different format.
23827 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
23828 @var{a} audio outputs.
23830 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
23831 segment, in the same order as the outputs, then the inputs for the second
23834 Related streams do not always have exactly the same duration, for various
23835 reasons including codec frame size or sloppy authoring. For that reason,
23836 related synchronized streams (e.g. a video and its audio track) should be
23837 concatenated at once. The concat filter will use the duration of the longest
23838 stream in each segment (except the last one), and if necessary pad shorter
23839 audio streams with silence.
23841 For this filter to work correctly, all segments must start at timestamp 0.
23843 All corresponding streams must have the same parameters in all segments; the
23844 filtering system will automatically select a common pixel format for video
23845 streams, and a common sample format, sample rate and channel layout for
23846 audio streams, but other settings, such as resolution, must be converted
23847 explicitly by the user.
23849 Different frame rates are acceptable but will result in variable frame rate
23850 at output; be sure to configure the output file to handle it.
23852 @subsection Examples
23856 Concatenate an opening, an episode and an ending, all in bilingual version
23857 (video in stream 0, audio in streams 1 and 2):
23859 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23860 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23861 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23862 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23866 Concatenate two parts, handling audio and video separately, using the
23867 (a)movie sources, and adjusting the resolution:
23869 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23870 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23871 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23873 Note that a desync will happen at the stitch if the audio and video streams
23874 do not have exactly the same duration in the first file.
23878 @subsection Commands
23880 This filter supports the following commands:
23883 Close the current segment and step to the next one
23889 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
23890 level. By default, it logs a message at a frequency of 10Hz with the
23891 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
23892 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
23894 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
23895 sample format is double-precision floating point. The input stream will be converted to
23896 this specification, if needed. Users may need to insert aformat and/or aresample filters
23897 after this filter to obtain the original parameters.
23899 The filter also has a video output (see the @var{video} option) with a real
23900 time graph to observe the loudness evolution. The graphic contains the logged
23901 message mentioned above, so it is not printed anymore when this option is set,
23902 unless the verbose logging is set. The main graphing area contains the
23903 short-term loudness (3 seconds of analysis), and the gauge on the right is for
23904 the momentary loudness (400 milliseconds), but can optionally be configured
23905 to instead display short-term loudness (see @var{gauge}).
23907 The green area marks a +/- 1LU target range around the target loudness
23908 (-23LUFS by default, unless modified through @var{target}).
23910 More information about the Loudness Recommendation EBU R128 on
23911 @url{http://tech.ebu.ch/loudness}.
23913 The filter accepts the following options:
23918 Activate the video output. The audio stream is passed unchanged whether this
23919 option is set or no. The video stream will be the first output stream if
23920 activated. Default is @code{0}.
23923 Set the video size. This option is for video only. For the syntax of this
23925 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23926 Default and minimum resolution is @code{640x480}.
23929 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
23930 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
23931 other integer value between this range is allowed.
23934 Set metadata injection. If set to @code{1}, the audio input will be segmented
23935 into 100ms output frames, each of them containing various loudness information
23936 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
23938 Default is @code{0}.
23941 Force the frame logging level.
23943 Available values are:
23946 information logging level
23948 verbose logging level
23951 By default, the logging level is set to @var{info}. If the @option{video} or
23952 the @option{metadata} options are set, it switches to @var{verbose}.
23957 Available modes can be cumulated (the option is a @code{flag} type). Possible
23961 Disable any peak mode (default).
23963 Enable sample-peak mode.
23965 Simple peak mode looking for the higher sample value. It logs a message
23966 for sample-peak (identified by @code{SPK}).
23968 Enable true-peak mode.
23970 If enabled, the peak lookup is done on an over-sampled version of the input
23971 stream for better peak accuracy. It logs a message for true-peak.
23972 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
23973 This mode requires a build with @code{libswresample}.
23977 Treat mono input files as "dual mono". If a mono file is intended for playback
23978 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
23979 If set to @code{true}, this option will compensate for this effect.
23980 Multi-channel input files are not affected by this option.
23983 Set a specific pan law to be used for the measurement of dual mono files.
23984 This parameter is optional, and has a default value of -3.01dB.
23987 Set a specific target level (in LUFS) used as relative zero in the visualization.
23988 This parameter is optional and has a default value of -23LUFS as specified
23989 by EBU R128. However, material published online may prefer a level of -16LUFS
23990 (e.g. for use with podcasts or video platforms).
23993 Set the value displayed by the gauge. Valid values are @code{momentary} and s
23994 @code{shortterm}. By default the momentary value will be used, but in certain
23995 scenarios it may be more useful to observe the short term value instead (e.g.
23999 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24000 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24001 video output, not the summary or continuous log output.
24004 @subsection Examples
24008 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24010 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24014 Run an analysis with @command{ffmpeg}:
24016 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24020 @section interleave, ainterleave
24022 Temporally interleave frames from several inputs.
24024 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24026 These filters read frames from several inputs and send the oldest
24027 queued frame to the output.
24029 Input streams must have well defined, monotonically increasing frame
24032 In order to submit one frame to output, these filters need to enqueue
24033 at least one frame for each input, so they cannot work in case one
24034 input is not yet terminated and will not receive incoming frames.
24036 For example consider the case when one input is a @code{select} filter
24037 which always drops input frames. The @code{interleave} filter will keep
24038 reading from that input, but it will never be able to send new frames
24039 to output until the input sends an end-of-stream signal.
24041 Also, depending on inputs synchronization, the filters will drop
24042 frames in case one input receives more frames than the other ones, and
24043 the queue is already filled.
