1 @chapter Filtering Introduction
2 @c man begin FILTERING INTRODUCTION
4 Filtering in FFmpeg is enabled through the libavfilter library.
6 In libavfilter, a filter can have multiple inputs and multiple
8 To illustrate the sorts of things that are possible, we consider the
13 input --> split ---------------------> overlay --> output
16 +-----> crop --> vflip -------+
19 This filtergraph splits the input stream in two streams, then sends one
20 stream through the crop filter and the vflip filter, before merging it
21 back with the other stream by overlaying it on top. You can use the
22 following command to achieve this:
25 ffmpeg -i INPUT -vf "split [main][tmp]; [tmp] crop=iw:ih/2:0:0, vflip [flip]; [main][flip] overlay=0:H/2" OUTPUT
28 The result will be that the top half of the video is mirrored
29 onto the bottom half of the output video.
31 Filters in the same linear chain are separated by commas, and distinct
32 linear chains of filters are separated by semicolons. In our example,
33 @var{crop,vflip} are in one linear chain, @var{split} and
34 @var{overlay} are separately in another. The points where the linear
35 chains join are labelled by names enclosed in square brackets. In the
36 example, the split filter generates two outputs that are associated to
37 the labels @var{[main]} and @var{[tmp]}.
39 The stream sent to the second output of @var{split}, labelled as
40 @var{[tmp]}, is processed through the @var{crop} filter, which crops
41 away the lower half part of the video, and then vertically flipped. The
42 @var{overlay} filter takes in input the first unchanged output of the
43 split filter (which was labelled as @var{[main]}), and overlay on its
44 lower half the output generated by the @var{crop,vflip} filterchain.
46 Some filters take in input a list of parameters: they are specified
47 after the filter name and an equal sign, and are separated from each other
50 There exist so-called @var{source filters} that do not have an
51 audio/video input, and @var{sink filters} that will not have audio/video
54 @c man end FILTERING INTRODUCTION
57 @c man begin GRAPH2DOT
59 The @file{graph2dot} program included in the FFmpeg @file{tools}
60 directory can be used to parse a filtergraph description and issue a
61 corresponding textual representation in the dot language.
68 to see how to use @file{graph2dot}.
70 You can then pass the dot description to the @file{dot} program (from
71 the graphviz suite of programs) and obtain a graphical representation
74 For example the sequence of commands:
76 echo @var{GRAPH_DESCRIPTION} | \
77 tools/graph2dot -o graph.tmp && \
78 dot -Tpng graph.tmp -o graph.png && \
82 can be used to create and display an image representing the graph
83 described by the @var{GRAPH_DESCRIPTION} string. Note that this string must be
84 a complete self-contained graph, with its inputs and outputs explicitly defined.
85 For example if your command line is of the form:
87 ffmpeg -i infile -vf scale=640:360 outfile
89 your @var{GRAPH_DESCRIPTION} string will need to be of the form:
91 nullsrc,scale=640:360,nullsink
93 you may also need to set the @var{nullsrc} parameters and add a @var{format}
94 filter in order to simulate a specific input file.
98 @chapter Filtergraph description
99 @c man begin FILTERGRAPH DESCRIPTION
101 A filtergraph is a directed graph of connected filters. It can contain
102 cycles, and there can be multiple links between a pair of
103 filters. Each link has one input pad on one side connecting it to one
104 filter from which it takes its input, and one output pad on the other
105 side connecting it to one filter accepting its output.
107 Each filter in a filtergraph is an instance of a filter class
108 registered in the application, which defines the features and the
109 number of input and output pads of the filter.
111 A filter with no input pads is called a "source", and a filter with no
112 output pads is called a "sink".
114 @anchor{Filtergraph syntax}
115 @section Filtergraph syntax
117 A filtergraph has a textual representation, which is recognized by the
118 @option{-filter}/@option{-vf}/@option{-af} and
119 @option{-filter_complex} options in @command{ffmpeg} and
120 @option{-vf}/@option{-af} in @command{ffplay}, and by the
121 @code{avfilter_graph_parse_ptr()} function defined in
122 @file{libavfilter/avfilter.h}.
124 A filterchain consists of a sequence of connected filters, each one
125 connected to the previous one in the sequence. A filterchain is
126 represented by a list of ","-separated filter descriptions.
128 A filtergraph consists of a sequence of filterchains. A sequence of
129 filterchains is represented by a list of ";"-separated filterchain
132 A filter is represented by a string of the form:
133 [@var{in_link_1}]...[@var{in_link_N}]@var{filter_name}@@@var{id}=@var{arguments}[@var{out_link_1}]...[@var{out_link_M}]
135 @var{filter_name} is the name of the filter class of which the
136 described filter is an instance of, and has to be the name of one of
137 the filter classes registered in the program optionally followed by "@@@var{id}".
138 The name of the filter class is optionally followed by a string
141 @var{arguments} is a string which contains the parameters used to
142 initialize the filter instance. It may have one of two forms:
146 A ':'-separated list of @var{key=value} pairs.
149 A ':'-separated list of @var{value}. In this case, the keys are assumed to be
150 the option names in the order they are declared. E.g. the @code{fade} filter
151 declares three options in this order -- @option{type}, @option{start_frame} and
152 @option{nb_frames}. Then the parameter list @var{in:0:30} means that the value
153 @var{in} is assigned to the option @option{type}, @var{0} to
154 @option{start_frame} and @var{30} to @option{nb_frames}.
157 A ':'-separated list of mixed direct @var{value} and long @var{key=value}
158 pairs. The direct @var{value} must precede the @var{key=value} pairs, and
159 follow the same constraints order of the previous point. The following
160 @var{key=value} pairs can be set in any preferred order.
164 If the option value itself is a list of items (e.g. the @code{format} filter
165 takes a list of pixel formats), the items in the list are usually separated by
168 The list of arguments can be quoted using the character @samp{'} as initial
169 and ending mark, and the character @samp{\} for escaping the characters
170 within the quoted text; otherwise the argument string is considered
171 terminated when the next special character (belonging to the set
172 @samp{[]=;,}) is encountered.
174 The name and arguments of the filter are optionally preceded and
175 followed by a list of link labels.
176 A link label allows one to name a link and associate it to a filter output
177 or input pad. The preceding labels @var{in_link_1}
178 ... @var{in_link_N}, are associated to the filter input pads,
179 the following labels @var{out_link_1} ... @var{out_link_M}, are
180 associated to the output pads.
182 When two link labels with the same name are found in the
183 filtergraph, a link between the corresponding input and output pad is
186 If an output pad is not labelled, it is linked by default to the first
187 unlabelled input pad of the next filter in the filterchain.
188 For example in the filterchain
190 nullsrc, split[L1], [L2]overlay, nullsink
192 the split filter instance has two output pads, and the overlay filter
193 instance two input pads. The first output pad of split is labelled
194 "L1", the first input pad of overlay is labelled "L2", and the second
195 output pad of split is linked to the second input pad of overlay,
196 which are both unlabelled.
198 In a filter description, if the input label of the first filter is not
199 specified, "in" is assumed; if the output label of the last filter is not
200 specified, "out" is assumed.
202 In a complete filterchain all the unlabelled filter input and output
203 pads must be connected. A filtergraph is considered valid if all the
204 filter input and output pads of all the filterchains are connected.
206 Libavfilter will automatically insert @ref{scale} filters where format
207 conversion is required. It is possible to specify swscale flags
208 for those automatically inserted scalers by prepending
209 @code{sws_flags=@var{flags};}
210 to the filtergraph description.
212 Here is a BNF description of the filtergraph syntax:
214 @var{NAME} ::= sequence of alphanumeric characters and '_'
215 @var{FILTER_NAME} ::= @var{NAME}["@@"@var{NAME}]
216 @var{LINKLABEL} ::= "[" @var{NAME} "]"
217 @var{LINKLABELS} ::= @var{LINKLABEL} [@var{LINKLABELS}]
218 @var{FILTER_ARGUMENTS} ::= sequence of chars (possibly quoted)
219 @var{FILTER} ::= [@var{LINKLABELS}] @var{FILTER_NAME} ["=" @var{FILTER_ARGUMENTS}] [@var{LINKLABELS}]
220 @var{FILTERCHAIN} ::= @var{FILTER} [,@var{FILTERCHAIN}]
221 @var{FILTERGRAPH} ::= [sws_flags=@var{flags};] @var{FILTERCHAIN} [;@var{FILTERGRAPH}]
224 @anchor{filtergraph escaping}
225 @section Notes on filtergraph escaping
227 Filtergraph description composition entails several levels of
228 escaping. See @ref{quoting_and_escaping,,the "Quoting and escaping"
229 section in the ffmpeg-utils(1) manual,ffmpeg-utils} for more
230 information about the employed escaping procedure.
232 A first level escaping affects the content of each filter option
233 value, which may contain the special character @code{:} used to
234 separate values, or one of the escaping characters @code{\'}.
236 A second level escaping affects the whole filter description, which
237 may contain the escaping characters @code{\'} or the special
238 characters @code{[],;} used by the filtergraph description.
240 Finally, when you specify a filtergraph on a shell commandline, you
241 need to perform a third level escaping for the shell special
242 characters contained within it.
244 For example, consider the following string to be embedded in
245 the @ref{drawtext} filter description @option{text} value:
247 this is a 'string': may contain one, or more, special characters
250 This string contains the @code{'} special escaping character, and the
251 @code{:} special character, so it needs to be escaped in this way:
253 text=this is a \'string\'\: may contain one, or more, special characters
256 A second level of escaping is required when embedding the filter
257 description in a filtergraph description, in order to escape all the
258 filtergraph special characters. Thus the example above becomes:
260 drawtext=text=this is a \\\'string\\\'\\: may contain one\, or more\, special characters
262 (note that in addition to the @code{\'} escaping special characters,
263 also @code{,} needs to be escaped).
265 Finally an additional level of escaping is needed when writing the
266 filtergraph description in a shell command, which depends on the
267 escaping rules of the adopted shell. For example, assuming that
268 @code{\} is special and needs to be escaped with another @code{\}, the
269 previous string will finally result in:
271 -vf "drawtext=text=this is a \\\\\\'string\\\\\\'\\\\: may contain one\\, or more\\, special characters"
274 @chapter Timeline editing
276 Some filters support a generic @option{enable} option. For the filters
277 supporting timeline editing, this option can be set to an expression which is
278 evaluated before sending a frame to the filter. If the evaluation is non-zero,
279 the filter will be enabled, otherwise the frame will be sent unchanged to the
280 next filter in the filtergraph.
282 The expression accepts the following values:
285 timestamp expressed in seconds, NAN if the input timestamp is unknown
288 sequential number of the input frame, starting from 0
291 the position in the file of the input frame, NAN if unknown
295 width and height of the input frame if video
298 Additionally, these filters support an @option{enable} command that can be used
299 to re-define the expression.
301 Like any other filtering option, the @option{enable} option follows the same
304 For example, to enable a blur filter (@ref{smartblur}) from 10 seconds to 3
305 minutes, and a @ref{curves} filter starting at 3 seconds:
307 smartblur = enable='between(t,10,3*60)',
308 curves = enable='gte(t,3)' : preset=cross_process
311 See @code{ffmpeg -filters} to view which filters have timeline support.
313 @c man end FILTERGRAPH DESCRIPTION
316 @chapter Changing options at runtime with a command
318 Some options can be changed during the operation of the filter using
319 a command. These options are marked 'T' on the output of
320 @command{ffmpeg} @option{-h filter=<name of filter>}.
321 The name of the command is the name of the option and the argument is
325 @chapter Options for filters with several inputs (framesync)
326 @c man begin OPTIONS FOR FILTERS WITH SEVERAL INPUTS
328 Some filters with several inputs support a common set of options.
329 These options can only be set by name, not with the short notation.
333 The action to take when EOF is encountered on the secondary input; it accepts
334 one of the following values:
338 Repeat the last frame (the default).
342 Pass the main input through.
346 If set to 1, force the output to terminate when the shortest input
347 terminates. Default value is 0.
350 If set to 1, force the filter to extend the last frame of secondary streams
351 until the end of the primary stream. A value of 0 disables this behavior.
355 @c man end OPTIONS FOR FILTERS WITH SEVERAL INPUTS
357 @chapter Audio Filters
358 @c man begin AUDIO FILTERS
360 When you configure your FFmpeg build, you can disable any of the
361 existing filters using @code{--disable-filters}.
362 The configure output will show the audio filters included in your
365 Below is a description of the currently available audio filters.
369 A compressor is mainly used to reduce the dynamic range of a signal.
370 Especially modern music is mostly compressed at a high ratio to
371 improve the overall loudness. It's done to get the highest attention
372 of a listener, "fatten" the sound and bring more "power" to the track.
373 If a signal is compressed too much it may sound dull or "dead"
374 afterwards or it may start to "pump" (which could be a powerful effect
375 but can also destroy a track completely).
376 The right compression is the key to reach a professional sound and is
377 the high art of mixing and mastering. Because of its complex settings
378 it may take a long time to get the right feeling for this kind of effect.
380 Compression is done by detecting the volume above a chosen level
381 @code{threshold} and dividing it by the factor set with @code{ratio}.
382 So if you set the threshold to -12dB and your signal reaches -6dB a ratio
383 of 2:1 will result in a signal at -9dB. Because an exact manipulation of
384 the signal would cause distortion of the waveform the reduction can be
385 levelled over the time. This is done by setting "Attack" and "Release".
386 @code{attack} determines how long the signal has to rise above the threshold
387 before any reduction will occur and @code{release} sets the time the signal
388 has to fall below the threshold to reduce the reduction again. Shorter signals
389 than the chosen attack time will be left untouched.
390 The overall reduction of the signal can be made up afterwards with the
391 @code{makeup} setting. So compressing the peaks of a signal about 6dB and
392 raising the makeup to this level results in a signal twice as loud than the
393 source. To gain a softer entry in the compression the @code{knee} flattens the
394 hard edge at the threshold in the range of the chosen decibels.
396 The filter accepts the following options:
400 Set input gain. Default is 1. Range is between 0.015625 and 64.
403 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
404 Default is @code{downward}.
407 If a signal of stream rises above this level it will affect the gain
409 By default it is 0.125. Range is between 0.00097563 and 1.
412 Set a ratio by which the signal is reduced. 1:2 means that if the level
413 rose 4dB above the threshold, it will be only 2dB above after the reduction.
414 Default is 2. Range is between 1 and 20.
417 Amount of milliseconds the signal has to rise above the threshold before gain
418 reduction starts. Default is 20. Range is between 0.01 and 2000.
421 Amount of milliseconds the signal has to fall below the threshold before
422 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
425 Set the amount by how much signal will be amplified after processing.
426 Default is 1. Range is from 1 to 64.
429 Curve the sharp knee around the threshold to enter gain reduction more softly.
430 Default is 2.82843. Range is between 1 and 8.
433 Choose if the @code{average} level between all channels of input stream
434 or the louder(@code{maximum}) channel of input stream affects the
435 reduction. Default is @code{average}.
438 Should the exact signal be taken in case of @code{peak} or an RMS one in case
439 of @code{rms}. Default is @code{rms} which is mostly smoother.
442 How much to use compressed signal in output. Default is 1.
443 Range is between 0 and 1.
448 This filter supports the all above options as @ref{commands}.
451 Simple audio dynamic range compression/expansion filter.
453 The filter accepts the following options:
457 Set contrast. Default is 33. Allowed range is between 0 and 100.
462 Copy the input audio source unchanged to the output. This is mainly useful for
467 Apply cross fade from one input audio stream to another input audio stream.
468 The cross fade is applied for specified duration near the end of first stream.
470 The filter accepts the following options:
474 Specify the number of samples for which the cross fade effect has to last.
475 At the end of the cross fade effect the first input audio will be completely
476 silent. Default is 44100.
479 Specify the duration of the cross fade effect. See
480 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
481 for the accepted syntax.
482 By default the duration is determined by @var{nb_samples}.
483 If set this option is used instead of @var{nb_samples}.
486 Should first stream end overlap with second stream start. Default is enabled.
489 Set curve for cross fade transition for first stream.
492 Set curve for cross fade transition for second stream.
494 For description of available curve types see @ref{afade} filter description.
501 Cross fade from one input to another:
503 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
507 Cross fade from one input to another but without overlapping:
509 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
514 Split audio stream into several bands.
516 This filter splits audio stream into two or more frequency ranges.
517 Summing all streams back will give flat output.
519 The filter accepts the following options:
523 Set split frequencies. Those must be positive and increasing.
526 Set filter order for each band split. This controls filter roll-off or steepness
527 of filter transfer function.
528 Available values are:
553 Default is @var{4th}.
556 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
559 Set output gain for each band. Default value is 1 for all bands.
566 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
567 each band will be in separate stream:
569 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
573 Same as above, but with higher filter order:
575 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
579 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
581 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=8th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
587 Reduce audio bit resolution.
589 This filter is bit crusher with enhanced functionality. A bit crusher
590 is used to audibly reduce number of bits an audio signal is sampled
591 with. This doesn't change the bit depth at all, it just produces the
592 effect. Material reduced in bit depth sounds more harsh and "digital".
593 This filter is able to even round to continuous values instead of discrete
595 Additionally it has a D/C offset which results in different crushing of
596 the lower and the upper half of the signal.
597 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
599 Another feature of this filter is the logarithmic mode.
600 This setting switches from linear distances between bits to logarithmic ones.
601 The result is a much more "natural" sounding crusher which doesn't gate low
602 signals for example. The human ear has a logarithmic perception,
603 so this kind of crushing is much more pleasant.
604 Logarithmic crushing is also able to get anti-aliased.
606 The filter accepts the following options:
622 Can be linear: @code{lin} or logarithmic: @code{log}.
631 Set sample reduction.
634 Enable LFO. By default disabled.
645 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
649 Remove impulsive noise from input audio.
651 Samples detected as impulsive noise are replaced by interpolated samples using
652 autoregressive modelling.
656 Set window size, in milliseconds. Allowed range is from @code{10} to
657 @code{100}. Default value is @code{55} milliseconds.
658 This sets size of window which will be processed at once.
661 Set window overlap, in percentage of window size. Allowed range is from
662 @code{50} to @code{95}. Default value is @code{75} percent.
663 Setting this to a very high value increases impulsive noise removal but makes
664 whole process much slower.
667 Set autoregression order, in percentage of window size. Allowed range is from
668 @code{0} to @code{25}. Default value is @code{2} percent. This option also
669 controls quality of interpolated samples using neighbour good samples.
672 Set threshold value. Allowed range is from @code{1} to @code{100}.
673 Default value is @code{2}.
674 This controls the strength of impulsive noise which is going to be removed.
675 The lower value, the more samples will be detected as impulsive noise.
678 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
679 @code{10}. Default value is @code{2}.
680 If any two samples detected as noise are spaced less than this value then any
681 sample between those two samples will be also detected as noise.
686 It accepts the following values:
689 Select overlap-add method. Even not interpolated samples are slightly
690 changed with this method.
693 Select overlap-save method. Not interpolated samples remain unchanged.
696 Default value is @code{a}.
700 Remove clipped samples from input audio.
702 Samples detected as clipped are replaced by interpolated samples using
703 autoregressive modelling.
707 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
708 Default value is @code{55} milliseconds.
709 This sets size of window which will be processed at once.
712 Set window overlap, in percentage of window size. Allowed range is from @code{50}
713 to @code{95}. Default value is @code{75} percent.
716 Set autoregression order, in percentage of window size. Allowed range is from
717 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
718 quality of interpolated samples using neighbour good samples.
721 Set threshold value. Allowed range is from @code{1} to @code{100}.
722 Default value is @code{10}. Higher values make clip detection less aggressive.
725 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
726 Default value is @code{1000}. Higher values make clip detection less aggressive.
731 It accepts the following values:
734 Select overlap-add method. Even not interpolated samples are slightly changed
738 Select overlap-save method. Not interpolated samples remain unchanged.
741 Default value is @code{a}.
746 Delay one or more audio channels.
748 Samples in delayed channel are filled with silence.
750 The filter accepts the following option:
754 Set list of delays in milliseconds for each channel separated by '|'.
755 Unused delays will be silently ignored. If number of given delays is
756 smaller than number of channels all remaining channels will not be delayed.
757 If you want to delay exact number of samples, append 'S' to number.
758 If you want instead to delay in seconds, append 's' to number.
761 Use last set delay for all remaining channels. By default is disabled.
762 This option if enabled changes how option @code{delays} is interpreted.
769 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
770 the second channel (and any other channels that may be present) unchanged.
776 Delay second channel by 500 samples, the third channel by 700 samples and leave
777 the first channel (and any other channels that may be present) unchanged.
783 Delay all channels by same number of samples:
785 adelay=delays=64S:all=1
790 Remedy denormals in audio by adding extremely low-level noise.
792 This filter shall be placed before any filter that can produce denormals.
794 A description of the accepted parameters follows.
798 Set level of added noise in dB. Default is @code{-351}.
799 Allowed range is from -451 to -90.
802 Set type of added noise.
815 Default is @code{dc}.
820 This filter supports the all above options as @ref{commands}.
822 @section aderivative, aintegral
824 Compute derivative/integral of audio stream.
826 Applying both filters one after another produces original audio.
830 Apply echoing to the input audio.
832 Echoes are reflected sound and can occur naturally amongst mountains
833 (and sometimes large buildings) when talking or shouting; digital echo
834 effects emulate this behaviour and are often used to help fill out the
835 sound of a single instrument or vocal. The time difference between the
836 original signal and the reflection is the @code{delay}, and the
837 loudness of the reflected signal is the @code{decay}.
838 Multiple echoes can have different delays and decays.
840 A description of the accepted parameters follows.
844 Set input gain of reflected signal. Default is @code{0.6}.
847 Set output gain of reflected signal. Default is @code{0.3}.
850 Set list of time intervals in milliseconds between original signal and reflections
851 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
852 Default is @code{1000}.
855 Set list of loudness of reflected signals separated by '|'.
856 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
857 Default is @code{0.5}.
864 Make it sound as if there are twice as many instruments as are actually playing:
866 aecho=0.8:0.88:60:0.4
870 If delay is very short, then it sounds like a (metallic) robot playing music:
876 A longer delay will sound like an open air concert in the mountains:
878 aecho=0.8:0.9:1000:0.3
882 Same as above but with one more mountain:
884 aecho=0.8:0.9:1000|1800:0.3|0.25
889 Audio emphasis filter creates or restores material directly taken from LPs or
890 emphased CDs with different filter curves. E.g. to store music on vinyl the
891 signal has to be altered by a filter first to even out the disadvantages of
892 this recording medium.
893 Once the material is played back the inverse filter has to be applied to
894 restore the distortion of the frequency response.
896 The filter accepts the following options:
906 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
907 use @code{production} mode. Default is @code{reproduction} mode.
910 Set filter type. Selects medium. Can be one of the following:
922 select Compact Disc (CD).
928 select 50µs (FM-KF).
930 select 75µs (FM-KF).
936 This filter supports the all above options as @ref{commands}.
940 Modify an audio signal according to the specified expressions.
942 This filter accepts one or more expressions (one for each channel),
943 which are evaluated and used to modify a corresponding audio signal.
945 It accepts the following parameters:
949 Set the '|'-separated expressions list for each separate channel. If
950 the number of input channels is greater than the number of
951 expressions, the last specified expression is used for the remaining
954 @item channel_layout, c
955 Set output channel layout. If not specified, the channel layout is
956 specified by the number of expressions. If set to @samp{same}, it will
957 use by default the same input channel layout.
960 Each expression in @var{exprs} can contain the following constants and functions:
964 channel number of the current expression
967 number of the evaluated sample, starting from 0
973 time of the evaluated sample expressed in seconds
976 @item nb_out_channels
977 input and output number of channels
980 the value of input channel with number @var{CH}
983 Note: this filter is slow. For faster processing you should use a
992 aeval=val(ch)/2:c=same
996 Invert phase of the second channel:
1005 Apply fade-in/out effect to input audio.
1007 A description of the accepted parameters follows.
1011 Specify the effect type, can be either @code{in} for fade-in, or
1012 @code{out} for a fade-out effect. Default is @code{in}.
1014 @item start_sample, ss
1015 Specify the number of the start sample for starting to apply the fade
1016 effect. Default is 0.
1018 @item nb_samples, ns
1019 Specify the number of samples for which the fade effect has to last. At
1020 the end of the fade-in effect the output audio will have the same
1021 volume as the input audio, at the end of the fade-out transition
1022 the output audio will be silence. Default is 44100.
1024 @item start_time, st
1025 Specify the start time of the fade effect. Default is 0.
1026 The value must be specified as a time duration; see
1027 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1028 for the accepted syntax.
1029 If set this option is used instead of @var{start_sample}.
1032 Specify the duration of the fade effect. See
1033 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1034 for the accepted syntax.
1035 At the end of the fade-in effect the output audio will have the same
1036 volume as the input audio, at the end of the fade-out transition
1037 the output audio will be silence.
1038 By default the duration is determined by @var{nb_samples}.
1039 If set this option is used instead of @var{nb_samples}.
1042 Set curve for fade transition.
1044 It accepts the following values:
1047 select triangular, linear slope (default)
1049 select quarter of sine wave
1051 select half of sine wave
1053 select exponential sine wave
1057 select inverted parabola
1071 select inverted quarter of sine wave
1073 select inverted half of sine wave
1075 select double-exponential seat
1077 select double-exponential sigmoid
1079 select logistic sigmoid
1081 select sine cardinal function
1083 select inverted sine cardinal function
1089 @subsection Commands
1091 This filter supports the all above options as @ref{commands}.
1093 @subsection Examples
1097 Fade in first 15 seconds of audio:
1099 afade=t=in:ss=0:d=15
1103 Fade out last 25 seconds of a 900 seconds audio:
1105 afade=t=out:st=875:d=25
1110 Denoise audio samples with FFT.
1112 A description of the accepted parameters follows.
1116 Set the noise reduction in dB, allowed range is 0.01 to 97.
1117 Default value is 12 dB.
1120 Set the noise floor in dB, allowed range is -80 to -20.
1121 Default value is -50 dB.
1126 It accepts the following values:
1135 Select shellac noise.
1138 Select custom noise, defined in @code{bn} option.
1140 Default value is white noise.
1144 Set custom band noise for every one of 15 bands.
1145 Bands are separated by ' ' or '|'.
1148 Set the residual floor in dB, allowed range is -80 to -20.
1149 Default value is -38 dB.
1152 Enable noise tracking. By default is disabled.
1153 With this enabled, noise floor is automatically adjusted.
1156 Enable residual tracking. By default is disabled.
1159 Set the output mode.
1161 It accepts the following values:
1164 Pass input unchanged.
1167 Pass noise filtered out.
1172 Default value is @var{o}.
1176 @subsection Commands
1178 This filter supports the following commands:
1180 @item sample_noise, sn
1181 Start or stop measuring noise profile.
1182 Syntax for the command is : "start" or "stop" string.
1183 After measuring noise profile is stopped it will be
1184 automatically applied in filtering.
1186 @item noise_reduction, nr
1187 Change noise reduction. Argument is single float number.
1188 Syntax for the command is : "@var{noise_reduction}"
1190 @item noise_floor, nf
1191 Change noise floor. Argument is single float number.
1192 Syntax for the command is : "@var{noise_floor}"
1194 @item output_mode, om
1195 Change output mode operation.
1196 Syntax for the command is : "i", "o" or "n" string.
1200 Apply arbitrary expressions to samples in frequency domain.
1204 Set frequency domain real expression for each separate channel separated
1205 by '|'. Default is "re".
1206 If the number of input channels is greater than the number of
1207 expressions, the last specified expression is used for the remaining
1211 Set frequency domain imaginary expression for each separate channel
1212 separated by '|'. Default is "im".
1214 Each expression in @var{real} and @var{imag} can contain the following
1215 constants and functions:
1222 current frequency bin number
1225 number of available bins
1228 channel number of the current expression
1237 current real part of frequency bin of current channel
1240 current imaginary part of frequency bin of current channel
1243 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1246 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1250 Set window size. Allowed range is from 16 to 131072.
1251 Default is @code{4096}
1254 Set window function. Default is @code{hann}.
1257 Set window overlap. If set to 1, the recommended overlap for selected
1258 window function will be picked. Default is @code{0.75}.
1261 @subsection Examples
1265 Leave almost only low frequencies in audio:
1267 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1271 Apply robotize effect:
1273 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1277 Apply whisper effect:
1279 afftfilt="real='hypot(re,im)*cos((random(0)*2-1)*2*3.14)':imag='hypot(re,im)*sin((random(1)*2-1)*2*3.14)':win_size=128:overlap=0.8"
1286 Apply an arbitrary Finite Impulse Response filter.
1288 This filter is designed for applying long FIR filters,
1289 up to 60 seconds long.
1291 It can be used as component for digital crossover filters,
1292 room equalization, cross talk cancellation, wavefield synthesis,
1293 auralization, ambiophonics, ambisonics and spatialization.
1295 This filter uses the streams higher than first one as FIR coefficients.
1296 If the non-first stream holds a single channel, it will be used
1297 for all input channels in the first stream, otherwise
1298 the number of channels in the non-first stream must be same as
1299 the number of channels in the first stream.
1301 It accepts the following parameters:
1305 Set dry gain. This sets input gain.
1308 Set wet gain. This sets final output gain.
1311 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1314 Enable applying gain measured from power of IR.
1316 Set which approach to use for auto gain measurement.
1320 Do not apply any gain.
1323 select peak gain, very conservative approach. This is default value.
1326 select DC gain, limited application.
1329 select gain to noise approach, this is most popular one.
1333 Set gain to be applied to IR coefficients before filtering.
1334 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1337 Set format of IR stream. Can be @code{mono} or @code{input}.
1338 Default is @code{input}.
1341 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1342 Allowed range is 0.1 to 60 seconds.
1345 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1346 By default it is disabled.
1349 Set for which IR channel to display frequency response. By default is first channel
1350 displayed. This option is used only when @var{response} is enabled.
1353 Set video stream size. This option is used only when @var{response} is enabled.
1356 Set video stream frame rate. This option is used only when @var{response} is enabled.
1359 Set minimal partition size used for convolution. Default is @var{8192}.
1360 Allowed range is from @var{1} to @var{32768}.
1361 Lower values decreases latency at cost of higher CPU usage.
1364 Set maximal partition size used for convolution. Default is @var{8192}.
1365 Allowed range is from @var{8} to @var{32768}.
1366 Lower values may increase CPU usage.
1369 Set number of input impulse responses streams which will be switchable at runtime.
1370 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1373 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1374 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1375 This option can be changed at runtime via @ref{commands}.
1378 @subsection Examples
1382 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1384 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1391 Set output format constraints for the input audio. The framework will
1392 negotiate the most appropriate format to minimize conversions.
1394 It accepts the following parameters:
1397 @item sample_fmts, f
1398 A '|'-separated list of requested sample formats.
1400 @item sample_rates, r
1401 A '|'-separated list of requested sample rates.
1403 @item channel_layouts, cl
1404 A '|'-separated list of requested channel layouts.
1406 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1407 for the required syntax.
1410 If a parameter is omitted, all values are allowed.
1412 Force the output to either unsigned 8-bit or signed 16-bit stereo
1414 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1418 Apply frequency shift to input audio samples.
1420 The filter accepts the following options:
1424 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1425 Default value is 0.0.
1428 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
1429 Default value is 1.0.
1432 @subsection Commands
1434 This filter supports the all above options as @ref{commands}.
1438 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1439 processing reduces disturbing noise between useful signals.
1441 Gating is done by detecting the volume below a chosen level @var{threshold}
1442 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1443 floor is set via @var{range}. Because an exact manipulation of the signal
1444 would cause distortion of the waveform the reduction can be levelled over
1445 time. This is done by setting @var{attack} and @var{release}.
1447 @var{attack} determines how long the signal has to fall below the threshold
1448 before any reduction will occur and @var{release} sets the time the signal
1449 has to rise above the threshold to reduce the reduction again.
1450 Shorter signals than the chosen attack time will be left untouched.
1454 Set input level before filtering.
1455 Default is 1. Allowed range is from 0.015625 to 64.
1458 Set the mode of operation. Can be @code{upward} or @code{downward}.
1459 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1460 will be amplified, expanding dynamic range in upward direction.
1461 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1464 Set the level of gain reduction when the signal is below the threshold.
1465 Default is 0.06125. Allowed range is from 0 to 1.
1466 Setting this to 0 disables reduction and then filter behaves like expander.
1469 If a signal rises above this level the gain reduction is released.
1470 Default is 0.125. Allowed range is from 0 to 1.
1473 Set a ratio by which the signal is reduced.
1474 Default is 2. Allowed range is from 1 to 9000.
1477 Amount of milliseconds the signal has to rise above the threshold before gain
1479 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1482 Amount of milliseconds the signal has to fall below the threshold before the
1483 reduction is increased again. Default is 250 milliseconds.
1484 Allowed range is from 0.01 to 9000.
1487 Set amount of amplification of signal after processing.
1488 Default is 1. Allowed range is from 1 to 64.
1491 Curve the sharp knee around the threshold to enter gain reduction more softly.
1492 Default is 2.828427125. Allowed range is from 1 to 8.
1495 Choose if exact signal should be taken for detection or an RMS like one.
1496 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1499 Choose if the average level between all channels or the louder channel affects
1501 Default is @code{average}. Can be @code{average} or @code{maximum}.
1504 @subsection Commands
1506 This filter supports the all above options as @ref{commands}.
1510 Apply an arbitrary Infinite Impulse Response filter.
1512 It accepts the following parameters:
1516 Set B/numerator/zeros/reflection coefficients.
1519 Set A/denominator/poles/ladder coefficients.
1531 Set coefficients format.
1535 lattice-ladder function
1537 analog transfer function
1539 digital transfer function
1541 Z-plane zeros/poles, cartesian (default)
1543 Z-plane zeros/poles, polar radians
1545 Z-plane zeros/poles, polar degrees
1551 Set type of processing.
1563 Set filtering precision.
1567 double-precision floating-point (default)
1569 single-precision floating-point
1577 Normalize filter coefficients, by default is enabled.
1578 Enabling it will normalize magnitude response at DC to 0dB.
1581 How much to use filtered signal in output. Default is 1.
1582 Range is between 0 and 1.
1585 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1586 By default it is disabled.
1589 Set for which IR channel to display frequency response. By default is first channel
1590 displayed. This option is used only when @var{response} is enabled.
1593 Set video stream size. This option is used only when @var{response} is enabled.
1596 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1599 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1600 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1603 Different coefficients and gains can be provided for every channel, in such case
1604 use '|' to separate coefficients or gains. Last provided coefficients will be
1605 used for all remaining channels.
1607 @subsection Examples
1611 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1613 aiir=k=1:z=7.957584807809675810E-1 -2.575128568908332300 3.674839853930788710 -2.57512875289799137 7.957586296317130880E-1:p=1 -2.86950072432325953 3.63022088054647218 -2.28075678147272232 6.361362326477423500E-1:f=tf:r=d
1617 Same as above but in @code{zp} format:
1619 aiir=k=0.79575848078096756:z=0.80918701+0.58773007i 0.80918701-0.58773007i 0.80884700+0.58784055i 0.80884700-0.58784055i:p=0.63892345+0.59951235i 0.63892345-0.59951235i 0.79582691+0.44198673i 0.79582691-0.44198673i:f=zp:r=s
1623 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1625 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1631 The limiter prevents an input signal from rising over a desired threshold.
1632 This limiter uses lookahead technology to prevent your signal from distorting.
1633 It means that there is a small delay after the signal is processed. Keep in mind
1634 that the delay it produces is the attack time you set.
1636 The filter accepts the following options:
1640 Set input gain. Default is 1.
1643 Set output gain. Default is 1.
1646 Don't let signals above this level pass the limiter. Default is 1.
1649 The limiter will reach its attenuation level in this amount of time in
1650 milliseconds. Default is 5 milliseconds.
1653 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1654 Default is 50 milliseconds.
1657 When gain reduction is always needed ASC takes care of releasing to an
1658 average reduction level rather than reaching a reduction of 0 in the release
1662 Select how much the release time is affected by ASC, 0 means nearly no changes
1663 in release time while 1 produces higher release times.
1666 Auto level output signal. Default is enabled.
1667 This normalizes audio back to 0dB if enabled.
1670 Depending on picked setting it is recommended to upsample input 2x or 4x times
1671 with @ref{aresample} before applying this filter.
1675 Apply a two-pole all-pass filter with central frequency (in Hz)
1676 @var{frequency}, and filter-width @var{width}.
1677 An all-pass filter changes the audio's frequency to phase relationship
1678 without changing its frequency to amplitude relationship.
1680 The filter accepts the following options:
1684 Set frequency in Hz.
1687 Set method to specify band-width of filter.
1702 Specify the band-width of a filter in width_type units.
1705 How much to use filtered signal in output. Default is 1.
1706 Range is between 0 and 1.
1709 Specify which channels to filter, by default all available are filtered.
1712 Normalize biquad coefficients, by default is disabled.
1713 Enabling it will normalize magnitude response at DC to 0dB.
1716 Set the filter order, can be 1 or 2. Default is 2.
1719 Set transform type of IIR filter.
1728 Set precison of filtering.
1731 Pick automatic sample format depending on surround filters.
1733 Always use signed 16-bit.
1735 Always use signed 32-bit.
1737 Always use float 32-bit.
1739 Always use float 64-bit.
1743 @subsection Commands
1745 This filter supports the following commands:
1748 Change allpass frequency.
1749 Syntax for the command is : "@var{frequency}"
1752 Change allpass width_type.
1753 Syntax for the command is : "@var{width_type}"
1756 Change allpass width.
1757 Syntax for the command is : "@var{width}"
1761 Syntax for the command is : "@var{mix}"
1768 The filter accepts the following options:
1772 Set the number of loops. Setting this value to -1 will result in infinite loops.
1776 Set maximal number of samples. Default is 0.
1779 Set first sample of loop. Default is 0.
1785 Merge two or more audio streams into a single multi-channel stream.
1787 The filter accepts the following options:
1792 Set the number of inputs. Default is 2.
1796 If the channel layouts of the inputs are disjoint, and therefore compatible,
1797 the channel layout of the output will be set accordingly and the channels
1798 will be reordered as necessary. If the channel layouts of the inputs are not
1799 disjoint, the output will have all the channels of the first input then all
1800 the channels of the second input, in that order, and the channel layout of
1801 the output will be the default value corresponding to the total number of
1804 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1805 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1806 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1807 first input, b1 is the first channel of the second input).
1809 On the other hand, if both input are in stereo, the output channels will be
1810 in the default order: a1, a2, b1, b2, and the channel layout will be
1811 arbitrarily set to 4.0, which may or may not be the expected value.
1813 All inputs must have the same sample rate, and format.
1815 If inputs do not have the same duration, the output will stop with the
1818 @subsection Examples
1822 Merge two mono files into a stereo stream:
1824 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1828 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1830 ffmpeg -i input.mkv -filter_complex "[0:1][0:2][0:3][0:4][0:5][0:6] amerge=inputs=6" -c:a pcm_s16le output.mkv
1836 Mixes multiple audio inputs into a single output.
1838 Note that this filter only supports float samples (the @var{amerge}
1839 and @var{pan} audio filters support many formats). If the @var{amix}
1840 input has integer samples then @ref{aresample} will be automatically
1841 inserted to perform the conversion to float samples.
1845 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1847 will mix 3 input audio streams to a single output with the same duration as the
1848 first input and a dropout transition time of 3 seconds.
1850 It accepts the following parameters:
1854 The number of inputs. If unspecified, it defaults to 2.
1857 How to determine the end-of-stream.
1861 The duration of the longest input. (default)
1864 The duration of the shortest input.
1867 The duration of the first input.
1871 @item dropout_transition
1872 The transition time, in seconds, for volume renormalization when an input
1873 stream ends. The default value is 2 seconds.
1876 Specify weight of each input audio stream as sequence.
1877 Each weight is separated by space. By default all inputs have same weight.
1880 @subsection Commands
1882 This filter supports the following commands:
1885 Syntax is same as option with same name.
1890 Multiply first audio stream with second audio stream and store result
1891 in output audio stream. Multiplication is done by multiplying each
1892 sample from first stream with sample at same position from second stream.
1894 With this element-wise multiplication one can create amplitude fades and
1895 amplitude modulations.
1897 @section anequalizer
1899 High-order parametric multiband equalizer for each channel.
1901 It accepts the following parameters:
1905 This option string is in format:
1906 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1907 Each equalizer band is separated by '|'.
1911 Set channel number to which equalization will be applied.
1912 If input doesn't have that channel the entry is ignored.
1915 Set central frequency for band.
1916 If input doesn't have that frequency the entry is ignored.
1919 Set band width in Hertz.
1922 Set band gain in dB.
1925 Set filter type for band, optional, can be:
1929 Butterworth, this is default.
1940 With this option activated frequency response of anequalizer is displayed
1944 Set video stream size. Only useful if curves option is activated.
1947 Set max gain that will be displayed. Only useful if curves option is activated.
1948 Setting this to a reasonable value makes it possible to display gain which is derived from
1949 neighbour bands which are too close to each other and thus produce higher gain
1950 when both are activated.
1953 Set frequency scale used to draw frequency response in video output.
1954 Can be linear or logarithmic. Default is logarithmic.
1957 Set color for each channel curve which is going to be displayed in video stream.
1958 This is list of color names separated by space or by '|'.
1959 Unrecognised or missing colors will be replaced by white color.
1962 @subsection Examples
1966 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1967 for first 2 channels using Chebyshev type 1 filter:
1969 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1973 @subsection Commands
1975 This filter supports the following commands:
1978 Alter existing filter parameters.
1979 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1981 @var{fN} is existing filter number, starting from 0, if no such filter is available
1983 @var{freq} set new frequency parameter.
1984 @var{width} set new width parameter in Hertz.
1985 @var{gain} set new gain parameter in dB.
1987 Full filter invocation with asendcmd may look like this:
1988 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1993 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1995 Each sample is adjusted by looking for other samples with similar contexts. This
1996 context similarity is defined by comparing their surrounding patches of size
1997 @option{p}. Patches are searched in an area of @option{r} around the sample.
1999 The filter accepts the following options:
2003 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
2006 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
2007 Default value is 2 milliseconds.
2010 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
2011 Default value is 6 milliseconds.
2014 Set the output mode.
2016 It accepts the following values:
2019 Pass input unchanged.
2022 Pass noise filtered out.
2027 Default value is @var{o}.
2031 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
2034 @subsection Commands
2036 This filter supports the all above options as @ref{commands}.
2039 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
2041 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
2042 relate to producing the least mean square of the error signal (difference between the desired,
2043 2nd input audio stream and the actual signal, the 1st input audio stream).
2045 A description of the accepted options follows.
2058 Set the filter leakage.
2061 It accepts the following values:
2070 Pass filtered samples.
2073 Pass difference between desired and filtered samples.
2075 Default value is @var{o}.
2079 @subsection Examples
2083 One of many usages of this filter is noise reduction, input audio is filtered
2084 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2086 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2090 @subsection Commands
2092 This filter supports the same commands as options, excluding option @code{order}.
2096 Pass the audio source unchanged to the output.
2100 Pad the end of an audio stream with silence.
2102 This can be used together with @command{ffmpeg} @option{-shortest} to
2103 extend audio streams to the same length as the video stream.
2105 A description of the accepted options follows.
2109 Set silence packet size. Default value is 4096.
2112 Set the number of samples of silence to add to the end. After the
2113 value is reached, the stream is terminated. This option is mutually
2114 exclusive with @option{whole_len}.
2117 Set the minimum total number of samples in the output audio stream. If
2118 the value is longer than the input audio length, silence is added to
2119 the end, until the value is reached. This option is mutually exclusive
2120 with @option{pad_len}.
2123 Specify the duration of samples of silence to add. See
2124 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2125 for the accepted syntax. Used only if set to non-zero value.
2128 Specify the minimum total duration in the output audio stream. See
2129 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2130 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2131 the input audio length, silence is added to the end, until the value is reached.
2132 This option is mutually exclusive with @option{pad_dur}
2135 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2136 nor @option{whole_dur} option is set, the filter will add silence to the end of
2137 the input stream indefinitely.
2139 @subsection Examples
2143 Add 1024 samples of silence to the end of the input:
2149 Make sure the audio output will contain at least 10000 samples, pad
2150 the input with silence if required:
2152 apad=whole_len=10000
2156 Use @command{ffmpeg} to pad the audio input with silence, so that the
2157 video stream will always result the shortest and will be converted
2158 until the end in the output file when using the @option{shortest}
2161 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2166 Add a phasing effect to the input audio.
2168 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2169 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2171 A description of the accepted parameters follows.
2175 Set input gain. Default is 0.4.
2178 Set output gain. Default is 0.74
2181 Set delay in milliseconds. Default is 3.0.
2184 Set decay. Default is 0.4.
2187 Set modulation speed in Hz. Default is 0.5.
2190 Set modulation type. Default is triangular.
2192 It accepts the following values:
2199 @section aphaseshift
2200 Apply phase shift to input audio samples.
2202 The filter accepts the following options:
2206 Specify phase shift. Allowed range is from -1.0 to 1.0.
2207 Default value is 0.0.
2210 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
2211 Default value is 1.0.
2214 @subsection Commands
2216 This filter supports the all above options as @ref{commands}.
2220 Audio pulsator is something between an autopanner and a tremolo.
2221 But it can produce funny stereo effects as well. Pulsator changes the volume
2222 of the left and right channel based on a LFO (low frequency oscillator) with
2223 different waveforms and shifted phases.
2224 This filter have the ability to define an offset between left and right
2225 channel. An offset of 0 means that both LFO shapes match each other.
2226 The left and right channel are altered equally - a conventional tremolo.
2227 An offset of 50% means that the shape of the right channel is exactly shifted
2228 in phase (or moved backwards about half of the frequency) - pulsator acts as
2229 an autopanner. At 1 both curves match again. Every setting in between moves the
2230 phase shift gapless between all stages and produces some "bypassing" sounds with
2231 sine and triangle waveforms. The more you set the offset near 1 (starting from
2232 the 0.5) the faster the signal passes from the left to the right speaker.
2234 The filter accepts the following options:
2238 Set input gain. By default it is 1. Range is [0.015625 - 64].
2241 Set output gain. By default it is 1. Range is [0.015625 - 64].
2244 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2245 sawup or sawdown. Default is sine.
2248 Set modulation. Define how much of original signal is affected by the LFO.
2251 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2254 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2257 Set pulse width. Default is 1. Allowed range is [0 - 2].
2260 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2263 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2267 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2271 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2272 if timing is set to hz.
2278 Resample the input audio to the specified parameters, using the
2279 libswresample library. If none are specified then the filter will
2280 automatically convert between its input and output.
2282 This filter is also able to stretch/squeeze the audio data to make it match
2283 the timestamps or to inject silence / cut out audio to make it match the
2284 timestamps, do a combination of both or do neither.
2286 The filter accepts the syntax
2287 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2288 expresses a sample rate and @var{resampler_options} is a list of
2289 @var{key}=@var{value} pairs, separated by ":". See the
2290 @ref{Resampler Options,,"Resampler Options" section in the
2291 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2292 for the complete list of supported options.
2294 @subsection Examples
2298 Resample the input audio to 44100Hz:
2304 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2305 samples per second compensation:
2307 aresample=async=1000
2313 Reverse an audio clip.
2315 Warning: This filter requires memory to buffer the entire clip, so trimming
2318 @subsection Examples
2322 Take the first 5 seconds of a clip, and reverse it.
2324 atrim=end=5,areverse
2330 Reduce noise from speech using Recurrent Neural Networks.
2332 This filter accepts the following options:
2336 Set train model file to load. This option is always required.
2339 Set how much to mix filtered samples into final output.
2340 Allowed range is from -1 to 1. Default value is 1.
2341 Negative values are special, they set how much to keep filtered noise
2342 in the final filter output. Set this option to -1 to hear actual
2343 noise removed from input signal.