24045 These filters accept the following options:
24049 Set the number of different inputs, it is 2 by default.
24052 How to determine the end-of-stream.
24056 The duration of the longest input. (default)
24059 The duration of the shortest input.
24062 The duration of the first input.
24067 @subsection Examples
24071 Interleave frames belonging to different streams using @command{ffmpeg}:
24073 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24077 Add flickering blur effect:
24079 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24083 @section metadata, ametadata
24085 Manipulate frame metadata.
24087 This filter accepts the following options:
24091 Set mode of operation of the filter.
24093 Can be one of the following:
24097 If both @code{value} and @code{key} is set, select frames
24098 which have such metadata. If only @code{key} is set, select
24099 every frame that has such key in metadata.
24102 Add new metadata @code{key} and @code{value}. If key is already available
24106 Modify value of already present key.
24109 If @code{value} is set, delete only keys that have such value.
24110 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24114 Print key and its value if metadata was found. If @code{key} is not set print all
24115 metadata values available in frame.
24119 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24122 Set metadata value which will be used. This option is mandatory for
24123 @code{modify} and @code{add} mode.
24126 Which function to use when comparing metadata value and @code{value}.
24128 Can be one of following:
24132 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24135 Values are interpreted as strings, returns true if metadata value starts with
24136 the @code{value} option string.
24139 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24142 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24145 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24148 Values are interpreted as floats, returns true if expression from option @code{expr}
24152 Values are interpreted as strings, returns true if metadata value ends with
24153 the @code{value} option string.
24157 Set expression which is used when @code{function} is set to @code{expr}.
24158 The expression is evaluated through the eval API and can contain the following
24163 Float representation of @code{value} from metadata key.
24166 Float representation of @code{value} as supplied by user in @code{value} option.
24170 If specified in @code{print} mode, output is written to the named file. Instead of
24171 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24172 for standard output. If @code{file} option is not set, output is written to the log
24173 with AV_LOG_INFO loglevel.
24176 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24180 @subsection Examples
24184 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24187 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24190 Print silencedetect output to file @file{metadata.txt}.
24192 silencedetect,ametadata=mode=print:file=metadata.txt
24195 Direct all metadata to a pipe with file descriptor 4.
24197 metadata=mode=print:file='pipe\:4'
24201 @section perms, aperms
24203 Set read/write permissions for the output frames.
24205 These filters are mainly aimed at developers to test direct path in the
24206 following filter in the filtergraph.
24208 The filters accept the following options:
24212 Select the permissions mode.
24214 It accepts the following values:
24217 Do nothing. This is the default.
24219 Set all the output frames read-only.
24221 Set all the output frames directly writable.
24223 Make the frame read-only if writable, and writable if read-only.
24225 Set each output frame read-only or writable randomly.
24229 Set the seed for the @var{random} mode, must be an integer included between
24230 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24231 @code{-1}, the filter will try to use a good random seed on a best effort
24235 Note: in case of auto-inserted filter between the permission filter and the
24236 following one, the permission might not be received as expected in that
24237 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24238 perms/aperms filter can avoid this problem.
24240 @section realtime, arealtime
24242 Slow down filtering to match real time approximately.
24244 These filters will pause the filtering for a variable amount of time to
24245 match the output rate with the input timestamps.
24246 They are similar to the @option{re} option to @code{ffmpeg}.
24248 They accept the following options:
24252 Time limit for the pauses. Any pause longer than that will be considered
24253 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24255 Speed factor for processing. The value must be a float larger than zero.
24256 Values larger than 1.0 will result in faster than realtime processing,
24257 smaller will slow processing down. The @var{limit} is automatically adapted
24258 accordingly. Default is 1.0.
24260 A processing speed faster than what is possible without these filters cannot
24265 @section select, aselect
24267 Select frames to pass in output.
24269 This filter accepts the following options:
24274 Set expression, which is evaluated for each input frame.
24276 If the expression is evaluated to zero, the frame is discarded.
24278 If the evaluation result is negative or NaN, the frame is sent to the
24279 first output; otherwise it is sent to the output with index
24280 @code{ceil(val)-1}, assuming that the input index starts from 0.
24282 For example a value of @code{1.2} corresponds to the output with index
24283 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24286 Set the number of outputs. The output to which to send the selected
24287 frame is based on the result of the evaluation. Default value is 1.
24290 The expression can contain the following constants:
24294 The (sequential) number of the filtered frame, starting from 0.
24297 The (sequential) number of the selected frame, starting from 0.
24299 @item prev_selected_n
24300 The sequential number of the last selected frame. It's NAN if undefined.
24303 The timebase of the input timestamps.
24306 The PTS (Presentation TimeStamp) of the filtered video frame,
24307 expressed in @var{TB} units. It's NAN if undefined.
24310 The PTS of the filtered video frame,
24311 expressed in seconds. It's NAN if undefined.
24314 The PTS of the previously filtered video frame. It's NAN if undefined.
24316 @item prev_selected_pts
24317 The PTS of the last previously filtered video frame. It's NAN if undefined.
24319 @item prev_selected_t
24320 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24323 The PTS of the first video frame in the video. It's NAN if undefined.
24326 The time of the first video frame in the video. It's NAN if undefined.
24328 @item pict_type @emph{(video only)}
24329 The type of the filtered frame. It can assume one of the following
24341 @item interlace_type @emph{(video only)}
24342 The frame interlace type. It can assume one of the following values:
24345 The frame is progressive (not interlaced).
24347 The frame is top-field-first.
24349 The frame is bottom-field-first.
24352 @item consumed_sample_n @emph{(audio only)}
24353 the number of selected samples before the current frame
24355 @item samples_n @emph{(audio only)}
24356 the number of samples in the current frame
24358 @item sample_rate @emph{(audio only)}
24359 the input sample rate
24362 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24365 the position in the file of the filtered frame, -1 if the information
24366 is not available (e.g. for synthetic video)
24368 @item scene @emph{(video only)}
24369 value between 0 and 1 to indicate a new scene; a low value reflects a low
24370 probability for the current frame to introduce a new scene, while a higher
24371 value means the current frame is more likely to be one (see the example below)
24373 @item concatdec_select
24374 The concat demuxer can select only part of a concat input file by setting an
24375 inpoint and an outpoint, but the output packets may not be entirely contained
24376 in the selected interval. By using this variable, it is possible to skip frames
24377 generated by the concat demuxer which are not exactly contained in the selected
24380 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24381 and the @var{lavf.concat.duration} packet metadata values which are also
24382 present in the decoded frames.
24384 The @var{concatdec_select} variable is -1 if the frame pts is at least
24385 start_time and either the duration metadata is missing or the frame pts is less
24386 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24389 That basically means that an input frame is selected if its pts is within the
24390 interval set by the concat demuxer.
24394 The default value of the select expression is "1".
24396 @subsection Examples
24400 Select all frames in input:
24405 The example above is the same as:
24417 Select only I-frames:
24419 select='eq(pict_type\,I)'
24423 Select one frame every 100:
24425 select='not(mod(n\,100))'
24429 Select only frames contained in the 10-20 time interval:
24431 select=between(t\,10\,20)
24435 Select only I-frames contained in the 10-20 time interval:
24437 select=between(t\,10\,20)*eq(pict_type\,I)
24441 Select frames with a minimum distance of 10 seconds:
24443 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24447 Use aselect to select only audio frames with samples number > 100:
24449 aselect='gt(samples_n\,100)'
24453 Create a mosaic of the first scenes:
24455 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24458 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24462 Send even and odd frames to separate outputs, and compose them:
24464 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24468 Select useful frames from an ffconcat file which is using inpoints and
24469 outpoints but where the source files are not intra frame only.
24471 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24475 @section sendcmd, asendcmd
24477 Send commands to filters in the filtergraph.
24479 These filters read commands to be sent to other filters in the
24482 @code{sendcmd} must be inserted between two video filters,
24483 @code{asendcmd} must be inserted between two audio filters, but apart
24484 from that they act the same way.
24486 The specification of commands can be provided in the filter arguments
24487 with the @var{commands} option, or in a file specified by the
24488 @var{filename} option.
24490 These filters accept the following options:
24493 Set the commands to be read and sent to the other filters.
24495 Set the filename of the commands to be read and sent to the other
24499 @subsection Commands syntax
24501 A commands description consists of a sequence of interval
24502 specifications, comprising a list of commands to be executed when a
24503 particular event related to that interval occurs. The occurring event
24504 is typically the current frame time entering or leaving a given time
24507 An interval is specified by the following syntax:
24509 @var{START}[-@var{END}] @var{COMMANDS};
24512 The time interval is specified by the @var{START} and @var{END} times.
24513 @var{END} is optional and defaults to the maximum time.
24515 The current frame time is considered within the specified interval if
24516 it is included in the interval [@var{START}, @var{END}), that is when
24517 the time is greater or equal to @var{START} and is lesser than
24520 @var{COMMANDS} consists of a sequence of one or more command
24521 specifications, separated by ",", relating to that interval. The
24522 syntax of a command specification is given by:
24524 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24527 @var{FLAGS} is optional and specifies the type of events relating to
24528 the time interval which enable sending the specified command, and must
24529 be a non-null sequence of identifier flags separated by "+" or "|" and
24530 enclosed between "[" and "]".
24532 The following flags are recognized:
24535 The command is sent when the current frame timestamp enters the
24536 specified interval. In other words, the command is sent when the
24537 previous frame timestamp was not in the given interval, and the
24541 The command is sent when the current frame timestamp leaves the
24542 specified interval. In other words, the command is sent when the
24543 previous frame timestamp was in the given interval, and the
24547 The command @var{ARG} is interpreted as expression and result of
24548 expression is passed as @var{ARG}.
24550 The expression is evaluated through the eval API and can contain the following
24555 Original position in the file of the frame, or undefined if undefined
24556 for the current frame.
24559 The presentation timestamp in input.
24562 The count of the input frame for video or audio, starting from 0.
24565 The time in seconds of the current frame.
24568 The start time in seconds of the current command interval.
24571 The end time in seconds of the current command interval.
24574 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24579 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24582 @var{TARGET} specifies the target of the command, usually the name of
24583 the filter class or a specific filter instance name.
24585 @var{COMMAND} specifies the name of the command for the target filter.
24587 @var{ARG} is optional and specifies the optional list of argument for
24588 the given @var{COMMAND}.
24590 Between one interval specification and another, whitespaces, or
24591 sequences of characters starting with @code{#} until the end of line,
24592 are ignored and can be used to annotate comments.
24594 A simplified BNF description of the commands specification syntax
24597 @var{COMMAND_FLAG} ::= "enter" | "leave"
24598 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24599 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24600 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24601 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24602 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24605 @subsection Examples
24609 Specify audio tempo change at second 4:
24611 asendcmd=c='4.0 atempo tempo 1.5',atempo
24615 Target a specific filter instance:
24617 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24621 Specify a list of drawtext and hue commands in a file.