2346 @section asetnsamples
2348 Set the number of samples per each output audio frame.
2350 The last output packet may contain a different number of samples, as
2351 the filter will flush all the remaining samples when the input audio
2354 The filter accepts the following options:
2358 @item nb_out_samples, n
2359 Set the number of frames per each output audio frame. The number is
2360 intended as the number of samples @emph{per each channel}.
2361 Default value is 1024.
2364 If set to 1, the filter will pad the last audio frame with zeroes, so
2365 that the last frame will contain the same number of samples as the
2366 previous ones. Default value is 1.
2369 For example, to set the number of per-frame samples to 1234 and
2370 disable padding for the last frame, use:
2372 asetnsamples=n=1234:p=0
2377 Set the sample rate without altering the PCM data.
2378 This will result in a change of speed and pitch.
2380 The filter accepts the following options:
2383 @item sample_rate, r
2384 Set the output sample rate. Default is 44100 Hz.
2389 Show a line containing various information for each input audio frame.
2390 The input audio is not modified.
2392 The shown line contains a sequence of key/value pairs of the form
2393 @var{key}:@var{value}.
2395 The following values are shown in the output:
2399 The (sequential) number of the input frame, starting from 0.
2402 The presentation timestamp of the input frame, in time base units; the time base
2403 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2406 The presentation timestamp of the input frame in seconds.
2409 position of the frame in the input stream, -1 if this information in
2410 unavailable and/or meaningless (for example in case of synthetic audio)
2419 The sample rate for the audio frame.
2422 The number of samples (per channel) in the frame.
2425 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2426 audio, the data is treated as if all the planes were concatenated.
2428 @item plane_checksums
2429 A list of Adler-32 checksums for each data plane.
2433 Apply audio soft clipping.
2435 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2436 along a smooth curve, rather than the abrupt shape of hard-clipping.
2438 This filter accepts the following options:
2442 Set type of soft-clipping.
2444 It accepts the following values:
2458 Set threshold from where to start clipping. Default value is 0dB or 1.
2461 Set gain applied to output. Default value is 0dB or 1.
2464 Set additional parameter which controls sigmoid function.
2467 Set oversampling factor.
2470 @subsection Commands
2472 This filter supports the all above options as @ref{commands}.
2475 Automatic Speech Recognition
2477 This filter uses PocketSphinx for speech recognition. To enable
2478 compilation of this filter, you need to configure FFmpeg with
2479 @code{--enable-pocketsphinx}.
2481 It accepts the following options:
2485 Set sampling rate of input audio. Defaults is @code{16000}.
2486 This need to match speech models, otherwise one will get poor results.
2489 Set dictionary containing acoustic model files.
2492 Set pronunciation dictionary.
2495 Set language model file.
2498 Set language model set.
2501 Set which language model to use.
2504 Set output for log messages.
2507 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2512 Display time domain statistical information about the audio channels.
2513 Statistics are calculated and displayed for each audio channel and,
2514 where applicable, an overall figure is also given.
2516 It accepts the following option:
2519 Short window length in seconds, used for peak and trough RMS measurement.
2520 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2524 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2525 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2528 Available keys for each channel are:
2574 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2575 this @code{lavfi.astats.Overall.Peak_count}.
2577 For description what each key means read below.
2580 Set number of frame after which stats are going to be recalculated.
2581 Default is disabled.
2583 @item measure_perchannel
2584 Select the entries which need to be measured per channel. The metadata keys can
2585 be used as flags, default is @option{all} which measures everything.
2586 @option{none} disables all per channel measurement.
2588 @item measure_overall
2589 Select the entries which need to be measured overall. The metadata keys can
2590 be used as flags, default is @option{all} which measures everything.
2591 @option{none} disables all overall measurement.
2595 A description of each shown parameter follows:
2599 Mean amplitude displacement from zero.
2602 Minimal sample level.
2605 Maximal sample level.
2607 @item Min difference
2608 Minimal difference between two consecutive samples.
2610 @item Max difference
2611 Maximal difference between two consecutive samples.
2613 @item Mean difference
2614 Mean difference between two consecutive samples.
2615 The average of each difference between two consecutive samples.
2617 @item RMS difference
2618 Root Mean Square difference between two consecutive samples.
2622 Standard peak and RMS level measured in dBFS.
2626 Peak and trough values for RMS level measured over a short window.
2629 Standard ratio of peak to RMS level (note: not in dB).
2632 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2633 (i.e. either @var{Min level} or @var{Max level}).
2636 Number of occasions (not the number of samples) that the signal attained either
2637 @var{Min level} or @var{Max level}.
2639 @item Noise floor dB
2640 Minimum local peak measured in dBFS over a short window.
2642 @item Noise floor count
2643 Number of occasions (not the number of samples) that the signal attained
2647 Overall bit depth of audio. Number of bits used for each sample.
2650 Measured dynamic range of audio in dB.
2652 @item Zero crossings
2653 Number of points where the waveform crosses the zero level axis.
2655 @item Zero crossings rate
2656 Rate of Zero crossings and number of audio samples.
2660 Boost subwoofer frequencies.
2662 The filter accepts the following options:
2666 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2667 Default value is 0.7.
2670 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2671 Default value is 0.7.
2674 Set delay line decay gain value. Allowed range is from 0 to 1.
2675 Default value is 0.7.
2678 Set delay line feedback gain value. Allowed range is from 0 to 1.
2679 Default value is 0.9.
2682 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2683 Default value is 100.
2686 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2687 Default value is 0.5.
2690 Set delay. Allowed range is from 1 to 100.
2691 Default value is 20.
2694 @subsection Commands
2696 This filter supports the all above options as @ref{commands}.
2699 Cut subwoofer frequencies.
2701 This filter allows to set custom, steeper
2702 roll off than highpass filter, and thus is able to more attenuate
2703 frequency content in stop-band.
2705 The filter accepts the following options:
2709 Set cutoff frequency in Hertz. Allowed range is 2 to 200.
2710 Default value is 20.
2713 Set filter order. Available values are from 3 to 20.
2714 Default value is 10.
2717 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2720 @subsection Commands
2722 This filter supports the all above options as @ref{commands}.
2725 Cut super frequencies.
2727 The filter accepts the following options:
2731 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2732 Default value is 20000.
2735 Set filter order. Available values are from 3 to 20.
2736 Default value is 10.
2739 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2742 @subsection Commands
2744 This filter supports the all above options as @ref{commands}.
2747 Apply high order Butterworth band-pass filter.
2749 The filter accepts the following options:
2753 Set center frequency in Hertz. Allowed range is 2 to 999999.
2754 Default value is 1000.
2757 Set filter order. Available values are from 4 to 20.
2761 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2764 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2767 @subsection Commands
2769 This filter supports the all above options as @ref{commands}.
2772 Apply high order Butterworth band-stop filter.
2774 The filter accepts the following options:
2778 Set center frequency in Hertz. Allowed range is 2 to 999999.
2779 Default value is 1000.
2782 Set filter order. Available values are from 4 to 20.
2786 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2789 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2792 @subsection Commands
2794 This filter supports the all above options as @ref{commands}.
2800 The filter accepts exactly one parameter, the audio tempo. If not
2801 specified then the filter will assume nominal 1.0 tempo. Tempo must
2802 be in the [0.5, 100.0] range.
2804 Note that tempo greater than 2 will skip some samples rather than
2805 blend them in. If for any reason this is a concern it is always
2806 possible to daisy-chain several instances of atempo to achieve the
2807 desired product tempo.
2809 @subsection Examples
2813 Slow down audio to 80% tempo:
2819 To speed up audio to 300% tempo:
2825 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2827 atempo=sqrt(3),atempo=sqrt(3)
2831 @subsection Commands
2833 This filter supports the following commands:
2836 Change filter tempo scale factor.
2837 Syntax for the command is : "@var{tempo}"
2842 Trim the input so that the output contains one continuous subpart of the input.
2844 It accepts the following parameters:
2847 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2848 sample with the timestamp @var{start} will be the first sample in the output.
2851 Specify time of the first audio sample that will be dropped, i.e. the
2852 audio sample immediately preceding the one with the timestamp @var{end} will be
2853 the last sample in the output.
2856 Same as @var{start}, except this option sets the start timestamp in samples
2860 Same as @var{end}, except this option sets the end timestamp in samples instead
2864 The maximum duration of the output in seconds.
2867 The number of the first sample that should be output.
2870 The number of the first sample that should be dropped.
2873 @option{start}, @option{end}, and @option{duration} are expressed as time
2874 duration specifications; see
2875 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2877 Note that the first two sets of the start/end options and the @option{duration}
2878 option look at the frame timestamp, while the _sample options simply count the
2879 samples that pass through the filter. So start/end_pts and start/end_sample will
2880 give different results when the timestamps are wrong, inexact or do not start at
2881 zero. Also note that this filter does not modify the timestamps. If you wish
2882 to have the output timestamps start at zero, insert the asetpts filter after the
2885 If multiple start or end options are set, this filter tries to be greedy and
2886 keep all samples that match at least one of the specified constraints. To keep
2887 only the part that matches all the constraints at once, chain multiple atrim
2890 The defaults are such that all the input is kept. So it is possible to set e.g.
2891 just the end values to keep everything before the specified time.
2896 Drop everything except the second minute of input:
2898 ffmpeg -i INPUT -af atrim=60:120
2902 Keep only the first 1000 samples:
2904 ffmpeg -i INPUT -af atrim=end_sample=1000
2909 @section axcorrelate
2910 Calculate normalized cross-correlation between two input audio streams.
2912 Resulted samples are always between -1 and 1 inclusive.
2913 If result is 1 it means two input samples are highly correlated in that selected segment.
2914 Result 0 means they are not correlated at all.
2915 If result is -1 it means two input samples are out of phase, which means they cancel each
2918 The filter accepts the following options:
2922 Set size of segment over which cross-correlation is calculated.
2923 Default is 256. Allowed range is from 2 to 131072.
2926 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2927 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2928 are always zero and thus need much less calculations to make.
2929 This is generally not true, but is valid for typical audio streams.
2932 @subsection Examples
2936 Calculate correlation between channels in stereo audio stream:
2938 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2944 Apply a two-pole Butterworth band-pass filter with central
2945 frequency @var{frequency}, and (3dB-point) band-width width.
2946 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2947 instead of the default: constant 0dB peak gain.
2948 The filter roll off at 6dB per octave (20dB per decade).
2950 The filter accepts the following options:
2954 Set the filter's central frequency. Default is @code{3000}.
2957 Constant skirt gain if set to 1. Defaults to 0.
2960 Set method to specify band-width of filter.
2975 Specify the band-width of a filter in width_type units.
2978 How much to use filtered signal in output. Default is 1.
2979 Range is between 0 and 1.
2982 Specify which channels to filter, by default all available are filtered.
2985 Normalize biquad coefficients, by default is disabled.
2986 Enabling it will normalize magnitude response at DC to 0dB.
2989 Set transform type of IIR filter.
2998 Set precison of filtering.
3001 Pick automatic sample format depending on surround filters.
3003 Always use signed 16-bit.
3005 Always use signed 32-bit.
3007 Always use float 32-bit.
3009 Always use float 64-bit.
3013 @subsection Commands
3015 This filter supports the following commands:
3018 Change bandpass frequency.
3019 Syntax for the command is : "@var{frequency}"
3022 Change bandpass width_type.
3023 Syntax for the command is : "@var{width_type}"
3026 Change bandpass width.
3027 Syntax for the command is : "@var{width}"
3030 Change bandpass mix.
3031 Syntax for the command is : "@var{mix}"
3036 Apply a two-pole Butterworth band-reject filter with central
3037 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
3038 The filter roll off at 6dB per octave (20dB per decade).
3040 The filter accepts the following options:
3044 Set the filter's central frequency. Default is @code{3000}.
3047 Set method to specify band-width of filter.
3062 Specify the band-width of a filter in width_type units.
3065 How much to use filtered signal in output. Default is 1.
3066 Range is between 0 and 1.
3069 Specify which channels to filter, by default all available are filtered.
3072 Normalize biquad coefficients, by default is disabled.
3073 Enabling it will normalize magnitude response at DC to 0dB.
3076 Set transform type of IIR filter.
3085 Set precison of filtering.
3088 Pick automatic sample format depending on surround filters.
3090 Always use signed 16-bit.
3092 Always use signed 32-bit.
3094 Always use float 32-bit.
3096 Always use float 64-bit.
3100 @subsection Commands
3102 This filter supports the following commands:
3105 Change bandreject frequency.
3106 Syntax for the command is : "@var{frequency}"
3109 Change bandreject width_type.
3110 Syntax for the command is : "@var{width_type}"
3113 Change bandreject width.
3114 Syntax for the command is : "@var{width}"
3117 Change bandreject mix.
3118 Syntax for the command is : "@var{mix}"
3121 @section bass, lowshelf
3123 Boost or cut the bass (lower) frequencies of the audio using a two-pole
3124 shelving filter with a response similar to that of a standard
3125 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
3127 The filter accepts the following options:
3131 Give the gain at 0 Hz. Its useful range is about -20
3132 (for a large cut) to +20 (for a large boost).
3133 Beware of clipping when using a positive gain.
3136 Set the filter's central frequency and so can be used
3137 to extend or reduce the frequency range to be boosted or cut.
3138 The default value is @code{100} Hz.
3141 Set method to specify band-width of filter.
3156 Determine how steep is the filter's shelf transition.
3159 Set number of poles. Default is 2.
3162 How much to use filtered signal in output. Default is 1.
3163 Range is between 0 and 1.
3166 Specify which channels to filter, by default all available are filtered.
3169 Normalize biquad coefficients, by default is disabled.
3170 Enabling it will normalize magnitude response at DC to 0dB.
3173 Set transform type of IIR filter.
3182 Set precison of filtering.
3185 Pick automatic sample format depending on surround filters.
3187 Always use signed 16-bit.
3189 Always use signed 32-bit.
3191 Always use float 32-bit.
3193 Always use float 64-bit.
3197 @subsection Commands
3199 This filter supports the following commands:
3202 Change bass frequency.
3203 Syntax for the command is : "@var{frequency}"
3206 Change bass width_type.
3207 Syntax for the command is : "@var{width_type}"
3211 Syntax for the command is : "@var{width}"
3215 Syntax for the command is : "@var{gain}"
3219 Syntax for the command is : "@var{mix}"
3224 Apply a biquad IIR filter with the given coefficients.
3225 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3226 are the numerator and denominator coefficients respectively.
3227 and @var{channels}, @var{c} specify which channels to filter, by default all
3228 available are filtered.
3230 @subsection Commands
3232 This filter supports the following commands:
3240 Change biquad parameter.
3241 Syntax for the command is : "@var{value}"
3244 How much to use filtered signal in output. Default is 1.
3245 Range is between 0 and 1.
3248 Specify which channels to filter, by default all available are filtered.
3251 Normalize biquad coefficients, by default is disabled.
3252 Enabling it will normalize magnitude response at DC to 0dB.
3255 Set transform type of IIR filter.
3264 Set precison of filtering.
3267 Pick automatic sample format depending on surround filters.
3269 Always use signed 16-bit.
3271 Always use signed 32-bit.
3273 Always use float 32-bit.
3275 Always use float 64-bit.
3280 Bauer stereo to binaural transformation, which improves headphone listening of
3281 stereo audio records.
3283 To enable compilation of this filter you need to configure FFmpeg with
3284 @code{--enable-libbs2b}.
3286 It accepts the following parameters:
3290 Pre-defined crossfeed level.
3294 Default level (fcut=700, feed=50).
3297 Chu Moy circuit (fcut=700, feed=60).
3300 Jan Meier circuit (fcut=650, feed=95).
3305 Cut frequency (in Hz).
3314 Remap input channels to new locations.
3316 It accepts the following parameters:
3319 Map channels from input to output. The argument is a '|'-separated list of
3320 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3321 @var{in_channel} form. @var{in_channel} can be either the name of the input
3322 channel (e.g. FL for front left) or its index in the input channel layout.
3323 @var{out_channel} is the name of the output channel or its index in the output
3324 channel layout. If @var{out_channel} is not given then it is implicitly an
3325 index, starting with zero and increasing by one for each mapping.
3327 @item channel_layout
3328 The channel layout of the output stream.
3331 If no mapping is present, the filter will implicitly map input channels to
3332 output channels, preserving indices.
3334 @subsection Examples
3338 For example, assuming a 5.1+downmix input MOV file,
3340 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3342 will create an output WAV file tagged as stereo from the downmix channels of
3346 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3348 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3352 @section channelsplit
3354 Split each channel from an input audio stream into a separate output stream.
3356 It accepts the following parameters:
3358 @item channel_layout
3359 The channel layout of the input stream. The default is "stereo".
3361 A channel layout describing the channels to be extracted as separate output streams
3362 or "all" to extract each input channel as a separate stream. The default is "all".
3364 Choosing channels not present in channel layout in the input will result in an error.
3367 @subsection Examples
3371 For example, assuming a stereo input MP3 file,
3373 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3375 will create an output Matroska file with two audio streams, one containing only
3376 the left channel and the other the right channel.
3379 Split a 5.1 WAV file into per-channel files:
3381 ffmpeg -i in.wav -filter_complex
3382 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3383 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3384 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3389 Extract only LFE from a 5.1 WAV file:
3391 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3392 -map '[LFE]' lfe.wav
3397 Add a chorus effect to the audio.
3399 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3401 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3402 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3403 The modulation depth defines the range the modulated delay is played before or after
3404 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3405 sound tuned around the original one, like in a chorus where some vocals are slightly
3408 It accepts the following parameters:
3411 Set input gain. Default is 0.4.
3414 Set output gain. Default is 0.4.
3417 Set delays. A typical delay is around 40ms to 60ms.
3429 @subsection Examples
3435 chorus=0.7:0.9:55:0.4:0.25:2
3441 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3445 Fuller sounding chorus with three delays:
3447 chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3
3452 Compress or expand the audio's dynamic range.
3454 It accepts the following parameters:
3460 A list of times in seconds for each channel over which the instantaneous level
3461 of the input signal is averaged to determine its volume. @var{attacks} refers to
3462 increase of volume and @var{decays} refers to decrease of volume. For most
3463 situations, the attack time (response to the audio getting louder) should be
3464 shorter than the decay time, because the human ear is more sensitive to sudden
3465 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3466 a typical value for decay is 0.8 seconds.
3467 If specified number of attacks & decays is lower than number of channels, the last
3468 set attack/decay will be used for all remaining channels.
3471 A list of points for the transfer function, specified in dB relative to the
3472 maximum possible signal amplitude. Each key points list must be defined using
3473 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3474 @code{x0/y0 x1/y1 x2/y2 ....}
3476 The input values must be in strictly increasing order but the transfer function
3477 does not have to be monotonically rising. The point @code{0/0} is assumed but
3478 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3479 function are @code{-70/-70|-60/-20|1/0}.
3482 Set the curve radius in dB for all joints. It defaults to 0.01.
3485 Set the additional gain in dB to be applied at all points on the transfer
3486 function. This allows for easy adjustment of the overall gain.
3490 Set an initial volume, in dB, to be assumed for each channel when filtering
3491 starts. This permits the user to supply a nominal level initially, so that, for
3492 example, a very large gain is not applied to initial signal levels before the
3493 companding has begun to operate. A typical value for audio which is initially
3494 quiet is -90 dB. It defaults to 0.
3497 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3498 delayed before being fed to the volume adjuster. Specifying a delay
3499 approximately equal to the attack/decay times allows the filter to effectively
3500 operate in predictive rather than reactive mode. It defaults to 0.
3504 @subsection Examples
3508 Make music with both quiet and loud passages suitable for listening to in a
3511 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3514 Another example for audio with whisper and explosion parts:
3516 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3520 A noise gate for when the noise is at a lower level than the signal:
3522 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3526 Here is another noise gate, this time for when the noise is at a higher level
3527 than the signal (making it, in some ways, similar to squelch):
3529 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3533 2:1 compression starting at -6dB:
3535 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3539 2:1 compression starting at -9dB:
3541 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3545 2:1 compression starting at -12dB:
3547 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3551 2:1 compression starting at -18dB:
3553 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3557 3:1 compression starting at -15dB:
3559 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3565 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3571 compand=attacks=0:points=-80/-169|-54/-80|-49.5/-64.6|-41.1/-41.1|-25.8/-15|-10.8/-4.5|0/0|20/8.3
3575 Hard limiter at -6dB:
3577 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3581 Hard limiter at -12dB:
3583 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3587 Hard noise gate at -35 dB:
3589 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3595 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3599 @section compensationdelay
3601 Compensation Delay Line is a metric based delay to compensate differing
3602 positions of microphones or speakers.
3604 For example, you have recorded guitar with two microphones placed in
3605 different locations. Because the front of sound wave has fixed speed in
3606 normal conditions, the phasing of microphones can vary and depends on
3607 their location and interposition. The best sound mix can be achieved when
3608 these microphones are in phase (synchronized). Note that a distance of
3609 ~30 cm between microphones makes one microphone capture the signal in
3610 antiphase to the other microphone. That makes the final mix sound moody.
3611 This filter helps to solve phasing problems by adding different delays
3612 to each microphone track and make them synchronized.
3614 The best result can be reached when you take one track as base and
3615 synchronize other tracks one by one with it.
3616 Remember that synchronization/delay tolerance depends on sample rate, too.
3617 Higher sample rates will give more tolerance.
3619 The filter accepts the following parameters:
3623 Set millimeters distance. This is compensation distance for fine tuning.
3627 Set cm distance. This is compensation distance for tightening distance setup.
3631 Set meters distance. This is compensation distance for hard distance setup.
3635 Set dry amount. Amount of unprocessed (dry) signal.
3639 Set wet amount. Amount of processed (wet) signal.
3643 Set temperature in degrees Celsius. This is the temperature of the environment.
3648 Apply headphone crossfeed filter.
3650 Crossfeed is the process of blending the left and right channels of stereo
3652 It is mainly used to reduce extreme stereo separation of low frequencies.
3654 The intent is to produce more speaker like sound to the listener.
3656 The filter accepts the following options:
3660 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3661 This sets gain of low shelf filter for side part of stereo image.
3662 Default is -6dB. Max allowed is -30db when strength is set to 1.
3665 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3666 This sets cut off frequency of low shelf filter. Default is cut off near
3667 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3670 Set curve slope of low shelf filter. Default is 0.5.
3671 Allowed range is from 0.01 to 1.
3674 Set input gain. Default is 0.9.
3677 Set output gain. Default is 1.
3680 @subsection Commands
3682 This filter supports the all above options as @ref{commands}.
3684 @section crystalizer
3685 Simple algorithm to expand audio dynamic range.
3687 The filter accepts the following options:
3691 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3692 (unchanged sound) to 10.0 (maximum effect).
3695 Enable clipping. By default is enabled.
3698 @subsection Commands
3700 This filter supports the all above options as @ref{commands}.
3703 Apply a DC shift to the audio.
3705 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3706 in the recording chain) from the audio. The effect of a DC offset is reduced
3707 headroom and hence volume. The @ref{astats} filter can be used to determine if
3708 a signal has a DC offset.
3712 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3716 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3717 used to prevent clipping.
3722 Apply de-essing to the audio samples.
3726 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3730 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3734 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3738 Set the output mode.
3740 It accepts the following values:
3743 Pass input unchanged.
3746 Pass ess filtered out.
3751 Default value is @var{o}.
3757 Measure audio dynamic range.
3759 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3760 is found in transition material. And anything less that 8 have very poor dynamics
3761 and is very compressed.
3763 The filter accepts the following options:
3767 Set window length in seconds used to split audio into segments of equal length.
3768 Default is 3 seconds.
3772 Dynamic Audio Normalizer.
3774 This filter applies a certain amount of gain to the input audio in order
3775 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3776 contrast to more "simple" normalization algorithms, the Dynamic Audio
3777 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3778 This allows for applying extra gain to the "quiet" sections of the audio
3779 while avoiding distortions or clipping the "loud" sections. In other words:
3780 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3781 sections, in the sense that the volume of each section is brought to the
3782 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3783 this goal *without* applying "dynamic range compressing". It will retain 100%
3784 of the dynamic range *within* each section of the audio file.
3788 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3789 Default is 500 milliseconds.
3790 The Dynamic Audio Normalizer processes the input audio in small chunks,
3791 referred to as frames. This is required, because a peak magnitude has no
3792 meaning for just a single sample value. Instead, we need to determine the
3793 peak magnitude for a contiguous sequence of sample values. While a "standard"
3794 normalizer would simply use the peak magnitude of the complete file, the
3795 Dynamic Audio Normalizer determines the peak magnitude individually for each
3796 frame. The length of a frame is specified in milliseconds. By default, the
3797 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3798 been found to give good results with most files.
3799 Note that the exact frame length, in number of samples, will be determined
3800 automatically, based on the sampling rate of the individual input audio file.
3803 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3804 number. Default is 31.
3805 Probably the most important parameter of the Dynamic Audio Normalizer is the
3806 @code{window size} of the Gaussian smoothing filter. The filter's window size
3807 is specified in frames, centered around the current frame. For the sake of
3808 simplicity, this must be an odd number. Consequently, the default value of 31
3809 takes into account the current frame, as well as the 15 preceding frames and
3810 the 15 subsequent frames. Using a larger window results in a stronger
3811 smoothing effect and thus in less gain variation, i.e. slower gain
3812 adaptation. Conversely, using a smaller window results in a weaker smoothing
3813 effect and thus in more gain variation, i.e. faster gain adaptation.
3814 In other words, the more you increase this value, the more the Dynamic Audio
3815 Normalizer will behave like a "traditional" normalization filter. On the
3816 contrary, the more you decrease this value, the more the Dynamic Audio
3817 Normalizer will behave like a dynamic range compressor.
3820 Set the target peak value. This specifies the highest permissible magnitude
3821 level for the normalized audio input. This filter will try to approach the
3822 target peak magnitude as closely as possible, but at the same time it also
3823 makes sure that the normalized signal will never exceed the peak magnitude.
3824 A frame's maximum local gain factor is imposed directly by the target peak
3825 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3826 It is not recommended to go above this value.
3829 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3830 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3831 factor for each input frame, i.e. the maximum gain factor that does not
3832 result in clipping or distortion. The maximum gain factor is determined by
3833 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3834 additionally bounds the frame's maximum gain factor by a predetermined
3835 (global) maximum gain factor. This is done in order to avoid excessive gain
3836 factors in "silent" or almost silent frames. By default, the maximum gain
3837 factor is 10.0, For most inputs the default value should be sufficient and
3838 it usually is not recommended to increase this value. Though, for input
3839 with an extremely low overall volume level, it may be necessary to allow even
3840 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3841 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3842 Instead, a "sigmoid" threshold function will be applied. This way, the
3843 gain factors will smoothly approach the threshold value, but never exceed that
3847 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3848 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3849 This means that the maximum local gain factor for each frame is defined
3850 (only) by the frame's highest magnitude sample. This way, the samples can
3851 be amplified as much as possible without exceeding the maximum signal
3852 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3853 Normalizer can also take into account the frame's root mean square,
3854 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3855 determine the power of a time-varying signal. It is therefore considered
3856 that the RMS is a better approximation of the "perceived loudness" than
3857 just looking at the signal's peak magnitude. Consequently, by adjusting all
3858 frames to a constant RMS value, a uniform "perceived loudness" can be
3859 established. If a target RMS value has been specified, a frame's local gain
3860 factor is defined as the factor that would result in exactly that RMS value.
3861 Note, however, that the maximum local gain factor is still restricted by the
3862 frame's highest magnitude sample, in order to prevent clipping.
3865 Enable channels coupling. By default is enabled.
3866 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3867 amount. This means the same gain factor will be applied to all channels, i.e.
3868 the maximum possible gain factor is determined by the "loudest" channel.
3869 However, in some recordings, it may happen that the volume of the different
3870 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3871 In this case, this option can be used to disable the channel coupling. This way,
3872 the gain factor will be determined independently for each channel, depending
3873 only on the individual channel's highest magnitude sample. This allows for
3874 harmonizing the volume of the different channels.
3877 Enable DC bias correction. By default is disabled.
3878 An audio signal (in the time domain) is a sequence of sample values.
3879 In the Dynamic Audio Normalizer these sample values are represented in the
3880 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3881 audio signal, or "waveform", should be centered around the zero point.
3882 That means if we calculate the mean value of all samples in a file, or in a
3883 single frame, then the result should be 0.0 or at least very close to that
3884 value. If, however, there is a significant deviation of the mean value from
3885 0.0, in either positive or negative direction, this is referred to as a
3886 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3887 Audio Normalizer provides optional DC bias correction.
3888 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3889 the mean value, or "DC correction" offset, of each input frame and subtract
3890 that value from all of the frame's sample values which ensures those samples
3891 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3892 boundaries, the DC correction offset values will be interpolated smoothly
3893 between neighbouring frames.
3895 @item altboundary, b
3896 Enable alternative boundary mode. By default is disabled.
3897 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3898 around each frame. This includes the preceding frames as well as the
3899 subsequent frames. However, for the "boundary" frames, located at the very
3900 beginning and at the very end of the audio file, not all neighbouring
3901 frames are available. In particular, for the first few frames in the audio
3902 file, the preceding frames are not known. And, similarly, for the last few
3903 frames in the audio file, the subsequent frames are not known. Thus, the
3904 question arises which gain factors should be assumed for the missing frames
3905 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3906 to deal with this situation. The default boundary mode assumes a gain factor
3907 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3908 "fade out" at the beginning and at the end of the input, respectively.
3911 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3912 By default, the Dynamic Audio Normalizer does not apply "traditional"
3913 compression. This means that signal peaks will not be pruned and thus the
3914 full dynamic range will be retained within each local neighbourhood. However,
3915 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3916 normalization algorithm with a more "traditional" compression.
3917 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3918 (thresholding) function. If (and only if) the compression feature is enabled,
3919 all input frames will be processed by a soft knee thresholding function prior
3920 to the actual normalization process. Put simply, the thresholding function is
3921 going to prune all samples whose magnitude exceeds a certain threshold value.
3922 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3923 value. Instead, the threshold value will be adjusted for each individual
3925 In general, smaller parameters result in stronger compression, and vice versa.
3926 Values below 3.0 are not recommended, because audible distortion may appear.
3929 Set the target threshold value. This specifies the lowest permissible
3930 magnitude level for the audio input which will be normalized.
3931 If input frame volume is above this value frame will be normalized.
3932 Otherwise frame may not be normalized at all. The default value is set
3933 to 0, which means all input frames will be normalized.
3934 This option is mostly useful if digital noise is not wanted to be amplified.
3937 @subsection Commands
3939 This filter supports the all above options as @ref{commands}.
3943 Make audio easier to listen to on headphones.
3945 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3946 so that when listened to on headphones the stereo image is moved from
3947 inside your head (standard for headphones) to outside and in front of
3948 the listener (standard for speakers).
3954 Apply a two-pole peaking equalisation (EQ) filter. With this
3955 filter, the signal-level at and around a selected frequency can
3956 be increased or decreased, whilst (unlike bandpass and bandreject
3957 filters) that at all other frequencies is unchanged.
3959 In order to produce complex equalisation curves, this filter can
3960 be given several times, each with a different central frequency.
3962 The filter accepts the following options:
3966 Set the filter's central frequency in Hz.
3969 Set method to specify band-width of filter.
3984 Specify the band-width of a filter in width_type units.
3987 Set the required gain or attenuation in dB.
3988 Beware of clipping when using a positive gain.
3991 How much to use filtered signal in output. Default is 1.
3992 Range is between 0 and 1.
3995 Specify which channels to filter, by default all available are filtered.
3998 Normalize biquad coefficients, by default is disabled.
3999 Enabling it will normalize magnitude response at DC to 0dB.
4002 Set transform type of IIR filter.
4011 Set precison of filtering.
4014 Pick automatic sample format depending on surround filters.
4016 Always use signed 16-bit.
4018 Always use signed 32-bit.
4020 Always use float 32-bit.
4022 Always use float 64-bit.
4026 @subsection Examples
4029 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
4031 equalizer=f=1000:t=h:width=200:g=-10
4035 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
4037 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
4041 @subsection Commands
4043 This filter supports the following commands:
4046 Change equalizer frequency.
4047 Syntax for the command is : "@var{frequency}"
4050 Change equalizer width_type.
4051 Syntax for the command is : "@var{width_type}"
4054 Change equalizer width.
4055 Syntax for the command is : "@var{width}"
4058 Change equalizer gain.
4059 Syntax for the command is : "@var{gain}"
4062 Change equalizer mix.
4063 Syntax for the command is : "@var{mix}"
4066 @section extrastereo
4068 Linearly increases the difference between left and right channels which
4069 adds some sort of "live" effect to playback.
4071 The filter accepts the following options:
4075 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
4076 (average of both channels), with 1.0 sound will be unchanged, with
4077 -1.0 left and right channels will be swapped.
4080 Enable clipping. By default is enabled.
4083 @subsection Commands
4085 This filter supports the all above options as @ref{commands}.
4087 @section firequalizer
4088 Apply FIR Equalization using arbitrary frequency response.
4090 The filter accepts the following option:
4094 Set gain curve equation (in dB). The expression can contain variables:
4097 the evaluated frequency
4101 channel number, set to 0 when multichannels evaluation is disabled
4103 channel id, see libavutil/channel_layout.h, set to the first channel id when
4104 multichannels evaluation is disabled
4108 channel_layout, see libavutil/channel_layout.h
4113 @item gain_interpolate(f)
4114 interpolate gain on frequency f based on gain_entry
4115 @item cubic_interpolate(f)
4116 same as gain_interpolate, but smoother
4118 This option is also available as command. Default is @code{gain_interpolate(f)}.
4121 Set gain entry for gain_interpolate function. The expression can
4125 store gain entry at frequency f with value g
4127 This option is also available as command.
4130 Set filter delay in seconds. Higher value means more accurate.
4131 Default is @code{0.01}.
4134 Set filter accuracy in Hz. Lower value means more accurate.
4135 Default is @code{5}.
4138 Set window function. Acceptable values are:
4141 rectangular window, useful when gain curve is already smooth
4143 hann window (default)
4149 3-terms continuous 1st derivative nuttall window
4151 minimum 3-terms discontinuous nuttall window
4153 4-terms continuous 1st derivative nuttall window
4155 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4157 blackman-harris window
4163 If enabled, use fixed number of audio samples. This improves speed when
4164 filtering with large delay. Default is disabled.
4167 Enable multichannels evaluation on gain. Default is disabled.
4170 Enable zero phase mode by subtracting timestamp to compensate delay.
4171 Default is disabled.
4174 Set scale used by gain. Acceptable values are:
4177 linear frequency, linear gain
4179 linear frequency, logarithmic (in dB) gain (default)
4181 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
4183 logarithmic frequency, logarithmic gain
4187 Set file for dumping, suitable for gnuplot.
4190 Set scale for dumpfile. Acceptable values are same with scale option.
4194 Enable 2-channel convolution using complex FFT. This improves speed significantly.
4195 Default is disabled.
4198 Enable minimum phase impulse response. Default is disabled.
4201 @subsection Examples
4206 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4209 lowpass at 1000 Hz with gain_entry:
4211 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4214 custom equalization:
4216 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4219 higher delay with zero phase to compensate delay:
4221 firequalizer=delay=0.1:fixed=on:zero_phase=on
4224 lowpass on left channel, highpass on right channel:
4226 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4227 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4232 Apply a flanging effect to the audio.
4234 The filter accepts the following options:
4238 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4241 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4244 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4248 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4249 Default value is 71.
4252 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4255 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4256 Default value is @var{sinusoidal}.
4259 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4260 Default value is 25.
4263 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4264 Default is @var{linear}.
4268 Apply Haas effect to audio.
4270 Note that this makes most sense to apply on mono signals.
4271 With this filter applied to mono signals it give some directionality and
4272 stretches its stereo image.
4274 The filter accepts the following options:
4278 Set input level. By default is @var{1}, or 0dB
4281 Set output level. By default is @var{1}, or 0dB.
4284 Set gain applied to side part of signal. By default is @var{1}.
4287 Set kind of middle source. Can be one of the following:
4297 Pick middle part signal of stereo image.
4300 Pick side part signal of stereo image.
4304 Change middle phase. By default is disabled.
4307 Set left channel delay. By default is @var{2.05} milliseconds.
4310 Set left channel balance. By default is @var{-1}.
4313 Set left channel gain. By default is @var{1}.
4316 Change left phase. By default is disabled.
4319 Set right channel delay. By defaults is @var{2.12} milliseconds.
4322 Set right channel balance. By default is @var{1}.
4325 Set right channel gain. By default is @var{1}.
4328 Change right phase. By default is enabled.
4333 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4334 embedded HDCD codes is expanded into a 20-bit PCM stream.
4336 The filter supports the Peak Extend and Low-level Gain Adjustment features
4337 of HDCD, and detects the Transient Filter flag.
4340 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4343 When using the filter with wav, note the default encoding for wav is 16-bit,
4344 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4345 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4347 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4348 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4351 The filter accepts the following options:
4354 @item disable_autoconvert
4355 Disable any automatic format conversion or resampling in the filter graph.
4357 @item process_stereo
4358 Process the stereo channels together. If target_gain does not match between
4359 channels, consider it invalid and use the last valid target_gain.
4362 Set the code detect timer period in ms.
4365 Always extend peaks above -3dBFS even if PE isn't signaled.
4368 Replace audio with a solid tone and adjust the amplitude to signal some
4369 specific aspect of the decoding process. The output file can be loaded in
4370 an audio editor alongside the original to aid analysis.
4372 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4379 Gain adjustment level at each sample
4381 Samples where peak extend occurs
4383 Samples where the code detect timer is active
4385 Samples where the target gain does not match between channels
4391 Apply head-related transfer functions (HRTFs) to create virtual
4392 loudspeakers around the user for binaural listening via headphones.
4393 The HRIRs are provided via additional streams, for each channel
4394 one stereo input stream is needed.
4396 The filter accepts the following options:
4400 Set mapping of input streams for convolution.
4401 The argument is a '|'-separated list of channel names in order as they
4402 are given as additional stream inputs for filter.
4403 This also specify number of input streams. Number of input streams
4404 must be not less than number of channels in first stream plus one.
4407 Set gain applied to audio. Value is in dB. Default is 0.
4410 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4411 processing audio in time domain which is slow.
4412 @var{freq} is processing audio in frequency domain which is fast.
4413 Default is @var{freq}.
4416 Set custom gain for LFE channels. Value is in dB. Default is 0.
4419 Set size of frame in number of samples which will be processed at once.
4420 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4423 Set format of hrir stream.
4424 Default value is @var{stereo}. Alternative value is @var{multich}.
4425 If value is set to @var{stereo}, number of additional streams should
4426 be greater or equal to number of input channels in first input stream.
4427 Also each additional stream should have stereo number of channels.
4428 If value is set to @var{multich}, number of additional streams should
4429 be exactly one. Also number of input channels of additional stream
4430 should be equal or greater than twice number of channels of first input
4434 @subsection Examples
4438 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4439 each amovie filter use stereo file with IR coefficients as input.
4440 The files give coefficients for each position of virtual loudspeaker:
4443 -filter_complex "amovie=azi_270_ele_0_DFC.wav[sr];amovie=azi_90_ele_0_DFC.wav[sl];amovie=azi_225_ele_0_DFC.wav[br];amovie=azi_135_ele_0_DFC.wav[bl];amovie=azi_0_ele_0_DFC.wav,asplit[fc][lfe];amovie=azi_35_ele_0_DFC.wav[fl];amovie=azi_325_ele_0_DFC.wav[fr];[0:a][fl][fr][fc][lfe][bl][br][sl][sr]headphone=FL|FR|FC|LFE|BL|BR|SL|SR"
4448 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4449 but now in @var{multich} @var{hrir} format.
4451 ffmpeg -i input.wav -filter_complex "amovie=minp.wav[hrirs];[0:a][hrirs]headphone=map=FL|FR|FC|LFE|BL|BR|SL|SR:hrir=multich"
4458 Apply a high-pass filter with 3dB point frequency.
4459 The filter can be either single-pole, or double-pole (the default).
4460 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4462 The filter accepts the following options:
4466 Set frequency in Hz. Default is 3000.
4469 Set number of poles. Default is 2.
4472 Set method to specify band-width of filter.
4487 Specify the band-width of a filter in width_type units.
4488 Applies only to double-pole filter.
4489 The default is 0.707q and gives a Butterworth response.
4492 How much to use filtered signal in output. Default is 1.
4493 Range is between 0 and 1.
4496 Specify which channels to filter, by default all available are filtered.
4499 Normalize biquad coefficients, by default is disabled.
4500 Enabling it will normalize magnitude response at DC to 0dB.
4503 Set transform type of IIR filter.
4512 Set precison of filtering.
4515 Pick automatic sample format depending on surround filters.
4517 Always use signed 16-bit.
4519 Always use signed 32-bit.
4521 Always use float 32-bit.
4523 Always use float 64-bit.
4527 @subsection Commands
4529 This filter supports the following commands:
4532 Change highpass frequency.
4533 Syntax for the command is : "@var{frequency}"
4536 Change highpass width_type.
4537 Syntax for the command is : "@var{width_type}"
4540 Change highpass width.
4541 Syntax for the command is : "@var{width}"
4544 Change highpass mix.
4545 Syntax for the command is : "@var{mix}"
4550 Join multiple input streams into one multi-channel stream.
4552 It accepts the following parameters:
4556 The number of input streams. It defaults to 2.
4558 @item channel_layout
4559 The desired output channel layout. It defaults to stereo.
4562 Map channels from inputs to output. The argument is a '|'-separated list of
4563 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4564 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4565 can be either the name of the input channel (e.g. FL for front left) or its
4566 index in the specified input stream. @var{out_channel} is the name of the output
4570 The filter will attempt to guess the mappings when they are not specified
4571 explicitly. It does so by first trying to find an unused matching input channel
4572 and if that fails it picks the first unused input channel.
4574 Join 3 inputs (with properly set channel layouts):
4576 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4579 Build a 5.1 output from 6 single-channel streams:
4581 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4582 'join=inputs=6:channel_layout=5.1:map=0.0-FL|1.0-FR|2.0-FC|3.0-SL|4.0-SR|5.0-LFE'
4588 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4590 To enable compilation of this filter you need to configure FFmpeg with
4591 @code{--enable-ladspa}.
4595 Specifies the name of LADSPA plugin library to load. If the environment
4596 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4597 each one of the directories specified by the colon separated list in
4598 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4599 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4600 @file{/usr/lib/ladspa/}.
4603 Specifies the plugin within the library. Some libraries contain only
4604 one plugin, but others contain many of them. If this is not set filter
4605 will list all available plugins within the specified library.
4608 Set the '|' separated list of controls which are zero or more floating point
4609 values that determine the behavior of the loaded plugin (for example delay,
4611 Controls need to be defined using the following syntax:
4612 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4613 @var{valuei} is the value set on the @var{i}-th control.
4614 Alternatively they can be also defined using the following syntax:
4615 @var{value0}|@var{value1}|@var{value2}|..., where
4616 @var{valuei} is the value set on the @var{i}-th control.
4617 If @option{controls} is set to @code{help}, all available controls and
4618 their valid ranges are printed.
4620 @item sample_rate, s
4621 Specify the sample rate, default to 44100. Only used if plugin have
4625 Set the number of samples per channel per each output frame, default
4626 is 1024. Only used if plugin have zero inputs.
4629 Set the minimum duration of the sourced audio. See
4630 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4631 for the accepted syntax.
4632 Note that the resulting duration may be greater than the specified duration,
4633 as the generated audio is always cut at the end of a complete frame.
4634 If not specified, or the expressed duration is negative, the audio is
4635 supposed to be generated forever.
4636 Only used if plugin have zero inputs.
4639 Enable latency compensation, by default is disabled.
4640 Only used if plugin have inputs.
4643 @subsection Examples
4647 List all available plugins within amp (LADSPA example plugin) library:
4653 List all available controls and their valid ranges for @code{vcf_notch}
4654 plugin from @code{VCF} library:
4656 ladspa=f=vcf:p=vcf_notch:c=help
4660 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4663 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4667 Add reverberation to the audio using TAP-plugins
4668 (Tom's Audio Processing plugins):
4670 ladspa=file=tap_reverb:tap_reverb
4674 Generate white noise, with 0.2 amplitude:
4676 ladspa=file=cmt:noise_source_white:c=c0=.2
4680 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4681 @code{C* Audio Plugin Suite} (CAPS) library:
4683 ladspa=file=caps:Click:c=c1=20'
4687 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4689 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4693 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4694 @code{SWH Plugins} collection:
4696 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4700 Attenuate low frequencies using Multiband EQ from Steve Harris
4701 @code{SWH Plugins} collection:
4703 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4707 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4710 ladspa=caps:Narrower
4714 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4716 ladspa=caps:White:.2
4720 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4722 ladspa=caps:Fractal:c=c1=1
4726 Dynamic volume normalization using @code{VLevel} plugin:
4728 ladspa=vlevel-ladspa:vlevel_mono
4732 @subsection Commands
4734 This filter supports the following commands:
4737 Modify the @var{N}-th control value.
4739 If the specified value is not valid, it is ignored and prior one is kept.
4744 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4745 Support for both single pass (livestreams, files) and double pass (files) modes.
4746 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4747 detect true peaks, the audio stream will be upsampled to 192 kHz.
4748 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4750 The filter accepts the following options:
4754 Set integrated loudness target.
4755 Range is -70.0 - -5.0. Default value is -24.0.
4758 Set loudness range target.
4759 Range is 1.0 - 20.0. Default value is 7.0.
4762 Set maximum true peak.
4763 Range is -9.0 - +0.0. Default value is -2.0.
4765 @item measured_I, measured_i
4766 Measured IL of input file.
4767 Range is -99.0 - +0.0.
4769 @item measured_LRA, measured_lra
4770 Measured LRA of input file.
4771 Range is 0.0 - 99.0.
4773 @item measured_TP, measured_tp
4774 Measured true peak of input file.
4775 Range is -99.0 - +99.0.
4777 @item measured_thresh
4778 Measured threshold of input file.
4779 Range is -99.0 - +0.0.
4782 Set offset gain. Gain is applied before the true-peak limiter.
4783 Range is -99.0 - +99.0. Default is +0.0.
4786 Normalize by linearly scaling the source audio.
4787 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4788 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4789 be lower than source LRA and the change in integrated loudness shouldn't
4790 result in a true peak which exceeds the target TP. If any of these
4791 conditions aren't met, normalization mode will revert to @var{dynamic}.
4792 Options are @code{true} or @code{false}. Default is @code{true}.
4795 Treat mono input files as "dual-mono". If a mono file is intended for playback
4796 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4797 If set to @code{true}, this option will compensate for this effect.
4798 Multi-channel input files are not affected by this option.
4799 Options are true or false. Default is false.
4802 Set print format for stats. Options are summary, json, or none.
4803 Default value is none.
4808 Apply a low-pass filter with 3dB point frequency.
4809 The filter can be either single-pole or double-pole (the default).
4810 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4812 The filter accepts the following options:
4816 Set frequency in Hz. Default is 500.
4819 Set number of poles. Default is 2.
4822 Set method to specify band-width of filter.
4837 Specify the band-width of a filter in width_type units.
4838 Applies only to double-pole filter.
4839 The default is 0.707q and gives a Butterworth response.
4842 How much to use filtered signal in output. Default is 1.
4843 Range is between 0 and 1.
4846 Specify which channels to filter, by default all available are filtered.
4849 Normalize biquad coefficients, by default is disabled.
4850 Enabling it will normalize magnitude response at DC to 0dB.
4853 Set transform type of IIR filter.
4862 Set precison of filtering.
4865 Pick automatic sample format depending on surround filters.
4867 Always use signed 16-bit.
4869 Always use signed 32-bit.
4871 Always use float 32-bit.
4873 Always use float 64-bit.
4877 @subsection Examples
4880 Lowpass only LFE channel, it LFE is not present it does nothing:
4886 @subsection Commands
4888 This filter supports the following commands:
4891 Change lowpass frequency.
4892 Syntax for the command is : "@var{frequency}"
4895 Change lowpass width_type.
4896 Syntax for the command is : "@var{width_type}"
4899 Change lowpass width.
4900 Syntax for the command is : "@var{width}"
4904 Syntax for the command is : "@var{mix}"
4909 Load a LV2 (LADSPA Version 2) plugin.
4911 To enable compilation of this filter you need to configure FFmpeg with
4912 @code{--enable-lv2}.
4916 Specifies the plugin URI. You may need to escape ':'.
4919 Set the '|' separated list of controls which are zero or more floating point
4920 values that determine the behavior of the loaded plugin (for example delay,
4922 If @option{controls} is set to @code{help}, all available controls and
4923 their valid ranges are printed.
4925 @item sample_rate, s
4926 Specify the sample rate, default to 44100. Only used if plugin have
4930 Set the number of samples per channel per each output frame, default
4931 is 1024. Only used if plugin have zero inputs.
4934 Set the minimum duration of the sourced audio. See
4935 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4936 for the accepted syntax.
4937 Note that the resulting duration may be greater than the specified duration,
4938 as the generated audio is always cut at the end of a complete frame.
4939 If not specified, or the expressed duration is negative, the audio is
4940 supposed to be generated forever.
4941 Only used if plugin have zero inputs.
4944 @subsection Examples
4948 Apply bass enhancer plugin from Calf:
4950 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4954 Apply vinyl plugin from Calf:
4956 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4960 Apply bit crusher plugin from ArtyFX:
4962 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4967 Multiband Compress or expand the audio's dynamic range.
4969 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4970 This is akin to the crossover of a loudspeaker, and results in flat frequency
4971 response when absent compander action.
4973 It accepts the following parameters:
4977 This option syntax is:
4978 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4979 For explanation of each item refer to compand filter documentation.
4985 Mix channels with specific gain levels. The filter accepts the output
4986 channel layout followed by a set of channels definitions.
4988 This filter is also designed to efficiently remap the channels of an audio
4991 The filter accepts parameters of the form:
4992 "@var{l}|@var{outdef}|@var{outdef}|..."
4996 output channel layout or number of channels
4999 output channel specification, of the form:
5000 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
5003 output channel to define, either a channel name (FL, FR, etc.) or a channel
5004 number (c0, c1, etc.)
5007 multiplicative coefficient for the channel, 1 leaving the volume unchanged
5010 input channel to use, see out_name for details; it is not possible to mix
5011 named and numbered input channels
5014 If the `=' in a channel specification is replaced by `<', then the gains for
5015 that specification will be renormalized so that the total is 1, thus
5016 avoiding clipping noise.
5018 @subsection Mixing examples
5020 For example, if you want to down-mix from stereo to mono, but with a bigger
5021 factor for the left channel:
5023 pan=1c|c0=0.9*c0+0.1*c1
5026 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
5027 7-channels surround:
5029 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
5032 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
5033 that should be preferred (see "-ac" option) unless you have very specific
5036 @subsection Remapping examples
5038 The channel remapping will be effective if, and only if:
5041 @item gain coefficients are zeroes or ones,
5042 @item only one input per channel output,
5045 If all these conditions are satisfied, the filter will notify the user ("Pure
5046 channel mapping detected"), and use an optimized and lossless method to do the
5049 For example, if you have a 5.1 source and want a stereo audio stream by
5050 dropping the extra channels:
5052 pan="stereo| c0=FL | c1=FR"
5055 Given the same source, you can also switch front left and front right channels
5056 and keep the input channel layout:
5058 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
5061 If the input is a stereo audio stream, you can mute the front left channel (and
5062 still keep the stereo channel layout) with:
5067 Still with a stereo audio stream input, you can copy the right channel in both
5068 front left and right:
5070 pan="stereo| c0=FR | c1=FR"
5075 ReplayGain scanner filter. This filter takes an audio stream as an input and
5076 outputs it unchanged.
5077 At end of filtering it displays @code{track_gain} and @code{track_peak}.
5081 Convert the audio sample format, sample rate and channel layout. It is
5082 not meant to be used directly.
5085 Apply time-stretching and pitch-shifting with librubberband.
5087 To enable compilation of this filter, you need to configure FFmpeg with
5088 @code{--enable-librubberband}.
5090 The filter accepts the following options:
5094 Set tempo scale factor.
5097 Set pitch scale factor.
5100 Set transients detector.
5101 Possible values are:
5110 Possible values are:
5119 Possible values are:
5126 Set processing window size.
5127 Possible values are:
5136 Possible values are:
5143 Enable formant preservation when shift pitching.
5144 Possible values are:
5152 Possible values are:
5161 Possible values are:
5168 @subsection Commands
5170 This filter supports the following commands:
5173 Change filter tempo scale factor.
5174 Syntax for the command is : "@var{tempo}"
5177 Change filter pitch scale factor.
5178 Syntax for the command is : "@var{pitch}"
5181 @section sidechaincompress
5183 This filter acts like normal compressor but has the ability to compress
5184 detected signal using second input signal.
5185 It needs two input streams and returns one output stream.
5186 First input stream will be processed depending on second stream signal.
5187 The filtered signal then can be filtered with other filters in later stages of
5188 processing. See @ref{pan} and @ref{amerge} filter.
5190 The filter accepts the following options:
5194 Set input gain. Default is 1. Range is between 0.015625 and 64.
5197 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
5198 Default is @code{downward}.
5201 If a signal of second stream raises above this level it will affect the gain
5202 reduction of first stream.
5203 By default is 0.125. Range is between 0.00097563 and 1.
5206 Set a ratio about which the signal is reduced. 1:2 means that if the level
5207 raised 4dB above the threshold, it will be only 2dB above after the reduction.
5208 Default is 2. Range is between 1 and 20.
5211 Amount of milliseconds the signal has to rise above the threshold before gain
5212 reduction starts. Default is 20. Range is between 0.01 and 2000.
5215 Amount of milliseconds the signal has to fall below the threshold before
5216 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
5219 Set the amount by how much signal will be amplified after processing.
5220 Default is 1. Range is from 1 to 64.
5223 Curve the sharp knee around the threshold to enter gain reduction more softly.
5224 Default is 2.82843. Range is between 1 and 8.
5227 Choose if the @code{average} level between all channels of side-chain stream
5228 or the louder(@code{maximum}) channel of side-chain stream affects the
5229 reduction. Default is @code{average}.
5232 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5233 of @code{rms}. Default is @code{rms} which is mainly smoother.
5236 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5239 How much to use compressed signal in output. Default is 1.
5240 Range is between 0 and 1.
5243 @subsection Commands
5245 This filter supports the all above options as @ref{commands}.
5247 @subsection Examples
5251 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5252 depending on the signal of 2nd input and later compressed signal to be
5253 merged with 2nd input:
5255 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5259 @section sidechaingate
5261 A sidechain gate acts like a normal (wideband) gate but has the ability to
5262 filter the detected signal before sending it to the gain reduction stage.
5263 Normally a gate uses the full range signal to detect a level above the
5265 For example: If you cut all lower frequencies from your sidechain signal
5266 the gate will decrease the volume of your track only if not enough highs
5267 appear. With this technique you are able to reduce the resonation of a
5268 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5270 It needs two input streams and returns one output stream.
5271 First input stream will be processed depending on second stream signal.
5273 The filter accepts the following options:
5277 Set input level before filtering.
5278 Default is 1. Allowed range is from 0.015625 to 64.
5281 Set the mode of operation. Can be @code{upward} or @code{downward}.
5282 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5283 will be amplified, expanding dynamic range in upward direction.
5284 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5287 Set the level of gain reduction when the signal is below the threshold.
5288 Default is 0.06125. Allowed range is from 0 to 1.
5289 Setting this to 0 disables reduction and then filter behaves like expander.
5292 If a signal rises above this level the gain reduction is released.
5293 Default is 0.125. Allowed range is from 0 to 1.
5296 Set a ratio about which the signal is reduced.
5297 Default is 2. Allowed range is from 1 to 9000.
5300 Amount of milliseconds the signal has to rise above the threshold before gain
5302 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5305 Amount of milliseconds the signal has to fall below the threshold before the
5306 reduction is increased again. Default is 250 milliseconds.
5307 Allowed range is from 0.01 to 9000.
5310 Set amount of amplification of signal after processing.
5311 Default is 1. Allowed range is from 1 to 64.
5314 Curve the sharp knee around the threshold to enter gain reduction more softly.
5315 Default is 2.828427125. Allowed range is from 1 to 8.
5318 Choose if exact signal should be taken for detection or an RMS like one.
5319 Default is rms. Can be peak or rms.
5322 Choose if the average level between all channels or the louder channel affects
5324 Default is average. Can be average or maximum.
5327 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5330 @subsection Commands
5332 This filter supports the all above options as @ref{commands}.
5334 @section silencedetect
5336 Detect silence in an audio stream.
5338 This filter logs a message when it detects that the input audio volume is less
5339 or equal to a noise tolerance value for a duration greater or equal to the
5340 minimum detected noise duration.
5342 The printed times and duration are expressed in seconds. The
5343 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5344 is set on the first frame whose timestamp equals or exceeds the detection
5345 duration and it contains the timestamp of the first frame of the silence.
5347 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5348 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5349 keys are set on the first frame after the silence. If @option{mono} is
5350 enabled, and each channel is evaluated separately, the @code{.X}
5351 suffixed keys are used, and @code{X} corresponds to the channel number.
5353 The filter accepts the following options:
5357 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5358 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5361 Set silence duration until notification (default is 2 seconds). See
5362 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5363 for the accepted syntax.
5366 Process each channel separately, instead of combined. By default is disabled.
5369 @subsection Examples
5373 Detect 5 seconds of silence with -50dB noise tolerance:
5375 silencedetect=n=-50dB:d=5
5379 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5380 tolerance in @file{silence.mp3}:
5382 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5386 @section silenceremove
5388 Remove silence from the beginning, middle or end of the audio.
5390 The filter accepts the following options:
5394 This value is used to indicate if audio should be trimmed at beginning of
5395 the audio. A value of zero indicates no silence should be trimmed from the
5396 beginning. When specifying a non-zero value, it trims audio up until it
5397 finds non-silence. Normally, when trimming silence from beginning of audio
5398 the @var{start_periods} will be @code{1} but it can be increased to higher
5399 values to trim all audio up to specific count of non-silence periods.
5400 Default value is @code{0}.
5402 @item start_duration
5403 Specify the amount of time that non-silence must be detected before it stops
5404 trimming audio. By increasing the duration, bursts of noises can be treated
5405 as silence and trimmed off. Default value is @code{0}.
5407 @item start_threshold
5408 This indicates what sample value should be treated as silence. For digital
5409 audio, a value of @code{0} may be fine but for audio recorded from analog,
5410 you may wish to increase the value to account for background noise.
5411 Can be specified in dB (in case "dB" is appended to the specified value)
5412 or amplitude ratio. Default value is @code{0}.
5415 Specify max duration of silence at beginning that will be kept after
5416 trimming. Default is 0, which is equal to trimming all samples detected
5420 Specify mode of detection of silence end in start of multi-channel audio.
5421 Can be @var{any} or @var{all}. Default is @var{any}.
5422 With @var{any}, any sample that is detected as non-silence will cause
5423 stopped trimming of silence.
5424 With @var{all}, only if all channels are detected as non-silence will cause
5425 stopped trimming of silence.
5428 Set the count for trimming silence from the end of audio.
5429 To remove silence from the middle of a file, specify a @var{stop_periods}
5430 that is negative. This value is then treated as a positive value and is
5431 used to indicate the effect should restart processing as specified by
5432 @var{start_periods}, making it suitable for removing periods of silence
5433 in the middle of the audio.
5434 Default value is @code{0}.
5437 Specify a duration of silence that must exist before audio is not copied any
5438 more. By specifying a higher duration, silence that is wanted can be left in
5440 Default value is @code{0}.
5442 @item stop_threshold
5443 This is the same as @option{start_threshold} but for trimming silence from
5445 Can be specified in dB (in case "dB" is appended to the specified value)
5446 or amplitude ratio. Default value is @code{0}.
5449 Specify max duration of silence at end that will be kept after
5450 trimming. Default is 0, which is equal to trimming all samples detected
5454 Specify mode of detection of silence start in end of multi-channel audio.
5455 Can be @var{any} or @var{all}. Default is @var{any}.
5456 With @var{any}, any sample that is detected as non-silence will cause
5457 stopped trimming of silence.
5458 With @var{all}, only if all channels are detected as non-silence will cause
5459 stopped trimming of silence.
5462 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5463 and works better with digital silence which is exactly 0.
5464 Default value is @code{rms}.
5467 Set duration in number of seconds used to calculate size of window in number
5468 of samples for detecting silence.
5469 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5472 @subsection Examples
5476 The following example shows how this filter can be used to start a recording
5477 that does not contain the delay at the start which usually occurs between
5478 pressing the record button and the start of the performance:
5480 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5484 Trim all silence encountered from beginning to end where there is more than 1
5485 second of silence in audio:
5487 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5491 Trim all digital silence samples, using peak detection, from beginning to end
5492 where there is more than 0 samples of digital silence in audio and digital
5493 silence is detected in all channels at same positions in stream:
5495 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5501 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5502 loudspeakers around the user for binaural listening via headphones (audio
5503 formats up to 9 channels supported).
5504 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5505 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5506 Austrian Academy of Sciences.
5508 To enable compilation of this filter you need to configure FFmpeg with
5509 @code{--enable-libmysofa}.
5511 The filter accepts the following options:
5515 Set the SOFA file used for rendering.
5518 Set gain applied to audio. Value is in dB. Default is 0.
5521 Set rotation of virtual loudspeakers in deg. Default is 0.
5524 Set elevation of virtual speakers in deg. Default is 0.
5527 Set distance in meters between loudspeakers and the listener with near-field
5528 HRTFs. Default is 1.
5531 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5532 processing audio in time domain which is slow.
5533 @var{freq} is processing audio in frequency domain which is fast.
5534 Default is @var{freq}.
5537 Set custom positions of virtual loudspeakers. Syntax for this option is:
5538 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5539 Each virtual loudspeaker is described with short channel name following with
5540 azimuth and elevation in degrees.
5541 Each virtual loudspeaker description is separated by '|'.
5542 For example to override front left and front right channel positions use:
5543 'speakers=FL 45 15|FR 345 15'.
5544 Descriptions with unrecognised channel names are ignored.
5547 Set custom gain for LFE channels. Value is in dB. Default is 0.
5550 Set custom frame size in number of samples. Default is 1024.
5551 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5552 is set to @var{freq}.
5555 Should all IRs be normalized upon importing SOFA file.
5556 By default is enabled.
5559 Should nearest IRs be interpolated with neighbor IRs if exact position
5560 does not match. By default is disabled.
5563 Minphase all IRs upon loading of SOFA file. By default is disabled.
5566 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5569 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5572 @subsection Examples
5576 Using ClubFritz6 sofa file:
5578 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5582 Using ClubFritz12 sofa file and bigger radius with small rotation:
5584 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5588 Similar as above but with custom speaker positions for front left, front right, back left and back right
5589 and also with custom gain:
5591 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5598 This filter expands or compresses each half-cycle of audio samples
5599 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5600 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5602 The filter accepts the following options:
5606 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5607 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5610 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5611 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5612 would be such that local peak value reaches target peak value but never to surpass it and that
5613 ratio between new and previous peak value does not surpass this option value.
5615 @item compression, c
5616 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5617 This option controls maximum local half-cycle of samples compression. This option is used
5618 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5619 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5620 that peak's half-cycle will be compressed by current compression factor.
5623 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5624 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5625 Any half-cycle samples with their local peak value below or same as this option value will be
5626 compressed by current compression factor, otherwise, if greater than threshold value they will be
5627 expanded with expansion factor so that it could reach peak target value but never surpass it.
5630 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5631 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5632 each new half-cycle until it reaches @option{expansion} value.
5633 Setting this options too high may lead to distortions.
5636 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5637 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5638 each new half-cycle until it reaches @option{compression} value.
5641 Specify which channels to filter, by default all available channels are filtered.
5644 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5645 option. When enabled any half-cycle of samples with their local peak value below or same as
5646 @option{threshold} option will be expanded otherwise it will be compressed.
5649 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5650 When disabled each filtered channel gain calculation is independent, otherwise when this option
5651 is enabled the minimum of all possible gains for each filtered channel is used.
5654 @subsection Commands
5656 This filter supports the all above options as @ref{commands}.
5658 @section stereotools
5660 This filter has some handy utilities to manage stereo signals, for converting
5661 M/S stereo recordings to L/R signal while having control over the parameters
5662 or spreading the stereo image of master track.
5664 The filter accepts the following options:
5668 Set input level before filtering for both channels. Defaults is 1.
5669 Allowed range is from 0.015625 to 64.
5672 Set output level after filtering for both channels. Defaults is 1.
5673 Allowed range is from 0.015625 to 64.
5676 Set input balance between both channels. Default is 0.
5677 Allowed range is from -1 to 1.
5680 Set output balance between both channels. Default is 0.
5681 Allowed range is from -1 to 1.
5684 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5685 clipping. Disabled by default.
5688 Mute the left channel. Disabled by default.
5691 Mute the right channel. Disabled by default.
5694 Change the phase of the left channel. Disabled by default.
5697 Change the phase of the right channel. Disabled by default.
5700 Set stereo mode. Available values are:
5704 Left/Right to Left/Right, this is default.
5707 Left/Right to Mid/Side.
5710 Mid/Side to Left/Right.
5713 Left/Right to Left/Left.
5716 Left/Right to Right/Right.
5719 Left/Right to Left + Right.
5722 Left/Right to Right/Left.
5725 Mid/Side to Left/Left.
5728 Mid/Side to Right/Right.
5731 Mid/Side to Right/Left.
5734 Left/Right to Left - Right.
5738 Set level of side signal. Default is 1.
5739 Allowed range is from 0.015625 to 64.
5742 Set balance of side signal. Default is 0.
5743 Allowed range is from -1 to 1.
5746 Set level of the middle signal. Default is 1.
5747 Allowed range is from 0.015625 to 64.
5750 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5753 Set stereo base between mono and inversed channels. Default is 0.
5754 Allowed range is from -1 to 1.
5757 Set delay in milliseconds how much to delay left from right channel and
5758 vice versa. Default is 0. Allowed range is from -20 to 20.
5761 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5764 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5766 @item bmode_in, bmode_out
5767 Set balance mode for balance_in/balance_out option.
5769 Can be one of the following:
5773 Classic balance mode. Attenuate one channel at time.
5774 Gain is raised up to 1.
5777 Similar as classic mode above but gain is raised up to 2.
5780 Equal power distribution, from -6dB to +6dB range.
5784 @subsection Commands
5786 This filter supports the all above options as @ref{commands}.
5788 @subsection Examples
5792 Apply karaoke like effect:
5794 stereotools=mlev=0.015625
5798 Convert M/S signal to L/R:
5800 "stereotools=mode=ms>lr"
5804 @section stereowiden
5806 This filter enhance the stereo effect by suppressing signal common to both
5807 channels and by delaying the signal of left into right and vice versa,
5808 thereby widening the stereo effect.
5810 The filter accepts the following options:
5814 Time in milliseconds of the delay of left signal into right and vice versa.
5815 Default is 20 milliseconds.
5818 Amount of gain in delayed signal into right and vice versa. Gives a delay
5819 effect of left signal in right output and vice versa which gives widening
5820 effect. Default is 0.3.
5823 Cross feed of left into right with inverted phase. This helps in suppressing
5824 the mono. If the value is 1 it will cancel all the signal common to both
5825 channels. Default is 0.3.
5828 Set level of input signal of original channel. Default is 0.8.
5831 @subsection Commands
5833 This filter supports the all above options except @code{delay} as @ref{commands}.
5835 @section superequalizer
5836 Apply 18 band equalizer.
5838 The filter accepts the following options:
5845 Set 131Hz band gain.
5847 Set 185Hz band gain.
5849 Set 262Hz band gain.
5851 Set 370Hz band gain.
5853 Set 523Hz band gain.
5855 Set 740Hz band gain.
5857 Set 1047Hz band gain.
5859 Set 1480Hz band gain.
5861 Set 2093Hz band gain.
5863 Set 2960Hz band gain.
5865 Set 4186Hz band gain.
5867 Set 5920Hz band gain.
5869 Set 8372Hz band gain.
5871 Set 11840Hz band gain.
5873 Set 16744Hz band gain.
5875 Set 20000Hz band gain.
5879 Apply audio surround upmix filter.
5881 This filter allows to produce multichannel output from audio stream.
5883 The filter accepts the following options:
5887 Set output channel layout. By default, this is @var{5.1}.
5889 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5890 for the required syntax.
5893 Set input channel layout. By default, this is @var{stereo}.
5895 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5896 for the required syntax.
5899 Set input volume level. By default, this is @var{1}.
5902 Set output volume level. By default, this is @var{1}.
5905 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5908 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5911 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5914 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5915 In @var{add} mode, LFE channel is created from input audio and added to output.
5916 In @var{sub} mode, LFE channel is created from input audio and added to output but
5917 also all non-LFE output channels are subtracted with output LFE channel.
5920 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5921 Default is @var{90}.
5924 Set front center input volume. By default, this is @var{1}.
5927 Set front center output volume. By default, this is @var{1}.
5930 Set front left input volume. By default, this is @var{1}.
5933 Set front left output volume. By default, this is @var{1}.
5936 Set front right input volume. By default, this is @var{1}.
5939 Set front right output volume. By default, this is @var{1}.
5942 Set side left input volume. By default, this is @var{1}.
5945 Set side left output volume. By default, this is @var{1}.
5948 Set side right input volume. By default, this is @var{1}.
5951 Set side right output volume. By default, this is @var{1}.
5954 Set back left input volume. By default, this is @var{1}.
5957 Set back left output volume. By default, this is @var{1}.
5960 Set back right input volume. By default, this is @var{1}.
5963 Set back right output volume. By default, this is @var{1}.
5966 Set back center input volume. By default, this is @var{1}.
5969 Set back center output volume. By default, this is @var{1}.
5972 Set LFE input volume. By default, this is @var{1}.
5975 Set LFE output volume. By default, this is @var{1}.
5978 Set spread usage of stereo image across X axis for all channels.
5981 Set spread usage of stereo image across Y axis for all channels.
5983 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5984 Set spread usage of stereo image across X axis for each channel.
5986 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5987 Set spread usage of stereo image across Y axis for each channel.
5990 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5993 Set window function.
5995 It accepts the following values:
6018 Default is @code{hann}.
6021 Set window overlap. If set to 1, the recommended overlap for selected
6022 window function will be picked. Default is @code{0.5}.
6025 @section treble, highshelf
6027 Boost or cut treble (upper) frequencies of the audio using a two-pole
6028 shelving filter with a response similar to that of a standard
6029 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
6031 The filter accepts the following options:
6035 Give the gain at whichever is the lower of ~22 kHz and the
6036 Nyquist frequency. Its useful range is about -20 (for a large cut)
6037 to +20 (for a large boost). Beware of clipping when using a positive gain.
6040 Set the filter's central frequency and so can be used
6041 to extend or reduce the frequency range to be boosted or cut.
6042 The default value is @code{3000} Hz.
6045 Set method to specify band-width of filter.
6060 Determine how steep is the filter's shelf transition.
6063 Set number of poles. Default is 2.
6066 How much to use filtered signal in output. Default is 1.
6067 Range is between 0 and 1.
6070 Specify which channels to filter, by default all available are filtered.
6073 Normalize biquad coefficients, by default is disabled.
6074 Enabling it will normalize magnitude response at DC to 0dB.
6077 Set transform type of IIR filter.
6086 Set precison of filtering.
6089 Pick automatic sample format depending on surround filters.
6091 Always use signed 16-bit.
6093 Always use signed 32-bit.
6095 Always use float 32-bit.
6097 Always use float 64-bit.
6101 @subsection Commands
6103 This filter supports the following commands:
6106 Change treble frequency.
6107 Syntax for the command is : "@var{frequency}"
6110 Change treble width_type.
6111 Syntax for the command is : "@var{width_type}"
6114 Change treble width.
6115 Syntax for the command is : "@var{width}"
6119 Syntax for the command is : "@var{gain}"
6123 Syntax for the command is : "@var{mix}"
6128 Sinusoidal amplitude modulation.
6130 The filter accepts the following options:
6134 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
6135 (20 Hz or lower) will result in a tremolo effect.
6136 This filter may also be used as a ring modulator by specifying
6137 a modulation frequency higher than 20 Hz.
6138 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6141 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6142 Default value is 0.5.
6147 Sinusoidal phase modulation.
6149 The filter accepts the following options:
6153 Modulation frequency in Hertz.
6154 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6157 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6158 Default value is 0.5.
6163 Adjust the input audio volume.
6165 It accepts the following parameters:
6169 Set audio volume expression.
6171 Output values are clipped to the maximum value.
6173 The output audio volume is given by the relation:
6175 @var{output_volume} = @var{volume} * @var{input_volume}
6178 The default value for @var{volume} is "1.0".
6181 This parameter represents the mathematical precision.
6183 It determines which input sample formats will be allowed, which affects the
6184 precision of the volume scaling.
6188 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
6190 32-bit floating-point; this limits input sample format to FLT. (default)
6192 64-bit floating-point; this limits input sample format to DBL.
6196 Choose the behaviour on encountering ReplayGain side data in input frames.
6200 Remove ReplayGain side data, ignoring its contents (the default).
6203 Ignore ReplayGain side data, but leave it in the frame.
6206 Prefer the track gain, if present.
6209 Prefer the album gain, if present.
6212 @item replaygain_preamp
6213 Pre-amplification gain in dB to apply to the selected replaygain gain.
6215 Default value for @var{replaygain_preamp} is 0.0.
6217 @item replaygain_noclip
6218 Prevent clipping by limiting the gain applied.
6220 Default value for @var{replaygain_noclip} is 1.
6223 Set when the volume expression is evaluated.
6225 It accepts the following values:
6228 only evaluate expression once during the filter initialization, or
6229 when the @samp{volume} command is sent
6232 evaluate expression for each incoming frame
6235 Default value is @samp{once}.
6238 The volume expression can contain the following parameters.
6242 frame number (starting at zero)
6245 @item nb_consumed_samples
6246 number of samples consumed by the filter
6248 number of samples in the current frame
6250 original frame position in the file
6256 PTS at start of stream
6258 time at start of stream
6264 last set volume value
6267 Note that when @option{eval} is set to @samp{once} only the
6268 @var{sample_rate} and @var{tb} variables are available, all other
6269 variables will evaluate to NAN.
6271 @subsection Commands
6273 This filter supports the following commands:
6276 Modify the volume expression.
6277 The command accepts the same syntax of the corresponding option.
6279 If the specified expression is not valid, it is kept at its current
6283 @subsection Examples
6287 Halve the input audio volume:
6291 volume=volume=-6.0206dB
6294 In all the above example the named key for @option{volume} can be
6295 omitted, for example like in:
6301 Increase input audio power by 6 decibels using fixed-point precision:
6303 volume=volume=6dB:precision=fixed
6307 Fade volume after time 10 with an annihilation period of 5 seconds:
6309 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6313 @section volumedetect
6315 Detect the volume of the input video.
6317 The filter has no parameters. The input is not modified. Statistics about
6318 the volume will be printed in the log when the input stream end is reached.
6320 In particular it will show the mean volume (root mean square), maximum
6321 volume (on a per-sample basis), and the beginning of a histogram of the
6322 registered volume values (from the maximum value to a cumulated 1/1000 of
6325 All volumes are in decibels relative to the maximum PCM value.
6327 @subsection Examples
6329 Here is an excerpt of the output:
6331 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6332 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6333 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6334 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6335 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6336 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6337 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6338 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6339 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6345 The mean square energy is approximately -27 dB, or 10^-2.7.
6347 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6349 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6352 In other words, raising the volume by +4 dB does not cause any clipping,
6353 raising it by +5 dB causes clipping for 6 samples, etc.
6355 @c man end AUDIO FILTERS
6357 @chapter Audio Sources
6358 @c man begin AUDIO SOURCES
6360 Below is a description of the currently available audio sources.
6364 Buffer audio frames, and make them available to the filter chain.
6366 This source is mainly intended for a programmatic use, in particular
6367 through the interface defined in @file{libavfilter/buffersrc.h}.
6369 It accepts the following parameters:
6373 The timebase which will be used for timestamps of submitted frames. It must be
6374 either a floating-point number or in @var{numerator}/@var{denominator} form.
6377 The sample rate of the incoming audio buffers.
6380 The sample format of the incoming audio buffers.
6381 Either a sample format name or its corresponding integer representation from
6382 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6384 @item channel_layout
6385 The channel layout of the incoming audio buffers.
6386 Either a channel layout name from channel_layout_map in
6387 @file{libavutil/channel_layout.c} or its corresponding integer representation
6388 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6391 The number of channels of the incoming audio buffers.
6392 If both @var{channels} and @var{channel_layout} are specified, then they
6397 @subsection Examples
6400 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6403 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6404 Since the sample format with name "s16p" corresponds to the number
6405 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6408 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6413 Generate an audio signal specified by an expression.
6415 This source accepts in input one or more expressions (one for each
6416 channel), which are evaluated and used to generate a corresponding
6419 This source accepts the following options:
6423 Set the '|'-separated expressions list for each separate channel. In case the
6424 @option{channel_layout} option is not specified, the selected channel layout
6425 depends on the number of provided expressions. Otherwise the last
6426 specified expression is applied to the remaining output channels.
6428 @item channel_layout, c
6429 Set the channel layout. The number of channels in the specified layout
6430 must be equal to the number of specified expressions.
6433 Set the minimum duration of the sourced audio. See
6434 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6435 for the accepted syntax.
6436 Note that the resulting duration may be greater than the specified
6437 duration, as the generated audio is always cut at the end of a
6440 If not specified, or the expressed duration is negative, the audio is
6441 supposed to be generated forever.
6444 Set the number of samples per channel per each output frame,
6447 @item sample_rate, s
6448 Specify the sample rate, default to 44100.
6451 Each expression in @var{exprs} can contain the following constants:
6455 number of the evaluated sample, starting from 0
6458 time of the evaluated sample expressed in seconds, starting from 0
6465 @subsection Examples
6475 Generate a sin signal with frequency of 440 Hz, set sample rate to
6478 aevalsrc="sin(440*2*PI*t):s=8000"
6482 Generate a two channels signal, specify the channel layout (Front
6483 Center + Back Center) explicitly:
6485 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6489 Generate white noise:
6491 aevalsrc="-2+random(0)"
6495 Generate an amplitude modulated signal:
6497 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6501 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6503 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6510 Generate a FIR coefficients using frequency sampling method.
6512 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6514 The filter accepts the following options:
6518 Set number of filter coefficents in output audio stream.
6519 Default value is 1025.
6522 Set frequency points from where magnitude and phase are set.
6523 This must be in non decreasing order, and first element must be 0, while last element
6524 must be 1. Elements are separated by white spaces.
6527 Set magnitude value for every frequency point set by @option{frequency}.
6528 Number of values must be same as number of frequency points.
6529 Values are separated by white spaces.
6532 Set phase value for every frequency point set by @option{frequency}.
6533 Number of values must be same as number of frequency points.
6534 Values are separated by white spaces.
6536 @item sample_rate, r
6537 Set sample rate, default is 44100.
6540 Set number of samples per each frame. Default is 1024.
6543 Set window function. Default is blackman.
6548 The null audio source, return unprocessed audio frames. It is mainly useful
6549 as a template and to be employed in analysis / debugging tools, or as
6550 the source for filters which ignore the input data (for example the sox
6553 This source accepts the following options:
6557 @item channel_layout, cl
6559 Specifies the channel layout, and can be either an integer or a string
6560 representing a channel layout. The default value of @var{channel_layout}
6563 Check the channel_layout_map definition in
6564 @file{libavutil/channel_layout.c} for the mapping between strings and
6565 channel layout values.
6567 @item sample_rate, r
6568 Specifies the sample rate, and defaults to 44100.
6571 Set the number of samples per requested frames.
6574 Set the duration of the sourced audio. See
6575 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6576 for the accepted syntax.
6578 If not specified, or the expressed duration is negative, the audio is
6579 supposed to be generated forever.
6582 @subsection Examples
6586 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6588 anullsrc=r=48000:cl=4
6592 Do the same operation with a more obvious syntax:
6594 anullsrc=r=48000:cl=mono
6598 All the parameters need to be explicitly defined.
6602 Synthesize a voice utterance using the libflite library.
6604 To enable compilation of this filter you need to configure FFmpeg with
6605 @code{--enable-libflite}.
6607 Note that versions of the flite library prior to 2.0 are not thread-safe.
6609 The filter accepts the following options:
6614 If set to 1, list the names of the available voices and exit
6615 immediately. Default value is 0.
6618 Set the maximum number of samples per frame. Default value is 512.
6621 Set the filename containing the text to speak.
6624 Set the text to speak.
6627 Set the voice to use for the speech synthesis. Default value is
6628 @code{kal}. See also the @var{list_voices} option.
6631 @subsection Examples
6635 Read from file @file{speech.txt}, and synthesize the text using the
6636 standard flite voice:
6638 flite=textfile=speech.txt
6642 Read the specified text selecting the @code{slt} voice:
6644 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6648 Input text to ffmpeg:
6650 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6654 Make @file{ffplay} speak the specified text, using @code{flite} and
6655 the @code{lavfi} device:
6657 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6661 For more information about libflite, check:
6662 @url{http://www.festvox.org/flite/}
6666 Generate a noise audio signal.
6668 The filter accepts the following options:
6671 @item sample_rate, r
6672 Specify the sample rate. Default value is 48000 Hz.
6675 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6679 Specify the duration of the generated audio stream. Not specifying this option
6680 results in noise with an infinite length.
6682 @item color, colour, c
6683 Specify the color of noise. Available noise colors are white, pink, brown,
6684 blue, violet and velvet. Default color is white.
6687 Specify a value used to seed the PRNG.
6690 Set the number of samples per each output frame, default is 1024.
6693 @subsection Examples
6698 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6700 anoisesrc=d=60:c=pink:r=44100:a=0.5
6706 Generate odd-tap Hilbert transform FIR coefficients.
6708 The resulting stream can be used with @ref{afir} filter for phase-shifting
6709 the signal by 90 degrees.
6711 This is used in many matrix coding schemes and for analytic signal generation.
6712 The process is often written as a multiplication by i (or j), the imaginary unit.
6714 The filter accepts the following options:
6718 @item sample_rate, s
6719 Set sample rate, default is 44100.
6722 Set length of FIR filter, default is 22051.
6725 Set number of samples per each frame.
6728 Set window function to be used when generating FIR coefficients.
6733 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6735 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6737 The filter accepts the following options:
6740 @item sample_rate, r
6741 Set sample rate, default is 44100.
6744 Set number of samples per each frame. Default is 1024.
6747 Set high-pass frequency. Default is 0.
6750 Set low-pass frequency. Default is 0.
6751 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6752 is higher than 0 then filter will create band-pass filter coefficients,
6753 otherwise band-reject filter coefficients.
6756 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6759 Set Kaiser window beta.
6762 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6765 Enable rounding, by default is disabled.
6768 Set number of taps for high-pass filter.
6771 Set number of taps for low-pass filter.
6776 Generate an audio signal made of a sine wave with amplitude 1/8.
6778 The audio signal is bit-exact.
6780 The filter accepts the following options:
6785 Set the carrier frequency. Default is 440 Hz.
6787 @item beep_factor, b
6788 Enable a periodic beep every second with frequency @var{beep_factor} times
6789 the carrier frequency. Default is 0, meaning the beep is disabled.
6791 @item sample_rate, r
6792 Specify the sample rate, default is 44100.
6795 Specify the duration of the generated audio stream.
6797 @item samples_per_frame
6798 Set the number of samples per output frame.
6800 The expression can contain the following constants:
6804 The (sequential) number of the output audio frame, starting from 0.
6807 The PTS (Presentation TimeStamp) of the output audio frame,
6808 expressed in @var{TB} units.
6811 The PTS of the output audio frame, expressed in seconds.
6814 The timebase of the output audio frames.
6817 Default is @code{1024}.
6820 @subsection Examples
6825 Generate a simple 440 Hz sine wave:
6831 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6835 sine=frequency=220:beep_factor=4:duration=5
6839 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6842 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6846 @c man end AUDIO SOURCES
6848 @chapter Audio Sinks
6849 @c man begin AUDIO SINKS
6851 Below is a description of the currently available audio sinks.
6853 @section abuffersink
6855 Buffer audio frames, and make them available to the end of filter chain.
6857 This sink is mainly intended for programmatic use, in particular
6858 through the interface defined in @file{libavfilter/buffersink.h}
6859 or the options system.
6861 It accepts a pointer to an AVABufferSinkContext structure, which
6862 defines the incoming buffers' formats, to be passed as the opaque
6863 parameter to @code{avfilter_init_filter} for initialization.
6866 Null audio sink; do absolutely nothing with the input audio. It is
6867 mainly useful as a template and for use in analysis / debugging
6870 @c man end AUDIO SINKS
6872 @chapter Video Filters
6873 @c man begin VIDEO FILTERS
6875 When you configure your FFmpeg build, you can disable any of the
6876 existing filters using @code{--disable-filters}.
6877 The configure output will show the video filters included in your
6880 Below is a description of the currently available video filters.
6884 Mark a region of interest in a video frame.
6886 The frame data is passed through unchanged, but metadata is attached
6887 to the frame indicating regions of interest which can affect the
6888 behaviour of later encoding. Multiple regions can be marked by
6889 applying the filter multiple times.
6893 Region distance in pixels from the left edge of the frame.
6895 Region distance in pixels from the top edge of the frame.
6897 Region width in pixels.
6899 Region height in pixels.
6901 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6902 and may contain the following variables:
6905 Width of the input frame.
6907 Height of the input frame.
6911 Quantisation offset to apply within the region.
6913 This must be a real value in the range -1 to +1. A value of zero
6914 indicates no quality change. A negative value asks for better quality
6915 (less quantisation), while a positive value asks for worse quality
6916 (greater quantisation).
6918 The range is calibrated so that the extreme values indicate the
6919 largest possible offset - if the rest of the frame is encoded with the
6920 worst possible quality, an offset of -1 indicates that this region
6921 should be encoded with the best possible quality anyway. Intermediate
6922 values are then interpolated in some codec-dependent way.
6924 For example, in 10-bit H.264 the quantisation parameter varies between
6925 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6926 this region should be encoded with a QP around one-tenth of the full
6927 range better than the rest of the frame. So, if most of the frame
6928 were to be encoded with a QP of around 30, this region would get a QP
6929 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6930 An extreme value of -1 would indicate that this region should be
6931 encoded with the best possible quality regardless of the treatment of
6932 the rest of the frame - that is, should be encoded at a QP of -12.
6934 If set to true, remove any existing regions of interest marked on the
6935 frame before adding the new one.
6938 @subsection Examples
6942 Mark the centre quarter of the frame as interesting.
6944 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6947 Mark the 100-pixel-wide region on the left edge of the frame as very
6948 uninteresting (to be encoded at much lower quality than the rest of
6951 addroi=0:0:100:ih:+1/5
6955 @section alphaextract
6957 Extract the alpha component from the input as a grayscale video. This
6958 is especially useful with the @var{alphamerge} filter.
6962 Add or replace the alpha component of the primary input with the
6963 grayscale value of a second input. This is intended for use with
6964 @var{alphaextract} to allow the transmission or storage of frame
6965 sequences that have alpha in a format that doesn't support an alpha
6968 For example, to reconstruct full frames from a normal YUV-encoded video
6969 and a separate video created with @var{alphaextract}, you might use:
6971 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6976 Amplify differences between current pixel and pixels of adjacent frames in
6977 same pixel location.
6979 This filter accepts the following options:
6983 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6984 For example radius of 3 will instruct filter to calculate average of 7 frames.
6987 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6990 Set threshold for difference amplification. Any difference greater or equal to
6991 this value will not alter source pixel. Default is 10.
6992 Allowed range is from 0 to 65535.
6995 Set tolerance for difference amplification. Any difference lower to
6996 this value will not alter source pixel. Default is 0.
6997 Allowed range is from 0 to 65535.
7000 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7001 This option controls maximum possible value that will decrease source pixel value.
7004 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7005 This option controls maximum possible value that will increase source pixel value.
7008 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
7011 @subsection Commands
7013 This filter supports the following @ref{commands} that corresponds to option of same name:
7025 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
7026 and libavformat to work. On the other hand, it is limited to ASS (Advanced
7027 Substation Alpha) subtitles files.
7029 This filter accepts the following option in addition to the common options from
7030 the @ref{subtitles} filter:
7034 Set the shaping engine
7036 Available values are:
7039 The default libass shaping engine, which is the best available.
7041 Fast, font-agnostic shaper that can do only substitutions
7043 Slower shaper using OpenType for substitutions and positioning
7046 The default is @code{auto}.
7050 Apply an Adaptive Temporal Averaging Denoiser to the video input.
7052 The filter accepts the following options:
7056 Set threshold A for 1st plane. Default is 0.02.
7057 Valid range is 0 to 0.3.
7060 Set threshold B for 1st plane. Default is 0.04.
7061 Valid range is 0 to 5.
7064 Set threshold A for 2nd plane. Default is 0.02.
7065 Valid range is 0 to 0.3.
7068 Set threshold B for 2nd plane. Default is 0.04.
7069 Valid range is 0 to 5.
7072 Set threshold A for 3rd plane. Default is 0.02.
7073 Valid range is 0 to 0.3.
7076 Set threshold B for 3rd plane. Default is 0.04.
7077 Valid range is 0 to 5.
7079 Threshold A is designed to react on abrupt changes in the input signal and
7080 threshold B is designed to react on continuous changes in the input signal.
7083 Set number of frames filter will use for averaging. Default is 9. Must be odd
7084 number in range [5, 129].
7087 Set what planes of frame filter will use for averaging. Default is all.
7090 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
7091 Alternatively can be set to @code{s} serial.
7093 Parallel can be faster then serial, while other way around is never true.
7094 Parallel will abort early on first change being greater then thresholds, while serial
7095 will continue processing other side of frames if they are equal or below thresholds.
7098 @subsection Commands
7099 This filter supports same @ref{commands} as options except option @code{s}.
7100 The command accepts the same syntax of the corresponding option.
7104 Apply average blur filter.
7106 The filter accepts the following options:
7110 Set horizontal radius size.
7113 Set which planes to filter. By default all planes are filtered.
7116 Set vertical radius size, if zero it will be same as @code{sizeX}.
7117 Default is @code{0}.
7120 @subsection Commands
7121 This filter supports same commands as options.
7122 The command accepts the same syntax of the corresponding option.
7124 If the specified expression is not valid, it is kept at its current
7129 Compute the bounding box for the non-black pixels in the input frame
7132 This filter computes the bounding box containing all the pixels with a
7133 luminance value greater than the minimum allowed value.
7134 The parameters describing the bounding box are printed on the filter
7137 The filter accepts the following option:
7141 Set the minimal luminance value. Default is @code{16}.
7145 Apply bilateral filter, spatial smoothing while preserving edges.
7147 The filter accepts the following options:
7150 Set sigma of gaussian function to calculate spatial weight.
7151 Allowed range is 0 to 512. Default is 0.1.
7154 Set sigma of gaussian function to calculate range weight.
7155 Allowed range is 0 to 1. Default is 0.1.
7158 Set planes to filter. Default is first only.
7161 @section bitplanenoise
7163 Show and measure bit plane noise.
7165 The filter accepts the following options:
7169 Set which plane to analyze. Default is @code{1}.
7172 Filter out noisy pixels from @code{bitplane} set above.
7173 Default is disabled.
7176 @section blackdetect
7178 Detect video intervals that are (almost) completely black. Can be
7179 useful to detect chapter transitions, commercials, or invalid
7182 The filter outputs its detection analysis to both the log as well as
7183 frame metadata. If a black segment of at least the specified minimum
7184 duration is found, a line with the start and end timestamps as well
7185 as duration is printed to the log with level @code{info}. In addition,
7186 a log line with level @code{debug} is printed per frame showing the
7187 black amount detected for that frame.
7189 The filter also attaches metadata to the first frame of a black
7190 segment with key @code{lavfi.black_start} and to the first frame
7191 after the black segment ends with key @code{lavfi.black_end}. The
7192 value is the frame's timestamp. This metadata is added regardless
7193 of the minimum duration specified.
7195 The filter accepts the following options:
7198 @item black_min_duration, d
7199 Set the minimum detected black duration expressed in seconds. It must
7200 be a non-negative floating point number.
7202 Default value is 2.0.
7204 @item picture_black_ratio_th, pic_th
7205 Set the threshold for considering a picture "black".
7206 Express the minimum value for the ratio:
7208 @var{nb_black_pixels} / @var{nb_pixels}
7211 for which a picture is considered black.
7212 Default value is 0.98.
7214 @item pixel_black_th, pix_th
7215 Set the threshold for considering a pixel "black".
7217 The threshold expresses the maximum pixel luminance value for which a
7218 pixel is considered "black". The provided value is scaled according to
7219 the following equation:
7221 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
7224 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
7225 the input video format, the range is [0-255] for YUV full-range
7226 formats and [16-235] for YUV non full-range formats.
7228 Default value is 0.10.
7231 The following example sets the maximum pixel threshold to the minimum
7232 value, and detects only black intervals of 2 or more seconds:
7234 blackdetect=d=2:pix_th=0.00
7239 Detect frames that are (almost) completely black. Can be useful to
7240 detect chapter transitions or commercials. Output lines consist of
7241 the frame number of the detected frame, the percentage of blackness,
7242 the position in the file if known or -1 and the timestamp in seconds.
7244 In order to display the output lines, you need to set the loglevel at
7245 least to the AV_LOG_INFO value.
7247 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
7248 The value represents the percentage of pixels in the picture that
7249 are below the threshold value.
7251 It accepts the following parameters:
7256 The percentage of the pixels that have to be below the threshold; it defaults to
7259 @item threshold, thresh
7260 The threshold below which a pixel value is considered black; it defaults to
7268 Blend two video frames into each other.
7270 The @code{blend} filter takes two input streams and outputs one
7271 stream, the first input is the "top" layer and second input is
7272 "bottom" layer. By default, the output terminates when the longest input terminates.
7274 The @code{tblend} (time blend) filter takes two consecutive frames
7275 from one single stream, and outputs the result obtained by blending
7276 the new frame on top of the old frame.
7278 A description of the accepted options follows.
7286 Set blend mode for specific pixel component or all pixel components in case
7287 of @var{all_mode}. Default value is @code{normal}.
7289 Available values for component modes are:
7331 Set blend opacity for specific pixel component or all pixel components in case
7332 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7339 Set blend expression for specific pixel component or all pixel components in case
7340 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7342 The expressions can use the following variables:
7346 The sequential number of the filtered frame, starting from @code{0}.
7350 the coordinates of the current sample
7354 the width and height of currently filtered plane
7358 Width and height scale for the plane being filtered. It is the
7359 ratio between the dimensions of the current plane to the luma plane,
7360 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7361 the luma plane and @code{0.5,0.5} for the chroma planes.
7364 Time of the current frame, expressed in seconds.
7367 Value of pixel component at current location for first video frame (top layer).
7370 Value of pixel component at current location for second video frame (bottom layer).
7374 The @code{blend} filter also supports the @ref{framesync} options.
7376 @subsection Examples
7380 Apply transition from bottom layer to top layer in first 10 seconds:
7382 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7386 Apply linear horizontal transition from top layer to bottom layer:
7388 blend=all_expr='A*(X/W)+B*(1-X/W)'
7392 Apply 1x1 checkerboard effect:
7394 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7398 Apply uncover left effect:
7400 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7404 Apply uncover down effect:
7406 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7410 Apply uncover up-left effect:
7412 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7416 Split diagonally video and shows top and bottom layer on each side:
7418 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7422 Display differences between the current and the previous frame:
7424 tblend=all_mode=grainextract
7430 Denoise frames using Block-Matching 3D algorithm.
7432 The filter accepts the following options.
7436 Set denoising strength. Default value is 1.
7437 Allowed range is from 0 to 999.9.
7438 The denoising algorithm is very sensitive to sigma, so adjust it
7439 according to the source.