24623 # show text in the interval 5-10
24624 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24625 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24627 # desaturate the image in the interval 15-20
24628 15.0-20.0 [enter] hue s 0,
24629 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24631 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24633 # apply an exponential saturation fade-out effect, starting from time 25
24634 25 [enter] hue s exp(25-t)
24637 A filtergraph allowing to read and process the above command list
24638 stored in a file @file{test.cmd}, can be specified with:
24640 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24645 @section setpts, asetpts
24647 Change the PTS (presentation timestamp) of the input frames.
24649 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24651 This filter accepts the following options:
24656 The expression which is evaluated for each frame to construct its timestamp.
24660 The expression is evaluated through the eval API and can contain the following
24664 @item FRAME_RATE, FR
24665 frame rate, only defined for constant frame-rate video
24668 The presentation timestamp in input
24671 The count of the input frame for video or the number of consumed samples,
24672 not including the current frame for audio, starting from 0.
24674 @item NB_CONSUMED_SAMPLES
24675 The number of consumed samples, not including the current frame (only
24678 @item NB_SAMPLES, S
24679 The number of samples in the current frame (only audio)
24681 @item SAMPLE_RATE, SR
24682 The audio sample rate.
24685 The PTS of the first frame.
24688 the time in seconds of the first frame
24691 State whether the current frame is interlaced.
24694 the time in seconds of the current frame
24697 original position in the file of the frame, or undefined if undefined
24698 for the current frame
24701 The previous input PTS.
24704 previous input time in seconds
24707 The previous output PTS.
24710 previous output time in seconds
24713 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
24717 The wallclock (RTC) time at the start of the movie in microseconds.
24720 The timebase of the input timestamps.
24724 @subsection Examples
24728 Start counting PTS from zero
24730 setpts=PTS-STARTPTS
24734 Apply fast motion effect:
24740 Apply slow motion effect:
24746 Set fixed rate of 25 frames per second:
24752 Set fixed rate 25 fps with some jitter:
24754 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
24758 Apply an offset of 10 seconds to the input PTS:
24764 Generate timestamps from a "live source" and rebase onto the current timebase:
24766 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
24770 Generate timestamps by counting samples:
24779 Force color range for the output video frame.
24781 The @code{setrange} filter marks the color range property for the
24782 output frames. It does not change the input frame, but only sets the
24783 corresponding property, which affects how the frame is treated by
24786 The filter accepts the following options:
24791 Available values are:
24795 Keep the same color range property.
24797 @item unspecified, unknown
24798 Set the color range as unspecified.
24800 @item limited, tv, mpeg
24801 Set the color range as limited.
24803 @item full, pc, jpeg
24804 Set the color range as full.
24808 @section settb, asettb
24810 Set the timebase to use for the output frames timestamps.
24811 It is mainly useful for testing timebase configuration.
24813 It accepts the following parameters:
24818 The expression which is evaluated into the output timebase.
24822 The value for @option{tb} is an arithmetic expression representing a
24823 rational. The expression can contain the constants "AVTB" (the default
24824 timebase), "intb" (the input timebase) and "sr" (the sample rate,
24825 audio only). Default value is "intb".
24827 @subsection Examples
24831 Set the timebase to 1/25:
24837 Set the timebase to 1/10:
24843 Set the timebase to 1001/1000:
24849 Set the timebase to 2*intb:
24855 Set the default timebase value:
24862 Convert input audio to a video output representing frequency spectrum
24863 logarithmically using Brown-Puckette constant Q transform algorithm with
24864 direct frequency domain coefficient calculation (but the transform itself
24865 is not really constant Q, instead the Q factor is actually variable/clamped),
24866 with musical tone scale, from E0 to D#10.
24868 The filter accepts the following options:
24872 Specify the video size for the output. It must be even. For the syntax of this option,
24873 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24874 Default value is @code{1920x1080}.
24877 Set the output frame rate. Default value is @code{25}.
24880 Set the bargraph height. It must be even. Default value is @code{-1} which
24881 computes the bargraph height automatically.
24884 Set the axis height. It must be even. Default value is @code{-1} which computes
24885 the axis height automatically.
24888 Set the sonogram height. It must be even. Default value is @code{-1} which
24889 computes the sonogram height automatically.
24892 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
24893 instead. Default value is @code{1}.
24895 @item sono_v, volume
24896 Specify the sonogram volume expression. It can contain variables:
24899 the @var{bar_v} evaluated expression
24900 @item frequency, freq, f
24901 the frequency where it is evaluated
24902 @item timeclamp, tc
24903 the value of @var{timeclamp} option
24907 @item a_weighting(f)
24908 A-weighting of equal loudness
24909 @item b_weighting(f)
24910 B-weighting of equal loudness
24911 @item c_weighting(f)
24912 C-weighting of equal loudness.
24914 Default value is @code{16}.
24916 @item bar_v, volume2
24917 Specify the bargraph volume expression. It can contain variables:
24920 the @var{sono_v} evaluated expression
24921 @item frequency, freq, f
24922 the frequency where it is evaluated
24923 @item timeclamp, tc
24924 the value of @var{timeclamp} option
24928 @item a_weighting(f)
24929 A-weighting of equal loudness
24930 @item b_weighting(f)
24931 B-weighting of equal loudness
24932 @item c_weighting(f)
24933 C-weighting of equal loudness.
24935 Default value is @code{sono_v}.
24937 @item sono_g, gamma
24938 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
24939 higher gamma makes the spectrum having more range. Default value is @code{3}.
24940 Acceptable range is @code{[1, 7]}.
24942 @item bar_g, gamma2
24943 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
24947 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
24948 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
24950 @item timeclamp, tc
24951 Specify the transform timeclamp. At low frequency, there is trade-off between
24952 accuracy in time domain and frequency domain. If timeclamp is lower,
24953 event in time domain is represented more accurately (such as fast bass drum),
24954 otherwise event in frequency domain is represented more accurately
24955 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
24958 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
24959 limits future samples by applying asymmetric windowing in time domain, useful
24960 when low latency is required. Accepted range is @code{[0, 1]}.