7442 Set local patch size. This sets dimensions in 2D.
7445 Set sliding step for processing blocks. Default value is 4.
7446 Allowed range is from 1 to 64.
7447 Smaller values allows processing more reference blocks and is slower.
7450 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7451 When set to 1, no block matching is done. Larger values allows more blocks
7453 Allowed range is from 1 to 256.
7456 Set radius for search block matching. Default is 9.
7457 Allowed range is from 1 to INT32_MAX.
7460 Set step between two search locations for block matching. Default is 1.
7461 Allowed range is from 1 to 64. Smaller is slower.
7464 Set threshold of mean square error for block matching. Valid range is 0 to
7468 Set thresholding parameter for hard thresholding in 3D transformed domain.
7469 Larger values results in stronger hard-thresholding filtering in frequency
7473 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7474 Default is @code{basic}.
7477 If enabled, filter will use 2nd stream for block matching.
7478 Default is disabled for @code{basic} value of @var{estim} option,
7479 and always enabled if value of @var{estim} is @code{final}.
7482 Set planes to filter. Default is all available except alpha.
7485 @subsection Examples
7489 Basic filtering with bm3d:
7491 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7495 Same as above, but filtering only luma:
7497 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7501 Same as above, but with both estimation modes:
7503 split[a][b],[a]bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7507 Same as above, but prefilter with @ref{nlmeans} filter instead:
7509 split[a][b],[a]nlmeans=s=3:r=7:p=3[a],[b][a]bm3d=sigma=3:block=4:bstep=2:group=16:estim=final:ref=1
7515 Apply a boxblur algorithm to the input video.
7517 It accepts the following parameters:
7521 @item luma_radius, lr
7522 @item luma_power, lp
7523 @item chroma_radius, cr
7524 @item chroma_power, cp
7525 @item alpha_radius, ar
7526 @item alpha_power, ap
7530 A description of the accepted options follows.
7533 @item luma_radius, lr
7534 @item chroma_radius, cr
7535 @item alpha_radius, ar
7536 Set an expression for the box radius in pixels used for blurring the
7537 corresponding input plane.
7539 The radius value must be a non-negative number, and must not be
7540 greater than the value of the expression @code{min(w,h)/2} for the
7541 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7544 Default value for @option{luma_radius} is "2". If not specified,
7545 @option{chroma_radius} and @option{alpha_radius} default to the
7546 corresponding value set for @option{luma_radius}.
7548 The expressions can contain the following constants:
7552 The input width and height in pixels.
7556 The input chroma image width and height in pixels.
7560 The horizontal and vertical chroma subsample values. For example, for the
7561 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7564 @item luma_power, lp
7565 @item chroma_power, cp
7566 @item alpha_power, ap
7567 Specify how many times the boxblur filter is applied to the
7568 corresponding plane.
7570 Default value for @option{luma_power} is 2. If not specified,
7571 @option{chroma_power} and @option{alpha_power} default to the
7572 corresponding value set for @option{luma_power}.
7574 A value of 0 will disable the effect.
7577 @subsection Examples
7581 Apply a boxblur filter with the luma, chroma, and alpha radii
7584 boxblur=luma_radius=2:luma_power=1
7589 Set the luma radius to 2, and alpha and chroma radius to 0:
7591 boxblur=2:1:cr=0:ar=0
7595 Set the luma and chroma radii to a fraction of the video dimension:
7597 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7603 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7604 Deinterlacing Filter").
7606 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7607 interpolation algorithms.
7608 It accepts the following parameters:
7612 The interlacing mode to adopt. It accepts one of the following values:
7616 Output one frame for each frame.
7618 Output one frame for each field.
7621 The default value is @code{send_field}.
7624 The picture field parity assumed for the input interlaced video. It accepts one
7625 of the following values:
7629 Assume the top field is first.
7631 Assume the bottom field is first.
7633 Enable automatic detection of field parity.
7636 The default value is @code{auto}.
7637 If the interlacing is unknown or the decoder does not export this information,
7638 top field first will be assumed.
7641 Specify which frames to deinterlace. Accepts one of the following
7646 Deinterlace all frames.
7648 Only deinterlace frames marked as interlaced.
7651 The default value is @code{all}.
7656 Apply Contrast Adaptive Sharpen filter to video stream.
7658 The filter accepts the following options:
7662 Set the sharpening strength. Default value is 0.
7665 Set planes to filter. Default value is to filter all
7666 planes except alpha plane.
7669 @subsection Commands
7670 This filter supports same @ref{commands} as options.
7673 Remove all color information for all colors except for certain one.
7675 The filter accepts the following options:
7679 The color which will not be replaced with neutral chroma.
7682 Similarity percentage with the above color.
7683 0.01 matches only the exact key color, while 1.0 matches everything.
7687 0.0 makes pixels either fully gray, or not gray at all.
7688 Higher values result in more preserved color.
7691 Signals that the color passed is already in YUV instead of RGB.
7693 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7694 This can be used to pass exact YUV values as hexadecimal numbers.
7697 @subsection Commands
7698 This filter supports same @ref{commands} as options.
7699 The command accepts the same syntax of the corresponding option.
7701 If the specified expression is not valid, it is kept at its current
7705 YUV colorspace color/chroma keying.
7707 The filter accepts the following options:
7711 The color which will be replaced with transparency.
7714 Similarity percentage with the key color.
7716 0.01 matches only the exact key color, while 1.0 matches everything.
7721 0.0 makes pixels either fully transparent, or not transparent at all.
7723 Higher values result in semi-transparent pixels, with a higher transparency
7724 the more similar the pixels color is to the key color.
7727 Signals that the color passed is already in YUV instead of RGB.
7729 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7730 This can be used to pass exact YUV values as hexadecimal numbers.
7733 @subsection Commands
7734 This filter supports same @ref{commands} as options.
7735 The command accepts the same syntax of the corresponding option.
7737 If the specified expression is not valid, it is kept at its current
7740 @subsection Examples
7744 Make every green pixel in the input image transparent:
7746 ffmpeg -i input.png -vf chromakey=green out.png
7750 Overlay a greenscreen-video on top of a static black background.
7752 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
7757 Reduce chrominance noise.
7759 The filter accepts the following options:
7763 Set threshold for averaging chrominance values.
7764 Sum of absolute difference of Y, U and V pixel components of current
7765 pixel and neighbour pixels lower than this threshold will be used in
7766 averaging. Luma component is left unchanged and is copied to output.
7767 Default value is 30. Allowed range is from 1 to 200.
7770 Set horizontal radius of rectangle used for averaging.
7771 Allowed range is from 1 to 100. Default value is 5.
7774 Set vertical radius of rectangle used for averaging.
7775 Allowed range is from 1 to 100. Default value is 5.
7778 Set horizontal step when averaging. Default value is 1.
7779 Allowed range is from 1 to 50.
7780 Mostly useful to speed-up filtering.
7783 Set vertical step when averaging. Default value is 1.
7784 Allowed range is from 1 to 50.
7785 Mostly useful to speed-up filtering.
7788 Set Y threshold for averaging chrominance values.
7789 Set finer control for max allowed difference between Y components
7790 of current pixel and neigbour pixels.
7791 Default value is 200. Allowed range is from 1 to 200.
7794 Set U threshold for averaging chrominance values.
7795 Set finer control for max allowed difference between U components
7796 of current pixel and neigbour pixels.
7797 Default value is 200. Allowed range is from 1 to 200.
7800 Set V threshold for averaging chrominance values.
7801 Set finer control for max allowed difference between V components
7802 of current pixel and neigbour pixels.
7803 Default value is 200. Allowed range is from 1 to 200.
7806 @subsection Commands
7807 This filter supports same @ref{commands} as options.
7808 The command accepts the same syntax of the corresponding option.
7810 @section chromashift
7811 Shift chroma pixels horizontally and/or vertically.
7813 The filter accepts the following options:
7816 Set amount to shift chroma-blue horizontally.
7818 Set amount to shift chroma-blue vertically.
7820 Set amount to shift chroma-red horizontally.
7822 Set amount to shift chroma-red vertically.
7824 Set edge mode, can be @var{smear}, default, or @var{warp}.
7827 @subsection Commands
7829 This filter supports the all above options as @ref{commands}.
7833 Display CIE color diagram with pixels overlaid onto it.
7835 The filter accepts the following options:
7850 @item uhdtv, rec2020
7864 Set what gamuts to draw.
7866 See @code{system} option for available values.
7869 Set ciescope size, by default set to 512.
7872 Set intensity used to map input pixel values to CIE diagram.
7875 Set contrast used to draw tongue colors that are out of active color system gamut.
7878 Correct gamma displayed on scope, by default enabled.
7881 Show white point on CIE diagram, by default disabled.
7884 Set input gamma. Used only with XYZ input color space.
7889 Visualize information exported by some codecs.
7891 Some codecs can export information through frames using side-data or other
7892 means. For example, some MPEG based codecs export motion vectors through the
7893 @var{export_mvs} flag in the codec @option{flags2} option.
7895 The filter accepts the following option:
7899 Set motion vectors to visualize.
7901 Available flags for @var{mv} are:
7905 forward predicted MVs of P-frames
7907 forward predicted MVs of B-frames
7909 backward predicted MVs of B-frames
7913 Display quantization parameters using the chroma planes.
7916 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7918 Available flags for @var{mv_type} are:
7922 forward predicted MVs
7924 backward predicted MVs
7927 @item frame_type, ft
7928 Set frame type to visualize motion vectors of.
7930 Available flags for @var{frame_type} are:
7934 intra-coded frames (I-frames)
7936 predicted frames (P-frames)
7938 bi-directionally predicted frames (B-frames)
7942 @subsection Examples
7946 Visualize forward predicted MVs of all frames using @command{ffplay}:
7948 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7952 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7954 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7958 @section colorbalance
7959 Modify intensity of primary colors (red, green and blue) of input frames.
7961 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7962 regions for the red-cyan, green-magenta or blue-yellow balance.
7964 A positive adjustment value shifts the balance towards the primary color, a negative
7965 value towards the complementary color.
7967 The filter accepts the following options:
7973 Adjust red, green and blue shadows (darkest pixels).
7978 Adjust red, green and blue midtones (medium pixels).
7983 Adjust red, green and blue highlights (brightest pixels).
7985 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7988 Preserve lightness when changing color balance. Default is disabled.
7991 @subsection Examples
7995 Add red color cast to shadows:
8001 @subsection Commands
8003 This filter supports the all above options as @ref{commands}.
8005 @section colorchannelmixer
8007 Adjust video input frames by re-mixing color channels.
8009 This filter modifies a color channel by adding the values associated to
8010 the other channels of the same pixels. For example if the value to
8011 modify is red, the output value will be:
8013 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
8016 The filter accepts the following options:
8023 Adjust contribution of input red, green, blue and alpha channels for output red channel.
8024 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
8030 Adjust contribution of input red, green, blue and alpha channels for output green channel.
8031 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
8037 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
8038 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
8044 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
8045 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
8047 Allowed ranges for options are @code{[-2.0, 2.0]}.
8050 @subsection Examples
8054 Convert source to grayscale:
8056 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
8059 Simulate sepia tones:
8061 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
8065 @subsection Commands
8067 This filter supports the all above options as @ref{commands}.
8070 RGB colorspace color keying.
8072 The filter accepts the following options:
8076 The color which will be replaced with transparency.
8079 Similarity percentage with the key color.
8081 0.01 matches only the exact key color, while 1.0 matches everything.
8086 0.0 makes pixels either fully transparent, or not transparent at all.
8088 Higher values result in semi-transparent pixels, with a higher transparency
8089 the more similar the pixels color is to the key color.
8092 @subsection Examples
8096 Make every green pixel in the input image transparent:
8098 ffmpeg -i input.png -vf colorkey=green out.png
8102 Overlay a greenscreen-video on top of a static background image.
8104 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
8108 @subsection Commands
8109 This filter supports same @ref{commands} as options.
8110 The command accepts the same syntax of the corresponding option.
8112 If the specified expression is not valid, it is kept at its current
8116 Remove all color information for all RGB colors except for certain one.
8118 The filter accepts the following options:
8122 The color which will not be replaced with neutral gray.
8125 Similarity percentage with the above color.
8126 0.01 matches only the exact key color, while 1.0 matches everything.
8129 Blend percentage. 0.0 makes pixels fully gray.
8130 Higher values result in more preserved color.
8133 @subsection Commands
8134 This filter supports same @ref{commands} as options.
8135 The command accepts the same syntax of the corresponding option.
8137 If the specified expression is not valid, it is kept at its current
8140 @section colorlevels
8142 Adjust video input frames using levels.
8144 The filter accepts the following options:
8151 Adjust red, green, blue and alpha input black point.
8152 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
8158 Adjust red, green, blue and alpha input white point.
8159 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
8161 Input levels are used to lighten highlights (bright tones), darken shadows
8162 (dark tones), change the balance of bright and dark tones.
8168 Adjust red, green, blue and alpha output black point.
8169 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
8175 Adjust red, green, blue and alpha output white point.
8176 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
8178 Output levels allows manual selection of a constrained output level range.
8181 @subsection Examples
8185 Make video output darker:
8187 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
8193 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
8197 Make video output lighter:
8199 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
8203 Increase brightness:
8205 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
8209 @subsection Commands
8211 This filter supports the all above options as @ref{commands}.
8213 @section colormatrix
8215 Convert color matrix.
8217 The filter accepts the following options:
8222 Specify the source and destination color matrix. Both values must be
8225 The accepted values are:
8253 For example to convert from BT.601 to SMPTE-240M, use the command:
8255 colormatrix=bt601:smpte240m
8260 Convert colorspace, transfer characteristics or color primaries.
8261 Input video needs to have an even size.
8263 The filter accepts the following options:
8268 Specify all color properties at once.
8270 The accepted values are:
8300 Specify output colorspace.
8302 The accepted values are:
8311 BT.470BG or BT.601-6 625
8314 SMPTE-170M or BT.601-6 525
8323 BT.2020 with non-constant luminance
8329 Specify output transfer characteristics.
8331 The accepted values are:
8343 Constant gamma of 2.2
8346 Constant gamma of 2.8
8349 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8367 BT.2020 for 10-bits content
8370 BT.2020 for 12-bits content
8376 Specify output color primaries.
8378 The accepted values are:
8387 BT.470BG or BT.601-6 625
8390 SMPTE-170M or BT.601-6 525
8414 Specify output color range.
8416 The accepted values are:
8419 TV (restricted) range
8422 MPEG (restricted) range
8433 Specify output color format.
8435 The accepted values are:
8438 YUV 4:2:0 planar 8-bits
8441 YUV 4:2:0 planar 10-bits
8444 YUV 4:2:0 planar 12-bits
8447 YUV 4:2:2 planar 8-bits
8450 YUV 4:2:2 planar 10-bits
8453 YUV 4:2:2 planar 12-bits
8456 YUV 4:4:4 planar 8-bits
8459 YUV 4:4:4 planar 10-bits
8462 YUV 4:4:4 planar 12-bits
8467 Do a fast conversion, which skips gamma/primary correction. This will take
8468 significantly less CPU, but will be mathematically incorrect. To get output
8469 compatible with that produced by the colormatrix filter, use fast=1.
8472 Specify dithering mode.
8474 The accepted values are:
8480 Floyd-Steinberg dithering
8484 Whitepoint adaptation mode.
8486 The accepted values are:
8489 Bradford whitepoint adaptation
8492 von Kries whitepoint adaptation
8495 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8499 Override all input properties at once. Same accepted values as @ref{all}.
8502 Override input colorspace. Same accepted values as @ref{space}.
8505 Override input color primaries. Same accepted values as @ref{primaries}.
8508 Override input transfer characteristics. Same accepted values as @ref{trc}.
8511 Override input color range. Same accepted values as @ref{range}.
8515 The filter converts the transfer characteristics, color space and color
8516 primaries to the specified user values. The output value, if not specified,
8517 is set to a default value based on the "all" property. If that property is
8518 also not specified, the filter will log an error. The output color range and
8519 format default to the same value as the input color range and format. The
8520 input transfer characteristics, color space, color primaries and color range
8521 should be set on the input data. If any of these are missing, the filter will
8522 log an error and no conversion will take place.
8524 For example to convert the input to SMPTE-240M, use the command:
8526 colorspace=smpte240m
8529 @section convolution
8531 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8533 The filter accepts the following options:
8540 Set matrix for each plane.
8541 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8542 and from 1 to 49 odd number of signed integers in @var{row} mode.
8548 Set multiplier for calculated value for each plane.
8549 If unset or 0, it will be sum of all matrix elements.
8555 Set bias for each plane. This value is added to the result of the multiplication.
8556 Useful for making the overall image brighter or darker. Default is 0.0.
8562 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8563 Default is @var{square}.
8566 @subsection Commands
8568 This filter supports the all above options as @ref{commands}.
8570 @subsection Examples
8576 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"
8582 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"
8588 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"
8594 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"
8598 Apply laplacian edge detector which includes diagonals:
8600 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"
8606 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"
8612 Apply 2D convolution of video stream in frequency domain using second stream
8615 The filter accepts the following options:
8619 Set which planes to process.
8622 Set which impulse video frames will be processed, can be @var{first}
8623 or @var{all}. Default is @var{all}.
8626 The @code{convolve} filter also supports the @ref{framesync} options.
8630 Copy the input video source unchanged to the output. This is mainly useful for
8635 Video filtering on GPU using Apple's CoreImage API on OSX.
8637 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8638 processed by video hardware. However, software-based OpenGL implementations
8639 exist which means there is no guarantee for hardware processing. It depends on
8642 There are many filters and image generators provided by Apple that come with a
8643 large variety of options. The filter has to be referenced by its name along
8646 The coreimage filter accepts the following options:
8649 List all available filters and generators along with all their respective
8650 options as well as possible minimum and maximum values along with the default
8657 Specify all filters by their respective name and options.
8658 Use @var{list_filters} to determine all valid filter names and options.
8659 Numerical options are specified by a float value and are automatically clamped
8660 to their respective value range. Vector and color options have to be specified
8661 by a list of space separated float values. Character escaping has to be done.
8662 A special option name @code{default} is available to use default options for a
8665 It is required to specify either @code{default} or at least one of the filter options.
8666 All omitted options are used with their default values.
8667 The syntax of the filter string is as follows:
8669 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8673 Specify a rectangle where the output of the filter chain is copied into the
8674 input image. It is given by a list of space separated float values:
8676 output_rect=x\ y\ width\ height
8678 If not given, the output rectangle equals the dimensions of the input image.
8679 The output rectangle is automatically cropped at the borders of the input
8680 image. Negative values are valid for each component.
8682 output_rect=25\ 25\ 100\ 100
8686 Several filters can be chained for successive processing without GPU-HOST
8687 transfers allowing for fast processing of complex filter chains.
8688 Currently, only filters with zero (generators) or exactly one (filters) input
8689 image and one output image are supported. Also, transition filters are not yet
8692 Some filters generate output images with additional padding depending on the
8693 respective filter kernel. The padding is automatically removed to ensure the
8694 filter output has the same size as the input image.
8696 For image generators, the size of the output image is determined by the
8697 previous output image of the filter chain or the input image of the whole
8698 filterchain, respectively. The generators do not use the pixel information of
8699 this image to generate their output. However, the generated output is
8700 blended onto this image, resulting in partial or complete coverage of the
8703 The @ref{coreimagesrc} video source can be used for generating input images
8704 which are directly fed into the filter chain. By using it, providing input
8705 images by another video source or an input video is not required.
8707 @subsection Examples
8712 List all filters available:
8714 coreimage=list_filters=true
8718 Use the CIBoxBlur filter with default options to blur an image:
8720 coreimage=filter=CIBoxBlur@@default
8724 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8725 its center at 100x100 and a radius of 50 pixels:
8727 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8731 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8732 given as complete and escaped command-line for Apple's standard bash shell:
8734 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8740 Cover a rectangular object
8742 It accepts the following options:
8746 Filepath of the optional cover image, needs to be in yuv420.
8751 It accepts the following values:
8754 cover it by the supplied image
8756 cover it by interpolating the surrounding pixels
8759 Default value is @var{blur}.
8762 @subsection Examples
8766 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8768 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8774 Crop the input video to given dimensions.
8776 It accepts the following parameters:
8780 The width of the output video. It defaults to @code{iw}.
8781 This expression is evaluated only once during the filter
8782 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8785 The height of the output video. It defaults to @code{ih}.
8786 This expression is evaluated only once during the filter
8787 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8790 The horizontal position, in the input video, of the left edge of the output
8791 video. It defaults to @code{(in_w-out_w)/2}.
8792 This expression is evaluated per-frame.
8795 The vertical position, in the input video, of the top edge of the output video.
8796 It defaults to @code{(in_h-out_h)/2}.
8797 This expression is evaluated per-frame.
8800 If set to 1 will force the output display aspect ratio
8801 to be the same of the input, by changing the output sample aspect
8802 ratio. It defaults to 0.
8805 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8806 width/height/x/y as specified and will not be rounded to nearest smaller value.
8810 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8811 expressions containing the following constants:
8816 The computed values for @var{x} and @var{y}. They are evaluated for
8821 The input width and height.
8825 These are the same as @var{in_w} and @var{in_h}.
8829 The output (cropped) width and height.
8833 These are the same as @var{out_w} and @var{out_h}.
8836 same as @var{iw} / @var{ih}
8839 input sample aspect ratio
8842 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8846 horizontal and vertical chroma subsample values. For example for the
8847 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8850 The number of the input frame, starting from 0.
8853 the position in the file of the input frame, NAN if unknown
8856 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8860 The expression for @var{out_w} may depend on the value of @var{out_h},
8861 and the expression for @var{out_h} may depend on @var{out_w}, but they
8862 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8863 evaluated after @var{out_w} and @var{out_h}.
8865 The @var{x} and @var{y} parameters specify the expressions for the
8866 position of the top-left corner of the output (non-cropped) area. They
8867 are evaluated for each frame. If the evaluated value is not valid, it
8868 is approximated to the nearest valid value.
8870 The expression for @var{x} may depend on @var{y}, and the expression
8871 for @var{y} may depend on @var{x}.
8873 @subsection Examples
8877 Crop area with size 100x100 at position (12,34).
8882 Using named options, the example above becomes:
8884 crop=w=100:h=100:x=12:y=34
8888 Crop the central input area with size 100x100:
8894 Crop the central input area with size 2/3 of the input video:
8896 crop=2/3*in_w:2/3*in_h
8900 Crop the input video central square:
8907 Delimit the rectangle with the top-left corner placed at position
8908 100:100 and the right-bottom corner corresponding to the right-bottom
8909 corner of the input image.
8911 crop=in_w-100:in_h-100:100:100
8915 Crop 10 pixels from the left and right borders, and 20 pixels from
8916 the top and bottom borders
8918 crop=in_w-2*10:in_h-2*20
8922 Keep only the bottom right quarter of the input image:
8924 crop=in_w/2:in_h/2:in_w/2:in_h/2
8928 Crop height for getting Greek harmony:
8930 crop=in_w:1/PHI*in_w
8934 Apply trembling effect:
8936 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)
8940 Apply erratic camera effect depending on timestamp:
8942 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)"
8946 Set x depending on the value of y:
8948 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8952 @subsection Commands
8954 This filter supports the following commands:
8960 Set width/height of the output video and the horizontal/vertical position
8962 The command accepts the same syntax of the corresponding option.
8964 If the specified expression is not valid, it is kept at its current
8970 Auto-detect the crop size.
8972 It calculates the necessary cropping parameters and prints the
8973 recommended parameters via the logging system. The detected dimensions
8974 correspond to the non-black area of the input video.
8976 It accepts the following parameters:
8981 Set higher black value threshold, which can be optionally specified
8982 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8983 value greater to the set value is considered non-black. It defaults to 24.
8984 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8985 on the bitdepth of the pixel format.
8988 The value which the width/height should be divisible by. It defaults to
8989 16. The offset is automatically adjusted to center the video. Use 2 to
8990 get only even dimensions (needed for 4:2:2 video). 16 is best when
8991 encoding to most video codecs.
8994 Set the number of initial frames for which evaluation is skipped.
8995 Default is 2. Range is 0 to INT_MAX.
8997 @item reset_count, reset
8998 Set the counter that determines after how many frames cropdetect will
8999 reset the previously detected largest video area and start over to
9000 detect the current optimal crop area. Default value is 0.
9002 This can be useful when channel logos distort the video area. 0
9003 indicates 'never reset', and returns the largest area encountered during
9010 Delay video filtering until a given wallclock timestamp. The filter first
9011 passes on @option{preroll} amount of frames, then it buffers at most
9012 @option{buffer} amount of frames and waits for the cue. After reaching the cue
9013 it forwards the buffered frames and also any subsequent frames coming in its
9016 The filter can be used synchronize the output of multiple ffmpeg processes for
9017 realtime output devices like decklink. By putting the delay in the filtering
9018 chain and pre-buffering frames the process can pass on data to output almost
9019 immediately after the target wallclock timestamp is reached.
9021 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
9027 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
9030 The duration of content to pass on as preroll expressed in seconds. Default is 0.
9033 The maximum duration of content to buffer before waiting for the cue expressed
9034 in seconds. Default is 0.
9041 Apply color adjustments using curves.
9043 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
9044 component (red, green and blue) has its values defined by @var{N} key points
9045 tied from each other using a smooth curve. The x-axis represents the pixel
9046 values from the input frame, and the y-axis the new pixel values to be set for
9049 By default, a component curve is defined by the two points @var{(0;0)} and
9050 @var{(1;1)}. This creates a straight line where each original pixel value is
9051 "adjusted" to its own value, which means no change to the image.
9053 The filter allows you to redefine these two points and add some more. A new
9054 curve (using a natural cubic spline interpolation) will be define to pass
9055 smoothly through all these new coordinates. The new defined points needs to be
9056 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
9057 be in the @var{[0;1]} interval. If the computed curves happened to go outside
9058 the vector spaces, the values will be clipped accordingly.
9060 The filter accepts the following options:
9064 Select one of the available color presets. This option can be used in addition
9065 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
9066 options takes priority on the preset values.
9067 Available presets are:
9070 @item color_negative
9073 @item increase_contrast
9075 @item linear_contrast
9076 @item medium_contrast
9078 @item strong_contrast
9081 Default is @code{none}.
9083 Set the master key points. These points will define a second pass mapping. It
9084 is sometimes called a "luminance" or "value" mapping. It can be used with
9085 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
9086 post-processing LUT.
9088 Set the key points for the red component.
9090 Set the key points for the green component.
9092 Set the key points for the blue component.
9094 Set the key points for all components (not including master).
9095 Can be used in addition to the other key points component
9096 options. In this case, the unset component(s) will fallback on this
9097 @option{all} setting.
9099 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
9101 Save Gnuplot script of the curves in specified file.
9104 To avoid some filtergraph syntax conflicts, each key points list need to be
9105 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
9107 @subsection Examples
9111 Increase slightly the middle level of blue:
9113 curves=blue='0/0 0.5/0.58 1/1'
9119 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'
9121 Here we obtain the following coordinates for each components:
9124 @code{(0;0.11) (0.42;0.51) (1;0.95)}
9126 @code{(0;0) (0.50;0.48) (1;1)}
9128 @code{(0;0.22) (0.49;0.44) (1;0.80)}
9132 The previous example can also be achieved with the associated built-in preset:
9134 curves=preset=vintage
9144 Use a Photoshop preset and redefine the points of the green component:
9146 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
9150 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
9151 and @command{gnuplot}:
9153 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
9154 gnuplot -p /tmp/curves.plt
9160 Video data analysis filter.
9162 This filter shows hexadecimal pixel values of part of video.
9164 The filter accepts the following options:
9168 Set output video size.
9171 Set x offset from where to pick pixels.
9174 Set y offset from where to pick pixels.
9177 Set scope mode, can be one of the following:
9180 Draw hexadecimal pixel values with white color on black background.
9183 Draw hexadecimal pixel values with input video pixel color on black
9187 Draw hexadecimal pixel values on color background picked from input video,
9188 the text color is picked in such way so its always visible.
9192 Draw rows and columns numbers on left and top of video.
9195 Set background opacity.
9198 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
9202 Apply Directional blur filter.
9204 The filter accepts the following options:
9208 Set angle of directional blur. Default is @code{45}.
9211 Set radius of directional blur. Default is @code{5}.
9214 Set which planes to filter. By default all planes are filtered.
9217 @subsection Commands
9218 This filter supports same @ref{commands} as options.
9219 The command accepts the same syntax of the corresponding option.
9221 If the specified expression is not valid, it is kept at its current
9226 Denoise frames using 2D DCT (frequency domain filtering).
9228 This filter is not designed for real time.
9230 The filter accepts the following options:
9234 Set the noise sigma constant.
9236 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
9237 coefficient (absolute value) below this threshold with be dropped.
9239 If you need a more advanced filtering, see @option{expr}.
9241 Default is @code{0}.
9244 Set number overlapping pixels for each block. Since the filter can be slow, you
9245 may want to reduce this value, at the cost of a less effective filter and the
9246 risk of various artefacts.
9248 If the overlapping value doesn't permit processing the whole input width or
9249 height, a warning will be displayed and according borders won't be denoised.
9251 Default value is @var{blocksize}-1, which is the best possible setting.
9254 Set the coefficient factor expression.
9256 For each coefficient of a DCT block, this expression will be evaluated as a
9257 multiplier value for the coefficient.
9259 If this is option is set, the @option{sigma} option will be ignored.
9261 The absolute value of the coefficient can be accessed through the @var{c}
9265 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
9266 @var{blocksize}, which is the width and height of the processed blocks.
9268 The default value is @var{3} (8x8) and can be raised to @var{4} for a
9269 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
9270 on the speed processing. Also, a larger block size does not necessarily means a
9274 @subsection Examples
9276 Apply a denoise with a @option{sigma} of @code{4.5}:
9281 The same operation can be achieved using the expression system:
9283 dctdnoiz=e='gte(c, 4.5*3)'
9286 Violent denoise using a block size of @code{16x16}:
9293 Remove banding artifacts from input video.
9294 It works by replacing banded pixels with average value of referenced pixels.
9296 The filter accepts the following options:
9303 Set banding detection threshold for each plane. Default is 0.02.
9304 Valid range is 0.00003 to 0.5.
9305 If difference between current pixel and reference pixel is less than threshold,
9306 it will be considered as banded.
9309 Banding detection range in pixels. Default is 16. If positive, random number
9310 in range 0 to set value will be used. If negative, exact absolute value
9312 The range defines square of four pixels around current pixel.
9315 Set direction in radians from which four pixel will be compared. If positive,
9316 random direction from 0 to set direction will be picked. If negative, exact of
9317 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9318 will pick only pixels on same row and -PI/2 will pick only pixels on same
9322 If enabled, current pixel is compared with average value of all four
9323 surrounding pixels. The default is enabled. If disabled current pixel is
9324 compared with all four surrounding pixels. The pixel is considered banded
9325 if only all four differences with surrounding pixels are less than threshold.
9328 If enabled, current pixel is changed if and only if all pixel components are banded,
9329 e.g. banding detection threshold is triggered for all color components.
9330 The default is disabled.
9335 Remove blocking artifacts from input video.
9337 The filter accepts the following options:
9341 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9342 This controls what kind of deblocking is applied.
9345 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9351 Set blocking detection thresholds. Allowed range is 0 to 1.
9352 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9353 Using higher threshold gives more deblocking strength.
9354 Setting @var{alpha} controls threshold detection at exact edge of block.
9355 Remaining options controls threshold detection near the edge. Each one for
9356 below/above or left/right. Setting any of those to @var{0} disables
9360 Set planes to filter. Default is to filter all available planes.
9363 @subsection Examples
9367 Deblock using weak filter and block size of 4 pixels.
9369 deblock=filter=weak:block=4
9373 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9374 deblocking more edges.
9376 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9380 Similar as above, but filter only first plane.
9382 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9386 Similar as above, but filter only second and third plane.
9388 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9395 Drop duplicated frames at regular intervals.
9397 The filter accepts the following options:
9401 Set the number of frames from which one will be dropped. Setting this to
9402 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9403 Default is @code{5}.
9406 Set the threshold for duplicate detection. If the difference metric for a frame
9407 is less than or equal to this value, then it is declared as duplicate. Default
9411 Set scene change threshold. Default is @code{15}.
9415 Set the size of the x and y-axis blocks used during metric calculations.
9416 Larger blocks give better noise suppression, but also give worse detection of
9417 small movements. Must be a power of two. Default is @code{32}.
9420 Mark main input as a pre-processed input and activate clean source input
9421 stream. This allows the input to be pre-processed with various filters to help
9422 the metrics calculation while keeping the frame selection lossless. When set to
9423 @code{1}, the first stream is for the pre-processed input, and the second
9424 stream is the clean source from where the kept frames are chosen. Default is
9428 Set whether or not chroma is considered in the metric calculations. Default is
9434 Apply 2D deconvolution of video stream in frequency domain using second stream
9437 The filter accepts the following options:
9441 Set which planes to process.
9444 Set which impulse video frames will be processed, can be @var{first}
9445 or @var{all}. Default is @var{all}.
9448 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9449 and height are not same and not power of 2 or if stream prior to convolving
9453 The @code{deconvolve} filter also supports the @ref{framesync} options.
9457 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9459 It accepts the following options:
9463 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9464 @var{rainbows} for cross-color reduction.
9467 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9470 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9473 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9476 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9481 Apply deflate effect to the video.
9483 This filter replaces the pixel by the local(3x3) average by taking into account
9484 only values lower than the pixel.
9486 It accepts the following options:
9493 Limit the maximum change for each plane, default is 65535.
9494 If 0, plane will remain unchanged.
9497 @subsection Commands
9499 This filter supports the all above options as @ref{commands}.
9503 Remove temporal frame luminance variations.
9505 It accepts the following options:
9509 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9512 Set averaging mode to smooth temporal luminance variations.
9514 Available values are:
9539 Do not actually modify frame. Useful when one only wants metadata.
9544 Remove judder produced by partially interlaced telecined content.
9546 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9547 source was partially telecined content then the output of @code{pullup,dejudder}
9548 will have a variable frame rate. May change the recorded frame rate of the
9549 container. Aside from that change, this filter will not affect constant frame
9552 The option available in this filter is:
9556 Specify the length of the window over which the judder repeats.
9558 Accepts any integer greater than 1. Useful values are:
9562 If the original was telecined from 24 to 30 fps (Film to NTSC).
9565 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9568 If a mixture of the two.
9571 The default is @samp{4}.
9576 Suppress a TV station logo by a simple interpolation of the surrounding
9577 pixels. Just set a rectangle covering the logo and watch it disappear
9578 (and sometimes something even uglier appear - your mileage may vary).
9580 It accepts the following parameters:
9585 Specify the top left corner coordinates of the logo. They must be
9590 Specify the width and height of the logo to clear. They must be
9594 Specify the thickness of the fuzzy edge of the rectangle (added to
9595 @var{w} and @var{h}). The default value is 1. This option is
9596 deprecated, setting higher values should no longer be necessary and
9600 When set to 1, a green rectangle is drawn on the screen to simplify
9601 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9602 The default value is 0.
9604 The rectangle is drawn on the outermost pixels which will be (partly)
9605 replaced with interpolated values. The values of the next pixels
9606 immediately outside this rectangle in each direction will be used to
9607 compute the interpolated pixel values inside the rectangle.
9611 @subsection Examples
9615 Set a rectangle covering the area with top left corner coordinates 0,0
9616 and size 100x77, and a band of size 10:
9618 delogo=x=0:y=0:w=100:h=77:band=10
9626 Remove the rain in the input image/video by applying the derain methods based on
9627 convolutional neural networks. Supported models:
9631 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9632 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9635 Training as well as model generation scripts are provided in
9636 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9638 Native model files (.model) can be generated from TensorFlow model
9639 files (.pb) by using tools/python/convert.py
9641 The filter accepts the following options:
9645 Specify which filter to use. This option accepts the following values:
9649 Derain filter. To conduct derain filter, you need to use a derain model.
9652 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9654 Default value is @samp{derain}.
9657 Specify which DNN backend to use for model loading and execution. This option accepts
9658 the following values:
9662 Native implementation of DNN loading and execution.
9665 TensorFlow backend. To enable this backend you
9666 need to install the TensorFlow for C library (see
9667 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9668 @code{--enable-libtensorflow}
9670 Default value is @samp{native}.
9673 Set path to model file specifying network architecture and its parameters.
9674 Note that different backends use different file formats. TensorFlow and native
9675 backend can load files for only its format.
9678 It can also be finished with @ref{dnn_processing} filter.
9682 Attempt to fix small changes in horizontal and/or vertical shift. This
9683 filter helps remove camera shake from hand-holding a camera, bumping a
9684 tripod, moving on a vehicle, etc.
9686 The filter accepts the following options:
9694 Specify a rectangular area where to limit the search for motion
9696 If desired the search for motion vectors can be limited to a
9697 rectangular area of the frame defined by its top left corner, width
9698 and height. These parameters have the same meaning as the drawbox
9699 filter which can be used to visualise the position of the bounding
9702 This is useful when simultaneous movement of subjects within the frame
9703 might be confused for camera motion by the motion vector search.
9705 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9706 then the full frame is used. This allows later options to be set
9707 without specifying the bounding box for the motion vector search.
9709 Default - search the whole frame.
9713 Specify the maximum extent of movement in x and y directions in the
9714 range 0-64 pixels. Default 16.
9717 Specify how to generate pixels to fill blanks at the edge of the
9718 frame. Available values are:
9721 Fill zeroes at blank locations
9723 Original image at blank locations
9725 Extruded edge value at blank locations
9727 Mirrored edge at blank locations
9729 Default value is @samp{mirror}.
9732 Specify the blocksize to use for motion search. Range 4-128 pixels,
9736 Specify the contrast threshold for blocks. Only blocks with more than
9737 the specified contrast (difference between darkest and lightest
9738 pixels) will be considered. Range 1-255, default 125.
9741 Specify the search strategy. Available values are:
9744 Set exhaustive search
9746 Set less exhaustive search.
9748 Default value is @samp{exhaustive}.
9751 If set then a detailed log of the motion search is written to the
9758 Remove unwanted contamination of foreground colors, caused by reflected color of
9759 greenscreen or bluescreen.
9761 This filter accepts the following options:
9765 Set what type of despill to use.
9768 Set how spillmap will be generated.
9771 Set how much to get rid of still remaining spill.
9774 Controls amount of red in spill area.
9777 Controls amount of green in spill area.
9778 Should be -1 for greenscreen.
9781 Controls amount of blue in spill area.
9782 Should be -1 for bluescreen.
9785 Controls brightness of spill area, preserving colors.
9788 Modify alpha from generated spillmap.
9791 @subsection Commands
9793 This filter supports the all above options as @ref{commands}.
9797 Apply an exact inverse of the telecine operation. It requires a predefined
9798 pattern specified using the pattern option which must be the same as that passed
9799 to the telecine filter.
9801 This filter accepts the following options:
9810 The default value is @code{top}.
9814 A string of numbers representing the pulldown pattern you wish to apply.
9815 The default value is @code{23}.
9818 A number representing position of the first frame with respect to the telecine
9819 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9824 Apply dilation effect to the video.
9826 This filter replaces the pixel by the local(3x3) maximum.
9828 It accepts the following options:
9835 Limit the maximum change for each plane, default is 65535.
9836 If 0, plane will remain unchanged.
9839 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9842 Flags to local 3x3 coordinates maps like this:
9849 @subsection Commands
9851 This filter supports the all above options as @ref{commands}.
9855 Displace pixels as indicated by second and third input stream.
9857 It takes three input streams and outputs one stream, the first input is the
9858 source, and second and third input are displacement maps.
9860 The second input specifies how much to displace pixels along the
9861 x-axis, while the third input specifies how much to displace pixels
9863 If one of displacement map streams terminates, last frame from that
9864 displacement map will be used.
9866 Note that once generated, displacements maps can be reused over and over again.
9868 A description of the accepted options follows.
9872 Set displace behavior for pixels that are out of range.
9874 Available values are:
9877 Missing pixels are replaced by black pixels.
9880 Adjacent pixels will spread out to replace missing pixels.
9883 Out of range pixels are wrapped so they point to pixels of other side.
9886 Out of range pixels will be replaced with mirrored pixels.
9888 Default is @samp{smear}.
9892 @subsection Examples
9896 Add ripple effect to rgb input of video size hd720:
9898 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
9902 Add wave effect to rgb input of video size hd720:
9904 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
9908 @anchor{dnn_processing}
9909 @section dnn_processing
9911 Do image processing with deep neural networks. It works together with another filter
9912 which converts the pixel format of the Frame to what the dnn network requires.
9914 The filter accepts the following options:
9918 Specify which DNN backend to use for model loading and execution. This option accepts
9919 the following values:
9923 Native implementation of DNN loading and execution.
9926 TensorFlow backend. To enable this backend you
9927 need to install the TensorFlow for C library (see
9928 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9929 @code{--enable-libtensorflow}
9932 OpenVINO backend. To enable this backend you
9933 need to build and install the OpenVINO for C library (see
9934 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9935 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9936 be needed if the header files and libraries are not installed into system path)
9940 Default value is @samp{native}.
9943 Set path to model file specifying network architecture and its parameters.
9944 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9945 backend can load files for only its format.
9947 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9950 Set the input name of the dnn network.
9953 Set the output name of the dnn network.
9957 @subsection Examples
9961 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9963 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9967 Halve the pixel value of the frame with format gray32f:
9969 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
9973 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9975 ./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
9979 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9981 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9988 Draw a colored box on the input image.
9990 It accepts the following parameters:
9995 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9999 The expressions which specify the width and height of the box; if 0 they are interpreted as
10000 the input width and height. It defaults to 0.
10003 Specify the color of the box to write. For the general syntax of this option,
10004 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10005 value @code{invert} is used, the box edge color is the same as the
10006 video with inverted luma.
10009 The expression which sets the thickness of the box edge.
10010 A value of @code{fill} will create a filled box. Default value is @code{3}.
10012 See below for the list of accepted constants.
10015 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
10016 will overwrite the video's color and alpha pixels.
10017 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
10020 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10021 following constants:
10025 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10029 horizontal and vertical chroma subsample values. For example for the
10030 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10034 The input width and height.
10037 The input sample aspect ratio.
10041 The x and y offset coordinates where the box is drawn.
10045 The width and height of the drawn box.
10048 The thickness of the drawn box.
10050 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10051 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10055 @subsection Examples
10059 Draw a black box around the edge of the input image:
10065 Draw a box with color red and an opacity of 50%:
10067 drawbox=10:20:200:60:red@@0.5
10070 The previous example can be specified as:
10072 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
10076 Fill the box with pink color:
10078 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
10082 Draw a 2-pixel red 2.40:1 mask:
10084 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
10088 @subsection Commands
10089 This filter supports same commands as options.
10090 The command accepts the same syntax of the corresponding option.
10092 If the specified expression is not valid, it is kept at its current
10097 Draw a graph using input video metadata.
10099 It accepts the following parameters:
10103 Set 1st frame metadata key from which metadata values will be used to draw a graph.
10106 Set 1st foreground color expression.
10109 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
10112 Set 2nd foreground color expression.
10115 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
10118 Set 3rd foreground color expression.
10121 Set 4th frame metadata key from which metadata values will be used to draw a graph.
10124 Set 4th foreground color expression.
10127 Set minimal value of metadata value.
10130 Set maximal value of metadata value.
10133 Set graph background color. Default is white.
10138 Available values for mode is:
10145 Default is @code{line}.
10150 Available values for slide is:
10153 Draw new frame when right border is reached.
10156 Replace old columns with new ones.
10159 Scroll from right to left.
10162 Scroll from left to right.
10165 Draw single picture.
10168 Default is @code{frame}.
10171 Set size of graph video. For the syntax of this option, check the
10172 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
10173 The default value is @code{900x256}.
10176 Set the output frame rate. Default value is @code{25}.
10178 The foreground color expressions can use the following variables:
10181 Minimal value of metadata value.
10184 Maximal value of metadata value.
10187 Current metadata key value.
10190 The color is defined as 0xAABBGGRR.
10193 Example using metadata from @ref{signalstats} filter:
10195 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
10198 Example using metadata from @ref{ebur128} filter:
10200 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
10205 Draw a grid on the input image.
10207 It accepts the following parameters:
10212 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
10216 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
10217 input width and height, respectively, minus @code{thickness}, so image gets
10218 framed. Default to 0.
10221 Specify the color of the grid. For the general syntax of this option,
10222 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10223 value @code{invert} is used, the grid color is the same as the
10224 video with inverted luma.
10227 The expression which sets the thickness of the grid line. Default value is @code{1}.
10229 See below for the list of accepted constants.
10232 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
10233 will overwrite the video's color and alpha pixels.
10234 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
10237 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10238 following constants:
10242 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10246 horizontal and vertical chroma subsample values. For example for the
10247 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10251 The input grid cell width and height.
10254 The input sample aspect ratio.
10258 The x and y coordinates of some point of grid intersection (meant to configure offset).
10262 The width and height of the drawn cell.
10265 The thickness of the drawn cell.
10267 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10268 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10272 @subsection Examples
10276 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
10278 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
10282 Draw a white 3x3 grid with an opacity of 50%:
10284 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10288 @subsection Commands
10289 This filter supports same commands as options.
10290 The command accepts the same syntax of the corresponding option.
10292 If the specified expression is not valid, it is kept at its current
10298 Draw a text string or text from a specified file on top of a video, using the
10299 libfreetype library.
10301 To enable compilation of this filter, you need to configure FFmpeg with
10302 @code{--enable-libfreetype}.
10303 To enable default font fallback and the @var{font} option you need to
10304 configure FFmpeg with @code{--enable-libfontconfig}.
10305 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10306 @code{--enable-libfribidi}.
10310 It accepts the following parameters:
10315 Used to draw a box around text using the background color.
10316 The value must be either 1 (enable) or 0 (disable).
10317 The default value of @var{box} is 0.
10320 Set the width of the border to be drawn around the box using @var{boxcolor}.
10321 The default value of @var{boxborderw} is 0.
10324 The color to be used for drawing box around text. For the syntax of this
10325 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10327 The default value of @var{boxcolor} is "white".
10330 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10331 The default value of @var{line_spacing} is 0.
10334 Set the width of the border to be drawn around the text using @var{bordercolor}.
10335 The default value of @var{borderw} is 0.
10338 Set the color to be used for drawing border around text. For the syntax of this
10339 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10341 The default value of @var{bordercolor} is "black".
10344 Select how the @var{text} is expanded. Can be either @code{none},
10345 @code{strftime} (deprecated) or
10346 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10350 Set a start time for the count. Value is in microseconds. Only applied
10351 in the deprecated strftime expansion mode. To emulate in normal expansion
10352 mode use the @code{pts} function, supplying the start time (in seconds)
10353 as the second argument.
10356 If true, check and fix text coords to avoid clipping.
10359 The color to be used for drawing fonts. For the syntax of this option, check
10360 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10362 The default value of @var{fontcolor} is "black".
10364 @item fontcolor_expr
10365 String which is expanded the same way as @var{text} to obtain dynamic
10366 @var{fontcolor} value. By default this option has empty value and is not
10367 processed. When this option is set, it overrides @var{fontcolor} option.
10370 The font family to be used for drawing text. By default Sans.
10373 The font file to be used for drawing text. The path must be included.
10374 This parameter is mandatory if the fontconfig support is disabled.
10377 Draw the text applying alpha blending. The value can
10378 be a number between 0.0 and 1.0.
10379 The expression accepts the same variables @var{x, y} as well.
10380 The default value is 1.
10381 Please see @var{fontcolor_expr}.
10384 The font size to be used for drawing text.
10385 The default value of @var{fontsize} is 16.
10388 If set to 1, attempt to shape the text (for example, reverse the order of
10389 right-to-left text and join Arabic characters) before drawing it.
10390 Otherwise, just draw the text exactly as given.
10391 By default 1 (if supported).
10393 @item ft_load_flags
10394 The flags to be used for loading the fonts.
10396 The flags map the corresponding flags supported by libfreetype, and are
10397 a combination of the following values:
10404 @item vertical_layout
10405 @item force_autohint
10408 @item ignore_global_advance_width
10410 @item ignore_transform
10412 @item linear_design
10416 Default value is "default".
10418 For more information consult the documentation for the FT_LOAD_*
10422 The color to be used for drawing a shadow behind the drawn text. For the
10423 syntax of this option, check the @ref{color syntax,,"Color" section in the
10424 ffmpeg-utils manual,ffmpeg-utils}.
10426 The default value of @var{shadowcolor} is "black".
10430 The x and y offsets for the text shadow position with respect to the
10431 position of the text. They can be either positive or negative
10432 values. The default value for both is "0".
10435 The starting frame number for the n/frame_num variable. The default value
10439 The size in number of spaces to use for rendering the tab.
10440 Default value is 4.
10443 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10444 format. It can be used with or without text parameter. @var{timecode_rate}
10445 option must be specified.
10447 @item timecode_rate, rate, r
10448 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10449 integer. Minimum value is "1".
10450 Drop-frame timecode is supported for frame rates 30 & 60.
10453 If set to 1, the output of the timecode option will wrap around at 24 hours.
10454 Default is 0 (disabled).
10457 The text string to be drawn. The text must be a sequence of UTF-8
10458 encoded characters.
10459 This parameter is mandatory if no file is specified with the parameter
10463 A text file containing text to be drawn. The text must be a sequence
10464 of UTF-8 encoded characters.
10466 This parameter is mandatory if no text string is specified with the
10467 parameter @var{text}.
10469 If both @var{text} and @var{textfile} are specified, an error is thrown.
10472 If set to 1, the @var{textfile} will be reloaded before each frame.
10473 Be sure to update it atomically, or it may be read partially, or even fail.
10477 The expressions which specify the offsets where text will be drawn
10478 within the video frame. They are relative to the top/left border of the
10481 The default value of @var{x} and @var{y} is "0".
10483 See below for the list of accepted constants and functions.
10486 The parameters for @var{x} and @var{y} are expressions containing the
10487 following constants and functions:
10491 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10495 horizontal and vertical chroma subsample values. For example for the
10496 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10499 the height of each text line
10507 @item max_glyph_a, ascent
10508 the maximum distance from the baseline to the highest/upper grid
10509 coordinate used to place a glyph outline point, for all the rendered
10511 It is a positive value, due to the grid's orientation with the Y axis
10514 @item max_glyph_d, descent
10515 the maximum distance from the baseline to the lowest grid coordinate
10516 used to place a glyph outline point, for all the rendered glyphs.
10517 This is a negative value, due to the grid's orientation, with the Y axis
10521 maximum glyph height, that is the maximum height for all the glyphs
10522 contained in the rendered text, it is equivalent to @var{ascent} -
10526 maximum glyph width, that is the maximum width for all the glyphs
10527 contained in the rendered text
10530 the number of input frame, starting from 0
10532 @item rand(min, max)
10533 return a random number included between @var{min} and @var{max}
10536 The input sample aspect ratio.
10539 timestamp expressed in seconds, NAN if the input timestamp is unknown
10542 the height of the rendered text
10545 the width of the rendered text
10549 the x and y offset coordinates where the text is drawn.
10551 These parameters allow the @var{x} and @var{y} expressions to refer
10552 to each other, so you can for example specify @code{y=x/dar}.
10555 A one character description of the current frame's picture type.
10558 The current packet's position in the input file or stream
10559 (in bytes, from the start of the input). A value of -1 indicates
10560 this info is not available.
10563 The current packet's duration, in seconds.
10566 The current packet's size (in bytes).
10569 @anchor{drawtext_expansion}
10570 @subsection Text expansion
10572 If @option{expansion} is set to @code{strftime},
10573 the filter recognizes strftime() sequences in the provided text and
10574 expands them accordingly. Check the documentation of strftime(). This
10575 feature is deprecated.
10577 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10579 If @option{expansion} is set to @code{normal} (which is the default),
10580 the following expansion mechanism is used.
10582 The backslash character @samp{\}, followed by any character, always expands to
10583 the second character.
10585 Sequences of the form @code{%@{...@}} are expanded. The text between the
10586 braces is a function name, possibly followed by arguments separated by ':'.
10587 If the arguments contain special characters or delimiters (':' or '@}'),
10588 they should be escaped.
10590 Note that they probably must also be escaped as the value for the
10591 @option{text} option in the filter argument string and as the filter
10592 argument in the filtergraph description, and possibly also for the shell,
10593 that makes up to four levels of escaping; using a text file avoids these
10596 The following functions are available:
10601 The expression evaluation result.
10603 It must take one argument specifying the expression to be evaluated,
10604 which accepts the same constants and functions as the @var{x} and
10605 @var{y} values. Note that not all constants should be used, for
10606 example the text size is not known when evaluating the expression, so
10607 the constants @var{text_w} and @var{text_h} will have an undefined
10610 @item expr_int_format, eif
10611 Evaluate the expression's value and output as formatted integer.
10613 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10614 The second argument specifies the output format. Allowed values are @samp{x},
10615 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10616 @code{printf} function.
10617 The third parameter is optional and sets the number of positions taken by the output.
10618 It can be used to add padding with zeros from the left.
10621 The time at which the filter is running, expressed in UTC.
10622 It can accept an argument: a strftime() format string.
10625 The time at which the filter is running, expressed in the local time zone.
10626 It can accept an argument: a strftime() format string.
10629 Frame metadata. Takes one or two arguments.
10631 The first argument is mandatory and specifies the metadata key.
10633 The second argument is optional and specifies a default value, used when the
10634 metadata key is not found or empty.
10636 Available metadata can be identified by inspecting entries
10637 starting with TAG included within each frame section
10638 printed by running @code{ffprobe -show_frames}.
10640 String metadata generated in filters leading to
10641 the drawtext filter are also available.
10644 The frame number, starting from 0.
10647 A one character description of the current picture type.
10650 The timestamp of the current frame.
10651 It can take up to three arguments.
10653 The first argument is the format of the timestamp; it defaults to @code{flt}
10654 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10655 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10656 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10657 @code{localtime} stands for the timestamp of the frame formatted as
10658 local time zone time.
10660 The second argument is an offset added to the timestamp.
10662 If the format is set to @code{hms}, a third argument @code{24HH} may be
10663 supplied to present the hour part of the formatted timestamp in 24h format
10666 If the format is set to @code{localtime} or @code{gmtime},
10667 a third argument may be supplied: a strftime() format string.
10668 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10671 @subsection Commands
10673 This filter supports altering parameters via commands:
10676 Alter existing filter parameters.
10678 Syntax for the argument is the same as for filter invocation, e.g.
10681 fontsize=56:fontcolor=green:text='Hello World'
10684 Full filter invocation with sendcmd would look like this:
10687 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10691 If the entire argument can't be parsed or applied as valid values then the filter will
10692 continue with its existing parameters.
10694 @subsection Examples
10698 Draw "Test Text" with font FreeSerif, using the default values for the
10699 optional parameters.
10702 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10706 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10707 and y=50 (counting from the top-left corner of the screen), text is
10708 yellow with a red box around it. Both the text and the box have an
10712 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10713 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10716 Note that the double quotes are not necessary if spaces are not used
10717 within the parameter list.
10720 Show the text at the center of the video frame:
10722 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10726 Show the text at a random position, switching to a new position every 30 seconds:
10728 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)"
10732 Show a text line sliding from right to left in the last row of the video
10733 frame. The file @file{LONG_LINE} is assumed to contain a single line
10736 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10740 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10742 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10746 Draw a single green letter "g", at the center of the input video.
10747 The glyph baseline is placed at half screen height.
10749 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10753 Show text for 1 second every 3 seconds:
10755 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10759 Use fontconfig to set the font. Note that the colons need to be escaped.
10761 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10765 Draw "Test Text" with font size dependent on height of the video.
10767 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10771 Print the date of a real-time encoding (see strftime(3)):
10773 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10777 Show text fading in and out (appearing/disappearing):
10780 DS=1.0 # display start
10781 DE=10.0 # display end
10782 FID=1.5 # fade in duration
10783 FOD=5 # fade out duration
10784 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 @}"
10788 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10789 and the @option{fontsize} value are included in the @option{y} offset.
10791 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10792 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10796 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10797 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10798 must have option @option{-export_path_metadata 1} for the special metadata fields
10799 to be available for filters.
10801 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10806 For more information about libfreetype, check:
10807 @url{http://www.freetype.org/}.
10809 For more information about fontconfig, check:
10810 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10812 For more information about libfribidi, check:
10813 @url{http://fribidi.org/}.
10815 @section edgedetect
10817 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10819 The filter accepts the following options:
10824 Set low and high threshold values used by the Canny thresholding
10827 The high threshold selects the "strong" edge pixels, which are then
10828 connected through 8-connectivity with the "weak" edge pixels selected
10829 by the low threshold.
10831 @var{low} and @var{high} threshold values must be chosen in the range
10832 [0,1], and @var{low} should be lesser or equal to @var{high}.
10834 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10838 Define the drawing mode.
10842 Draw white/gray wires on black background.
10845 Mix the colors to create a paint/cartoon effect.
10848 Apply Canny edge detector on all selected planes.
10850 Default value is @var{wires}.
10853 Select planes for filtering. By default all available planes are filtered.
10856 @subsection Examples
10860 Standard edge detection with custom values for the hysteresis thresholding:
10862 edgedetect=low=0.1:high=0.4
10866 Painting effect without thresholding:
10868 edgedetect=mode=colormix:high=0
10874 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10876 For each input image, the filter will compute the optimal mapping from
10877 the input to the output given the codebook length, that is the number
10878 of distinct output colors.
10880 This filter accepts the following options.
10883 @item codebook_length, l
10884 Set codebook length. The value must be a positive integer, and
10885 represents the number of distinct output colors. Default value is 256.
10888 Set the maximum number of iterations to apply for computing the optimal
10889 mapping. The higher the value the better the result and the higher the
10890 computation time. Default value is 1.
10893 Set a random seed, must be an integer included between 0 and
10894 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10895 will try to use a good random seed on a best effort basis.
10898 Set pal8 output pixel format. This option does not work with codebook
10899 length greater than 256.
10904 Measure graylevel entropy in histogram of color channels of video frames.
10906 It accepts the following parameters:
10910 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10912 @var{diff} mode measures entropy of histogram delta values, absolute differences
10913 between neighbour histogram values.
10917 Set brightness, contrast, saturation and approximate gamma adjustment.
10919 The filter accepts the following options:
10923 Set the contrast expression. The value must be a float value in range
10924 @code{-1000.0} to @code{1000.0}. The default value is "1".
10927 Set the brightness expression. The value must be a float value in
10928 range @code{-1.0} to @code{1.0}. The default value is "0".
10931 Set the saturation expression. The value must be a float in
10932 range @code{0.0} to @code{3.0}. The default value is "1".
10935 Set the gamma expression. The value must be a float in range
10936 @code{0.1} to @code{10.0}. The default value is "1".
10939 Set the gamma expression for red. The value must be a float in
10940 range @code{0.1} to @code{10.0}. The default value is "1".
10943 Set the gamma expression for green. The value must be a float in range
10944 @code{0.1} to @code{10.0}. The default value is "1".
10947 Set the gamma expression for blue. The value must be a float in range
10948 @code{0.1} to @code{10.0}. The default value is "1".
10951 Set the gamma weight expression. It can be used to reduce the effect
10952 of a high gamma value on bright image areas, e.g. keep them from
10953 getting overamplified and just plain white. The value must be a float
10954 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10955 gamma correction all the way down while @code{1.0} leaves it at its
10956 full strength. Default is "1".
10959 Set when the expressions for brightness, contrast, saturation and
10960 gamma expressions are evaluated.
10962 It accepts the following values:
10965 only evaluate expressions once during the filter initialization or
10966 when a command is processed
10969 evaluate expressions for each incoming frame
10972 Default value is @samp{init}.
10975 The expressions accept the following parameters:
10978 frame count of the input frame starting from 0
10981 byte position of the corresponding packet in the input file, NAN if
10985 frame rate of the input video, NAN if the input frame rate is unknown
10988 timestamp expressed in seconds, NAN if the input timestamp is unknown
10991 @subsection Commands
10992 The filter supports the following commands:
10996 Set the contrast expression.
10999 Set the brightness expression.
11002 Set the saturation expression.
11005 Set the gamma expression.
11008 Set the gamma_r expression.
11011 Set gamma_g expression.
11014 Set gamma_b expression.
11017 Set gamma_weight expression.
11019 The command accepts the same syntax of the corresponding option.
11021 If the specified expression is not valid, it is kept at its current
11028 Apply erosion effect to the video.
11030 This filter replaces the pixel by the local(3x3) minimum.
11032 It accepts the following options:
11039 Limit the maximum change for each plane, default is 65535.
11040 If 0, plane will remain unchanged.
11043 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
11046 Flags to local 3x3 coordinates maps like this:
11053 @subsection Commands
11055 This filter supports the all above options as @ref{commands}.
11057 @section extractplanes
11059 Extract color channel components from input video stream into
11060 separate grayscale video streams.
11062 The filter accepts the following option:
11066 Set plane(s) to extract.
11068 Available values for planes are:
11079 Choosing planes not available in the input will result in an error.
11080 That means you cannot select @code{r}, @code{g}, @code{b} planes
11081 with @code{y}, @code{u}, @code{v} planes at same time.
11084 @subsection Examples
11088 Extract luma, u and v color channel component from input video frame
11089 into 3 grayscale outputs:
11091 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
11097 Apply a fade-in/out effect to the input video.
11099 It accepts the following parameters:
11103 The effect type can be either "in" for a fade-in, or "out" for a fade-out
11105 Default is @code{in}.
11107 @item start_frame, s
11108 Specify the number of the frame to start applying the fade
11109 effect at. Default is 0.
11112 The number of frames that the fade effect lasts. At the end of the
11113 fade-in effect, the output video will have the same intensity as the input video.
11114 At the end of the fade-out transition, the output video will be filled with the
11115 selected @option{color}.
11119 If set to 1, fade only alpha channel, if one exists on the input.
11120 Default value is 0.
11122 @item start_time, st
11123 Specify the timestamp (in seconds) of the frame to start to apply the fade
11124 effect. If both start_frame and start_time are specified, the fade will start at
11125 whichever comes last. Default is 0.
11128 The number of seconds for which the fade effect has to last. At the end of the
11129 fade-in effect the output video will have the same intensity as the input video,
11130 at the end of the fade-out transition the output video will be filled with the
11131 selected @option{color}.
11132 If both duration and nb_frames are specified, duration is used. Default is 0
11133 (nb_frames is used by default).
11136 Specify the color of the fade. Default is "black".
11139 @subsection Examples
11143 Fade in the first 30 frames of video:
11148 The command above is equivalent to:
11154 Fade out the last 45 frames of a 200-frame video:
11157 fade=type=out:start_frame=155:nb_frames=45
11161 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
11163 fade=in:0:25, fade=out:975:25
11167 Make the first 5 frames yellow, then fade in from frame 5-24:
11169 fade=in:5:20:color=yellow
11173 Fade in alpha over first 25 frames of video:
11175 fade=in:0:25:alpha=1
11179 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
11181 fade=t=in:st=5.5:d=0.5
11187 Denoise frames using 3D FFT (frequency domain filtering).
11189 The filter accepts the following options:
11193 Set the noise sigma constant. This sets denoising strength.
11194 Default value is 1. Allowed range is from 0 to 30.
11195 Using very high sigma with low overlap may give blocking artifacts.
11198 Set amount of denoising. By default all detected noise is reduced.
11199 Default value is 1. Allowed range is from 0 to 1.
11202 Set size of block, Default is 4, can be 3, 4, 5 or 6.
11203 Actual size of block in pixels is 2 to power of @var{block}, so by default
11204 block size in pixels is 2^4 which is 16.
11207 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
11210 Set number of previous frames to use for denoising. By default is set to 0.
11213 Set number of next frames to to use for denoising. By default is set to 0.
11216 Set planes which will be filtered, by default are all available filtered
11221 Apply arbitrary expressions to samples in frequency domain
11225 Adjust the dc value (gain) of the luma plane of the image. The filter
11226 accepts an integer value in range @code{0} to @code{1000}. The default
11227 value is set to @code{0}.
11230 Adjust the dc value (gain) of the 1st chroma plane of the image. The
11231 filter accepts an integer value in range @code{0} to @code{1000}. The
11232 default value is set to @code{0}.
11235 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
11236 filter accepts an integer value in range @code{0} to @code{1000}. The
11237 default value is set to @code{0}.
11240 Set the frequency domain weight expression for the luma plane.
11243 Set the frequency domain weight expression for the 1st chroma plane.
11246 Set the frequency domain weight expression for the 2nd chroma plane.
11249 Set when the expressions are evaluated.
11251 It accepts the following values:
11254 Only evaluate expressions once during the filter initialization.
11257 Evaluate expressions for each incoming frame.
11260 Default value is @samp{init}.
11262 The filter accepts the following variables:
11265 The coordinates of the current sample.
11269 The width and height of the image.
11272 The number of input frame, starting from 0.
11275 @subsection Examples
11281 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11287 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11293 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11299 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11306 Extract a single field from an interlaced image using stride
11307 arithmetic to avoid wasting CPU time. The output frames are marked as
11310 The filter accepts the following options:
11314 Specify whether to extract the top (if the value is @code{0} or
11315 @code{top}) or the bottom field (if the value is @code{1} or
11321 Create new frames by copying the top and bottom fields from surrounding frames
11322 supplied as numbers by the hint file.
11326 Set file containing hints: absolute/relative frame numbers.
11328 There must be one line for each frame in a clip. Each line must contain two
11329 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11330 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11331 is current frame number for @code{absolute} mode or out of [-1, 1] range
11332 for @code{relative} mode. First number tells from which frame to pick up top
11333 field and second number tells from which frame to pick up bottom field.
11335 If optionally followed by @code{+} output frame will be marked as interlaced,
11336 else if followed by @code{-} output frame will be marked as progressive, else
11337 it will be marked same as input frame.
11338 If optionally followed by @code{t} output frame will use only top field, or in
11339 case of @code{b} it will use only bottom field.
11340 If line starts with @code{#} or @code{;} that line is skipped.
11343 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11346 Example of first several lines of @code{hint} file for @code{relative} mode:
11348 0,0 - # first frame
11349 1,0 - # second frame, use third's frame top field and second's frame bottom field
11350 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11365 @section fieldmatch
11367 Field matching filter for inverse telecine. It is meant to reconstruct the
11368 progressive frames from a telecined stream. The filter does not drop duplicated
11369 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11370 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11372 The separation of the field matching and the decimation is notably motivated by
11373 the possibility of inserting a de-interlacing filter fallback between the two.
11374 If the source has mixed telecined and real interlaced content,
11375 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11376 But these remaining combed frames will be marked as interlaced, and thus can be
11377 de-interlaced by a later filter such as @ref{yadif} before decimation.
11379 In addition to the various configuration options, @code{fieldmatch} can take an
11380 optional second stream, activated through the @option{ppsrc} option. If
11381 enabled, the frames reconstruction will be based on the fields and frames from
11382 this second stream. This allows the first input to be pre-processed in order to
11383 help the various algorithms of the filter, while keeping the output lossless
11384 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11385 or brightness/contrast adjustments can help.
11387 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11388 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11389 which @code{fieldmatch} is based on. While the semantic and usage are very
11390 close, some behaviour and options names can differ.
11392 The @ref{decimate} filter currently only works for constant frame rate input.
11393 If your input has mixed telecined (30fps) and progressive content with a lower
11394 framerate like 24fps use the following filterchain to produce the necessary cfr
11395 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11397 The filter accepts the following options:
11401 Specify the assumed field order of the input stream. Available values are:
11405 Auto detect parity (use FFmpeg's internal parity value).
11407 Assume bottom field first.
11409 Assume top field first.
11412 Note that it is sometimes recommended not to trust the parity announced by the
11415 Default value is @var{auto}.
11418 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11419 sense that it won't risk creating jerkiness due to duplicate frames when
11420 possible, but if there are bad edits or blended fields it will end up
11421 outputting combed frames when a good match might actually exist. On the other
11422 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11423 but will almost always find a good frame if there is one. The other values are
11424 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11425 jerkiness and creating duplicate frames versus finding good matches in sections
11426 with bad edits, orphaned fields, blended fields, etc.
11428 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11430 Available values are:
11434 2-way matching (p/c)
11436 2-way matching, and trying 3rd match if still combed (p/c + n)
11438 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11440 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11441 still combed (p/c + n + u/b)
11443 3-way matching (p/c/n)
11445 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11446 detected as combed (p/c/n + u/b)
11449 The parenthesis at the end indicate the matches that would be used for that
11450 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11453 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11456 Default value is @var{pc_n}.
11459 Mark the main input stream as a pre-processed input, and enable the secondary
11460 input stream as the clean source to pick the fields from. See the filter
11461 introduction for more details. It is similar to the @option{clip2} feature from
11464 Default value is @code{0} (disabled).
11467 Set the field to match from. It is recommended to set this to the same value as
11468 @option{order} unless you experience matching failures with that setting. In
11469 certain circumstances changing the field that is used to match from can have a
11470 large impact on matching performance. Available values are:
11474 Automatic (same value as @option{order}).
11476 Match from the bottom field.
11478 Match from the top field.
11481 Default value is @var{auto}.
11484 Set whether or not chroma is included during the match comparisons. In most
11485 cases it is recommended to leave this enabled. You should set this to @code{0}
11486 only if your clip has bad chroma problems such as heavy rainbowing or other
11487 artifacts. Setting this to @code{0} could also be used to speed things up at
11488 the cost of some accuracy.
11490 Default value is @code{1}.
11494 These define an exclusion band which excludes the lines between @option{y0} and
11495 @option{y1} from being included in the field matching decision. An exclusion
11496 band can be used to ignore subtitles, a logo, or other things that may
11497 interfere with the matching. @option{y0} sets the starting scan line and
11498 @option{y1} sets the ending line; all lines in between @option{y0} and
11499 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11500 @option{y0} and @option{y1} to the same value will disable the feature.
11501 @option{y0} and @option{y1} defaults to @code{0}.
11504 Set the scene change detection threshold as a percentage of maximum change on
11505 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11506 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11507 @option{scthresh} is @code{[0.0, 100.0]}.
11509 Default value is @code{12.0}.
11512 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11513 account the combed scores of matches when deciding what match to use as the
11514 final match. Available values are:
11518 No final matching based on combed scores.
11520 Combed scores are only used when a scene change is detected.
11522 Use combed scores all the time.
11525 Default is @var{sc}.
11528 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11529 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11530 Available values are:
11534 No forced calculation.
11536 Force p/c/n calculations.
11538 Force p/c/n/u/b calculations.
11541 Default value is @var{none}.
11544 This is the area combing threshold used for combed frame detection. This
11545 essentially controls how "strong" or "visible" combing must be to be detected.
11546 Larger values mean combing must be more visible and smaller values mean combing
11547 can be less visible or strong and still be detected. Valid settings are from
11548 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11549 be detected as combed). This is basically a pixel difference value. A good
11550 range is @code{[8, 12]}.
11552 Default value is @code{9}.
11555 Sets whether or not chroma is considered in the combed frame decision. Only
11556 disable this if your source has chroma problems (rainbowing, etc.) that are
11557 causing problems for the combed frame detection with chroma enabled. Actually,
11558 using @option{chroma}=@var{0} is usually more reliable, except for the case
11559 where there is chroma only combing in the source.
11561 Default value is @code{0}.
11565 Respectively set the x-axis and y-axis size of the window used during combed
11566 frame detection. This has to do with the size of the area in which
11567 @option{combpel} pixels are required to be detected as combed for a frame to be
11568 declared combed. See the @option{combpel} parameter description for more info.
11569 Possible values are any number that is a power of 2 starting at 4 and going up
11572 Default value is @code{16}.
11575 The number of combed pixels inside any of the @option{blocky} by
11576 @option{blockx} size blocks on the frame for the frame to be detected as
11577 combed. While @option{cthresh} controls how "visible" the combing must be, this
11578 setting controls "how much" combing there must be in any localized area (a
11579 window defined by the @option{blockx} and @option{blocky} settings) on the
11580 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11581 which point no frames will ever be detected as combed). This setting is known
11582 as @option{MI} in TFM/VFM vocabulary.
11584 Default value is @code{80}.
11587 @anchor{p/c/n/u/b meaning}
11588 @subsection p/c/n/u/b meaning
11590 @subsubsection p/c/n
11592 We assume the following telecined stream:
11595 Top fields: 1 2 2 3 4
11596 Bottom fields: 1 2 3 4 4
11599 The numbers correspond to the progressive frame the fields relate to. Here, the
11600 first two frames are progressive, the 3rd and 4th are combed, and so on.
11602 When @code{fieldmatch} is configured to run a matching from bottom
11603 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11608 B 1 2 3 4 4 <-- matching reference
11617 As a result of the field matching, we can see that some frames get duplicated.
11618 To perform a complete inverse telecine, you need to rely on a decimation filter
11619 after this operation. See for instance the @ref{decimate} filter.
11621 The same operation now matching from top fields (@option{field}=@var{top})
11626 T 1 2 2 3 4 <-- matching reference
11636 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11637 basically, they refer to the frame and field of the opposite parity:
11640 @item @var{p} matches the field of the opposite parity in the previous frame
11641 @item @var{c} matches the field of the opposite parity in the current frame
11642 @item @var{n} matches the field of the opposite parity in the next frame
11647 The @var{u} and @var{b} matching are a bit special in the sense that they match
11648 from the opposite parity flag. In the following examples, we assume that we are
11649 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11650 'x' is placed above and below each matched fields.
11652 With bottom matching (@option{field}=@var{bottom}):
11657 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11658 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11666 With top matching (@option{field}=@var{top}):
11671 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11672 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11680 @subsection Examples
11682 Simple IVTC of a top field first telecined stream:
11684 fieldmatch=order=tff:combmatch=none, decimate
11687 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11689 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11692 @section fieldorder
11694 Transform the field order of the input video.
11696 It accepts the following parameters:
11701 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11702 for bottom field first.
11705 The default value is @samp{tff}.
11707 The transformation is done by shifting the picture content up or down
11708 by one line, and filling the remaining line with appropriate picture content.
11709 This method is consistent with most broadcast field order converters.
11711 If the input video is not flagged as being interlaced, or it is already
11712 flagged as being of the required output field order, then this filter does
11713 not alter the incoming video.
11715 It is very useful when converting to or from PAL DV material,
11716 which is bottom field first.
11720 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11723 @section fifo, afifo
11725 Buffer input images and send them when they are requested.
11727 It is mainly useful when auto-inserted by the libavfilter
11730 It does not take parameters.
11732 @section fillborders
11734 Fill borders of the input video, without changing video stream dimensions.
11735 Sometimes video can have garbage at the four edges and you may not want to
11736 crop video input to keep size multiple of some number.
11738 This filter accepts the following options:
11742 Number of pixels to fill from left border.
11745 Number of pixels to fill from right border.
11748 Number of pixels to fill from top border.
11751 Number of pixels to fill from bottom border.
11756 It accepts the following values:
11759 fill pixels using outermost pixels
11762 fill pixels using mirroring (half sample symmetric)
11765 fill pixels with constant value
11768 fill pixels using reflecting (whole sample symmetric)
11771 fill pixels using wrapping
11774 fade pixels to constant value
11777 Default is @var{smear}.
11780 Set color for pixels in fixed or fade mode. Default is @var{black}.
11783 @subsection Commands
11784 This filter supports same @ref{commands} as options.
11785 The command accepts the same syntax of the corresponding option.
11787 If the specified expression is not valid, it is kept at its current
11792 Find a rectangular object
11794 It accepts the following options:
11798 Filepath of the object image, needs to be in gray8.
11801 Detection threshold, default is 0.5.
11804 Number of mipmaps, default is 3.
11806 @item xmin, ymin, xmax, ymax
11807 Specifies the rectangle in which to search.
11810 @subsection Examples
11814 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11816 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11822 Flood area with values of same pixel components with another values.
11824 It accepts the following options:
11827 Set pixel x coordinate.
11830 Set pixel y coordinate.
11833 Set source #0 component value.
11836 Set source #1 component value.
11839 Set source #2 component value.
11842 Set source #3 component value.
11845 Set destination #0 component value.
11848 Set destination #1 component value.
11851 Set destination #2 component value.
11854 Set destination #3 component value.
11860 Convert the input video to one of the specified pixel formats.
11861 Libavfilter will try to pick one that is suitable as input to
11864 It accepts the following parameters:
11868 A '|'-separated list of pixel format names, such as
11869 "pix_fmts=yuv420p|monow|rgb24".
11873 @subsection Examples
11877 Convert the input video to the @var{yuv420p} format
11879 format=pix_fmts=yuv420p
11882 Convert the input video to any of the formats in the list
11884 format=pix_fmts=yuv420p|yuv444p|yuv410p
11891 Convert the video to specified constant frame rate by duplicating or dropping
11892 frames as necessary.
11894 It accepts the following parameters:
11898 The desired output frame rate. The default is @code{25}.
11901 Assume the first PTS should be the given value, in seconds. This allows for
11902 padding/trimming at the start of stream. By default, no assumption is made
11903 about the first frame's expected PTS, so no padding or trimming is done.
11904 For example, this could be set to 0 to pad the beginning with duplicates of
11905 the first frame if a video stream starts after the audio stream or to trim any
11906 frames with a negative PTS.
11909 Timestamp (PTS) rounding method.
11911 Possible values are:
11918 round towards -infinity
11920 round towards +infinity
11924 The default is @code{near}.
11927 Action performed when reading the last frame.
11929 Possible values are:
11932 Use same timestamp rounding method as used for other frames.
11934 Pass through last frame if input duration has not been reached yet.
11936 The default is @code{round}.
11940 Alternatively, the options can be specified as a flat string:
11941 @var{fps}[:@var{start_time}[:@var{round}]].
11943 See also the @ref{setpts} filter.
11945 @subsection Examples
11949 A typical usage in order to set the fps to 25:
11955 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11957 fps=fps=film:round=near
11963 Pack two different video streams into a stereoscopic video, setting proper
11964 metadata on supported codecs. The two views should have the same size and
11965 framerate and processing will stop when the shorter video ends. Please note
11966 that you may conveniently adjust view properties with the @ref{scale} and
11969 It accepts the following parameters:
11973 The desired packing format. Supported values are:
11978 The views are next to each other (default).
11981 The views are on top of each other.
11984 The views are packed by line.
11987 The views are packed by column.
11990 The views are temporally interleaved.
11999 # Convert left and right views into a frame-sequential video
12000 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
12002 # Convert views into a side-by-side video with the same output resolution as the input
12003 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
12008 Change the frame rate by interpolating new video output frames from the source
12011 This filter is not designed to function correctly with interlaced media. If
12012 you wish to change the frame rate of interlaced media then you are required
12013 to deinterlace before this filter and re-interlace after this filter.
12015 A description of the accepted options follows.
12019 Specify the output frames per second. This option can also be specified
12020 as a value alone. The default is @code{50}.
12023 Specify the start of a range where the output frame will be created as a
12024 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12025 the default is @code{15}.
12028 Specify the end of a range where the output frame will be created as a
12029 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12030 the default is @code{240}.
12033 Specify the level at which a scene change is detected as a value between
12034 0 and 100 to indicate a new scene; a low value reflects a low
12035 probability for the current frame to introduce a new scene, while a higher
12036 value means the current frame is more likely to be one.
12037 The default is @code{8.2}.
12040 Specify flags influencing the filter process.
12042 Available value for @var{flags} is:
12045 @item scene_change_detect, scd
12046 Enable scene change detection using the value of the option @var{scene}.
12047 This flag is enabled by default.
12053 Select one frame every N-th frame.
12055 This filter accepts the following option:
12058 Select frame after every @code{step} frames.
12059 Allowed values are positive integers higher than 0. Default value is @code{1}.
12062 @section freezedetect
12064 Detect frozen video.
12066 This filter logs a message and sets frame metadata when it detects that the
12067 input video has no significant change in content during a specified duration.
12068 Video freeze detection calculates the mean average absolute difference of all
12069 the components of video frames and compares it to a noise floor.
12071 The printed times and duration are expressed in seconds. The
12072 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
12073 whose timestamp equals or exceeds the detection duration and it contains the
12074 timestamp of the first frame of the freeze. The
12075 @code{lavfi.freezedetect.freeze_duration} and
12076 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
12079 The filter accepts the following options:
12083 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
12084 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
12088 Set freeze duration until notification (default is 2 seconds).
12091 @section freezeframes
12093 Freeze video frames.
12095 This filter freezes video frames using frame from 2nd input.
12097 The filter accepts the following options:
12101 Set number of first frame from which to start freeze.
12104 Set number of last frame from which to end freeze.
12107 Set number of frame from 2nd input which will be used instead of replaced frames.
12113 Apply a frei0r effect to the input video.
12115 To enable the compilation of this filter, you need to install the frei0r
12116 header and configure FFmpeg with @code{--enable-frei0r}.
12118 It accepts the following parameters:
12123 The name of the frei0r effect to load. If the environment variable
12124 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
12125 directories specified by the colon-separated list in @env{FREI0R_PATH}.
12126 Otherwise, the standard frei0r paths are searched, in this order:
12127 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
12128 @file{/usr/lib/frei0r-1/}.
12130 @item filter_params
12131 A '|'-separated list of parameters to pass to the frei0r effect.
12135 A frei0r effect parameter can be a boolean (its value is either
12136 "y" or "n"), a double, a color (specified as
12137 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
12138 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
12139 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
12140 a position (specified as @var{X}/@var{Y}, where
12141 @var{X} and @var{Y} are floating point numbers) and/or a string.
12143 The number and types of parameters depend on the loaded effect. If an
12144 effect parameter is not specified, the default value is set.
12146 @subsection Examples
12150 Apply the distort0r effect, setting the first two double parameters:
12152 frei0r=filter_name=distort0r:filter_params=0.5|0.01
12156 Apply the colordistance effect, taking a color as the first parameter:
12158 frei0r=colordistance:0.2/0.3/0.4
12159 frei0r=colordistance:violet
12160 frei0r=colordistance:0x112233
12164 Apply the perspective effect, specifying the top left and top right image
12167 frei0r=perspective:0.2/0.2|0.8/0.2
12171 For more information, see
12172 @url{http://frei0r.dyne.org}
12174 @subsection Commands
12176 This filter supports the @option{filter_params} option as @ref{commands}.
12180 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
12182 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
12183 processing filter, one of them is performed once per block, not per pixel.
12184 This allows for much higher speed.
12186 The filter accepts the following options:
12190 Set quality. This option defines the number of levels for averaging. It accepts
12191 an integer in the range 4-5. Default value is @code{4}.
12194 Force a constant quantization parameter. It accepts an integer in range 0-63.
12195 If not set, the filter will use the QP from the video stream (if available).
12198 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
12199 more details but also more artifacts, while higher values make the image smoother
12200 but also blurrier. Default value is @code{0} − PSNR optimal.
12202 @item use_bframe_qp
12203 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
12204 option may cause flicker since the B-Frames have often larger QP. Default is
12205 @code{0} (not enabled).
12211 Apply Gaussian blur filter.
12213 The filter accepts the following options:
12217 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
12220 Set number of steps for Gaussian approximation. Default is @code{1}.
12223 Set which planes to filter. By default all planes are filtered.
12226 Set vertical sigma, if negative it will be same as @code{sigma}.
12227 Default is @code{-1}.
12230 @subsection Commands
12231 This filter supports same commands as options.
12232 The command accepts the same syntax of the corresponding option.
12234 If the specified expression is not valid, it is kept at its current
12239 Apply generic equation to each pixel.
12241 The filter accepts the following options:
12244 @item lum_expr, lum
12245 Set the luminance expression.
12247 Set the chrominance blue expression.
12249 Set the chrominance red expression.
12250 @item alpha_expr, a
12251 Set the alpha expression.
12253 Set the red expression.
12254 @item green_expr, g
12255 Set the green expression.
12257 Set the blue expression.
12260 The colorspace is selected according to the specified options. If one
12261 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
12262 options is specified, the filter will automatically select a YCbCr
12263 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
12264 @option{blue_expr} options is specified, it will select an RGB
12267 If one of the chrominance expression is not defined, it falls back on the other
12268 one. If no alpha expression is specified it will evaluate to opaque value.
12269 If none of chrominance expressions are specified, they will evaluate
12270 to the luminance expression.
12272 The expressions can use the following variables and functions:
12276 The sequential number of the filtered frame, starting from @code{0}.
12280 The coordinates of the current sample.
12284 The width and height of the image.
12288 Width and height scale depending on the currently filtered plane. It is the
12289 ratio between the corresponding luma plane number of pixels and the current
12290 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
12291 @code{0.5,0.5} for chroma planes.
12294 Time of the current frame, expressed in seconds.
12297 Return the value of the pixel at location (@var{x},@var{y}) of the current
12301 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12305 Return the value of the pixel at location (@var{x},@var{y}) of the
12306 blue-difference chroma plane. Return 0 if there is no such plane.
12309 Return the value of the pixel at location (@var{x},@var{y}) of the
12310 red-difference chroma plane. Return 0 if there is no such plane.
12315 Return the value of the pixel at location (@var{x},@var{y}) of the
12316 red/green/blue component. Return 0 if there is no such component.
12319 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12320 plane. Return 0 if there is no such plane.
12322 @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)
12323 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12324 sums of samples within a rectangle. See the functions without the sum postfix.
12326 @item interpolation
12327 Set one of interpolation methods:
12332 Default is bilinear.
12335 For functions, if @var{x} and @var{y} are outside the area, the value will be
12336 automatically clipped to the closer edge.
12338 Please note that this filter can use multiple threads in which case each slice
12339 will have its own expression state. If you want to use only a single expression
12340 state because your expressions depend on previous state then you should limit
12341 the number of filter threads to 1.
12343 @subsection Examples
12347 Flip the image horizontally:
12353 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12354 wavelength of 100 pixels:
12356 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12360 Generate a fancy enigmatic moving light:
12362 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
12366 Generate a quick emboss effect:
12368 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12372 Modify RGB components depending on pixel position:
12374 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12378 Create a radial gradient that is the same size as the input (also see
12379 the @ref{vignette} filter):
12381 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12387 Fix the banding artifacts that are sometimes introduced into nearly flat
12388 regions by truncation to 8-bit color depth.
12389 Interpolate the gradients that should go where the bands are, and
12392 It is designed for playback only. Do not use it prior to
12393 lossy compression, because compression tends to lose the dither and
12394 bring back the bands.
12396 It accepts the following parameters:
12401 The maximum amount by which the filter will change any one pixel. This is also
12402 the threshold for detecting nearly flat regions. Acceptable values range from
12403 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12407 The neighborhood to fit the gradient to. A larger radius makes for smoother
12408 gradients, but also prevents the filter from modifying the pixels near detailed
12409 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12410 values will be clipped to the valid range.
12414 Alternatively, the options can be specified as a flat string:
12415 @var{strength}[:@var{radius}]
12417 @subsection Examples
12421 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12427 Specify radius, omitting the strength (which will fall-back to the default
12435 @anchor{graphmonitor}
12436 @section graphmonitor
12437 Show various filtergraph stats.
12439 With this filter one can debug complete filtergraph.
12440 Especially issues with links filling with queued frames.
12442 The filter accepts the following options:
12446 Set video output size. Default is @var{hd720}.
12449 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12452 Set output mode, can be @var{fulll} or @var{compact}.
12453 In @var{compact} mode only filters with some queued frames have displayed stats.
12456 Set flags which enable which stats are shown in video.
12458 Available values for flags are:
12461 Display number of queued frames in each link.
12463 @item frame_count_in
12464 Display number of frames taken from filter.
12466 @item frame_count_out
12467 Display number of frames given out from filter.
12470 Display current filtered frame pts.
12473 Display current filtered frame time.
12476 Display time base for filter link.
12479 Display used format for filter link.
12482 Display video size or number of audio channels in case of audio used by filter link.
12485 Display video frame rate or sample rate in case of audio used by filter link.
12488 Display link output status.
12492 Set upper limit for video rate of output stream, Default value is @var{25}.
12493 This guarantee that output video frame rate will not be higher than this value.
12497 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12498 and corrects the scene colors accordingly.
12500 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12502 The filter accepts the following options:
12506 The order of differentiation to be applied on the scene. Must be chosen in the range
12507 [0,2] and default value is 1.
12510 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12511 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12512 max value instead of calculating Minkowski distance.
12515 The standard deviation of Gaussian blur to be applied on the scene. Must be
12516 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12517 can't be equal to 0 if @var{difford} is greater than 0.
12520 @subsection Examples
12526 greyedge=difford=1:minknorm=5:sigma=2
12532 greyedge=difford=1:minknorm=0:sigma=2
12540 Apply a Hald CLUT to a video stream.
12542 First input is the video stream to process, and second one is the Hald CLUT.
12543 The Hald CLUT input can be a simple picture or a complete video stream.
12545 The filter accepts the following options:
12549 Force termination when the shortest input terminates. Default is @code{0}.
12551 Continue applying the last CLUT after the end of the stream. A value of
12552 @code{0} disable the filter after the last frame of the CLUT is reached.
12553 Default is @code{1}.
12556 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12557 filters share the same internals).
12559 This filter also supports the @ref{framesync} options.
12561 More information about the Hald CLUT can be found on Eskil Steenberg's website
12562 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12564 @subsection Workflow examples
12566 @subsubsection Hald CLUT video stream
12568 Generate an identity Hald CLUT stream altered with various effects:
12570 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
12573 Note: make sure you use a lossless codec.
12575 Then use it with @code{haldclut} to apply it on some random stream:
12577 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12580 The Hald CLUT will be applied to the 10 first seconds (duration of
12581 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12582 to the remaining frames of the @code{mandelbrot} stream.
12584 @subsubsection Hald CLUT with preview
12586 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12587 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12588 biggest possible square starting at the top left of the picture. The remaining
12589 padding pixels (bottom or right) will be ignored. This area can be used to add
12590 a preview of the Hald CLUT.
12592 Typically, the following generated Hald CLUT will be supported by the
12593 @code{haldclut} filter:
12596 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12597 pad=iw+320 [padded_clut];
12598 smptebars=s=320x256, split [a][b];
12599 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12600 [main][b] overlay=W-320" -frames:v 1 clut.png
12603 It contains the original and a preview of the effect of the CLUT: SMPTE color
12604 bars are displayed on the right-top, and below the same color bars processed by
12607 Then, the effect of this Hald CLUT can be visualized with:
12609 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12614 Flip the input video horizontally.
12616 For example, to horizontally flip the input video with @command{ffmpeg}:
12618 ffmpeg -i in.avi -vf "hflip" out.avi
12622 This filter applies a global color histogram equalization on a
12625 It can be used to correct video that has a compressed range of pixel
12626 intensities. The filter redistributes the pixel intensities to
12627 equalize their distribution across the intensity range. It may be
12628 viewed as an "automatically adjusting contrast filter". This filter is
12629 useful only for correcting degraded or poorly captured source
12632 The filter accepts the following options:
12636 Determine the amount of equalization to be applied. As the strength
12637 is reduced, the distribution of pixel intensities more-and-more
12638 approaches that of the input frame. The value must be a float number
12639 in the range [0,1] and defaults to 0.200.
12642 Set the maximum intensity that can generated and scale the output
12643 values appropriately. The strength should be set as desired and then
12644 the intensity can be limited if needed to avoid washing-out. The value
12645 must be a float number in the range [0,1] and defaults to 0.210.
12648 Set the antibanding level. If enabled the filter will randomly vary
12649 the luminance of output pixels by a small amount to avoid banding of
12650 the histogram. Possible values are @code{none}, @code{weak} or
12651 @code{strong}. It defaults to @code{none}.
12657 Compute and draw a color distribution histogram for the input video.
12659 The computed histogram is a representation of the color component
12660 distribution in an image.
12662 Standard histogram displays the color components distribution in an image.
12663 Displays color graph for each color component. Shows distribution of
12664 the Y, U, V, A or R, G, B components, depending on input format, in the
12665 current frame. Below each graph a color component scale meter is shown.
12667 The filter accepts the following options:
12671 Set height of level. Default value is @code{200}.
12672 Allowed range is [50, 2048].
12675 Set height of color scale. Default value is @code{12}.
12676 Allowed range is [0, 40].
12680 It accepts the following values:
12683 Per color component graphs are placed below each other.
12686 Per color component graphs are placed side by side.
12689 Presents information identical to that in the @code{parade}, except
12690 that the graphs representing color components are superimposed directly
12693 Default is @code{stack}.
12696 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12697 Default is @code{linear}.
12700 Set what color components to display.
12701 Default is @code{7}.
12704 Set foreground opacity. Default is @code{0.7}.
12707 Set background opacity. Default is @code{0.5}.
12710 @subsection Examples
12715 Calculate and draw histogram:
12717 ffplay -i input -vf histogram
12725 This is a high precision/quality 3d denoise filter. It aims to reduce
12726 image noise, producing smooth images and making still images really
12727 still. It should enhance compressibility.
12729 It accepts the following optional parameters:
12733 A non-negative floating point number which specifies spatial luma strength.
12734 It defaults to 4.0.
12736 @item chroma_spatial
12737 A non-negative floating point number which specifies spatial chroma strength.
12738 It defaults to 3.0*@var{luma_spatial}/4.0.
12741 A floating point number which specifies luma temporal strength. It defaults to
12742 6.0*@var{luma_spatial}/4.0.
12745 A floating point number which specifies chroma temporal strength. It defaults to
12746 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12749 @subsection Commands
12750 This filter supports same @ref{commands} as options.
12751 The command accepts the same syntax of the corresponding option.
12753 If the specified expression is not valid, it is kept at its current
12756 @anchor{hwdownload}
12757 @section hwdownload
12759 Download hardware frames to system memory.
12761 The input must be in hardware frames, and the output a non-hardware format.
12762 Not all formats will be supported on the output - it may be necessary to insert
12763 an additional @option{format} filter immediately following in the graph to get
12764 the output in a supported format.
12768 Map hardware frames to system memory or to another device.
12770 This filter has several different modes of operation; which one is used depends
12771 on the input and output formats:
12774 Hardware frame input, normal frame output
12776 Map the input frames to system memory and pass them to the output. If the
12777 original hardware frame is later required (for example, after overlaying
12778 something else on part of it), the @option{hwmap} filter can be used again
12779 in the next mode to retrieve it.
12781 Normal frame input, hardware frame output
12783 If the input is actually a software-mapped hardware frame, then unmap it -
12784 that is, return the original hardware frame.
12786 Otherwise, a device must be provided. Create new hardware surfaces on that
12787 device for the output, then map them back to the software format at the input
12788 and give those frames to the preceding filter. This will then act like the
12789 @option{hwupload} filter, but may be able to avoid an additional copy when
12790 the input is already in a compatible format.
12792 Hardware frame input and output
12794 A device must be supplied for the output, either directly or with the
12795 @option{derive_device} option. The input and output devices must be of
12796 different types and compatible - the exact meaning of this is
12797 system-dependent, but typically it means that they must refer to the same
12798 underlying hardware context (for example, refer to the same graphics card).
12800 If the input frames were originally created on the output device, then unmap
12801 to retrieve the original frames.
12803 Otherwise, map the frames to the output device - create new hardware frames
12804 on the output corresponding to the frames on the input.
12807 The following additional parameters are accepted:
12811 Set the frame mapping mode. Some combination of:
12814 The mapped frame should be readable.
12816 The mapped frame should be writeable.
12818 The mapping will always overwrite the entire frame.
12820 This may improve performance in some cases, as the original contents of the
12821 frame need not be loaded.
12823 The mapping must not involve any copying.
12825 Indirect mappings to copies of frames are created in some cases where either
12826 direct mapping is not possible or it would have unexpected properties.
12827 Setting this flag ensures that the mapping is direct and will fail if that is
12830 Defaults to @var{read+write} if not specified.
12832 @item derive_device @var{type}
12833 Rather than using the device supplied at initialisation, instead derive a new
12834 device of type @var{type} from the device the input frames exist on.
12837 In a hardware to hardware mapping, map in reverse - create frames in the sink
12838 and map them back to the source. This may be necessary in some cases where
12839 a mapping in one direction is required but only the opposite direction is
12840 supported by the devices being used.