24963 Specify the transform base frequency. Default value is @code{20.01523126408007475},
24964 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
24967 Specify the transform end frequency. Default value is @code{20495.59681441799654},
24968 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
24971 This option is deprecated and ignored.
24974 Specify the transform length in time domain. Use this option to control accuracy
24975 trade-off between time domain and frequency domain at every frequency sample.
24976 It can contain variables:
24978 @item frequency, freq, f
24979 the frequency where it is evaluated
24980 @item timeclamp, tc
24981 the value of @var{timeclamp} option.
24983 Default value is @code{384*tc/(384+tc*f)}.
24986 Specify the transform count for every video frame. Default value is @code{6}.
24987 Acceptable range is @code{[1, 30]}.
24990 Specify the transform count for every single pixel. Default value is @code{0},
24991 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
24994 Specify font file for use with freetype to draw the axis. If not specified,
24995 use embedded font. Note that drawing with font file or embedded font is not
24996 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25000 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25001 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25005 Specify font color expression. This is arithmetic expression that should return
25006 integer value 0xRRGGBB. It can contain variables:
25008 @item frequency, freq, f
25009 the frequency where it is evaluated
25010 @item timeclamp, tc
25011 the value of @var{timeclamp} option
25016 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25017 @item r(x), g(x), b(x)
25018 red, green, and blue value of intensity x.
25020 Default value is @code{st(0, (midi(f)-59.5)/12);
25021 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25022 r(1-ld(1)) + b(ld(1))}.
25025 Specify image file to draw the axis. This option override @var{fontfile} and
25026 @var{fontcolor} option.
25029 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25030 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25031 Default value is @code{1}.
25034 Set colorspace. The accepted values are:
25037 Unspecified (default)
25046 BT.470BG or BT.601-6 625
25049 SMPTE-170M or BT.601-6 525
25055 BT.2020 with non-constant luminance
25060 Set spectrogram color scheme. This is list of floating point values with format
25061 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25062 The default is @code{1|0.5|0|0|0.5|1}.
25066 @subsection Examples
25070 Playing audio while showing the spectrum:
25072 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25076 Same as above, but with frame rate 30 fps:
25078 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25082 Playing at 1280x720:
25084 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25088 Disable sonogram display:
25094 A1 and its harmonics: A1, A2, (near)E3, A3:
25096 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),
25097 asplit[a][out1]; [a] showcqt [out0]'
25101 Same as above, but with more accuracy in frequency domain:
25103 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),
25104 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25110 bar_v=10:sono_v=bar_v*a_weighting(f)
25114 Custom gamma, now spectrum is linear to the amplitude.
25120 Custom tlength equation:
25122 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)))'
25126 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25128 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25132 Custom font using fontconfig:
25134 font='Courier New,Monospace,mono|bold'
25138 Custom frequency range with custom axis using image file:
25140 axisfile=myaxis.png:basefreq=40:endfreq=10000
25146 Convert input audio to video output representing the audio power spectrum.
25147 Audio amplitude is on Y-axis while frequency is on X-axis.
25149 The filter accepts the following options:
25153 Specify size of video. For the syntax of this option, check the
25154 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25155 Default is @code{1024x512}.
25159 This set how each frequency bin will be represented.
25161 It accepts the following values:
25167 Default is @code{bar}.
25170 Set amplitude scale.
25172 It accepts the following values:
25186 Default is @code{log}.
25189 Set frequency scale.
25191 It accepts the following values:
25200 Reverse logarithmic scale.
25202 Default is @code{lin}.
25205 Set window size. Allowed range is from 16 to 65536.
25207 Default is @code{2048}
25210 Set windowing function.
25212 It accepts the following values:
25235 Default is @code{hanning}.
25238 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25239 which means optimal overlap for selected window function will be picked.
25242 Set time averaging. Setting this to 0 will display current maximal peaks.
25243 Default is @code{1}, which means time averaging is disabled.
25246 Specify list of colors separated by space or by '|' which will be used to
25247 draw channel frequencies. Unrecognized or missing colors will be replaced
25251 Set channel display mode.
25253 It accepts the following values:
25258 Default is @code{combined}.
25261 Set minimum amplitude used in @code{log} amplitude scaler.
25264 Set data display mode.
25266 It accepts the following values:
25272 Default is @code{magnitude}.
25275 @section showspatial
25277 Convert stereo input audio to a video output, representing the spatial relationship
25278 between two channels.
25280 The filter accepts the following options:
25284 Specify the video size for the output. For the syntax of this option, check the
25285 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25286 Default value is @code{512x512}.
25289 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25292 Set window function.
25294 It accepts the following values:
25319 Default value is @code{hann}.
25322 Set ratio of overlap window. Default value is @code{0.5}.
25323 When value is @code{1} overlap is set to recommended size for specific
25324 window function currently used.
25327 @anchor{showspectrum}
25328 @section showspectrum
25330 Convert input audio to a video output, representing the audio frequency
25333 The filter accepts the following options:
25337 Specify the video size for the output. For the syntax of this option, check the
25338 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25339 Default value is @code{640x512}.