12842 This option is dangerous - it may break the preceding filter in undefined
12843 ways if there are any additional constraints on that filter's output.
12844 Do not use it without fully understanding the implications of its use.
12850 Upload system memory frames to hardware surfaces.
12852 The device to upload to must be supplied when the filter is initialised. If
12853 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12854 option or with the @option{derive_device} option. The input and output devices
12855 must be of different types and compatible - the exact meaning of this is
12856 system-dependent, but typically it means that they must refer to the same
12857 underlying hardware context (for example, refer to the same graphics card).
12859 The following additional parameters are accepted:
12862 @item derive_device @var{type}
12863 Rather than using the device supplied at initialisation, instead derive a new
12864 device of type @var{type} from the device the input frames exist on.
12867 @anchor{hwupload_cuda}
12868 @section hwupload_cuda
12870 Upload system memory frames to a CUDA device.
12872 It accepts the following optional parameters:
12876 The number of the CUDA device to use
12881 Apply a high-quality magnification filter designed for pixel art. This filter
12882 was originally created by Maxim Stepin.
12884 It accepts the following option:
12888 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12889 @code{hq3x} and @code{4} for @code{hq4x}.
12890 Default is @code{3}.
12894 Stack input videos horizontally.
12896 All streams must be of same pixel format and of same height.
12898 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12899 to create same output.
12901 The filter accepts the following option:
12905 Set number of input streams. Default is 2.
12908 If set to 1, force the output to terminate when the shortest input
12909 terminates. Default value is 0.
12914 Modify the hue and/or the saturation of the input.
12916 It accepts the following parameters:
12920 Specify the hue angle as a number of degrees. It accepts an expression,
12921 and defaults to "0".
12924 Specify the saturation in the [-10,10] range. It accepts an expression and
12928 Specify the hue angle as a number of radians. It accepts an
12929 expression, and defaults to "0".
12932 Specify the brightness in the [-10,10] range. It accepts an expression and
12936 @option{h} and @option{H} are mutually exclusive, and can't be
12937 specified at the same time.
12939 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12940 expressions containing the following constants:
12944 frame count of the input frame starting from 0
12947 presentation timestamp of the input frame expressed in time base units
12950 frame rate of the input video, NAN if the input frame rate is unknown
12953 timestamp expressed in seconds, NAN if the input timestamp is unknown
12956 time base of the input video
12959 @subsection Examples
12963 Set the hue to 90 degrees and the saturation to 1.0:
12969 Same command but expressing the hue in radians:
12975 Rotate hue and make the saturation swing between 0
12976 and 2 over a period of 1 second:
12978 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12982 Apply a 3 seconds saturation fade-in effect starting at 0:
12984 hue="s=min(t/3\,1)"
12987 The general fade-in expression can be written as:
12989 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12993 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12995 hue="s=max(0\, min(1\, (8-t)/3))"
12998 The general fade-out expression can be written as:
13000 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
13005 @subsection Commands
13007 This filter supports the following commands:
13013 Modify the hue and/or the saturation and/or brightness of the input video.
13014 The command accepts the same syntax of the corresponding option.
13016 If the specified expression is not valid, it is kept at its current
13020 @section hysteresis
13022 Grow first stream into second stream by connecting components.
13023 This makes it possible to build more robust edge masks.
13025 This filter accepts the following options:
13029 Set which planes will be processed as bitmap, unprocessed planes will be
13030 copied from first stream.
13031 By default value 0xf, all planes will be processed.
13034 Set threshold which is used in filtering. If pixel component value is higher than
13035 this value filter algorithm for connecting components is activated.
13036 By default value is 0.
13039 The @code{hysteresis} filter also supports the @ref{framesync} options.
13043 Detect video interlacing type.
13045 This filter tries to detect if the input frames are interlaced, progressive,
13046 top or bottom field first. It will also try to detect fields that are
13047 repeated between adjacent frames (a sign of telecine).
13049 Single frame detection considers only immediately adjacent frames when classifying each frame.
13050 Multiple frame detection incorporates the classification history of previous frames.
13052 The filter will log these metadata values:
13055 @item single.current_frame
13056 Detected type of current frame using single-frame detection. One of:
13057 ``tff'' (top field first), ``bff'' (bottom field first),
13058 ``progressive'', or ``undetermined''
13061 Cumulative number of frames detected as top field first using single-frame detection.
13064 Cumulative number of frames detected as top field first using multiple-frame detection.
13067 Cumulative number of frames detected as bottom field first using single-frame detection.
13069 @item multiple.current_frame
13070 Detected type of current frame using multiple-frame detection. One of:
13071 ``tff'' (top field first), ``bff'' (bottom field first),
13072 ``progressive'', or ``undetermined''
13075 Cumulative number of frames detected as bottom field first using multiple-frame detection.
13077 @item single.progressive
13078 Cumulative number of frames detected as progressive using single-frame detection.
13080 @item multiple.progressive
13081 Cumulative number of frames detected as progressive using multiple-frame detection.
13083 @item single.undetermined
13084 Cumulative number of frames that could not be classified using single-frame detection.
13086 @item multiple.undetermined
13087 Cumulative number of frames that could not be classified using multiple-frame detection.
13089 @item repeated.current_frame
13090 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
13092 @item repeated.neither
13093 Cumulative number of frames with no repeated field.
13096 Cumulative number of frames with the top field repeated from the previous frame's top field.
13098 @item repeated.bottom
13099 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
13102 The filter accepts the following options:
13106 Set interlacing threshold.
13108 Set progressive threshold.
13110 Threshold for repeated field detection.
13112 Number of frames after which a given frame's contribution to the
13113 statistics is halved (i.e., it contributes only 0.5 to its
13114 classification). The default of 0 means that all frames seen are given
13115 full weight of 1.0 forever.
13116 @item analyze_interlaced_flag
13117 When this is not 0 then idet will use the specified number of frames to determine
13118 if the interlaced flag is accurate, it will not count undetermined frames.
13119 If the flag is found to be accurate it will be used without any further
13120 computations, if it is found to be inaccurate it will be cleared without any
13121 further computations. This allows inserting the idet filter as a low computational
13122 method to clean up the interlaced flag
13127 Deinterleave or interleave fields.
13129 This filter allows one to process interlaced images fields without
13130 deinterlacing them. Deinterleaving splits the input frame into 2
13131 fields (so called half pictures). Odd lines are moved to the top
13132 half of the output image, even lines to the bottom half.
13133 You can process (filter) them independently and then re-interleave them.
13135 The filter accepts the following options:
13139 @item chroma_mode, c
13140 @item alpha_mode, a
13141 Available values for @var{luma_mode}, @var{chroma_mode} and
13142 @var{alpha_mode} are:
13148 @item deinterleave, d
13149 Deinterleave fields, placing one above the other.
13151 @item interleave, i
13152 Interleave fields. Reverse the effect of deinterleaving.
13154 Default value is @code{none}.
13156 @item luma_swap, ls
13157 @item chroma_swap, cs
13158 @item alpha_swap, as
13159 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
13162 @subsection Commands
13164 This filter supports the all above options as @ref{commands}.
13168 Apply inflate effect to the video.
13170 This filter replaces the pixel by the local(3x3) average by taking into account
13171 only values higher than the pixel.
13173 It accepts the following options:
13180 Limit the maximum change for each plane, default is 65535.
13181 If 0, plane will remain unchanged.
13184 @subsection Commands
13186 This filter supports the all above options as @ref{commands}.
13190 Simple interlacing filter from progressive contents. This interleaves upper (or
13191 lower) lines from odd frames with lower (or upper) lines from even frames,
13192 halving the frame rate and preserving image height.
13195 Original Original New Frame
13196 Frame 'j' Frame 'j+1' (tff)
13197 ========== =========== ==================
13198 Line 0 --------------------> Frame 'j' Line 0
13199 Line 1 Line 1 ----> Frame 'j+1' Line 1
13200 Line 2 ---------------------> Frame 'j' Line 2
13201 Line 3 Line 3 ----> Frame 'j+1' Line 3
13203 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
13206 It accepts the following optional parameters:
13210 This determines whether the interlaced frame is taken from the even
13211 (tff - default) or odd (bff) lines of the progressive frame.
13214 Vertical lowpass filter to avoid twitter interlacing and
13215 reduce moire patterns.
13219 Disable vertical lowpass filter
13222 Enable linear filter (default)
13225 Enable complex filter. This will slightly less reduce twitter and moire
13226 but better retain detail and subjective sharpness impression.
13233 Deinterlace input video by applying Donald Graft's adaptive kernel
13234 deinterling. Work on interlaced parts of a video to produce
13235 progressive frames.
13237 The description of the accepted parameters follows.
13241 Set the threshold which affects the filter's tolerance when
13242 determining if a pixel line must be processed. It must be an integer
13243 in the range [0,255] and defaults to 10. A value of 0 will result in
13244 applying the process on every pixels.
13247 Paint pixels exceeding the threshold value to white if set to 1.
13251 Set the fields order. Swap fields if set to 1, leave fields alone if
13255 Enable additional sharpening if set to 1. Default is 0.
13258 Enable twoway sharpening if set to 1. Default is 0.
13261 @subsection Examples
13265 Apply default values:
13267 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
13271 Enable additional sharpening:
13277 Paint processed pixels in white:
13285 Slowly update darker pixels.
13287 This filter makes short flashes of light appear longer.
13288 This filter accepts the following options:
13292 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13295 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13298 @section lenscorrection
13300 Correct radial lens distortion
13302 This filter can be used to correct for radial distortion as can result from the use
13303 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13304 one can use tools available for example as part of opencv or simply trial-and-error.
13305 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13306 and extract the k1 and k2 coefficients from the resulting matrix.
13308 Note that effectively the same filter is available in the open-source tools Krita and
13309 Digikam from the KDE project.
13311 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13312 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13313 brightness distribution, so you may want to use both filters together in certain
13314 cases, though you will have to take care of ordering, i.e. whether vignetting should
13315 be applied before or after lens correction.
13317 @subsection Options
13319 The filter accepts the following options:
13323 Relative x-coordinate of the focal point of the image, and thereby the center of the
13324 distortion. This value has a range [0,1] and is expressed as fractions of the image
13325 width. Default is 0.5.
13327 Relative y-coordinate of the focal point of the image, and thereby the center of the
13328 distortion. This value has a range [0,1] and is expressed as fractions of the image
13329 height. Default is 0.5.
13331 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13332 no correction. Default is 0.
13334 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13335 0 means no correction. Default is 0.
13338 The formula that generates the correction is:
13340 @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)
13342 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13343 distances from the focal point in the source and target images, respectively.
13347 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13349 The @code{lensfun} filter requires the camera make, camera model, and lens model
13350 to apply the lens correction. The filter will load the lensfun database and
13351 query it to find the corresponding camera and lens entries in the database. As
13352 long as these entries can be found with the given options, the filter can
13353 perform corrections on frames. Note that incomplete strings will result in the
13354 filter choosing the best match with the given options, and the filter will
13355 output the chosen camera and lens models (logged with level "info"). You must
13356 provide the make, camera model, and lens model as they are required.
13358 The filter accepts the following options:
13362 The make of the camera (for example, "Canon"). This option is required.
13365 The model of the camera (for example, "Canon EOS 100D"). This option is
13369 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13370 option is required.
13373 The type of correction to apply. The following values are valid options:
13377 Enables fixing lens vignetting.
13380 Enables fixing lens geometry. This is the default.
13383 Enables fixing chromatic aberrations.
13386 Enables fixing lens vignetting and lens geometry.
13389 Enables fixing lens vignetting and chromatic aberrations.
13392 Enables fixing both lens geometry and chromatic aberrations.
13395 Enables all possible corrections.
13399 The focal length of the image/video (zoom; expected constant for video). For
13400 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13401 range should be chosen when using that lens. Default 18.
13404 The aperture of the image/video (expected constant for video). Note that
13405 aperture is only used for vignetting correction. Default 3.5.
13407 @item focus_distance
13408 The focus distance of the image/video (expected constant for video). Note that
13409 focus distance is only used for vignetting and only slightly affects the
13410 vignetting correction process. If unknown, leave it at the default value (which
13414 The scale factor which is applied after transformation. After correction the
13415 video is no longer necessarily rectangular. This parameter controls how much of
13416 the resulting image is visible. The value 0 means that a value will be chosen
13417 automatically such that there is little or no unmapped area in the output
13418 image. 1.0 means that no additional scaling is done. Lower values may result
13419 in more of the corrected image being visible, while higher values may avoid
13420 unmapped areas in the output.
13422 @item target_geometry
13423 The target geometry of the output image/video. The following values are valid
13427 @item rectilinear (default)
13430 @item equirectangular
13431 @item fisheye_orthographic
13432 @item fisheye_stereographic
13433 @item fisheye_equisolid
13434 @item fisheye_thoby
13437 Apply the reverse of image correction (instead of correcting distortion, apply
13440 @item interpolation
13441 The type of interpolation used when correcting distortion. The following values
13446 @item linear (default)
13451 @subsection Examples
13455 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13456 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13460 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
13464 Apply the same as before, but only for the first 5 seconds of video.
13467 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
13474 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13475 score between two input videos.
13477 The obtained VMAF score is printed through the logging system.
13479 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13480 After installing the library it can be enabled using:
13481 @code{./configure --enable-libvmaf}.
13482 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13484 The filter has following options:
13488 Set the model path which is to be used for SVM.
13489 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13492 Set the file path to be used to store logs.
13495 Set the format of the log file (csv, json or xml).
13497 @item enable_transform
13498 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13499 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13500 Default value: @code{false}
13503 Invokes the phone model which will generate VMAF scores higher than in the
13504 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13505 Default value: @code{false}
13508 Enables computing psnr along with vmaf.
13509 Default value: @code{false}
13512 Enables computing ssim along with vmaf.
13513 Default value: @code{false}
13516 Enables computing ms_ssim along with vmaf.
13517 Default value: @code{false}
13520 Set the pool method to be used for computing vmaf.
13521 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13524 Set number of threads to be used when computing vmaf.
13525 Default value: @code{0}, which makes use of all available logical processors.
13528 Set interval for frame subsampling used when computing vmaf.
13529 Default value: @code{1}
13531 @item enable_conf_interval
13532 Enables confidence interval.
13533 Default value: @code{false}
13536 This filter also supports the @ref{framesync} options.
13538 @subsection Examples
13541 On the below examples the input file @file{main.mpg} being processed is
13542 compared with the reference file @file{ref.mpg}.
13545 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13549 Example with options:
13551 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13555 Example with options and different containers:
13557 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 -
13563 Limits the pixel components values to the specified range [min, max].
13565 The filter accepts the following options:
13569 Lower bound. Defaults to the lowest allowed value for the input.
13572 Upper bound. Defaults to the highest allowed value for the input.
13575 Specify which planes will be processed. Defaults to all available.
13578 @subsection Commands
13580 This filter supports the all above options as @ref{commands}.
13586 The filter accepts the following options:
13590 Set the number of loops. Setting this value to -1 will result in infinite loops.
13594 Set maximal size in number of frames. Default is 0.
13597 Set first frame of loop. Default is 0.
13600 @subsection Examples
13604 Loop single first frame infinitely:
13606 loop=loop=-1:size=1:start=0
13610 Loop single first frame 10 times:
13612 loop=loop=10:size=1:start=0
13616 Loop 10 first frames 5 times:
13618 loop=loop=5:size=10:start=0
13624 Apply a 1D LUT to an input video.
13626 The filter accepts the following options:
13630 Set the 1D LUT file name.
13632 Currently supported formats:
13641 Select interpolation mode.
13643 Available values are:
13647 Use values from the nearest defined point.
13649 Interpolate values using the linear interpolation.
13651 Interpolate values using the cosine interpolation.
13653 Interpolate values using the cubic interpolation.
13655 Interpolate values using the spline interpolation.
13662 Apply a 3D LUT to an input video.
13664 The filter accepts the following options:
13668 Set the 3D LUT file name.
13670 Currently supported formats:
13684 Select interpolation mode.
13686 Available values are:
13690 Use values from the nearest defined point.
13692 Interpolate values using the 8 points defining a cube.
13694 Interpolate values using a tetrahedron.
13700 Turn certain luma values into transparency.
13702 The filter accepts the following options:
13706 Set the luma which will be used as base for transparency.
13707 Default value is @code{0}.
13710 Set the range of luma values to be keyed out.
13711 Default value is @code{0.01}.
13714 Set the range of softness. Default value is @code{0}.
13715 Use this to control gradual transition from zero to full transparency.
13718 @subsection Commands
13719 This filter supports same @ref{commands} as options.
13720 The command accepts the same syntax of the corresponding option.
13722 If the specified expression is not valid, it is kept at its current
13725 @section lut, lutrgb, lutyuv
13727 Compute a look-up table for binding each pixel component input value
13728 to an output value, and apply it to the input video.
13730 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13731 to an RGB input video.
13733 These filters accept the following parameters:
13736 set first pixel component expression
13738 set second pixel component expression
13740 set third pixel component expression
13742 set fourth pixel component expression, corresponds to the alpha component
13745 set red component expression
13747 set green component expression
13749 set blue component expression
13751 alpha component expression
13754 set Y/luminance component expression
13756 set U/Cb component expression
13758 set V/Cr component expression
13761 Each of them specifies the expression to use for computing the lookup table for
13762 the corresponding pixel component values.
13764 The exact component associated to each of the @var{c*} options depends on the
13767 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13768 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13770 The expressions can contain the following constants and functions:
13775 The input width and height.
13778 The input value for the pixel component.
13781 The input value, clipped to the @var{minval}-@var{maxval} range.
13784 The maximum value for the pixel component.
13787 The minimum value for the pixel component.
13790 The negated value for the pixel component value, clipped to the
13791 @var{minval}-@var{maxval} range; it corresponds to the expression
13792 "maxval-clipval+minval".
13795 The computed value in @var{val}, clipped to the
13796 @var{minval}-@var{maxval} range.
13798 @item gammaval(gamma)
13799 The computed gamma correction value of the pixel component value,
13800 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13802 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13806 All expressions default to "val".
13808 @subsection Examples
13812 Negate input video:
13814 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13815 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13818 The above is the same as:
13820 lutrgb="r=negval:g=negval:b=negval"
13821 lutyuv="y=negval:u=negval:v=negval"
13831 Remove chroma components, turning the video into a graytone image:
13833 lutyuv="u=128:v=128"
13837 Apply a luma burning effect:
13843 Remove green and blue components:
13849 Set a constant alpha channel value on input:
13851 format=rgba,lutrgb=a="maxval-minval/2"
13855 Correct luminance gamma by a factor of 0.5:
13857 lutyuv=y=gammaval(0.5)
13861 Discard least significant bits of luma:
13863 lutyuv=y='bitand(val, 128+64+32)'
13867 Technicolor like effect:
13869 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13873 @section lut2, tlut2
13875 The @code{lut2} filter takes two input streams and outputs one
13878 The @code{tlut2} (time lut2) filter takes two consecutive frames
13879 from one single stream.
13881 This filter accepts the following parameters:
13884 set first pixel component expression
13886 set second pixel component expression
13888 set third pixel component expression
13890 set fourth pixel component expression, corresponds to the alpha component
13893 set output bit depth, only available for @code{lut2} filter. By default is 0,
13894 which means bit depth is automatically picked from first input format.
13897 The @code{lut2} filter also supports the @ref{framesync} options.
13899 Each of them specifies the expression to use for computing the lookup table for
13900 the corresponding pixel component values.
13902 The exact component associated to each of the @var{c*} options depends on the
13905 The expressions can contain the following constants:
13910 The input width and height.
13913 The first input value for the pixel component.
13916 The second input value for the pixel component.
13919 The first input video bit depth.
13922 The second input video bit depth.
13925 All expressions default to "x".
13927 @subsection Examples
13931 Highlight differences between two RGB video streams:
13933 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)'
13937 Highlight differences between two YUV video streams:
13939 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)'
13943 Show max difference between two video streams:
13945 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)))'
13949 @section maskedclamp
13951 Clamp the first input stream with the second input and third input stream.
13953 Returns the value of first stream to be between second input
13954 stream - @code{undershoot} and third input stream + @code{overshoot}.
13956 This filter accepts the following options:
13959 Default value is @code{0}.
13962 Default value is @code{0}.
13965 Set which planes will be processed as bitmap, unprocessed planes will be
13966 copied from first stream.
13967 By default value 0xf, all planes will be processed.
13972 Merge the second and third input stream into output stream using absolute differences
13973 between second input stream and first input stream and absolute difference between
13974 third input stream and first input stream. The picked value will be from second input
13975 stream if second absolute difference is greater than first one or from third input stream
13978 This filter accepts the following options:
13981 Set which planes will be processed as bitmap, unprocessed planes will be
13982 copied from first stream.
13983 By default value 0xf, all planes will be processed.
13986 @section maskedmerge
13988 Merge the first input stream with the second input stream using per pixel
13989 weights in the third input stream.
13991 A value of 0 in the third stream pixel component means that pixel component
13992 from first stream is returned unchanged, while maximum value (eg. 255 for
13993 8-bit videos) means that pixel component from second stream is returned
13994 unchanged. Intermediate values define the amount of merging between both
13995 input stream's pixel components.
13997 This filter accepts the following options:
14000 Set which planes will be processed as bitmap, unprocessed planes will be
14001 copied from first stream.
14002 By default value 0xf, all planes will be processed.
14007 Merge the second and third input stream into output stream using absolute differences
14008 between second input stream and first input stream and absolute difference between
14009 third input stream and first input stream. The picked value will be from second input
14010 stream if second absolute difference is less than first one or from third input stream
14013 This filter accepts the following options:
14016 Set which planes will be processed as bitmap, unprocessed planes will be
14017 copied from first stream.
14018 By default value 0xf, all planes will be processed.
14021 @section maskedthreshold
14022 Pick pixels comparing absolute difference of two video streams with fixed
14025 If absolute difference between pixel component of first and second video
14026 stream is equal or lower than user supplied threshold than pixel component
14027 from first video stream is picked, otherwise pixel component from second
14028 video stream is picked.
14030 This filter accepts the following options:
14033 Set threshold used when picking pixels from absolute difference from two input
14037 Set which planes will be processed as bitmap, unprocessed planes will be
14038 copied from second stream.
14039 By default value 0xf, all planes will be processed.
14043 Create mask from input video.
14045 For example it is useful to create motion masks after @code{tblend} filter.
14047 This filter accepts the following options:
14051 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
14054 Set high threshold. Any pixel component higher than this value will be set to max value
14055 allowed for current pixel format.
14058 Set planes to filter, by default all available planes are filtered.
14061 Fill all frame pixels with this value.
14064 Set max average pixel value for frame. If sum of all pixel components is higher that this
14065 average, output frame will be completely filled with value set by @var{fill} option.
14066 Typically useful for scene changes when used in combination with @code{tblend} filter.
14071 Apply motion-compensation deinterlacing.
14073 It needs one field per frame as input and must thus be used together
14074 with yadif=1/3 or equivalent.
14076 This filter accepts the following options:
14079 Set the deinterlacing mode.
14081 It accepts one of the following values:
14086 use iterative motion estimation
14088 like @samp{slow}, but use multiple reference frames.
14090 Default value is @samp{fast}.
14093 Set the picture field parity assumed for the input video. It must be
14094 one of the following values:
14098 assume top field first
14100 assume bottom field first
14103 Default value is @samp{bff}.
14106 Set per-block quantization parameter (QP) used by the internal
14109 Higher values should result in a smoother motion vector field but less
14110 optimal individual vectors. Default value is 1.
14115 Pick median pixel from certain rectangle defined by radius.
14117 This filter accepts the following options:
14121 Set horizontal radius size. Default value is @code{1}.
14122 Allowed range is integer from 1 to 127.
14125 Set which planes to process. Default is @code{15}, which is all available planes.
14128 Set vertical radius size. Default value is @code{0}.
14129 Allowed range is integer from 0 to 127.
14130 If it is 0, value will be picked from horizontal @code{radius} option.
14133 Set median percentile. Default value is @code{0.5}.
14134 Default value of @code{0.5} will pick always median values, while @code{0} will pick
14135 minimum values, and @code{1} maximum values.
14138 @subsection Commands
14139 This filter supports same @ref{commands} as options.
14140 The command accepts the same syntax of the corresponding option.
14142 If the specified expression is not valid, it is kept at its current
14145 @section mergeplanes
14147 Merge color channel components from several video streams.
14149 The filter accepts up to 4 input streams, and merge selected input
14150 planes to the output video.
14152 This filter accepts the following options:
14155 Set input to output plane mapping. Default is @code{0}.
14157 The mappings is specified as a bitmap. It should be specified as a
14158 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
14159 mapping for the first plane of the output stream. 'A' sets the number of
14160 the input stream to use (from 0 to 3), and 'a' the plane number of the
14161 corresponding input to use (from 0 to 3). The rest of the mappings is
14162 similar, 'Bb' describes the mapping for the output stream second
14163 plane, 'Cc' describes the mapping for the output stream third plane and
14164 'Dd' describes the mapping for the output stream fourth plane.
14167 Set output pixel format. Default is @code{yuva444p}.
14170 @subsection Examples
14174 Merge three gray video streams of same width and height into single video stream:
14176 [a0][a1][a2]mergeplanes=0x001020:yuv444p
14180 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
14182 [a0][a1]mergeplanes=0x00010210:yuva444p
14186 Swap Y and A plane in yuva444p stream:
14188 format=yuva444p,mergeplanes=0x03010200:yuva444p
14192 Swap U and V plane in yuv420p stream:
14194 format=yuv420p,mergeplanes=0x000201:yuv420p
14198 Cast a rgb24 clip to yuv444p:
14200 format=rgb24,mergeplanes=0x000102:yuv444p
14206 Estimate and export motion vectors using block matching algorithms.
14207 Motion vectors are stored in frame side data to be used by other filters.
14209 This filter accepts the following options:
14212 Specify the motion estimation method. Accepts one of the following values:
14216 Exhaustive search algorithm.
14218 Three step search algorithm.
14220 Two dimensional logarithmic search algorithm.
14222 New three step search algorithm.
14224 Four step search algorithm.
14226 Diamond search algorithm.
14228 Hexagon-based search algorithm.
14230 Enhanced predictive zonal search algorithm.
14232 Uneven multi-hexagon search algorithm.
14234 Default value is @samp{esa}.
14237 Macroblock size. Default @code{16}.
14240 Search parameter. Default @code{7}.
14243 @section midequalizer
14245 Apply Midway Image Equalization effect using two video streams.
14247 Midway Image Equalization adjusts a pair of images to have the same
14248 histogram, while maintaining their dynamics as much as possible. It's
14249 useful for e.g. matching exposures from a pair of stereo cameras.
14251 This filter has two inputs and one output, which must be of same pixel format, but
14252 may be of different sizes. The output of filter is first input adjusted with
14253 midway histogram of both inputs.
14255 This filter accepts the following option:
14259 Set which planes to process. Default is @code{15}, which is all available planes.
14262 @section minterpolate
14264 Convert the video to specified frame rate using motion interpolation.
14266 This filter accepts the following options:
14269 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}.
14272 Motion interpolation mode. Following values are accepted:
14275 Duplicate previous or next frame for interpolating new ones.
14277 Blend source frames. Interpolated frame is mean of previous and next frames.
14279 Motion compensated interpolation. Following options are effective when this mode is selected:
14283 Motion compensation mode. Following values are accepted:
14286 Overlapped block motion compensation.
14288 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
14290 Default mode is @samp{obmc}.
14293 Motion estimation mode. Following values are accepted:
14296 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14298 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14300 Default mode is @samp{bilat}.
14303 The algorithm to be used for motion estimation. Following values are accepted:
14306 Exhaustive search algorithm.
14308 Three step search algorithm.
14310 Two dimensional logarithmic search algorithm.
14312 New three step search algorithm.
14314 Four step search algorithm.
14316 Diamond search algorithm.
14318 Hexagon-based search algorithm.
14320 Enhanced predictive zonal search algorithm.
14322 Uneven multi-hexagon search algorithm.
14324 Default algorithm is @samp{epzs}.
14327 Macroblock size. Default @code{16}.
14330 Motion estimation search parameter. Default @code{32}.
14333 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).
14338 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:
14341 Disable scene change detection.
14343 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14345 Default method is @samp{fdiff}.
14347 @item scd_threshold
14348 Scene change detection threshold. Default is @code{10.}.
14353 Mix several video input streams into one video stream.
14355 A description of the accepted options follows.
14359 The number of inputs. If unspecified, it defaults to 2.
14362 Specify weight of each input video stream as sequence.
14363 Each weight is separated by space. If number of weights
14364 is smaller than number of @var{frames} last specified
14365 weight will be used for all remaining unset weights.
14368 Specify scale, if it is set it will be multiplied with sum
14369 of each weight multiplied with pixel values to give final destination
14370 pixel value. By default @var{scale} is auto scaled to sum of weights.
14373 Specify how end of stream is determined.
14376 The duration of the longest input. (default)
14379 The duration of the shortest input.
14382 The duration of the first input.
14386 @section mpdecimate
14388 Drop frames that do not differ greatly from the previous frame in
14389 order to reduce frame rate.
14391 The main use of this filter is for very-low-bitrate encoding
14392 (e.g. streaming over dialup modem), but it could in theory be used for
14393 fixing movies that were inverse-telecined incorrectly.
14395 A description of the accepted options follows.
14399 Set the maximum number of consecutive frames which can be dropped (if
14400 positive), or the minimum interval between dropped frames (if
14401 negative). If the value is 0, the frame is dropped disregarding the
14402 number of previous sequentially dropped frames.
14404 Default value is 0.
14409 Set the dropping threshold values.
14411 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14412 represent actual pixel value differences, so a threshold of 64
14413 corresponds to 1 unit of difference for each pixel, or the same spread
14414 out differently over the block.
14416 A frame is a candidate for dropping if no 8x8 blocks differ by more
14417 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14418 meaning the whole image) differ by more than a threshold of @option{lo}.
14420 Default value for @option{hi} is 64*12, default value for @option{lo} is
14421 64*5, and default value for @option{frac} is 0.33.
14427 Negate (invert) the input video.
14429 It accepts the following option:
14434 With value 1, it negates the alpha component, if present. Default value is 0.
14440 Denoise frames using Non-Local Means algorithm.
14442 Each pixel is adjusted by looking for other pixels with similar contexts. This
14443 context similarity is defined by comparing their surrounding patches of size
14444 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14447 Note that the research area defines centers for patches, which means some
14448 patches will be made of pixels outside that research area.
14450 The filter accepts the following options.
14454 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14457 Set patch size. Default is 7. Must be odd number in range [0, 99].
14460 Same as @option{p} but for chroma planes.
14462 The default value is @var{0} and means automatic.
14465 Set research size. Default is 15. Must be odd number in range [0, 99].
14468 Same as @option{r} but for chroma planes.
14470 The default value is @var{0} and means automatic.
14475 Deinterlace video using neural network edge directed interpolation.
14477 This filter accepts the following options:
14481 Mandatory option, without binary file filter can not work.
14482 Currently file can be found here:
14483 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14486 Set which frames to deinterlace, by default it is @code{all}.
14487 Can be @code{all} or @code{interlaced}.
14490 Set mode of operation.
14492 Can be one of the following:
14496 Use frame flags, both fields.
14498 Use frame flags, single field.
14500 Use top field only.
14502 Use bottom field only.
14504 Use both fields, top first.
14506 Use both fields, bottom first.
14510 Set which planes to process, by default filter process all frames.
14513 Set size of local neighborhood around each pixel, used by the predictor neural
14516 Can be one of the following:
14529 Set the number of neurons in predictor neural network.
14530 Can be one of the following:
14541 Controls the number of different neural network predictions that are blended
14542 together to compute the final output value. Can be @code{fast}, default or
14546 Set which set of weights to use in the predictor.
14547 Can be one of the following:
14551 weights trained to minimize absolute error
14553 weights trained to minimize squared error
14557 Controls whether or not the prescreener neural network is used to decide
14558 which pixels should be processed by the predictor neural network and which
14559 can be handled by simple cubic interpolation.
14560 The prescreener is trained to know whether cubic interpolation will be
14561 sufficient for a pixel or whether it should be predicted by the predictor nn.
14562 The computational complexity of the prescreener nn is much less than that of
14563 the predictor nn. Since most pixels can be handled by cubic interpolation,
14564 using the prescreener generally results in much faster processing.
14565 The prescreener is pretty accurate, so the difference between using it and not
14566 using it is almost always unnoticeable.
14568 Can be one of the following:
14576 Default is @code{new}.
14579 Set various debugging flags.
14584 Force libavfilter not to use any of the specified pixel formats for the
14585 input to the next filter.
14587 It accepts the following parameters:
14591 A '|'-separated list of pixel format names, such as
14592 pix_fmts=yuv420p|monow|rgb24".
14596 @subsection Examples
14600 Force libavfilter to use a format different from @var{yuv420p} for the
14601 input to the vflip filter:
14603 noformat=pix_fmts=yuv420p,vflip
14607 Convert the input video to any of the formats not contained in the list:
14609 noformat=yuv420p|yuv444p|yuv410p
14615 Add noise on video input frame.
14617 The filter accepts the following options:
14625 Set noise seed for specific pixel component or all pixel components in case
14626 of @var{all_seed}. Default value is @code{123457}.
14628 @item all_strength, alls
14629 @item c0_strength, c0s
14630 @item c1_strength, c1s
14631 @item c2_strength, c2s
14632 @item c3_strength, c3s
14633 Set noise strength for specific pixel component or all pixel components in case
14634 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14636 @item all_flags, allf
14637 @item c0_flags, c0f
14638 @item c1_flags, c1f
14639 @item c2_flags, c2f
14640 @item c3_flags, c3f
14641 Set pixel component flags or set flags for all components if @var{all_flags}.
14642 Available values for component flags are:
14645 averaged temporal noise (smoother)
14647 mix random noise with a (semi)regular pattern
14649 temporal noise (noise pattern changes between frames)
14651 uniform noise (gaussian otherwise)
14655 @subsection Examples
14657 Add temporal and uniform noise to input video:
14659 noise=alls=20:allf=t+u
14664 Normalize RGB video (aka histogram stretching, contrast stretching).
14665 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14667 For each channel of each frame, the filter computes the input range and maps
14668 it linearly to the user-specified output range. The output range defaults
14669 to the full dynamic range from pure black to pure white.
14671 Temporal smoothing can be used on the input range to reduce flickering (rapid
14672 changes in brightness) caused when small dark or bright objects enter or leave
14673 the scene. This is similar to the auto-exposure (automatic gain control) on a
14674 video camera, and, like a video camera, it may cause a period of over- or
14675 under-exposure of the video.
14677 The R,G,B channels can be normalized independently, which may cause some
14678 color shifting, or linked together as a single channel, which prevents
14679 color shifting. Linked normalization preserves hue. Independent normalization
14680 does not, so it can be used to remove some color casts. Independent and linked
14681 normalization can be combined in any ratio.
14683 The normalize filter accepts the following options:
14688 Colors which define the output range. The minimum input value is mapped to
14689 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14690 The defaults are black and white respectively. Specifying white for
14691 @var{blackpt} and black for @var{whitept} will give color-inverted,
14692 normalized video. Shades of grey can be used to reduce the dynamic range
14693 (contrast). Specifying saturated colors here can create some interesting
14697 The number of previous frames to use for temporal smoothing. The input range
14698 of each channel is smoothed using a rolling average over the current frame
14699 and the @var{smoothing} previous frames. The default is 0 (no temporal
14703 Controls the ratio of independent (color shifting) channel normalization to
14704 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14705 independent. Defaults to 1.0 (fully independent).
14708 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14709 expensive no-op. Defaults to 1.0 (full strength).
14713 @subsection Commands
14714 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14715 The command accepts the same syntax of the corresponding option.
14717 If the specified expression is not valid, it is kept at its current
14720 @subsection Examples
14722 Stretch video contrast to use the full dynamic range, with no temporal
14723 smoothing; may flicker depending on the source content:
14725 normalize=blackpt=black:whitept=white:smoothing=0
14728 As above, but with 50 frames of temporal smoothing; flicker should be
14729 reduced, depending on the source content:
14731 normalize=blackpt=black:whitept=white:smoothing=50
14734 As above, but with hue-preserving linked channel normalization:
14736 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14739 As above, but with half strength:
14741 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14744 Map the darkest input color to red, the brightest input color to cyan:
14746 normalize=blackpt=red:whitept=cyan
14751 Pass the video source unchanged to the output.
14754 Optical Character Recognition
14756 This filter uses Tesseract for optical character recognition. To enable
14757 compilation of this filter, you need to configure FFmpeg with
14758 @code{--enable-libtesseract}.
14760 It accepts the following options:
14764 Set datapath to tesseract data. Default is to use whatever was
14765 set at installation.
14768 Set language, default is "eng".
14771 Set character whitelist.
14774 Set character blacklist.
14777 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14778 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14782 Apply a video transform using libopencv.
14784 To enable this filter, install the libopencv library and headers and
14785 configure FFmpeg with @code{--enable-libopencv}.
14787 It accepts the following parameters:
14792 The name of the libopencv filter to apply.
14794 @item filter_params
14795 The parameters to pass to the libopencv filter. If not specified, the default
14796 values are assumed.
14800 Refer to the official libopencv documentation for more precise
14802 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14804 Several libopencv filters are supported; see the following subsections.
14809 Dilate an image by using a specific structuring element.
14810 It corresponds to the libopencv function @code{cvDilate}.
14812 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14814 @var{struct_el} represents a structuring element, and has the syntax:
14815 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14817 @var{cols} and @var{rows} represent the number of columns and rows of
14818 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14819 point, and @var{shape} the shape for the structuring element. @var{shape}
14820 must be "rect", "cross", "ellipse", or "custom".
14822 If the value for @var{shape} is "custom", it must be followed by a
14823 string of the form "=@var{filename}". The file with name
14824 @var{filename} is assumed to represent a binary image, with each
14825 printable character corresponding to a bright pixel. When a custom
14826 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14827 or columns and rows of the read file are assumed instead.
14829 The default value for @var{struct_el} is "3x3+0x0/rect".
14831 @var{nb_iterations} specifies the number of times the transform is
14832 applied to the image, and defaults to 1.
14836 # Use the default values
14839 # Dilate using a structuring element with a 5x5 cross, iterating two times
14840 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14842 # Read the shape from the file diamond.shape, iterating two times.
14843 # The file diamond.shape may contain a pattern of characters like this
14849 # The specified columns and rows are ignored
14850 # but the anchor point coordinates are not
14851 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14856 Erode an image by using a specific structuring element.
14857 It corresponds to the libopencv function @code{cvErode}.
14859 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14860 with the same syntax and semantics as the @ref{dilate} filter.
14864 Smooth the input video.
14866 The filter takes the following parameters:
14867 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14869 @var{type} is the type of smooth filter to apply, and must be one of
14870 the following values: "blur", "blur_no_scale", "median", "gaussian",
14871 or "bilateral". The default value is "gaussian".
14873 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14874 depends on the smooth type. @var{param1} and
14875 @var{param2} accept integer positive values or 0. @var{param3} and
14876 @var{param4} accept floating point values.
14878 The default value for @var{param1} is 3. The default value for the
14879 other parameters is 0.
14881 These parameters correspond to the parameters assigned to the
14882 libopencv function @code{cvSmooth}.
14884 @section oscilloscope
14886 2D Video Oscilloscope.
14888 Useful to measure spatial impulse, step responses, chroma delays, etc.
14890 It accepts the following parameters:
14894 Set scope center x position.
14897 Set scope center y position.
14900 Set scope size, relative to frame diagonal.
14903 Set scope tilt/rotation.
14909 Set trace center x position.
14912 Set trace center y position.
14915 Set trace width, relative to width of frame.
14918 Set trace height, relative to height of frame.
14921 Set which components to trace. By default it traces first three components.
14924 Draw trace grid. By default is enabled.
14927 Draw some statistics. By default is enabled.
14930 Draw scope. By default is enabled.
14933 @subsection Commands
14934 This filter supports same @ref{commands} as options.
14935 The command accepts the same syntax of the corresponding option.
14937 If the specified expression is not valid, it is kept at its current
14940 @subsection Examples
14944 Inspect full first row of video frame.
14946 oscilloscope=x=0.5:y=0:s=1
14950 Inspect full last row of video frame.
14952 oscilloscope=x=0.5:y=1:s=1
14956 Inspect full 5th line of video frame of height 1080.
14958 oscilloscope=x=0.5:y=5/1080:s=1
14962 Inspect full last column of video frame.
14964 oscilloscope=x=1:y=0.5:s=1:t=1
14972 Overlay one video on top of another.
14974 It takes two inputs and has one output. The first input is the "main"
14975 video on which the second input is overlaid.
14977 It accepts the following parameters:
14979 A description of the accepted options follows.
14984 Set the expression for the x and y coordinates of the overlaid video
14985 on the main video. Default value is "0" for both expressions. In case
14986 the expression is invalid, it is set to a huge value (meaning that the
14987 overlay will not be displayed within the output visible area).
14990 See @ref{framesync}.
14993 Set when the expressions for @option{x}, and @option{y} are evaluated.
14995 It accepts the following values:
14998 only evaluate expressions once during the filter initialization or
14999 when a command is processed
15002 evaluate expressions for each incoming frame
15005 Default value is @samp{frame}.
15008 See @ref{framesync}.
15011 Set the format for the output video.
15013 It accepts the following values:
15016 force YUV420 output
15019 force YUV420p10 output
15022 force YUV422 output
15025 force YUV422p10 output
15028 force YUV444 output
15031 force packed RGB output
15034 force planar RGB output
15037 automatically pick format
15040 Default value is @samp{yuv420}.
15043 See @ref{framesync}.
15046 Set format of alpha of the overlaid video, it can be @var{straight} or
15047 @var{premultiplied}. Default is @var{straight}.
15050 The @option{x}, and @option{y} expressions can contain the following
15056 The main input width and height.
15060 The overlay input width and height.
15064 The computed values for @var{x} and @var{y}. They are evaluated for
15069 horizontal and vertical chroma subsample values of the output
15070 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
15074 the number of input frame, starting from 0
15077 the position in the file of the input frame, NAN if unknown
15080 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
15084 This filter also supports the @ref{framesync} options.
15086 Note that the @var{n}, @var{pos}, @var{t} variables are available only
15087 when evaluation is done @emph{per frame}, and will evaluate to NAN
15088 when @option{eval} is set to @samp{init}.
15090 Be aware that frames are taken from each input video in timestamp
15091 order, hence, if their initial timestamps differ, it is a good idea
15092 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
15093 have them begin in the same zero timestamp, as the example for
15094 the @var{movie} filter does.
15096 You can chain together more overlays but you should test the
15097 efficiency of such approach.
15099 @subsection Commands
15101 This filter supports the following commands:
15105 Modify the x and y of the overlay input.
15106 The command accepts the same syntax of the corresponding option.
15108 If the specified expression is not valid, it is kept at its current
15112 @subsection Examples
15116 Draw the overlay at 10 pixels from the bottom right corner of the main
15119 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15122 Using named options the example above becomes:
15124 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15128 Insert a transparent PNG logo in the bottom left corner of the input,
15129 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
15131 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15135 Insert 2 different transparent PNG logos (second logo on bottom
15136 right corner) using the @command{ffmpeg} tool:
15138 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
15142 Add a transparent color layer on top of the main video; @code{WxH}
15143 must specify the size of the main input to the overlay filter:
15145 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
15149 Play an original video and a filtered version (here with the deshake
15150 filter) side by side using the @command{ffplay} tool:
15152 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15155 The above command is the same as:
15157 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15161 Make a sliding overlay appearing from the left to the right top part of the
15162 screen starting since time 2:
15164 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15168 Compose output by putting two input videos side to side:
15170 ffmpeg -i left.avi -i right.avi -filter_complex "
15171 nullsrc=size=200x100 [background];
15172 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15173 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15174 [background][left] overlay=shortest=1 [background+left];
15175 [background+left][right] overlay=shortest=1:x=100 [left+right]
15180 Mask 10-20 seconds of a video by applying the delogo filter to a section
15182 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15183 -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]'
15188 Chain several overlays in cascade:
15190 nullsrc=s=200x200 [bg];
15191 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15192 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15193 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15194 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15195 [in3] null, [mid2] overlay=100:100 [out0]
15200 @anchor{overlay_cuda}
15201 @section overlay_cuda
15203 Overlay one video on top of another.
15205 This is the CUDA variant of the @ref{overlay} filter.
15206 It only accepts CUDA frames. The underlying input pixel formats have to match.
15208 It takes two inputs and has one output. The first input is the "main"
15209 video on which the second input is overlaid.
15211 It accepts the following parameters:
15216 Set the x and y coordinates of the overlaid video on the main video.
15217 Default value is "0" for both expressions.
15220 See @ref{framesync}.
15223 See @ref{framesync}.
15226 See @ref{framesync}.
15230 This filter also supports the @ref{framesync} options.
15234 Apply Overcomplete Wavelet denoiser.
15236 The filter accepts the following options:
15242 Larger depth values will denoise lower frequency components more, but
15243 slow down filtering.
15245 Must be an int in the range 8-16, default is @code{8}.
15247 @item luma_strength, ls
15250 Must be a double value in the range 0-1000, default is @code{1.0}.
15252 @item chroma_strength, cs
15253 Set chroma strength.
15255 Must be a double value in the range 0-1000, default is @code{1.0}.
15261 Add paddings to the input image, and place the original input at the
15262 provided @var{x}, @var{y} coordinates.
15264 It accepts the following parameters:
15269 Specify an expression for the size of the output image with the
15270 paddings added. If the value for @var{width} or @var{height} is 0, the
15271 corresponding input size is used for the output.
15273 The @var{width} expression can reference the value set by the
15274 @var{height} expression, and vice versa.
15276 The default value of @var{width} and @var{height} is 0.
15280 Specify the offsets to place the input image at within the padded area,
15281 with respect to the top/left border of the output image.
15283 The @var{x} expression can reference the value set by the @var{y}
15284 expression, and vice versa.
15286 The default value of @var{x} and @var{y} is 0.
15288 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15289 so the input image is centered on the padded area.
15292 Specify the color of the padded area. For the syntax of this option,
15293 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15294 manual,ffmpeg-utils}.
15296 The default value of @var{color} is "black".
15299 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15301 It accepts the following values:
15305 Only evaluate expressions once during the filter initialization or when
15306 a command is processed.
15309 Evaluate expressions for each incoming frame.
15313 Default value is @samp{init}.
15316 Pad to aspect instead to a resolution.
15320 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15321 options are expressions containing the following constants:
15326 The input video width and height.
15330 These are the same as @var{in_w} and @var{in_h}.
15334 The output width and height (the size of the padded area), as
15335 specified by the @var{width} and @var{height} expressions.
15339 These are the same as @var{out_w} and @var{out_h}.
15343 The x and y offsets as specified by the @var{x} and @var{y}
15344 expressions, or NAN if not yet specified.
15347 same as @var{iw} / @var{ih}
15350 input sample aspect ratio
15353 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15357 The horizontal and vertical chroma subsample values. For example for the
15358 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15361 @subsection Examples
15365 Add paddings with the color "violet" to the input video. The output video
15366 size is 640x480, and the top-left corner of the input video is placed at
15369 pad=640:480:0:40:violet
15372 The example above is equivalent to the following command:
15374 pad=width=640:height=480:x=0:y=40:color=violet
15378 Pad the input to get an output with dimensions increased by 3/2,
15379 and put the input video at the center of the padded area:
15381 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15385 Pad the input to get a squared output with size equal to the maximum
15386 value between the input width and height, and put the input video at
15387 the center of the padded area:
15389 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15393 Pad the input to get a final w/h ratio of 16:9:
15395 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15399 In case of anamorphic video, in order to set the output display aspect
15400 correctly, it is necessary to use @var{sar} in the expression,
15401 according to the relation:
15403 (ih * X / ih) * sar = output_dar
15404 X = output_dar / sar
15407 Thus the previous example needs to be modified to:
15409 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15413 Double the output size and put the input video in the bottom-right
15414 corner of the output padded area:
15416 pad="2*iw:2*ih:ow-iw:oh-ih"
15420 @anchor{palettegen}
15421 @section palettegen
15423 Generate one palette for a whole video stream.
15425 It accepts the following options:
15429 Set the maximum number of colors to quantize in the palette.
15430 Note: the palette will still contain 256 colors; the unused palette entries
15433 @item reserve_transparent
15434 Create a palette of 255 colors maximum and reserve the last one for
15435 transparency. Reserving the transparency color is useful for GIF optimization.
15436 If not set, the maximum of colors in the palette will be 256. You probably want
15437 to disable this option for a standalone image.
15440 @item transparency_color
15441 Set the color that will be used as background for transparency.
15444 Set statistics mode.
15446 It accepts the following values:
15449 Compute full frame histograms.
15451 Compute histograms only for the part that differs from previous frame. This
15452 might be relevant to give more importance to the moving part of your input if
15453 the background is static.
15455 Compute new histogram for each frame.
15458 Default value is @var{full}.
15461 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15462 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15463 color quantization of the palette. This information is also visible at
15464 @var{info} logging level.
15466 @subsection Examples
15470 Generate a representative palette of a given video using @command{ffmpeg}:
15472 ffmpeg -i input.mkv -vf palettegen palette.png
15476 @section paletteuse
15478 Use a palette to downsample an input video stream.
15480 The filter takes two inputs: one video stream and a palette. The palette must
15481 be a 256 pixels image.
15483 It accepts the following options:
15487 Select dithering mode. Available algorithms are:
15490 Ordered 8x8 bayer dithering (deterministic)
15492 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15493 Note: this dithering is sometimes considered "wrong" and is included as a
15495 @item floyd_steinberg
15496 Floyd and Steingberg dithering (error diffusion)
15498 Frankie Sierra dithering v2 (error diffusion)
15500 Frankie Sierra dithering v2 "Lite" (error diffusion)
15503 Default is @var{sierra2_4a}.
15506 When @var{bayer} dithering is selected, this option defines the scale of the
15507 pattern (how much the crosshatch pattern is visible). A low value means more
15508 visible pattern for less banding, and higher value means less visible pattern
15509 at the cost of more banding.
15511 The option must be an integer value in the range [0,5]. Default is @var{2}.
15514 If set, define the zone to process
15518 Only the changing rectangle will be reprocessed. This is similar to GIF
15519 cropping/offsetting compression mechanism. This option can be useful for speed
15520 if only a part of the image is changing, and has use cases such as limiting the
15521 scope of the error diffusal @option{dither} to the rectangle that bounds the
15522 moving scene (it leads to more deterministic output if the scene doesn't change
15523 much, and as a result less moving noise and better GIF compression).
15526 Default is @var{none}.
15529 Take new palette for each output frame.
15531 @item alpha_threshold
15532 Sets the alpha threshold for transparency. Alpha values above this threshold
15533 will be treated as completely opaque, and values below this threshold will be
15534 treated as completely transparent.
15536 The option must be an integer value in the range [0,255]. Default is @var{128}.
15539 @subsection Examples
15543 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15544 using @command{ffmpeg}:
15546 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15550 @section perspective
15552 Correct perspective of video not recorded perpendicular to the screen.
15554 A description of the accepted parameters follows.
15565 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15566 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15567 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15568 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15569 then the corners of the source will be sent to the specified coordinates.
15571 The expressions can use the following variables:
15576 the width and height of video frame.
15580 Output frame count.
15583 @item interpolation
15584 Set interpolation for perspective correction.
15586 It accepts the following values:
15592 Default value is @samp{linear}.
15595 Set interpretation of coordinate options.
15597 It accepts the following values:
15601 Send point in the source specified by the given coordinates to
15602 the corners of the destination.
15604 @item 1, destination
15606 Send the corners of the source to the point in the destination specified
15607 by the given coordinates.
15609 Default value is @samp{source}.
15613 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15615 It accepts the following values:
15618 only evaluate expressions once during the filter initialization or
15619 when a command is processed
15622 evaluate expressions for each incoming frame
15625 Default value is @samp{init}.
15630 Delay interlaced video by one field time so that the field order changes.
15632 The intended use is to fix PAL movies that have been captured with the
15633 opposite field order to the film-to-video transfer.
15635 A description of the accepted parameters follows.
15641 It accepts the following values:
15644 Capture field order top-first, transfer bottom-first.
15645 Filter will delay the bottom field.
15648 Capture field order bottom-first, transfer top-first.
15649 Filter will delay the top field.
15652 Capture and transfer with the same field order. This mode only exists
15653 for the documentation of the other options to refer to, but if you
15654 actually select it, the filter will faithfully do nothing.
15657 Capture field order determined automatically by field flags, transfer
15659 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15660 basis using field flags. If no field information is available,
15661 then this works just like @samp{u}.
15664 Capture unknown or varying, transfer opposite.
15665 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15666 analyzing the images and selecting the alternative that produces best
15667 match between the fields.
15670 Capture top-first, transfer unknown or varying.
15671 Filter selects among @samp{t} and @samp{p} using image analysis.
15674 Capture bottom-first, transfer unknown or varying.
15675 Filter selects among @samp{b} and @samp{p} using image analysis.
15678 Capture determined by field flags, transfer unknown or varying.
15679 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15680 image analysis. If no field information is available, then this works just
15681 like @samp{U}. This is the default mode.
15684 Both capture and transfer unknown or varying.
15685 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15689 @section photosensitivity
15690 Reduce various flashes in video, so to help users with epilepsy.
15692 It accepts the following options:
15695 Set how many frames to use when filtering. Default is 30.
15698 Set detection threshold factor. Default is 1.
15702 Set how many pixels to skip when sampling frames. Default is 1.
15703 Allowed range is from 1 to 1024.
15706 Leave frames unchanged. Default is disabled.
15709 @section pixdesctest
15711 Pixel format descriptor test filter, mainly useful for internal
15712 testing. The output video should be equal to the input video.
15716 format=monow, pixdesctest
15719 can be used to test the monowhite pixel format descriptor definition.
15723 Display sample values of color channels. Mainly useful for checking color
15724 and levels. Minimum supported resolution is 640x480.
15726 The filters accept the following options:
15730 Set scope X position, relative offset on X axis.
15733 Set scope Y position, relative offset on Y axis.
15742 Set window opacity. This window also holds statistics about pixel area.
15745 Set window X position, relative offset on X axis.
15748 Set window Y position, relative offset on Y axis.
15753 Enable the specified chain of postprocessing subfilters using libpostproc. This
15754 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15755 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15756 Each subfilter and some options have a short and a long name that can be used
15757 interchangeably, i.e. dr/dering are the same.
15759 The filters accept the following options:
15763 Set postprocessing subfilters string.
15766 All subfilters share common options to determine their scope:
15770 Honor the quality commands for this subfilter.
15773 Do chrominance filtering, too (default).
15776 Do luminance filtering only (no chrominance).
15779 Do chrominance filtering only (no luminance).
15782 These options can be appended after the subfilter name, separated by a '|'.
15784 Available subfilters are:
15787 @item hb/hdeblock[|difference[|flatness]]
15788 Horizontal deblocking filter
15791 Difference factor where higher values mean more deblocking (default: @code{32}).
15793 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15796 @item vb/vdeblock[|difference[|flatness]]
15797 Vertical deblocking filter
15800 Difference factor where higher values mean more deblocking (default: @code{32}).
15802 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15805 @item ha/hadeblock[|difference[|flatness]]
15806 Accurate horizontal deblocking filter
15809 Difference factor where higher values mean more deblocking (default: @code{32}).
15811 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15814 @item va/vadeblock[|difference[|flatness]]
15815 Accurate vertical deblocking filter
15818 Difference factor where higher values mean more deblocking (default: @code{32}).
15820 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15824 The horizontal and vertical deblocking filters share the difference and
15825 flatness values so you cannot set different horizontal and vertical
15829 @item h1/x1hdeblock
15830 Experimental horizontal deblocking filter
15832 @item v1/x1vdeblock
15833 Experimental vertical deblocking filter
15838 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15841 larger -> stronger filtering
15843 larger -> stronger filtering
15845 larger -> stronger filtering
15848 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15851 Stretch luminance to @code{0-255}.
15854 @item lb/linblenddeint
15855 Linear blend deinterlacing filter that deinterlaces the given block by
15856 filtering all lines with a @code{(1 2 1)} filter.
15858 @item li/linipoldeint
15859 Linear interpolating deinterlacing filter that deinterlaces the given block by
15860 linearly interpolating every second line.
15862 @item ci/cubicipoldeint
15863 Cubic interpolating deinterlacing filter deinterlaces the given block by
15864 cubically interpolating every second line.
15866 @item md/mediandeint
15867 Median deinterlacing filter that deinterlaces the given block by applying a
15868 median filter to every second line.
15870 @item fd/ffmpegdeint
15871 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15872 second line with a @code{(-1 4 2 4 -1)} filter.
15875 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15876 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15878 @item fq/forceQuant[|quantizer]
15879 Overrides the quantizer table from the input with the constant quantizer you
15887 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15890 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15893 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15896 @subsection Examples
15900 Apply horizontal and vertical deblocking, deringing and automatic
15901 brightness/contrast:
15907 Apply default filters without brightness/contrast correction:
15913 Apply default filters and temporal denoiser:
15915 pp=default/tmpnoise|1|2|3
15919 Apply deblocking on luminance only, and switch vertical deblocking on or off
15920 automatically depending on available CPU time:
15927 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15928 similar to spp = 6 with 7 point DCT, where only the center sample is
15931 The filter accepts the following options:
15935 Force a constant quantization parameter. It accepts an integer in range
15936 0 to 63. If not set, the filter will use the QP from the video stream
15940 Set thresholding mode. Available modes are:
15944 Set hard thresholding.
15946 Set soft thresholding (better de-ringing effect, but likely blurrier).
15948 Set medium thresholding (good results, default).
15952 @section premultiply
15953 Apply alpha premultiply effect to input video stream using first plane
15954 of second stream as alpha.
15956 Both streams must have same dimensions and same pixel format.
15958 The filter accepts the following option:
15962 Set which planes will be processed, unprocessed planes will be copied.
15963 By default value 0xf, all planes will be processed.
15966 Do not require 2nd input for processing, instead use alpha plane from input stream.
15970 Apply prewitt operator to input video stream.
15972 The filter accepts the following option:
15976 Set which planes will be processed, unprocessed planes will be copied.
15977 By default value 0xf, all planes will be processed.
15980 Set value which will be multiplied with filtered result.
15983 Set value which will be added to filtered result.
15986 @subsection Commands
15988 This filter supports the all above options as @ref{commands}.
15990 @section pseudocolor
15992 Alter frame colors in video with pseudocolors.
15994 This filter accepts the following options:
15998 set pixel first component expression
16001 set pixel second component expression
16004 set pixel third component expression
16007 set pixel fourth component expression, corresponds to the alpha component
16010 set component to use as base for altering colors
16013 Each of them specifies the expression to use for computing the lookup table for
16014 the corresponding pixel component values.
16016 The expressions can contain the following constants and functions:
16021 The input width and height.
16024 The input value for the pixel component.
16026 @item ymin, umin, vmin, amin
16027 The minimum allowed component value.
16029 @item ymax, umax, vmax, amax
16030 The maximum allowed component value.
16033 All expressions default to "val".
16035 @subsection Examples
16039 Change too high luma values to gradient:
16041 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'"
16047 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16048 Ratio) between two input videos.
16050 This filter takes in input two input videos, the first input is
16051 considered the "main" source and is passed unchanged to the
16052 output. The second input is used as a "reference" video for computing
16055 Both video inputs must have the same resolution and pixel format for
16056 this filter to work correctly. Also it assumes that both inputs
16057 have the same number of frames, which are compared one by one.
16059 The obtained average PSNR is printed through the logging system.
16061 The filter stores the accumulated MSE (mean squared error) of each
16062 frame, and at the end of the processing it is averaged across all frames
16063 equally, and the following formula is applied to obtain the PSNR:
16066 PSNR = 10*log10(MAX^2/MSE)
16069 Where MAX is the average of the maximum values of each component of the
16072 The description of the accepted parameters follows.
16075 @item stats_file, f
16076 If specified the filter will use the named file to save the PSNR of
16077 each individual frame. When filename equals "-" the data is sent to
16080 @item stats_version
16081 Specifies which version of the stats file format to use. Details of
16082 each format are written below.
16083 Default value is 1.
16085 @item stats_add_max
16086 Determines whether the max value is output to the stats log.
16087 Default value is 0.
16088 Requires stats_version >= 2. If this is set and stats_version < 2,
16089 the filter will return an error.
16092 This filter also supports the @ref{framesync} options.
16094 The file printed if @var{stats_file} is selected, contains a sequence of
16095 key/value pairs of the form @var{key}:@var{value} for each compared
16098 If a @var{stats_version} greater than 1 is specified, a header line precedes
16099 the list of per-frame-pair stats, with key value pairs following the frame
16100 format with the following parameters:
16103 @item psnr_log_version
16104 The version of the log file format. Will match @var{stats_version}.
16107 A comma separated list of the per-frame-pair parameters included in
16111 A description of each shown per-frame-pair parameter follows:
16115 sequential number of the input frame, starting from 1
16118 Mean Square Error pixel-by-pixel average difference of the compared
16119 frames, averaged over all the image components.
16121 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16122 Mean Square Error pixel-by-pixel average difference of the compared
16123 frames for the component specified by the suffix.
16125 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16126 Peak Signal to Noise ratio of the compared frames for the component
16127 specified by the suffix.
16129 @item max_avg, max_y, max_u, max_v
16130 Maximum allowed value for each channel, and average over all
16134 @subsection Examples
16139 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16140 [main][ref] psnr="stats_file=stats.log" [out]
16143 On this example the input file being processed is compared with the
16144 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
16145 is stored in @file{stats.log}.
16148 Another example with different containers:
16150 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 -
16157 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16158 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
16161 The pullup filter is designed to take advantage of future context in making
16162 its decisions. This filter is stateless in the sense that it does not lock
16163 onto a pattern to follow, but it instead looks forward to the following
16164 fields in order to identify matches and rebuild progressive frames.
16166 To produce content with an even framerate, insert the fps filter after
16167 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
16168 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
16170 The filter accepts the following options:
16177 These options set the amount of "junk" to ignore at the left, right, top, and
16178 bottom of the image, respectively. Left and right are in units of 8 pixels,
16179 while top and bottom are in units of 2 lines.
16180 The default is 8 pixels on each side.
16183 Set the strict breaks. Setting this option to 1 will reduce the chances of
16184 filter generating an occasional mismatched frame, but it may also cause an
16185 excessive number of frames to be dropped during high motion sequences.
16186 Conversely, setting it to -1 will make filter match fields more easily.
16187 This may help processing of video where there is slight blurring between
16188 the fields, but may also cause there to be interlaced frames in the output.
16189 Default value is @code{0}.
16192 Set the metric plane to use. It accepts the following values:
16198 Use chroma blue plane.
16201 Use chroma red plane.
16204 This option may be set to use chroma plane instead of the default luma plane
16205 for doing filter's computations. This may improve accuracy on very clean
16206 source material, but more likely will decrease accuracy, especially if there
16207 is chroma noise (rainbow effect) or any grayscale video.
16208 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
16209 load and make pullup usable in realtime on slow machines.
16212 For best results (without duplicated frames in the output file) it is
16213 necessary to change the output frame rate. For example, to inverse
16214 telecine NTSC input:
16216 ffmpeg -i input -vf pullup -r 24000/1001 ...
16221 Change video quantization parameters (QP).
16223 The filter accepts the following option:
16227 Set expression for quantization parameter.
16230 The expression is evaluated through the eval API and can contain, among others,
16231 the following constants:
16235 1 if index is not 129, 0 otherwise.
16238 Sequential index starting from -129 to 128.
16241 @subsection Examples
16245 Some equation like:
16253 Flush video frames from internal cache of frames into a random order.
16254 No frame is discarded.
16255 Inspired by @ref{frei0r} nervous filter.
16259 Set size in number of frames of internal cache, in range from @code{2} to
16260 @code{512}. Default is @code{30}.
16263 Set seed for random number generator, must be an integer included between
16264 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
16265 less than @code{0}, the filter will try to use a good random seed on a
16269 @section readeia608
16271 Read closed captioning (EIA-608) information from the top lines of a video frame.
16273 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
16274 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
16275 with EIA-608 data (starting from 0). A description of each metadata value follows:
16278 @item lavfi.readeia608.X.cc
16279 The two bytes stored as EIA-608 data (printed in hexadecimal).
16281 @item lavfi.readeia608.X.line
16282 The number of the line on which the EIA-608 data was identified and read.
16285 This filter accepts the following options:
16289 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16292 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16295 Set the ratio of width reserved for sync code detection.
16296 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16299 Enable checking the parity bit. In the event of a parity error, the filter will output
16300 @code{0x00} for that character. Default is false.
16303 Lowpass lines prior to further processing. Default is enabled.
16306 @subsection Commands
16308 This filter supports the all above options as @ref{commands}.
16310 @subsection Examples
16314 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16316 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
16322 Read vertical interval timecode (VITC) information from the top lines of a
16325 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16326 timecode value, if a valid timecode has been detected. Further metadata key
16327 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16328 timecode data has been found or not.
16330 This filter accepts the following options:
16334 Set the maximum number of lines to scan for VITC data. If the value is set to
16335 @code{-1} the full video frame is scanned. Default is @code{45}.
16338 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16339 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16342 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16343 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16346 @subsection Examples
16350 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16351 draw @code{--:--:--:--} as a placeholder:
16353 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16359 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16361 Destination pixel at position (X, Y) will be picked from source (x, y) position
16362 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16363 value for pixel will be used for destination pixel.
16365 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16366 will have Xmap/Ymap video stream dimensions.
16367 Xmap and Ymap input video streams are 16bit depth, single channel.
16371 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16372 Default is @code{color}.
16375 Specify the color of the unmapped pixels. For the syntax of this option,
16376 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16377 manual,ffmpeg-utils}. Default color is @code{black}.
16380 @section removegrain
16382 The removegrain filter is a spatial denoiser for progressive video.
16386 Set mode for the first plane.
16389 Set mode for the second plane.
16392 Set mode for the third plane.
16395 Set mode for the fourth plane.
16398 Range of mode is from 0 to 24. Description of each mode follows:
16402 Leave input plane unchanged. Default.
16405 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16408 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16411 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16414 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16415 This is equivalent to a median filter.
16418 Line-sensitive clipping giving the minimal change.
16421 Line-sensitive clipping, intermediate.
16424 Line-sensitive clipping, intermediate.
16427 Line-sensitive clipping, intermediate.
16430 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16433 Replaces the target pixel with the closest neighbour.
16436 [1 2 1] horizontal and vertical kernel blur.
16442 Bob mode, interpolates top field from the line where the neighbours
16443 pixels are the closest.
16446 Bob mode, interpolates bottom field from the line where the neighbours
16447 pixels are the closest.
16450 Bob mode, interpolates top field. Same as 13 but with a more complicated
16451 interpolation formula.
16454 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16455 interpolation formula.
16458 Clips the pixel with the minimum and maximum of respectively the maximum and
16459 minimum of each pair of opposite neighbour pixels.
16462 Line-sensitive clipping using opposite neighbours whose greatest distance from
16463 the current pixel is minimal.
16466 Replaces the pixel with the average of its 8 neighbours.
16469 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16472 Clips pixels using the averages of opposite neighbour.
16475 Same as mode 21 but simpler and faster.
16478 Small edge and halo removal, but reputed useless.
16484 @section removelogo
16486 Suppress a TV station logo, using an image file to determine which
16487 pixels comprise the logo. It works by filling in the pixels that
16488 comprise the logo with neighboring pixels.
16490 The filter accepts the following options:
16494 Set the filter bitmap file, which can be any image format supported by
16495 libavformat. The width and height of the image file must match those of the
16496 video stream being processed.
16499 Pixels in the provided bitmap image with a value of zero are not
16500 considered part of the logo, non-zero pixels are considered part of
16501 the logo. If you use white (255) for the logo and black (0) for the
16502 rest, you will be safe. For making the filter bitmap, it is
16503 recommended to take a screen capture of a black frame with the logo
16504 visible, and then using a threshold filter followed by the erode
16505 filter once or twice.
16507 If needed, little splotches can be fixed manually. Remember that if
16508 logo pixels are not covered, the filter quality will be much
16509 reduced. Marking too many pixels as part of the logo does not hurt as
16510 much, but it will increase the amount of blurring needed to cover over
16511 the image and will destroy more information than necessary, and extra
16512 pixels will slow things down on a large logo.
16514 @section repeatfields
16516 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16517 fields based on its value.
16521 Reverse a video clip.
16523 Warning: This filter requires memory to buffer the entire clip, so trimming
16526 @subsection Examples
16530 Take the first 5 seconds of a clip, and reverse it.
16537 Shift R/G/B/A pixels horizontally and/or vertically.
16539 The filter accepts the following options:
16542 Set amount to shift red horizontally.
16544 Set amount to shift red vertically.
16546 Set amount to shift green horizontally.
16548 Set amount to shift green vertically.
16550 Set amount to shift blue horizontally.
16552 Set amount to shift blue vertically.
16554 Set amount to shift alpha horizontally.
16556 Set amount to shift alpha vertically.
16558 Set edge mode, can be @var{smear}, default, or @var{warp}.
16561 @subsection Commands
16563 This filter supports the all above options as @ref{commands}.
16566 Apply roberts cross operator to input video stream.
16568 The filter accepts the following option:
16572 Set which planes will be processed, unprocessed planes will be copied.
16573 By default value 0xf, all planes will be processed.
16576 Set value which will be multiplied with filtered result.
16579 Set value which will be added to filtered result.
16582 @subsection Commands
16584 This filter supports the all above options as @ref{commands}.
16588 Rotate video by an arbitrary angle expressed in radians.
16590 The filter accepts the following options:
16592 A description of the optional parameters follows.
16595 Set an expression for the angle by which to rotate the input video
16596 clockwise, expressed as a number of radians. A negative value will
16597 result in a counter-clockwise rotation. By default it is set to "0".
16599 This expression is evaluated for each frame.
16602 Set the output width expression, default value is "iw".
16603 This expression is evaluated just once during configuration.
16606 Set the output height expression, default value is "ih".
16607 This expression is evaluated just once during configuration.
16610 Enable bilinear interpolation if set to 1, a value of 0 disables
16611 it. Default value is 1.
16614 Set the color used to fill the output area not covered by the rotated
16615 image. For the general syntax of this option, check the
16616 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16617 If the special value "none" is selected then no
16618 background is printed (useful for example if the background is never shown).
16620 Default value is "black".
16623 The expressions for the angle and the output size can contain the
16624 following constants and functions:
16628 sequential number of the input frame, starting from 0. It is always NAN
16629 before the first frame is filtered.
16632 time in seconds of the input frame, it is set to 0 when the filter is
16633 configured. It is always NAN before the first frame is filtered.
16637 horizontal and vertical chroma subsample values. For example for the
16638 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16642 the input video width and height
16646 the output width and height, that is the size of the padded area as
16647 specified by the @var{width} and @var{height} expressions
16651 the minimal width/height required for completely containing the input
16652 video rotated by @var{a} radians.
16654 These are only available when computing the @option{out_w} and
16655 @option{out_h} expressions.
16658 @subsection Examples
16662 Rotate the input by PI/6 radians clockwise:
16668 Rotate the input by PI/6 radians counter-clockwise:
16674 Rotate the input by 45 degrees clockwise:
16680 Apply a constant rotation with period T, starting from an angle of PI/3:
16682 rotate=PI/3+2*PI*t/T
16686 Make the input video rotation oscillating with a period of T
16687 seconds and an amplitude of A radians:
16689 rotate=A*sin(2*PI/T*t)
16693 Rotate the video, output size is chosen so that the whole rotating
16694 input video is always completely contained in the output:
16696 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16700 Rotate the video, reduce the output size so that no background is ever
16703 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16707 @subsection Commands
16709 The filter supports the following commands:
16713 Set the angle expression.
16714 The command accepts the same syntax of the corresponding option.
16716 If the specified expression is not valid, it is kept at its current
16722 Apply Shape Adaptive Blur.
16724 The filter accepts the following options:
16727 @item luma_radius, lr
16728 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16729 value is 1.0. A greater value will result in a more blurred image, and
16730 in slower processing.
16732 @item luma_pre_filter_radius, lpfr
16733 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16736 @item luma_strength, ls
16737 Set luma maximum difference between pixels to still be considered, must
16738 be a value in the 0.1-100.0 range, default value is 1.0.
16740 @item chroma_radius, cr
16741 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16742 greater value will result in a more blurred image, and in slower
16745 @item chroma_pre_filter_radius, cpfr
16746 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16748 @item chroma_strength, cs
16749 Set chroma maximum difference between pixels to still be considered,
16750 must be a value in the -0.9-100.0 range.
16753 Each chroma option value, if not explicitly specified, is set to the
16754 corresponding luma option value.
16759 Scale (resize) the input video, using the libswscale library.
16761 The scale filter forces the output display aspect ratio to be the same
16762 of the input, by changing the output sample aspect ratio.
16764 If the input image format is different from the format requested by
16765 the next filter, the scale filter will convert the input to the
16768 @subsection Options
16769 The filter accepts the following options, or any of the options
16770 supported by the libswscale scaler.
16772 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16773 the complete list of scaler options.
16778 Set the output video dimension expression. Default value is the input
16781 If the @var{width} or @var{w} value is 0, the input width is used for
16782 the output. If the @var{height} or @var{h} value is 0, the input height
16783 is used for the output.
16785 If one and only one of the values is -n with n >= 1, the scale filter
16786 will use a value that maintains the aspect ratio of the input image,
16787 calculated from the other specified dimension. After that it will,
16788 however, make sure that the calculated dimension is divisible by n and
16789 adjust the value if necessary.
16791 If both values are -n with n >= 1, the behavior will be identical to
16792 both values being set to 0 as previously detailed.
16794 See below for the list of accepted constants for use in the dimension
16798 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16802 Only evaluate expressions once during the filter initialization or when a command is processed.
16805 Evaluate expressions for each incoming frame.
16809 Default value is @samp{init}.
16813 Set the interlacing mode. It accepts the following values:
16817 Force interlaced aware scaling.
16820 Do not apply interlaced scaling.
16823 Select interlaced aware scaling depending on whether the source frames
16824 are flagged as interlaced or not.
16827 Default value is @samp{0}.
16830 Set libswscale scaling flags. See
16831 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16832 complete list of values. If not explicitly specified the filter applies
16836 @item param0, param1
16837 Set libswscale input parameters for scaling algorithms that need them. See
16838 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16839 complete documentation. If not explicitly specified the filter applies
16845 Set the video size. For the syntax of this option, check the
16846 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16848 @item in_color_matrix
16849 @item out_color_matrix
16850 Set in/output YCbCr color space type.
16852 This allows the autodetected value to be overridden as well as allows forcing
16853 a specific value used for the output and encoder.
16855 If not specified, the color space type depends on the pixel format.
16861 Choose automatically.
16864 Format conforming to International Telecommunication Union (ITU)
16865 Recommendation BT.709.
16868 Set color space conforming to the United States Federal Communications
16869 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16874 Set color space conforming to:
16878 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16881 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16884 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16889 Set color space conforming to SMPTE ST 240:1999.
16892 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16897 Set in/output YCbCr sample range.
16899 This allows the autodetected value to be overridden as well as allows forcing
16900 a specific value used for the output and encoder. If not specified, the
16901 range depends on the pixel format. Possible values:
16905 Choose automatically.
16908 Set full range (0-255 in case of 8-bit luma).
16910 @item mpeg/limited/tv
16911 Set "MPEG" range (16-235 in case of 8-bit luma).
16914 @item force_original_aspect_ratio
16915 Enable decreasing or increasing output video width or height if necessary to
16916 keep the original aspect ratio. Possible values:
16920 Scale the video as specified and disable this feature.
16923 The output video dimensions will automatically be decreased if needed.
16926 The output video dimensions will automatically be increased if needed.
16930 One useful instance of this option is that when you know a specific device's
16931 maximum allowed resolution, you can use this to limit the output video to
16932 that, while retaining the aspect ratio. For example, device A allows
16933 1280x720 playback, and your video is 1920x800. Using this option (set it to
16934 decrease) and specifying 1280x720 to the command line makes the output
16937 Please note that this is a different thing than specifying -1 for @option{w}
16938 or @option{h}, you still need to specify the output resolution for this option
16941 @item force_divisible_by
16942 Ensures that both the output dimensions, width and height, are divisible by the
16943 given integer when used together with @option{force_original_aspect_ratio}. This
16944 works similar to using @code{-n} in the @option{w} and @option{h} options.
16946 This option respects the value set for @option{force_original_aspect_ratio},
16947 increasing or decreasing the resolution accordingly. The video's aspect ratio
16948 may be slightly modified.
16950 This option can be handy if you need to have a video fit within or exceed
16951 a defined resolution using @option{force_original_aspect_ratio} but also have
16952 encoder restrictions on width or height divisibility.
16956 The values of the @option{w} and @option{h} options are expressions
16957 containing the following constants:
16962 The input width and height
16966 These are the same as @var{in_w} and @var{in_h}.
16970 The output (scaled) width and height
16974 These are the same as @var{out_w} and @var{out_h}
16977 The same as @var{iw} / @var{ih}
16980 input sample aspect ratio
16983 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16987 horizontal and vertical input chroma subsample values. For example for the
16988 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16992 horizontal and vertical output chroma subsample values. For example for the
16993 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16996 The (sequential) number of the input frame, starting from 0.
16997 Only available with @code{eval=frame}.
17000 The presentation timestamp of the input frame, expressed as a number of
17001 seconds. Only available with @code{eval=frame}.
17004 The position (byte offset) of the frame in the input stream, or NaN if
17005 this information is unavailable and/or meaningless (for example in case of synthetic video).
17006 Only available with @code{eval=frame}.
17009 @subsection Examples
17013 Scale the input video to a size of 200x100
17018 This is equivalent to:
17029 Specify a size abbreviation for the output size:
17034 which can also be written as:
17040 Scale the input to 2x:
17042 scale=w=2*iw:h=2*ih
17046 The above is the same as:
17048 scale=2*in_w:2*in_h
17052 Scale the input to 2x with forced interlaced scaling:
17054 scale=2*iw:2*ih:interl=1
17058 Scale the input to half size:
17060 scale=w=iw/2:h=ih/2
17064 Increase the width, and set the height to the same size:
17070 Seek Greek harmony:
17077 Increase the height, and set the width to 3/2 of the height:
17079 scale=w=3/2*oh:h=3/5*ih
17083 Increase the size, making the size a multiple of the chroma
17086 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17090 Increase the width to a maximum of 500 pixels,
17091 keeping the same aspect ratio as the input:
17093 scale=w='min(500\, iw*3/2):h=-1'
17097 Make pixels square by combining scale and setsar:
17099 scale='trunc(ih*dar):ih',setsar=1/1
17103 Make pixels square by combining scale and setsar,
17104 making sure the resulting resolution is even (required by some codecs):
17106 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17110 @subsection Commands
17112 This filter supports the following commands:
17116 Set the output video dimension expression.
17117 The command accepts the same syntax of the corresponding option.
17119 If the specified expression is not valid, it is kept at its current
17125 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
17126 format conversion on CUDA video frames. Setting the output width and height
17127 works in the same way as for the @var{scale} filter.
17129 The following additional options are accepted:
17132 The pixel format of the output CUDA frames. If set to the string "same" (the
17133 default), the input format will be kept. Note that automatic format negotiation
17134 and conversion is not yet supported for hardware frames
17137 The interpolation algorithm used for resizing. One of the following:
17144 @item cubic2p_bspline
17145 2-parameter cubic (B=1, C=0)
17147 @item cubic2p_catmullrom
17148 2-parameter cubic (B=0, C=1/2)
17150 @item cubic2p_b05c03
17151 2-parameter cubic (B=1/2, C=3/10)
17159 @item force_original_aspect_ratio
17160 Enable decreasing or increasing output video width or height if necessary to
17161 keep the original aspect ratio. Possible values:
17165 Scale the video as specified and disable this feature.
17168 The output video dimensions will automatically be decreased if needed.
17171 The output video dimensions will automatically be increased if needed.
17175 One useful instance of this option is that when you know a specific device's
17176 maximum allowed resolution, you can use this to limit the output video to
17177 that, while retaining the aspect ratio. For example, device A allows
17178 1280x720 playback, and your video is 1920x800. Using this option (set it to
17179 decrease) and specifying 1280x720 to the command line makes the output
17182 Please note that this is a different thing than specifying -1 for @option{w}
17183 or @option{h}, you still need to specify the output resolution for this option
17186 @item force_divisible_by
17187 Ensures that both the output dimensions, width and height, are divisible by the
17188 given integer when used together with @option{force_original_aspect_ratio}. This
17189 works similar to using @code{-n} in the @option{w} and @option{h} options.
17191 This option respects the value set for @option{force_original_aspect_ratio},
17192 increasing or decreasing the resolution accordingly. The video's aspect ratio
17193 may be slightly modified.
17195 This option can be handy if you need to have a video fit within or exceed
17196 a defined resolution using @option{force_original_aspect_ratio} but also have
17197 encoder restrictions on width or height divisibility.
17203 Scale (resize) the input video, based on a reference video.
17205 See the scale filter for available options, scale2ref supports the same but
17206 uses the reference video instead of the main input as basis. scale2ref also
17207 supports the following additional constants for the @option{w} and
17208 @option{h} options:
17213 The main input video's width and height
17216 The same as @var{main_w} / @var{main_h}
17219 The main input video's sample aspect ratio
17221 @item main_dar, mdar
17222 The main input video's display aspect ratio. Calculated from
17223 @code{(main_w / main_h) * main_sar}.
17227 The main input video's horizontal and vertical chroma subsample values.
17228 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
17232 The (sequential) number of the main input frame, starting from 0.
17233 Only available with @code{eval=frame}.
17236 The presentation timestamp of the main input frame, expressed as a number of
17237 seconds. Only available with @code{eval=frame}.
17240 The position (byte offset) of the frame in the main input stream, or NaN if
17241 this information is unavailable and/or meaningless (for example in case of synthetic video).
17242 Only available with @code{eval=frame}.
17245 @subsection Examples
17249 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
17251 'scale2ref[b][a];[a][b]overlay'
17255 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
17257 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17261 @subsection Commands
17263 This filter supports the following commands:
17267 Set the output video dimension expression.
17268 The command accepts the same syntax of the corresponding option.
17270 If the specified expression is not valid, it is kept at its current
17275 Scroll input video horizontally and/or vertically by constant speed.
17277 The filter accepts the following options:
17279 @item horizontal, h
17280 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
17281 Negative values changes scrolling direction.
17284 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
17285 Negative values changes scrolling direction.
17288 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
17291 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
17294 @subsection Commands
17296 This filter supports the following @ref{commands}:
17298 @item horizontal, h
17299 Set the horizontal scrolling speed.
17301 Set the vertical scrolling speed.
17307 Detect video scene change.
17309 This filter sets frame metadata with mafd between frame, the scene score, and
17310 forward the frame to the next filter, so they can use these metadata to detect
17311 scene change or others.
17313 In addition, this filter logs a message and sets frame metadata when it detects
17314 a scene change by @option{threshold}.
17316 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17318 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17319 to detect scene change.
17321 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17322 detect scene change with @option{threshold}.
17324 The filter accepts the following options:
17328 Set the scene change detection threshold as a percentage of maximum change. Good
17329 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17332 Default value is @code{10.}.
17335 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17336 You can enable it if you want to get snapshot of scene change frames only.
17339 @anchor{selectivecolor}
17340 @section selectivecolor
17342 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17343 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17344 by the "purity" of the color (that is, how saturated it already is).
17346 This filter is similar to the Adobe Photoshop Selective Color tool.
17348 The filter accepts the following options:
17351 @item correction_method
17352 Select color correction method.
17354 Available values are:
17357 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17360 Specified adjustments are relative to the original component value.
17362 Default is @code{absolute}.
17364 Adjustments for red pixels (pixels where the red component is the maximum)
17366 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17368 Adjustments for green pixels (pixels where the green component is the maximum)
17370 Adjustments for cyan pixels (pixels where the red component is the minimum)
17372 Adjustments for blue pixels (pixels where the blue component is the maximum)
17374 Adjustments for magenta pixels (pixels where the green component is the minimum)
17376 Adjustments for white pixels (pixels where all components are greater than 128)
17378 Adjustments for all pixels except pure black and pure white
17380 Adjustments for black pixels (pixels where all components are lesser than 128)
17382 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17385 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17386 4 space separated floating point adjustment values in the [-1,1] range,
17387 respectively to adjust the amount of cyan, magenta, yellow and black for the
17388 pixels of its range.
17390 @subsection Examples
17394 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17395 increase magenta by 27% in blue areas:
17397 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17401 Use a Photoshop selective color preset:
17403 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17407 @anchor{separatefields}
17408 @section separatefields
17410 The @code{separatefields} takes a frame-based video input and splits
17411 each frame into its components fields, producing a new half height clip
17412 with twice the frame rate and twice the frame count.
17414 This filter use field-dominance information in frame to decide which
17415 of each pair of fields to place first in the output.
17416 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17418 @section setdar, setsar
17420 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17423 This is done by changing the specified Sample (aka Pixel) Aspect
17424 Ratio, according to the following equation:
17426 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17429 Keep in mind that the @code{setdar} filter does not modify the pixel
17430 dimensions of the video frame. Also, the display aspect ratio set by
17431 this filter may be changed by later filters in the filterchain,
17432 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17435 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17436 the filter output video.
17438 Note that as a consequence of the application of this filter, the
17439 output display aspect ratio will change according to the equation
17442 Keep in mind that the sample aspect ratio set by the @code{setsar}
17443 filter may be changed by later filters in the filterchain, e.g. if
17444 another "setsar" or a "setdar" filter is applied.
17446 It accepts the following parameters:
17449 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17450 Set the aspect ratio used by the filter.
17452 The parameter can be a floating point number string, an expression, or
17453 a string of the form @var{num}:@var{den}, where @var{num} and
17454 @var{den} are the numerator and denominator of the aspect ratio. If
17455 the parameter is not specified, it is assumed the value "0".
17456 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17460 Set the maximum integer value to use for expressing numerator and
17461 denominator when reducing the expressed aspect ratio to a rational.
17462 Default value is @code{100}.
17466 The parameter @var{sar} is an expression containing
17467 the following constants:
17471 These are approximated values for the mathematical constants e
17472 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17475 The input width and height.
17478 These are the same as @var{w} / @var{h}.
17481 The input sample aspect ratio.
17484 The input display aspect ratio. It is the same as
17485 (@var{w} / @var{h}) * @var{sar}.
17488 Horizontal and vertical chroma subsample values. For example, for the
17489 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17492 @subsection Examples
17497 To change the display aspect ratio to 16:9, specify one of the following:
17504 To change the sample aspect ratio to 10:11, specify:
17510 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17511 1000 in the aspect ratio reduction, use the command:
17513 setdar=ratio=16/9:max=1000
17521 Force field for the output video frame.
17523 The @code{setfield} filter marks the interlace type field for the
17524 output frames. It does not change the input frame, but only sets the
17525 corresponding property, which affects how the frame is treated by
17526 following filters (e.g. @code{fieldorder} or @code{yadif}).
17528 The filter accepts the following options:
17533 Available values are:
17537 Keep the same field property.
17540 Mark the frame as bottom-field-first.
17543 Mark the frame as top-field-first.
17546 Mark the frame as progressive.
17553 Force frame parameter for the output video frame.
17555 The @code{setparams} filter marks interlace and color range for the
17556 output frames. It does not change the input frame, but only sets the
17557 corresponding property, which affects how the frame is treated by
17562 Available values are:
17566 Keep the same field property (default).
17569 Mark the frame as bottom-field-first.
17572 Mark the frame as top-field-first.
17575 Mark the frame as progressive.
17579 Available values are:
17583 Keep the same color range property (default).
17585 @item unspecified, unknown
17586 Mark the frame as unspecified color range.
17588 @item limited, tv, mpeg
17589 Mark the frame as limited range.
17591 @item full, pc, jpeg
17592 Mark the frame as full range.
17595 @item color_primaries
17596 Set the color primaries.
17597 Available values are:
17601 Keep the same color primaries property (default).
17618 Set the color transfer.
17619 Available values are:
17623 Keep the same color trc property (default).
17645 Set the colorspace.
17646 Available values are:
17650 Keep the same colorspace property (default).
17663 @item chroma-derived-nc
17664 @item chroma-derived-c
17671 Show a line containing various information for each input video frame.
17672 The input video is not modified.
17674 This filter supports the following options:
17678 Calculate checksums of each plane. By default enabled.
17681 The shown line contains a sequence of key/value pairs of the form
17682 @var{key}:@var{value}.
17684 The following values are shown in the output:
17688 The (sequential) number of the input frame, starting from 0.
17691 The Presentation TimeStamp of the input frame, expressed as a number of
17692 time base units. The time base unit depends on the filter input pad.
17695 The Presentation TimeStamp of the input frame, expressed as a number of
17699 The position of the frame in the input stream, or -1 if this information is
17700 unavailable and/or meaningless (for example in case of synthetic video).
17703 The pixel format name.
17706 The sample aspect ratio of the input frame, expressed in the form
17707 @var{num}/@var{den}.
17710 The size of the input frame. For the syntax of this option, check the
17711 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17714 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17715 for bottom field first).
17718 This is 1 if the frame is a key frame, 0 otherwise.
17721 The picture type of the input frame ("I" for an I-frame, "P" for a
17722 P-frame, "B" for a B-frame, or "?" for an unknown type).
17723 Also refer to the documentation of the @code{AVPictureType} enum and of
17724 the @code{av_get_picture_type_char} function defined in
17725 @file{libavutil/avutil.h}.
17728 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17730 @item plane_checksum
17731 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17732 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17735 The mean value of pixels in each plane of the input frame, expressed in the form
17736 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17739 The standard deviation of pixel values in each plane of the input frame, expressed
17740 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17744 @section showpalette
17746 Displays the 256 colors palette of each frame. This filter is only relevant for
17747 @var{pal8} pixel format frames.
17749 It accepts the following option:
17753 Set the size of the box used to represent one palette color entry. Default is
17754 @code{30} (for a @code{30x30} pixel box).
17757 @section shuffleframes
17759 Reorder and/or duplicate and/or drop video frames.
17761 It accepts the following parameters:
17765 Set the destination indexes of input frames.
17766 This is space or '|' separated list of indexes that maps input frames to output
17767 frames. Number of indexes also sets maximal value that each index may have.
17768 '-1' index have special meaning and that is to drop frame.
17771 The first frame has the index 0. The default is to keep the input unchanged.
17773 @subsection Examples
17777 Swap second and third frame of every three frames of the input:
17779 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17783 Swap 10th and 1st frame of every ten frames of the input:
17785 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17789 @section shuffleplanes
17791 Reorder and/or duplicate video planes.
17793 It accepts the following parameters:
17798 The index of the input plane to be used as the first output plane.
17801 The index of the input plane to be used as the second output plane.
17804 The index of the input plane to be used as the third output plane.
17807 The index of the input plane to be used as the fourth output plane.
17811 The first plane has the index 0. The default is to keep the input unchanged.
17813 @subsection Examples
17817 Swap the second and third planes of the input:
17819 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17823 @anchor{signalstats}
17824 @section signalstats
17825 Evaluate various visual metrics that assist in determining issues associated
17826 with the digitization of analog video media.
17828 By default the filter will log these metadata values:
17832 Display the minimal Y value contained within the input frame. Expressed in
17836 Display the Y value at the 10% percentile within the input frame. Expressed in
17840 Display the average Y value within the input frame. Expressed in range of
17844 Display the Y value at the 90% percentile within the input frame. Expressed in
17848 Display the maximum Y value contained within the input frame. Expressed in
17852 Display the minimal U value contained within the input frame. Expressed in
17856 Display the U value at the 10% percentile within the input frame. Expressed in
17860 Display the average U value within the input frame. Expressed in range of
17864 Display the U value at the 90% percentile within the input frame. Expressed in
17868 Display the maximum U value contained within the input frame. Expressed in
17872 Display the minimal V value contained within the input frame. Expressed in
17876 Display the V value at the 10% percentile within the input frame. Expressed in
17880 Display the average V value within the input frame. Expressed in range of
17884 Display the V value at the 90% percentile within the input frame. Expressed in
17888 Display the maximum V value contained within the input frame. Expressed in
17892 Display the minimal saturation value contained within the input frame.
17893 Expressed in range of [0-~181.02].
17896 Display the saturation value at the 10% percentile within the input frame.
17897 Expressed in range of [0-~181.02].
17900 Display the average saturation value within the input frame. Expressed in range
17904 Display the saturation value at the 90% percentile within the input frame.
17905 Expressed in range of [0-~181.02].
17908 Display the maximum saturation value contained within the input frame.
17909 Expressed in range of [0-~181.02].
17912 Display the median value for hue within the input frame. Expressed in range of
17916 Display the average value for hue within the input frame. Expressed in range of
17920 Display the average of sample value difference between all values of the Y
17921 plane in the current frame and corresponding values of the previous input frame.
17922 Expressed in range of [0-255].
17925 Display the average of sample value difference between all values of the U
17926 plane in the current frame and corresponding values of the previous input frame.
17927 Expressed in range of [0-255].
17930 Display the average of sample value difference between all values of the V
17931 plane in the current frame and corresponding values of the previous input frame.
17932 Expressed in range of [0-255].
17935 Display bit depth of Y plane in current frame.
17936 Expressed in range of [0-16].
17939 Display bit depth of U plane in current frame.
17940 Expressed in range of [0-16].
17943 Display bit depth of V plane in current frame.
17944 Expressed in range of [0-16].
17947 The filter accepts the following options:
17953 @option{stat} specify an additional form of image analysis.
17954 @option{out} output video with the specified type of pixel highlighted.
17956 Both options accept the following values:
17960 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17961 unlike the neighboring pixels of the same field. Examples of temporal outliers
17962 include the results of video dropouts, head clogs, or tape tracking issues.
17965 Identify @var{vertical line repetition}. Vertical line repetition includes
17966 similar rows of pixels within a frame. In born-digital video vertical line
17967 repetition is common, but this pattern is uncommon in video digitized from an
17968 analog source. When it occurs in video that results from the digitization of an
17969 analog source it can indicate concealment from a dropout compensator.
17972 Identify pixels that fall outside of legal broadcast range.
17976 Set the highlight color for the @option{out} option. The default color is
17980 @subsection Examples
17984 Output data of various video metrics:
17986 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17990 Output specific data about the minimum and maximum values of the Y plane per frame:
17992 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17996 Playback video while highlighting pixels that are outside of broadcast range in red.
17998 ffplay example.mov -vf signalstats="out=brng:color=red"
18002 Playback video with signalstats metadata drawn over the frame.
18004 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18007 The contents of signalstat_drawtext.txt used in the command are:
18010 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
18011 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
18012 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
18013 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
18021 Calculates the MPEG-7 Video Signature. The filter can handle more than one
18022 input. In this case the matching between the inputs can be calculated additionally.
18023 The filter always passes through the first input. The signature of each stream can
18024 be written into a file.
18026 It accepts the following options:
18030 Enable or disable the matching process.
18032 Available values are:
18036 Disable the calculation of a matching (default).
18038 Calculate the matching for the whole video and output whether the whole video
18039 matches or only parts.
18041 Calculate only until a matching is found or the video ends. Should be faster in
18046 Set the number of inputs. The option value must be a non negative integer.
18047 Default value is 1.
18050 Set the path to which the output is written. If there is more than one input,
18051 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
18052 integer), that will be replaced with the input number. If no filename is
18053 specified, no output will be written. This is the default.
18056 Choose the output format.
18058 Available values are:
18062 Use the specified binary representation (default).
18064 Use the specified xml representation.
18068 Set threshold to detect one word as similar. The option value must be an integer
18069 greater than zero. The default value is 9000.
18072 Set threshold to detect all words as similar. The option value must be an integer
18073 greater than zero. The default value is 60000.