25342 Specify how the spectrum should slide along the window.
25344 It accepts the following values:
25347 the samples start again on the left when they reach the right
25349 the samples scroll from right to left
25351 frames are only produced when the samples reach the right
25353 the samples scroll from left to right
25356 Default value is @code{replace}.
25359 Specify display mode.
25361 It accepts the following values:
25364 all channels are displayed in the same row
25366 all channels are displayed in separate rows
25369 Default value is @samp{combined}.
25372 Specify display color mode.
25374 It accepts the following values:
25377 each channel is displayed in a separate color
25379 each channel is displayed using the same color scheme
25381 each channel is displayed using the rainbow color scheme
25383 each channel is displayed using the moreland color scheme
25385 each channel is displayed using the nebulae color scheme
25387 each channel is displayed using the fire color scheme
25389 each channel is displayed using the fiery color scheme
25391 each channel is displayed using the fruit color scheme
25393 each channel is displayed using the cool color scheme
25395 each channel is displayed using the magma color scheme
25397 each channel is displayed using the green color scheme
25399 each channel is displayed using the viridis color scheme
25401 each channel is displayed using the plasma color scheme
25403 each channel is displayed using the cividis color scheme
25405 each channel is displayed using the terrain color scheme
25408 Default value is @samp{channel}.
25411 Specify scale used for calculating intensity color values.
25413 It accepts the following values:
25418 square root, default
25429 Default value is @samp{sqrt}.
25432 Specify frequency scale.
25434 It accepts the following values:
25442 Default value is @samp{lin}.
25445 Set saturation modifier for displayed colors. Negative values provide
25446 alternative color scheme. @code{0} is no saturation at all.
25447 Saturation must be in [-10.0, 10.0] range.
25448 Default value is @code{1}.
25451 Set window function.
25453 It accepts the following values:
25478 Default value is @code{hann}.
25481 Set orientation of time vs frequency axis. Can be @code{vertical} or
25482 @code{horizontal}. Default is @code{vertical}.
25485 Set ratio of overlap window. Default value is @code{0}.
25486 When value is @code{1} overlap is set to recommended size for specific
25487 window function currently used.
25490 Set scale gain for calculating intensity color values.
25491 Default value is @code{1}.
25494 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25497 Set color rotation, must be in [-1.0, 1.0] range.
25498 Default value is @code{0}.
25501 Set start frequency from which to display spectrogram. Default is @code{0}.
25504 Set stop frequency to which to display spectrogram. Default is @code{0}.
25507 Set upper frame rate limit. Default is @code{auto}, unlimited.
25510 Draw time and frequency axes and legends. Default is disabled.
25513 The usage is very similar to the showwaves filter; see the examples in that
25516 @subsection Examples
25520 Large window with logarithmic color scaling:
25522 showspectrum=s=1280x480:scale=log
25526 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25528 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25529 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25533 @section showspectrumpic
25535 Convert input audio to a single video frame, representing the audio frequency
25538 The filter accepts the following options:
25542 Specify the video size for the output. For the syntax of this option, check the
25543 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25544 Default value is @code{4096x2048}.
25547 Specify display mode.
25549 It accepts the following values:
25552 all channels are displayed in the same row
25554 all channels are displayed in separate rows
25556 Default value is @samp{combined}.
25559 Specify display color mode.
25561 It accepts the following values:
25564 each channel is displayed in a separate color
25566 each channel is displayed using the same color scheme
25568 each channel is displayed using the rainbow color scheme
25570 each channel is displayed using the moreland color scheme
25572 each channel is displayed using the nebulae color scheme
25574 each channel is displayed using the fire color scheme
25576 each channel is displayed using the fiery color scheme
25578 each channel is displayed using the fruit color scheme
25580 each channel is displayed using the cool color scheme
25582 each channel is displayed using the magma color scheme
25584 each channel is displayed using the green color scheme
25586 each channel is displayed using the viridis color scheme
25588 each channel is displayed using the plasma color scheme
25590 each channel is displayed using the cividis color scheme
25592 each channel is displayed using the terrain color scheme
25594 Default value is @samp{intensity}.
25597 Specify scale used for calculating intensity color values.
25599 It accepts the following values:
25604 square root, default
25614 Default value is @samp{log}.
25617 Specify frequency scale.
25619 It accepts the following values:
25627 Default value is @samp{lin}.
25630 Set saturation modifier for displayed colors. Negative values provide
25631 alternative color scheme. @code{0} is no saturation at all.
25632 Saturation must be in [-10.0, 10.0] range.
25633 Default value is @code{1}.
25636 Set window function.
25638 It accepts the following values:
25662 Default value is @code{hann}.
25665 Set orientation of time vs frequency axis. Can be @code{vertical} or
25666 @code{horizontal}. Default is @code{vertical}.
25669 Set scale gain for calculating intensity color values.
25670 Default value is @code{1}.
25673 Draw time and frequency axes and legends. Default is enabled.
25676 Set color rotation, must be in [-1.0, 1.0] range.
25677 Default value is @code{0}.
25680 Set start frequency from which to display spectrogram. Default is @code{0}.
25683 Set stop frequency to which to display spectrogram. Default is @code{0}.
25686 @subsection Examples
25690 Extract an audio spectrogram of a whole audio track
25691 in a 1024x1024 picture using @command{ffmpeg}:
25693 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25697 @section showvolume
25699 Convert input audio volume to a video output.
25701 The filter accepts the following options:
25708 Set border width, allowed range is [0, 5]. Default is 1.
25711 Set channel width, allowed range is [80, 8192]. Default is 400.
25714 Set channel height, allowed range is [1, 900]. Default is 20.
25717 Set fade, allowed range is [0, 1]. Default is 0.95.
25720 Set volume color expression.
25722 The expression can use the following variables:
25726 Current max volume of channel in dB.
25732 Current channel number, starting from 0.
25736 If set, displays channel names. Default is enabled.
25739 If set, displays volume values. Default is enabled.
25742 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
25743 default is @code{h}.
25746 Set step size, allowed range is [0, 5]. Default is 0, which means
25750 Set background opacity, allowed range is [0, 1]. Default is 0.
25753 Set metering mode, can be peak: @code{p} or rms: @code{r},
25754 default is @code{p}.
25757 Set display scale, can be linear: @code{lin} or log: @code{log},
25758 default is @code{lin}.
25762 If set to > 0., display a line for the max level
25763 in the previous seconds.
25764 default is disabled: @code{0.}
25767 The color of the max line. Use when @code{dm} option is set to > 0.
25768 default is: @code{orange}
25773 Convert input audio to a video output, representing the samples waves.
25775 The filter accepts the following options:
25779 Specify the video size for the output. For the syntax of this option, check the
25780 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25781 Default value is @code{600x240}.
25786 Available values are:
25789 Draw a point for each sample.
25792 Draw a vertical line for each sample.
25795 Draw a point for each sample and a line between them.
25798 Draw a centered vertical line for each sample.
25801 Default value is @code{point}.
25804 Set the number of samples which are printed on the same column. A
25805 larger value will decrease the frame rate. Must be a positive
25806 integer. This option can be set only if the value for @var{rate}
25807 is not explicitly specified.
25810 Set the (approximate) output frame rate. This is done by setting the
25811 option @var{n}. Default value is "25".
25813 @item split_channels
25814 Set if channels should be drawn separately or overlap. Default value is 0.
25817 Set colors separated by '|' which are going to be used for drawing of each channel.
25820 Set amplitude scale.
25822 Available values are:
25840 Set the draw mode. This is mostly useful to set for high @var{n}.
25842 Available values are:
25845 Scale pixel values for each drawn sample.
25848 Draw every sample directly.
25851 Default value is @code{scale}.
25854 @subsection Examples
25858 Output the input file audio and the corresponding video representation
25861 amovie=a.mp3,asplit[out0],showwaves[out1]
25865 Create a synthetic signal and show it with showwaves, forcing a
25866 frame rate of 30 frames per second:
25868 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
25872 @section showwavespic
25874 Convert input audio to a single video frame, representing the samples waves.
25876 The filter accepts the following options:
25880 Specify the video size for the output. For the syntax of this option, check the
25881 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25882 Default value is @code{600x240}.
25884 @item split_channels
25885 Set if channels should be drawn separately or overlap. Default value is 0.
25888 Set colors separated by '|' which are going to be used for drawing of each channel.
25891 Set amplitude scale.
25893 Available values are:
25913 Available values are:
25916 Scale pixel values for each drawn sample.
25919 Draw every sample directly.
25922 Default value is @code{scale}.
25925 Set the filter mode.
25927 Available values are:
25930 Use average samples values for each drawn sample.
25933 Use peak samples values for each drawn sample.
25936 Default value is @code{average}.
25939 @subsection Examples
25943 Extract a channel split representation of the wave form of a whole audio track
25944 in a 1024x800 picture using @command{ffmpeg}:
25946 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25950 @section sidedata, asidedata
25952 Delete frame side data, or select frames based on it.
25954 This filter accepts the following options:
25958 Set mode of operation of the filter.
25960 Can be one of the following:
25964 Select every frame with side data of @code{type}.
25967 Delete side data of @code{type}. If @code{type} is not set, delete all side
25973 Set side data type used with all modes. Must be set for @code{select} mode. For
25974 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
25975 in @file{libavutil/frame.h}. For example, to choose
25976 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
25980 @section spectrumsynth
25982 Synthesize audio from 2 input video spectrums, first input stream represents
25983 magnitude across time and second represents phase across time.
25984 The filter will transform from frequency domain as displayed in videos back
25985 to time domain as presented in audio output.
25987 This filter is primarily created for reversing processed @ref{showspectrum}
25988 filter outputs, but can synthesize sound from other spectrograms too.
25989 But in such case results are going to be poor if the phase data is not
25990 available, because in such cases phase data need to be recreated, usually
25991 it's just recreated from random noise.
25992 For best results use gray only output (@code{channel} color mode in
25993 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
25994 @code{lin} scale for phase video. To produce phase, for 2nd video, use
25995 @code{data} option. Inputs videos should generally use @code{fullframe}
25996 slide mode as that saves resources needed for decoding video.
25998 The filter accepts the following options:
26002 Specify sample rate of output audio, the sample rate of audio from which
26003 spectrum was generated may differ.
26006 Set number of channels represented in input video spectrums.
26009 Set scale which was used when generating magnitude input spectrum.
26010 Can be @code{lin} or @code{log}. Default is @code{log}.
26013 Set slide which was used when generating inputs spectrums.
26014 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26015 Default is @code{fullframe}.
26018 Set window function used for resynthesis.
26021 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26022 which means optimal overlap for selected window function will be picked.
26025 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26026 Default is @code{vertical}.
26029 @subsection Examples
26033 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26034 then resynthesize videos back to audio with spectrumsynth:
26036 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
26037 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
26038 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26042 @section split, asplit
26044 Split input into several identical outputs.
26046 @code{asplit} works with audio input, @code{split} with video.
26048 The filter accepts a single parameter which specifies the number of outputs. If
26049 unspecified, it defaults to 2.
26051 @subsection Examples
26055 Create two separate outputs from the same input:
26057 [in] split [out0][out1]
26061 To create 3 or more outputs, you need to specify the number of
26064 [in] asplit=3 [out0][out1][out2]
26068 Create two separate outputs from the same input, one cropped and
26071 [in] split [splitout1][splitout2];
26072 [splitout1] crop=100:100:0:0 [cropout];
26073 [splitout2] pad=200:200:100:100 [padout];
26077 Create 5 copies of the input audio with @command{ffmpeg}:
26079 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26085 Receive commands sent through a libzmq client, and forward them to
26086 filters in the filtergraph.
26088 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26089 must be inserted between two video filters, @code{azmq} between two
26090 audio filters. Both are capable to send messages to any filter type.
26092 To enable these filters you need to install the libzmq library and
26093 headers and configure FFmpeg with @code{--enable-libzmq}.
26095 For more information about libzmq see:
26096 @url{http://www.zeromq.org/}
26098 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26099 receives messages sent through a network interface defined by the
26100 @option{bind_address} (or the abbreviation "@option{b}") option.
26101 Default value of this option is @file{tcp://localhost:5555}. You may
26102 want to alter this value to your needs, but do not forget to escape any
26103 ':' signs (see @ref{filtergraph escaping}).
26105 The received message must be in the form:
26107 @var{TARGET} @var{COMMAND} [@var{ARG}]
26110 @var{TARGET} specifies the target of the command, usually the name of
26111 the filter class or a specific filter instance name. The default
26112 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26113 but you can override this by using the @samp{filter_name@@id} syntax
26114 (see @ref{Filtergraph syntax}).
26116 @var{COMMAND} specifies the name of the command for the target filter.
26118 @var{ARG} is optional and specifies the optional argument list for the
26119 given @var{COMMAND}.
26121 Upon reception, the message is processed and the corresponding command
26122 is injected into the filtergraph. Depending on the result, the filter
26123 will send a reply to the client, adopting the format:
26125 @var{ERROR_CODE} @var{ERROR_REASON}
26129 @var{MESSAGE} is optional.
26131 @subsection Examples
26133 Look at @file{tools/zmqsend} for an example of a zmq client which can
26134 be used to send commands processed by these filters.
26136 Consider the following filtergraph generated by @command{ffplay}.
26137 In this example the last overlay filter has an instance name. All other
26138 filters will have default instance names.
26141 ffplay -dumpgraph 1 -f lavfi "
26142 color=s=100x100:c=red [l];
26143 color=s=100x100:c=blue [r];
26144 nullsrc=s=200x100, zmq [bg];
26145 [bg][l] overlay [bg+l];
26146 [bg+l][r] overlay@@my=x=100 "
26149 To change the color of the left side of the video, the following
26150 command can be used:
26152 echo Parsed_color_0 c yellow | tools/zmqsend
26155 To change the right side:
26157 echo Parsed_color_1 c pink | tools/zmqsend
26160 To change the position of the right side:
26162 echo overlay@@my x 150 | tools/zmqsend
26166 @c man end MULTIMEDIA FILTERS
26168 @chapter Multimedia Sources
26169 @c man begin MULTIMEDIA SOURCES
26171 Below is a description of the currently available multimedia sources.
26175 This is the same as @ref{movie} source, except it selects an audio
26181 Read audio and/or video stream(s) from a movie container.
26183 It accepts the following parameters:
26187 The name of the resource to read (not necessarily a file; it can also be a
26188 device or a stream accessed through some protocol).
26190 @item format_name, f
26191 Specifies the format assumed for the movie to read, and can be either
26192 the name of a container or an input device. If not specified, the
26193 format is guessed from @var{movie_name} or by probing.
26195 @item seek_point, sp
26196 Specifies the seek point in seconds. The frames will be output
26197 starting from this seek point. The parameter is evaluated with
26198 @code{av_strtod}, so the numerical value may be suffixed by an IS
26199 postfix. The default value is "0".
26202 Specifies the streams to read. Several streams can be specified,
26203 separated by "+". The source will then have as many outputs, in the
26204 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26205 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26206 respectively the default (best suited) video and audio stream. Default
26207 is "dv", or "da" if the filter is called as "amovie".
26209 @item stream_index, si
26210 Specifies the index of the video stream to read. If the value is -1,
26211 the most suitable video stream will be automatically selected. The default
26212 value is "-1". Deprecated. If the filter is called "amovie", it will select
26213 audio instead of video.
26216 Specifies how many times to read the stream in sequence.
26217 If the value is 0, the stream will be looped infinitely.
26218 Default value is "1".
26220 Note that when the movie is looped the source timestamps are not
26221 changed, so it will generate non monotonically increasing timestamps.
26223 @item discontinuity
26224 Specifies the time difference between frames above which the point is
26225 considered a timestamp discontinuity which is removed by adjusting the later
26229 It allows overlaying a second video on top of the main input of
26230 a filtergraph, as shown in this graph:
26232 input -----------> deltapts0 --> overlay --> output
26235 movie --> scale--> deltapts1 -------+
26237 @subsection Examples
26241 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26242 on top of the input labelled "in":
26244 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26245 [in] setpts=PTS-STARTPTS [main];
26246 [main][over] overlay=16:16 [out]
26250 Read from a video4linux2 device, and overlay it on top of the input
26253 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26254 [in] setpts=PTS-STARTPTS [main];
26255 [main][over] overlay=16:16 [out]
26259 Read the first video stream and the audio stream with id 0x81 from
26260 dvd.vob; the video is connected to the pad named "video" and the audio is
26261 connected to the pad named "audio":
26263 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26267 @subsection Commands
26269 Both movie and amovie support the following commands:
26272 Perform seek using "av_seek_frame".
26273 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26276 @var{stream_index}: If stream_index is -1, a default
26277 stream is selected, and @var{timestamp} is automatically converted
26278 from AV_TIME_BASE units to the stream specific time_base.
26280 @var{timestamp}: Timestamp in AVStream.time_base units
26281 or, if no stream is specified, in AV_TIME_BASE units.
26283 @var{flags}: Flags which select direction and seeking mode.
26287 Get movie duration in AV_TIME_BASE units.
26291 @c man end MULTIMEDIA SOURCES