18076 Set threshold to detect frames as similar. The option value must be an integer
18077 greater than zero. The default value is 116.
18080 Set the minimum length of a sequence in frames to recognize it as matching
18081 sequence. The option value must be a non negative integer value.
18082 The default value is 0.
18085 Set the minimum relation, that matching frames to all frames must have.
18086 The option value must be a double value between 0 and 1. The default value is 0.5.
18089 @subsection Examples
18093 To calculate the signature of an input video and store it in signature.bin:
18095 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18099 To detect whether two videos match and store the signatures in XML format in
18100 signature0.xml and signature1.xml:
18102 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 -
18110 Blur the input video without impacting the outlines.
18112 It accepts the following options:
18115 @item luma_radius, lr
18116 Set the luma radius. The option value must be a float number in
18117 the range [0.1,5.0] that specifies the variance of the gaussian filter
18118 used to blur the image (slower if larger). Default value is 1.0.
18120 @item luma_strength, ls
18121 Set the luma strength. The option value must be a float number
18122 in the range [-1.0,1.0] that configures the blurring. A value included
18123 in [0.0,1.0] will blur the image whereas a value included in
18124 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18126 @item luma_threshold, lt
18127 Set the luma threshold used as a coefficient to determine
18128 whether a pixel should be blurred or not. The option value must be an
18129 integer in the range [-30,30]. A value of 0 will filter all the image,
18130 a value included in [0,30] will filter flat areas and a value included
18131 in [-30,0] will filter edges. Default value is 0.
18133 @item chroma_radius, cr
18134 Set the chroma radius. The option value must be a float number in
18135 the range [0.1,5.0] that specifies the variance of the gaussian filter
18136 used to blur the image (slower if larger). Default value is @option{luma_radius}.
18138 @item chroma_strength, cs
18139 Set the chroma strength. The option value must be a float number
18140 in the range [-1.0,1.0] that configures the blurring. A value included
18141 in [0.0,1.0] will blur the image whereas a value included in
18142 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
18144 @item chroma_threshold, ct
18145 Set the chroma threshold used as a coefficient to determine
18146 whether a pixel should be blurred or not. The option value must be an
18147 integer in the range [-30,30]. A value of 0 will filter all the image,
18148 a value included in [0,30] will filter flat areas and a value included
18149 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
18152 If a chroma option is not explicitly set, the corresponding luma value
18156 Apply sobel operator to input video stream.
18158 The filter accepts the following option:
18162 Set which planes will be processed, unprocessed planes will be copied.
18163 By default value 0xf, all planes will be processed.
18166 Set value which will be multiplied with filtered result.
18169 Set value which will be added to filtered result.
18172 @subsection Commands
18174 This filter supports the all above options as @ref{commands}.
18179 Apply a simple postprocessing filter that compresses and decompresses the image
18180 at several (or - in the case of @option{quality} level @code{6} - all) shifts
18181 and average the results.
18183 The filter accepts the following options:
18187 Set quality. This option defines the number of levels for averaging. It accepts
18188 an integer in the range 0-6. If set to @code{0}, the filter will have no
18189 effect. A value of @code{6} means the higher quality. For each increment of
18190 that value the speed drops by a factor of approximately 2. Default value is
18194 Force a constant quantization parameter. If not set, the filter will use the QP
18195 from the video stream (if available).
18198 Set thresholding mode. Available modes are:
18202 Set hard thresholding (default).
18204 Set soft thresholding (better de-ringing effect, but likely blurrier).
18207 @item use_bframe_qp
18208 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
18209 option may cause flicker since the B-Frames have often larger QP. Default is
18210 @code{0} (not enabled).
18213 @subsection Commands
18215 This filter supports the following commands:
18217 @item quality, level
18218 Set quality level. The value @code{max} can be used to set the maximum level,
18219 currently @code{6}.
18225 Scale the input by applying one of the super-resolution methods based on
18226 convolutional neural networks. Supported models:
18230 Super-Resolution Convolutional Neural Network model (SRCNN).
18231 See @url{https://arxiv.org/abs/1501.00092}.
18234 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18235 See @url{https://arxiv.org/abs/1609.05158}.
18238 Training scripts as well as scripts for model file (.pb) saving can be found at
18239 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
18240 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
18242 Native model files (.model) can be generated from TensorFlow model
18243 files (.pb) by using tools/python/convert.py
18245 The filter accepts the following options:
18249 Specify which DNN backend to use for model loading and execution. This option accepts
18250 the following values:
18254 Native implementation of DNN loading and execution.
18257 TensorFlow backend. To enable this backend you
18258 need to install the TensorFlow for C library (see
18259 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
18260 @code{--enable-libtensorflow}
18263 Default value is @samp{native}.
18266 Set path to model file specifying network architecture and its parameters.
18267 Note that different backends use different file formats. TensorFlow backend
18268 can load files for both formats, while native backend can load files for only
18272 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
18273 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
18274 input upscaled using bicubic upscaling with proper scale factor.
18277 This feature can also be finished with @ref{dnn_processing} filter.
18281 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
18283 This filter takes in input two input videos, the first input is
18284 considered the "main" source and is passed unchanged to the
18285 output. The second input is used as a "reference" video for computing
18288 Both video inputs must have the same resolution and pixel format for
18289 this filter to work correctly. Also it assumes that both inputs
18290 have the same number of frames, which are compared one by one.
18292 The filter stores the calculated SSIM of each frame.
18294 The description of the accepted parameters follows.
18297 @item stats_file, f
18298 If specified the filter will use the named file to save the SSIM of
18299 each individual frame. When filename equals "-" the data is sent to
18303 The file printed if @var{stats_file} is selected, contains a sequence of
18304 key/value pairs of the form @var{key}:@var{value} for each compared
18307 A description of each shown parameter follows:
18311 sequential number of the input frame, starting from 1
18313 @item Y, U, V, R, G, B
18314 SSIM of the compared frames for the component specified by the suffix.
18317 SSIM of the compared frames for the whole frame.
18320 Same as above but in dB representation.
18323 This filter also supports the @ref{framesync} options.
18325 @subsection Examples
18330 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18331 [main][ref] ssim="stats_file=stats.log" [out]
18334 On this example the input file being processed is compared with the
18335 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18336 is stored in @file{stats.log}.
18339 Another example with both psnr and ssim at same time:
18341 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18345 Another example with different containers:
18347 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 -
18353 Convert between different stereoscopic image formats.
18355 The filters accept the following options:
18359 Set stereoscopic image format of input.
18361 Available values for input image formats are:
18364 side by side parallel (left eye left, right eye right)
18367 side by side crosseye (right eye left, left eye right)
18370 side by side parallel with half width resolution
18371 (left eye left, right eye right)
18374 side by side crosseye with half width resolution
18375 (right eye left, left eye right)
18379 above-below (left eye above, right eye below)
18383 above-below (right eye above, left eye below)
18387 above-below with half height resolution
18388 (left eye above, right eye below)
18392 above-below with half height resolution
18393 (right eye above, left eye below)
18396 alternating frames (left eye first, right eye second)
18399 alternating frames (right eye first, left eye second)
18402 interleaved rows (left eye has top row, right eye starts on next row)
18405 interleaved rows (right eye has top row, left eye starts on next row)
18408 interleaved columns, left eye first
18411 interleaved columns, right eye first
18413 Default value is @samp{sbsl}.
18417 Set stereoscopic image format of output.
18421 side by side parallel (left eye left, right eye right)
18424 side by side crosseye (right eye left, left eye right)
18427 side by side parallel with half width resolution
18428 (left eye left, right eye right)
18431 side by side crosseye with half width resolution
18432 (right eye left, left eye right)
18436 above-below (left eye above, right eye below)
18440 above-below (right eye above, left eye below)
18444 above-below with half height resolution
18445 (left eye above, right eye below)
18449 above-below with half height resolution
18450 (right eye above, left eye below)
18453 alternating frames (left eye first, right eye second)
18456 alternating frames (right eye first, left eye second)
18459 interleaved rows (left eye has top row, right eye starts on next row)
18462 interleaved rows (right eye has top row, left eye starts on next row)
18465 anaglyph red/blue gray
18466 (red filter on left eye, blue filter on right eye)
18469 anaglyph red/green gray
18470 (red filter on left eye, green filter on right eye)
18473 anaglyph red/cyan gray
18474 (red filter on left eye, cyan filter on right eye)
18477 anaglyph red/cyan half colored
18478 (red filter on left eye, cyan filter on right eye)
18481 anaglyph red/cyan color
18482 (red filter on left eye, cyan filter on right eye)
18485 anaglyph red/cyan color optimized with the least squares projection of dubois
18486 (red filter on left eye, cyan filter on right eye)
18489 anaglyph green/magenta gray
18490 (green filter on left eye, magenta filter on right eye)
18493 anaglyph green/magenta half colored
18494 (green filter on left eye, magenta filter on right eye)
18497 anaglyph green/magenta colored
18498 (green filter on left eye, magenta filter on right eye)
18501 anaglyph green/magenta color optimized with the least squares projection of dubois
18502 (green filter on left eye, magenta filter on right eye)
18505 anaglyph yellow/blue gray
18506 (yellow filter on left eye, blue filter on right eye)
18509 anaglyph yellow/blue half colored
18510 (yellow filter on left eye, blue filter on right eye)
18513 anaglyph yellow/blue colored
18514 (yellow filter on left eye, blue filter on right eye)
18517 anaglyph yellow/blue color optimized with the least squares projection of dubois
18518 (yellow filter on left eye, blue filter on right eye)
18521 mono output (left eye only)
18524 mono output (right eye only)
18527 checkerboard, left eye first
18530 checkerboard, right eye first
18533 interleaved columns, left eye first
18536 interleaved columns, right eye first
18542 Default value is @samp{arcd}.
18545 @subsection Examples
18549 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18555 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18561 @section streamselect, astreamselect
18562 Select video or audio streams.
18564 The filter accepts the following options:
18568 Set number of inputs. Default is 2.
18571 Set input indexes to remap to outputs.
18574 @subsection Commands
18576 The @code{streamselect} and @code{astreamselect} filter supports the following
18581 Set input indexes to remap to outputs.
18584 @subsection Examples
18588 Select first 5 seconds 1st stream and rest of time 2nd stream:
18590 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18594 Same as above, but for audio:
18596 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18603 Draw subtitles on top of input video using the libass library.
18605 To enable compilation of this filter you need to configure FFmpeg with
18606 @code{--enable-libass}. This filter also requires a build with libavcodec and
18607 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18608 Alpha) subtitles format.
18610 The filter accepts the following options:
18614 Set the filename of the subtitle file to read. It must be specified.
18616 @item original_size
18617 Specify the size of the original video, the video for which the ASS file
18618 was composed. For the syntax of this option, check the
18619 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18620 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18621 correctly scale the fonts if the aspect ratio has been changed.
18624 Set a directory path containing fonts that can be used by the filter.
18625 These fonts will be used in addition to whatever the font provider uses.
18628 Process alpha channel, by default alpha channel is untouched.
18631 Set subtitles input character encoding. @code{subtitles} filter only. Only
18632 useful if not UTF-8.
18634 @item stream_index, si
18635 Set subtitles stream index. @code{subtitles} filter only.
18638 Override default style or script info parameters of the subtitles. It accepts a
18639 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18642 If the first key is not specified, it is assumed that the first value
18643 specifies the @option{filename}.
18645 For example, to render the file @file{sub.srt} on top of the input
18646 video, use the command:
18651 which is equivalent to:
18653 subtitles=filename=sub.srt
18656 To render the default subtitles stream from file @file{video.mkv}, use:
18658 subtitles=video.mkv
18661 To render the second subtitles stream from that file, use:
18663 subtitles=video.mkv:si=1
18666 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18667 @code{DejaVu Serif}, use:
18669 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18672 @section super2xsai
18674 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18675 Interpolate) pixel art scaling algorithm.
18677 Useful for enlarging pixel art images without reducing sharpness.
18681 Swap two rectangular objects in video.
18683 This filter accepts the following options:
18693 Set 1st rect x coordinate.
18696 Set 1st rect y coordinate.
18699 Set 2nd rect x coordinate.
18702 Set 2nd rect y coordinate.
18704 All expressions are evaluated once for each frame.
18707 The all options are expressions containing the following constants:
18712 The input width and height.
18715 same as @var{w} / @var{h}
18718 input sample aspect ratio
18721 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18724 The number of the input frame, starting from 0.
18727 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18730 the position in the file of the input frame, NAN if unknown
18737 Blend successive video frames.
18743 Apply telecine process to the video.
18745 This filter accepts the following options:
18754 The default value is @code{top}.
18758 A string of numbers representing the pulldown pattern you wish to apply.
18759 The default value is @code{23}.
18763 Some typical patterns:
18768 24p: 2332 (preferred)
18775 24p: 222222222223 ("Euro pulldown")
18780 @section thistogram
18782 Compute and draw a color distribution histogram for the input video across time.
18784 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18785 at certain time, this filter shows also past histograms of number of frames defined
18786 by @code{width} option.
18788 The computed histogram is a representation of the color component
18789 distribution in an image.
18791 The filter accepts the following options:
18795 Set width of single color component output. Default value is @code{0}.
18796 Value of @code{0} means width will be picked from input video.
18797 This also set number of passed histograms to keep.
18798 Allowed range is [0, 8192].
18800 @item display_mode, d
18802 It accepts the following values:
18805 Per color component graphs are placed below each other.
18808 Per color component graphs are placed side by side.
18811 Presents information identical to that in the @code{parade}, except
18812 that the graphs representing color components are superimposed directly
18815 Default is @code{stack}.
18817 @item levels_mode, m
18818 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18819 Default is @code{linear}.
18821 @item components, c
18822 Set what color components to display.
18823 Default is @code{7}.
18826 Set background opacity. Default is @code{0.9}.
18829 Show envelope. Default is disabled.
18832 Set envelope color. Default is @code{gold}.
18837 Available values for slide is:
18840 Draw new frame when right border is reached.
18843 Replace old columns with new ones.
18846 Scroll from right to left.
18849 Scroll from left to right.
18852 Draw single picture.
18855 Default is @code{replace}.
18860 Apply threshold effect to video stream.
18862 This filter needs four video streams to perform thresholding.
18863 First stream is stream we are filtering.
18864 Second stream is holding threshold values, third stream is holding min values,
18865 and last, fourth stream is holding max values.
18867 The filter accepts the following option:
18871 Set which planes will be processed, unprocessed planes will be copied.
18872 By default value 0xf, all planes will be processed.
18875 For example if first stream pixel's component value is less then threshold value
18876 of pixel component from 2nd threshold stream, third stream value will picked,
18877 otherwise fourth stream pixel component value will be picked.
18879 Using color source filter one can perform various types of thresholding:
18881 @subsection Examples
18885 Binary threshold, using gray color as threshold:
18887 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18891 Inverted binary threshold, using gray color as threshold:
18893 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18897 Truncate binary threshold, using gray color as threshold:
18899 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18903 Threshold to zero, using gray color as threshold:
18905 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18909 Inverted threshold to zero, using gray color as threshold:
18911 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18916 Select the most representative frame in a given sequence of consecutive frames.
18918 The filter accepts the following options:
18922 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18923 will pick one of them, and then handle the next batch of @var{n} frames until
18924 the end. Default is @code{100}.
18927 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18928 value will result in a higher memory usage, so a high value is not recommended.
18930 @subsection Examples
18934 Extract one picture each 50 frames:
18940 Complete example of a thumbnail creation with @command{ffmpeg}:
18942 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18949 Tile several successive frames together.
18951 The @ref{untile} filter can do the reverse.
18953 The filter accepts the following options:
18958 Set the grid size (i.e. the number of lines and columns). For the syntax of
18959 this option, check the
18960 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18963 Set the maximum number of frames to render in the given area. It must be less
18964 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18965 the area will be used.
18968 Set the outer border margin in pixels.
18971 Set the inner border thickness (i.e. the number of pixels between frames). For
18972 more advanced padding options (such as having different values for the edges),
18973 refer to the pad video filter.
18976 Specify the color of the unused area. For the syntax of this option, check the
18977 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18978 The default value of @var{color} is "black".
18981 Set the number of frames to overlap when tiling several successive frames together.
18982 The value must be between @code{0} and @var{nb_frames - 1}.
18985 Set the number of frames to initially be empty before displaying first output frame.
18986 This controls how soon will one get first output frame.
18987 The value must be between @code{0} and @var{nb_frames - 1}.
18990 @subsection Examples
18994 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18996 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18998 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18999 duplicating each output frame to accommodate the originally detected frame
19003 Display @code{5} pictures in an area of @code{3x2} frames,
19004 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
19005 mixed flat and named options:
19007 tile=3x2:nb_frames=5:padding=7:margin=2
19011 @section tinterlace
19013 Perform various types of temporal field interlacing.
19015 Frames are counted starting from 1, so the first input frame is
19018 The filter accepts the following options:
19023 Specify the mode of the interlacing. This option can also be specified
19024 as a value alone. See below for a list of values for this option.
19026 Available values are:
19030 Move odd frames into the upper field, even into the lower field,
19031 generating a double height frame at half frame rate.
19035 Frame 1 Frame 2 Frame 3 Frame 4
19037 11111 22222 33333 44444
19038 11111 22222 33333 44444
19039 11111 22222 33333 44444
19040 11111 22222 33333 44444
19054 Only output odd frames, even frames are dropped, generating a frame with
19055 unchanged height at half frame rate.
19060 Frame 1 Frame 2 Frame 3 Frame 4
19062 11111 22222 33333 44444
19063 11111 22222 33333 44444
19064 11111 22222 33333 44444
19065 11111 22222 33333 44444
19075 Only output even frames, odd frames are dropped, generating a frame with
19076 unchanged height at half frame rate.
19081 Frame 1 Frame 2 Frame 3 Frame 4
19083 11111 22222 33333 44444
19084 11111 22222 33333 44444
19085 11111 22222 33333 44444
19086 11111 22222 33333 44444
19096 Expand each frame to full height, but pad alternate lines with black,
19097 generating a frame with double height at the same input frame rate.
19102 Frame 1 Frame 2 Frame 3 Frame 4
19104 11111 22222 33333 44444
19105 11111 22222 33333 44444
19106 11111 22222 33333 44444
19107 11111 22222 33333 44444
19110 11111 ..... 33333 .....
19111 ..... 22222 ..... 44444
19112 11111 ..... 33333 .....
19113 ..... 22222 ..... 44444
19114 11111 ..... 33333 .....
19115 ..... 22222 ..... 44444
19116 11111 ..... 33333 .....
19117 ..... 22222 ..... 44444
19121 @item interleave_top, 4
19122 Interleave the upper field from odd frames with the lower field from
19123 even frames, generating a frame with unchanged height at half frame rate.
19128 Frame 1 Frame 2 Frame 3 Frame 4
19130 11111<- 22222 33333<- 44444
19131 11111 22222<- 33333 44444<-
19132 11111<- 22222 33333<- 44444
19133 11111 22222<- 33333 44444<-
19143 @item interleave_bottom, 5
19144 Interleave the lower field from odd frames with the upper field from
19145 even frames, generating a frame with unchanged height at half frame rate.
19150 Frame 1 Frame 2 Frame 3 Frame 4
19152 11111 22222<- 33333 44444<-
19153 11111<- 22222 33333<- 44444
19154 11111 22222<- 33333 44444<-
19155 11111<- 22222 33333<- 44444
19165 @item interlacex2, 6
19166 Double frame rate with unchanged height. Frames are inserted each
19167 containing the second temporal field from the previous input frame and
19168 the first temporal field from the next input frame. This mode relies on
19169 the top_field_first flag. Useful for interlaced video displays with no
19170 field synchronisation.
19175 Frame 1 Frame 2 Frame 3 Frame 4
19177 11111 22222 33333 44444
19178 11111 22222 33333 44444
19179 11111 22222 33333 44444
19180 11111 22222 33333 44444
19183 11111 22222 22222 33333 33333 44444 44444
19184 11111 11111 22222 22222 33333 33333 44444
19185 11111 22222 22222 33333 33333 44444 44444
19186 11111 11111 22222 22222 33333 33333 44444
19191 Move odd frames into the upper field, even into the lower field,
19192 generating a double height frame at same frame rate.
19197 Frame 1 Frame 2 Frame 3 Frame 4
19199 11111 22222 33333 44444
19200 11111 22222 33333 44444
19201 11111 22222 33333 44444
19202 11111 22222 33333 44444
19205 11111 33333 33333 55555
19206 22222 22222 44444 44444
19207 11111 33333 33333 55555
19208 22222 22222 44444 44444
19209 11111 33333 33333 55555
19210 22222 22222 44444 44444
19211 11111 33333 33333 55555
19212 22222 22222 44444 44444
19217 Numeric values are deprecated but are accepted for backward
19218 compatibility reasons.
19220 Default mode is @code{merge}.
19223 Specify flags influencing the filter process.
19225 Available value for @var{flags} is:
19228 @item low_pass_filter, vlpf
19229 Enable linear vertical low-pass filtering in the filter.
19230 Vertical low-pass filtering is required when creating an interlaced
19231 destination from a progressive source which contains high-frequency
19232 vertical detail. Filtering will reduce interlace 'twitter' and Moire
19235 @item complex_filter, cvlpf
19236 Enable complex vertical low-pass filtering.
19237 This will slightly less reduce interlace 'twitter' and Moire
19238 patterning but better retain detail and subjective sharpness impression.
19241 Bypass already interlaced frames, only adjust the frame rate.
19244 Vertical low-pass filtering and bypassing already interlaced frames can only be
19245 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
19250 Pick median pixels from several successive input video frames.
19252 The filter accepts the following options:
19256 Set radius of median filter.
19257 Default is 1. Allowed range is from 1 to 127.
19260 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19263 Set median percentile. Default value is @code{0.5}.
19264 Default value of @code{0.5} will pick always median values, while @code{0} will pick
19265 minimum values, and @code{1} maximum values.
19270 Mix successive video frames.
19272 A description of the accepted options follows.
19276 The number of successive frames to mix. If unspecified, it defaults to 3.
19279 Specify weight of each input video frame.
19280 Each weight is separated by space. If number of weights is smaller than
19281 number of @var{frames} last specified weight will be used for all remaining
19285 Specify scale, if it is set it will be multiplied with sum
19286 of each weight multiplied with pixel values to give final destination
19287 pixel value. By default @var{scale} is auto scaled to sum of weights.
19290 @subsection Examples
19294 Average 7 successive frames:
19296 tmix=frames=7:weights="1 1 1 1 1 1 1"
19300 Apply simple temporal convolution:
19302 tmix=frames=3:weights="-1 3 -1"
19306 Similar as above but only showing temporal differences:
19308 tmix=frames=3:weights="-1 2 -1":scale=1
19314 Tone map colors from different dynamic ranges.
19316 This filter expects data in single precision floating point, as it needs to
19317 operate on (and can output) out-of-range values. Another filter, such as
19318 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19320 The tonemapping algorithms implemented only work on linear light, so input
19321 data should be linearized beforehand (and possibly correctly tagged).
19324 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19327 @subsection Options
19328 The filter accepts the following options.
19332 Set the tone map algorithm to use.
19334 Possible values are:
19337 Do not apply any tone map, only desaturate overbright pixels.
19340 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19341 in-range values, while distorting out-of-range values.
19344 Stretch the entire reference gamut to a linear multiple of the display.
19347 Fit a logarithmic transfer between the tone curves.
19350 Preserve overall image brightness with a simple curve, using nonlinear
19351 contrast, which results in flattening details and degrading color accuracy.
19354 Preserve both dark and bright details better than @var{reinhard}, at the cost
19355 of slightly darkening everything. Use it when detail preservation is more
19356 important than color and brightness accuracy.
19359 Smoothly map out-of-range values, while retaining contrast and colors for
19360 in-range material as much as possible. Use it when color accuracy is more
19361 important than detail preservation.
19367 Tune the tone mapping algorithm.
19369 This affects the following algorithms:
19375 Specifies the scale factor to use while stretching.
19379 Specifies the exponent of the function.
19383 Specify an extra linear coefficient to multiply into the signal before clipping.
19387 Specify the local contrast coefficient at the display peak.
19388 Default to 0.5, which means that in-gamut values will be about half as bright
19395 Specify the transition point from linear to mobius transform. Every value
19396 below this point is guaranteed to be mapped 1:1. The higher the value, the
19397 more accurate the result will be, at the cost of losing bright details.
19398 Default to 0.3, which due to the steep initial slope still preserves in-range
19399 colors fairly accurately.
19403 Apply desaturation for highlights that exceed this level of brightness. The
19404 higher the parameter, the more color information will be preserved. This
19405 setting helps prevent unnaturally blown-out colors for super-highlights, by
19406 (smoothly) turning into white instead. This makes images feel more natural,
19407 at the cost of reducing information about out-of-range colors.
19409 The default of 2.0 is somewhat conservative and will mostly just apply to
19410 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19412 This option works only if the input frame has a supported color tag.
19415 Override signal/nominal/reference peak with this value. Useful when the
19416 embedded peak information in display metadata is not reliable or when tone
19417 mapping from a lower range to a higher range.
19422 Temporarily pad video frames.
19424 The filter accepts the following options:
19428 Specify number of delay frames before input video stream. Default is 0.
19431 Specify number of padding frames after input video stream.
19432 Set to -1 to pad indefinitely. Default is 0.
19435 Set kind of frames added to beginning of stream.
19436 Can be either @var{add} or @var{clone}.
19437 With @var{add} frames of solid-color are added.
19438 With @var{clone} frames are clones of first frame.
19439 Default is @var{add}.
19442 Set kind of frames added to end of stream.
19443 Can be either @var{add} or @var{clone}.
19444 With @var{add} frames of solid-color are added.
19445 With @var{clone} frames are clones of last frame.
19446 Default is @var{add}.
19448 @item start_duration, stop_duration
19449 Specify the duration of the start/stop delay. See
19450 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19451 for the accepted syntax.
19452 These options override @var{start} and @var{stop}. Default is 0.
19455 Specify the color of the padded area. For the syntax of this option,
19456 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19457 manual,ffmpeg-utils}.
19459 The default value of @var{color} is "black".
19465 Transpose rows with columns in the input video and optionally flip it.
19467 It accepts the following parameters:
19472 Specify the transposition direction.
19474 Can assume the following values:
19476 @item 0, 4, cclock_flip
19477 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19485 Rotate by 90 degrees clockwise, that is:
19493 Rotate by 90 degrees counterclockwise, that is:
19500 @item 3, 7, clock_flip
19501 Rotate by 90 degrees clockwise and vertically flip, that is:
19509 For values between 4-7, the transposition is only done if the input
19510 video geometry is portrait and not landscape. These values are
19511 deprecated, the @code{passthrough} option should be used instead.
19513 Numerical values are deprecated, and should be dropped in favor of
19514 symbolic constants.
19517 Do not apply the transposition if the input geometry matches the one
19518 specified by the specified value. It accepts the following values:
19521 Always apply transposition.
19523 Preserve portrait geometry (when @var{height} >= @var{width}).
19525 Preserve landscape geometry (when @var{width} >= @var{height}).
19528 Default value is @code{none}.
19531 For example to rotate by 90 degrees clockwise and preserve portrait
19534 transpose=dir=1:passthrough=portrait
19537 The command above can also be specified as:
19539 transpose=1:portrait
19542 @section transpose_npp
19544 Transpose rows with columns in the input video and optionally flip it.
19545 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19547 It accepts the following parameters:
19552 Specify the transposition direction.
19554 Can assume the following values:
19557 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19560 Rotate by 90 degrees clockwise.
19563 Rotate by 90 degrees counterclockwise.
19566 Rotate by 90 degrees clockwise and vertically flip.
19570 Do not apply the transposition if the input geometry matches the one
19571 specified by the specified value. It accepts the following values:
19574 Always apply transposition. (default)
19576 Preserve portrait geometry (when @var{height} >= @var{width}).
19578 Preserve landscape geometry (when @var{width} >= @var{height}).
19584 Trim the input so that the output contains one continuous subpart of the input.
19586 It accepts the following parameters:
19589 Specify the time of the start of the kept section, i.e. the frame with the
19590 timestamp @var{start} will be the first frame in the output.
19593 Specify the time of the first frame that will be dropped, i.e. the frame
19594 immediately preceding the one with the timestamp @var{end} will be the last
19595 frame in the output.
19598 This is the same as @var{start}, except this option sets the start timestamp
19599 in timebase units instead of seconds.
19602 This is the same as @var{end}, except this option sets the end timestamp
19603 in timebase units instead of seconds.
19606 The maximum duration of the output in seconds.
19609 The number of the first frame that should be passed to the output.
19612 The number of the first frame that should be dropped.
19615 @option{start}, @option{end}, and @option{duration} are expressed as time
19616 duration specifications; see
19617 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19618 for the accepted syntax.
19620 Note that the first two sets of the start/end options and the @option{duration}
19621 option look at the frame timestamp, while the _frame variants simply count the
19622 frames that pass through the filter. Also note that this filter does not modify
19623 the timestamps. If you wish for the output timestamps to start at zero, insert a
19624 setpts filter after the trim filter.
19626 If multiple start or end options are set, this filter tries to be greedy and
19627 keep all the frames that match at least one of the specified constraints. To keep
19628 only the part that matches all the constraints at once, chain multiple trim
19631 The defaults are such that all the input is kept. So it is possible to set e.g.
19632 just the end values to keep everything before the specified time.
19637 Drop everything except the second minute of input:
19639 ffmpeg -i INPUT -vf trim=60:120
19643 Keep only the first second:
19645 ffmpeg -i INPUT -vf trim=duration=1
19650 @section unpremultiply
19651 Apply alpha unpremultiply effect to input video stream using first plane
19652 of second stream as alpha.
19654 Both streams must have same dimensions and same pixel format.
19656 The filter accepts the following option:
19660 Set which planes will be processed, unprocessed planes will be copied.
19661 By default value 0xf, all planes will be processed.
19663 If the format has 1 or 2 components, then luma is bit 0.
19664 If the format has 3 or 4 components:
19665 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19666 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19667 If present, the alpha channel is always the last bit.
19670 Do not require 2nd input for processing, instead use alpha plane from input stream.
19676 Sharpen or blur the input video.
19678 It accepts the following parameters:
19681 @item luma_msize_x, lx
19682 Set the luma matrix horizontal size. It must be an odd integer between
19683 3 and 23. The default value is 5.
19685 @item luma_msize_y, ly
19686 Set the luma matrix vertical size. It must be an odd integer between 3
19687 and 23. The default value is 5.
19689 @item luma_amount, la
19690 Set the luma effect strength. It must be a floating point number, reasonable
19691 values lay between -1.5 and 1.5.
19693 Negative values will blur the input video, while positive values will
19694 sharpen it, a value of zero will disable the effect.
19696 Default value is 1.0.
19698 @item chroma_msize_x, cx
19699 Set the chroma matrix horizontal size. It must be an odd integer
19700 between 3 and 23. The default value is 5.
19702 @item chroma_msize_y, cy
19703 Set the chroma matrix vertical size. It must be an odd integer
19704 between 3 and 23. The default value is 5.
19706 @item chroma_amount, ca
19707 Set the chroma effect strength. It must be a floating point number, reasonable
19708 values lay between -1.5 and 1.5.
19710 Negative values will blur the input video, while positive values will
19711 sharpen it, a value of zero will disable the effect.
19713 Default value is 0.0.
19717 All parameters are optional and default to the equivalent of the
19718 string '5:5:1.0:5:5:0.0'.
19720 @subsection Examples
19724 Apply strong luma sharpen effect:
19726 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19730 Apply a strong blur of both luma and chroma parameters:
19732 unsharp=7:7:-2:7:7:-2
19739 Decompose a video made of tiled images into the individual images.
19741 The frame rate of the output video is the frame rate of the input video
19742 multiplied by the number of tiles.
19744 This filter does the reverse of @ref{tile}.
19746 The filter accepts the following options:
19751 Set the grid size (i.e. the number of lines and columns). For the syntax of
19752 this option, check the
19753 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19756 @subsection Examples
19760 Produce a 1-second video from a still image file made of 25 frames stacked
19761 vertically, like an analogic film reel:
19763 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19769 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19770 the image at several (or - in the case of @option{quality} level @code{8} - all)
19771 shifts and average the results.
19773 The way this differs from the behavior of spp is that uspp actually encodes &
19774 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19775 DCT similar to MJPEG.
19777 The filter accepts the following options:
19781 Set quality. This option defines the number of levels for averaging. It accepts
19782 an integer in the range 0-8. If set to @code{0}, the filter will have no
19783 effect. A value of @code{8} means the higher quality. For each increment of
19784 that value the speed drops by a factor of approximately 2. Default value is
19788 Force a constant quantization parameter. If not set, the filter will use the QP
19789 from the video stream (if available).
19794 Convert 360 videos between various formats.
19796 The filter accepts the following options:
19802 Set format of the input/output video.
19810 Equirectangular projection.
19815 Cubemap with 3x2/6x1/1x6 layout.
19817 Format specific options:
19822 Set padding proportion for the input/output cubemap. Values in decimals.
19829 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)
19832 Default value is @b{@samp{0}}.
19833 Maximum value is @b{@samp{0.1}}.
19837 Set fixed padding for the input/output cubemap. Values in pixels.
19839 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19843 Set order of faces for the input/output cubemap. Choose one direction for each position.
19845 Designation of directions:
19861 Default value is @b{@samp{rludfb}}.
19865 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19867 Designation of angles:
19870 0 degrees clockwise
19872 90 degrees clockwise
19874 180 degrees clockwise
19876 270 degrees clockwise
19879 Default value is @b{@samp{000000}}.
19883 Equi-Angular Cubemap.
19890 Format specific options:
19895 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19897 If diagonal field of view is set it overrides horizontal and vertical field of view.
19902 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19904 If diagonal field of view is set it overrides horizontal and vertical field of view.
19910 Format specific options:
19915 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19917 If diagonal field of view is set it overrides horizontal and vertical field of view.
19922 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19924 If diagonal field of view is set it overrides horizontal and vertical field of view.
19930 Facebook's 360 formats.
19933 Stereographic format.
19935 Format specific options:
19940 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19942 If diagonal field of view is set it overrides horizontal and vertical field of view.
19947 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19949 If diagonal field of view is set it overrides horizontal and vertical field of view.
19956 Ball format, gives significant distortion toward the back.
19959 Hammer-Aitoff map projection format.
19962 Sinusoidal map projection format.
19965 Fisheye projection.
19967 Format specific options:
19972 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19974 If diagonal field of view is set it overrides horizontal and vertical field of view.
19979 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19981 If diagonal field of view is set it overrides horizontal and vertical field of view.
19985 Pannini projection.
19987 Format specific options:
19990 Set output pannini parameter.
19993 Set input pannini parameter.
19997 Cylindrical projection.
19999 Format specific options:
20004 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20006 If diagonal field of view is set it overrides horizontal and vertical field of view.
20011 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20013 If diagonal field of view is set it overrides horizontal and vertical field of view.
20017 Perspective projection. @i{(output only)}
20019 Format specific options:
20022 Set perspective parameter.
20026 Tetrahedron projection.
20029 Truncated square pyramid projection.
20033 Half equirectangular projection.
20038 Format specific options:
20043 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20045 If diagonal field of view is set it overrides horizontal and vertical field of view.
20050 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20052 If diagonal field of view is set it overrides horizontal and vertical field of view.
20056 Orthographic format.
20058 Format specific options:
20063 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20065 If diagonal field of view is set it overrides horizontal and vertical field of view.
20070 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20072 If diagonal field of view is set it overrides horizontal and vertical field of view.
20076 Octahedron projection.
20080 Set interpolation method.@*
20081 @i{Note: more complex interpolation methods require much more memory to run.}
20091 Bilinear interpolation.
20093 Lagrange9 interpolation.
20096 Bicubic interpolation.
20099 Lanczos interpolation.
20102 Spline16 interpolation.
20105 Gaussian interpolation.
20107 Mitchell interpolation.
20110 Default value is @b{@samp{line}}.
20114 Set the output video resolution.
20116 Default resolution depends on formats.
20120 Set the input/output stereo format.
20131 Default value is @b{@samp{2d}} for input and output format.
20136 Set rotation for the output video. Values in degrees.
20139 Set rotation order for the output video. Choose one item for each position.
20150 Default value is @b{@samp{ypr}}.
20155 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
20159 Set if input video is flipped horizontally/vertically. Boolean values.
20162 Set if input video is transposed. Boolean value, by default disabled.
20165 Set if output video needs to be transposed. Boolean value, by default disabled.
20168 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
20171 @subsection Examples
20175 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
20177 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20180 Extract back view of Equi-Angular Cubemap:
20182 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20185 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
20187 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20191 @subsection Commands
20193 This filter supports subset of above options as @ref{commands}.
20195 @section vaguedenoiser
20197 Apply a wavelet based denoiser.
20199 It transforms each frame from the video input into the wavelet domain,
20200 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20201 the obtained coefficients. It does an inverse wavelet transform after.
20202 Due to wavelet properties, it should give a nice smoothed result, and
20203 reduced noise, without blurring picture features.
20205 This filter accepts the following options:
20209 The filtering strength. The higher, the more filtered the video will be.
20210 Hard thresholding can use a higher threshold than soft thresholding
20211 before the video looks overfiltered. Default value is 2.
20214 The filtering method the filter will use.
20216 It accepts the following values:
20219 All values under the threshold will be zeroed.
20222 All values under the threshold will be zeroed. All values above will be
20223 reduced by the threshold.
20226 Scales or nullifies coefficients - intermediary between (more) soft and
20227 (less) hard thresholding.
20230 Default is garrote.
20233 Number of times, the wavelet will decompose the picture. Picture can't
20234 be decomposed beyond a particular point (typically, 8 for a 640x480
20235 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
20238 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
20241 A list of the planes to process. By default all planes are processed.
20244 The threshold type the filter will use.
20246 It accepts the following values:
20249 Threshold used is same for all decompositions.
20252 Threshold used depends also on each decomposition coefficients.
20255 Default is universal.
20258 @section vectorscope
20260 Display 2 color component values in the two dimensional graph (which is called
20263 This filter accepts the following options:
20267 Set vectorscope mode.
20269 It accepts the following values:
20273 Gray values are displayed on graph, higher brightness means more pixels have
20274 same component color value on location in graph. This is the default mode.
20277 Gray values are displayed on graph. Surrounding pixels values which are not
20278 present in video frame are drawn in gradient of 2 color components which are
20279 set by option @code{x} and @code{y}. The 3rd color component is static.
20282 Actual color components values present in video frame are displayed on graph.
20285 Similar as color2 but higher frequency of same values @code{x} and @code{y}
20286 on graph increases value of another color component, which is luminance by
20287 default values of @code{x} and @code{y}.
20290 Actual colors present in video frame are displayed on graph. If two different
20291 colors map to same position on graph then color with higher value of component
20292 not present in graph is picked.
20295 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
20296 component picked from radial gradient.
20300 Set which color component will be represented on X-axis. Default is @code{1}.
20303 Set which color component will be represented on Y-axis. Default is @code{2}.
20306 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20307 of color component which represents frequency of (X, Y) location in graph.
20312 No envelope, this is default.
20315 Instant envelope, even darkest single pixel will be clearly highlighted.
20318 Hold maximum and minimum values presented in graph over time. This way you
20319 can still spot out of range values without constantly looking at vectorscope.
20322 Peak and instant envelope combined together.
20326 Set what kind of graticule to draw.
20335 Set graticule opacity.
20338 Set graticule flags.
20342 Draw graticule for white point.
20345 Draw graticule for black point.
20348 Draw color points short names.
20352 Set background opacity.
20354 @item lthreshold, l
20355 Set low threshold for color component not represented on X or Y axis.
20356 Values lower than this value will be ignored. Default is 0.
20357 Note this value is multiplied with actual max possible value one pixel component
20358 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20361 @item hthreshold, h
20362 Set high threshold for color component not represented on X or Y axis.
20363 Values higher than this value will be ignored. Default is 1.
20364 Note this value is multiplied with actual max possible value one pixel component
20365 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20366 is 0.9 * 255 = 230.
20368 @item colorspace, c
20369 Set what kind of colorspace to use when drawing graticule.
20379 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20380 This means no tint, and output will remain gray.
20383 @anchor{vidstabdetect}
20384 @section vidstabdetect
20386 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20387 @ref{vidstabtransform} for pass 2.
20389 This filter generates a file with relative translation and rotation
20390 transform information about subsequent frames, which is then used by
20391 the @ref{vidstabtransform} filter.
20393 To enable compilation of this filter you need to configure FFmpeg with
20394 @code{--enable-libvidstab}.
20396 This filter accepts the following options:
20400 Set the path to the file used to write the transforms information.
20401 Default value is @file{transforms.trf}.
20404 Set how shaky the video is and how quick the camera is. It accepts an
20405 integer in the range 1-10, a value of 1 means little shakiness, a
20406 value of 10 means strong shakiness. Default value is 5.
20409 Set the accuracy of the detection process. It must be a value in the
20410 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20411 accuracy. Default value is 15.
20414 Set stepsize of the search process. The region around minimum is
20415 scanned with 1 pixel resolution. Default value is 6.
20418 Set minimum contrast. Below this value a local measurement field is
20419 discarded. Must be a floating point value in the range 0-1. Default
20423 Set reference frame number for tripod mode.
20425 If enabled, the motion of the frames is compared to a reference frame
20426 in the filtered stream, identified by the specified number. The idea
20427 is to compensate all movements in a more-or-less static scene and keep
20428 the camera view absolutely still.
20430 If set to 0, it is disabled. The frames are counted starting from 1.
20433 Show fields and transforms in the resulting frames. It accepts an
20434 integer in the range 0-2. Default value is 0, which disables any
20438 @subsection Examples
20442 Use default values:
20448 Analyze strongly shaky movie and put the results in file
20449 @file{mytransforms.trf}:
20451 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20455 Visualize the result of internal transformations in the resulting
20458 vidstabdetect=show=1
20462 Analyze a video with medium shakiness using @command{ffmpeg}:
20464 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20468 @anchor{vidstabtransform}
20469 @section vidstabtransform
20471 Video stabilization/deshaking: pass 2 of 2,
20472 see @ref{vidstabdetect} for pass 1.
20474 Read a file with transform information for each frame and
20475 apply/compensate them. Together with the @ref{vidstabdetect}
20476 filter this can be used to deshake videos. See also
20477 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20478 the @ref{unsharp} filter, see below.
20480 To enable compilation of this filter you need to configure FFmpeg with
20481 @code{--enable-libvidstab}.
20483 @subsection Options
20487 Set path to the file used to read the transforms. Default value is
20488 @file{transforms.trf}.
20491 Set the number of frames (value*2 + 1) used for lowpass filtering the
20492 camera movements. Default value is 10.
20494 For example a number of 10 means that 21 frames are used (10 in the
20495 past and 10 in the future) to smoothen the motion in the video. A
20496 larger value leads to a smoother video, but limits the acceleration of
20497 the camera (pan/tilt movements). 0 is a special case where a static
20498 camera is simulated.
20501 Set the camera path optimization algorithm.
20503 Accepted values are:
20506 gaussian kernel low-pass filter on camera motion (default)
20508 averaging on transformations
20512 Set maximal number of pixels to translate frames. Default value is -1,
20516 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20517 value is -1, meaning no limit.
20520 Specify how to deal with borders that may be visible due to movement
20523 Available values are:
20526 keep image information from previous frame (default)
20528 fill the border black
20532 Invert transforms if set to 1. Default value is 0.
20535 Consider transforms as relative to previous frame if set to 1,
20536 absolute if set to 0. Default value is 0.
20539 Set percentage to zoom. A positive value will result in a zoom-in
20540 effect, a negative value in a zoom-out effect. Default value is 0 (no
20544 Set optimal zooming to avoid borders.
20546 Accepted values are:
20551 optimal static zoom value is determined (only very strong movements
20552 will lead to visible borders) (default)
20554 optimal adaptive zoom value is determined (no borders will be
20555 visible), see @option{zoomspeed}
20558 Note that the value given at zoom is added to the one calculated here.
20561 Set percent to zoom maximally each frame (enabled when
20562 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20566 Specify type of interpolation.
20568 Available values are:
20573 linear only horizontal
20575 linear in both directions (default)
20577 cubic in both directions (slow)
20581 Enable virtual tripod mode if set to 1, which is equivalent to
20582 @code{relative=0:smoothing=0}. Default value is 0.
20584 Use also @code{tripod} option of @ref{vidstabdetect}.
20587 Increase log verbosity if set to 1. Also the detected global motions
20588 are written to the temporary file @file{global_motions.trf}. Default
20592 @subsection Examples
20596 Use @command{ffmpeg} for a typical stabilization with default values:
20598 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20601 Note the use of the @ref{unsharp} filter which is always recommended.
20604 Zoom in a bit more and load transform data from a given file:
20606 vidstabtransform=zoom=5:input="mytransforms.trf"
20610 Smoothen the video even more:
20612 vidstabtransform=smoothing=30
20618 Flip the input video vertically.
20620 For example, to vertically flip a video with @command{ffmpeg}:
20622 ffmpeg -i in.avi -vf "vflip" out.avi
20627 Detect variable frame rate video.
20629 This filter tries to detect if the input is variable or constant frame rate.
20631 At end it will output number of frames detected as having variable delta pts,
20632 and ones with constant delta pts.
20633 If there was frames with variable delta, than it will also show min, max and
20634 average delta encountered.
20638 Boost or alter saturation.
20640 The filter accepts the following options:
20643 Set strength of boost if positive value or strength of alter if negative value.
20644 Default is 0. Allowed range is from -2 to 2.
20647 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20650 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20653 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20656 Set the red luma coefficient.
20659 Set the green luma coefficient.
20662 Set the blue luma coefficient.
20665 If @code{intensity} is negative and this is set to 1, colors will change,
20666 otherwise colors will be less saturated, more towards gray.
20669 @subsection Commands
20671 This filter supports the all above options as @ref{commands}.
20676 Make or reverse a natural vignetting effect.
20678 The filter accepts the following options:
20682 Set lens angle expression as a number of radians.
20684 The value is clipped in the @code{[0,PI/2]} range.
20686 Default value: @code{"PI/5"}
20690 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20694 Set forward/backward mode.
20696 Available modes are:
20699 The larger the distance from the central point, the darker the image becomes.
20702 The larger the distance from the central point, the brighter the image becomes.
20703 This can be used to reverse a vignette effect, though there is no automatic
20704 detection to extract the lens @option{angle} and other settings (yet). It can
20705 also be used to create a burning effect.
20708 Default value is @samp{forward}.
20711 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20713 It accepts the following values:
20716 Evaluate expressions only once during the filter initialization.
20719 Evaluate expressions for each incoming frame. This is way slower than the
20720 @samp{init} mode since it requires all the scalers to be re-computed, but it
20721 allows advanced dynamic expressions.
20724 Default value is @samp{init}.
20727 Set dithering to reduce the circular banding effects. Default is @code{1}
20731 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20732 Setting this value to the SAR of the input will make a rectangular vignetting
20733 following the dimensions of the video.
20735 Default is @code{1/1}.
20738 @subsection Expressions
20740 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20741 following parameters.
20746 input width and height
20749 the number of input frame, starting from 0
20752 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20753 @var{TB} units, NAN if undefined
20756 frame rate of the input video, NAN if the input frame rate is unknown
20759 the PTS (Presentation TimeStamp) of the filtered video frame,
20760 expressed in seconds, NAN if undefined
20763 time base of the input video
20767 @subsection Examples
20771 Apply simple strong vignetting effect:
20777 Make a flickering vignetting:
20779 vignette='PI/4+random(1)*PI/50':eval=frame
20784 @section vmafmotion
20786 Obtain the average VMAF motion score of a video.
20787 It is one of the component metrics of VMAF.
20789 The obtained average motion score is printed through the logging system.
20791 The filter accepts the following options:
20795 If specified, the filter will use the named file to save the motion score of
20796 each frame with respect to the previous frame.
20797 When filename equals "-" the data is sent to standard output.
20802 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20806 Stack input videos vertically.
20808 All streams must be of same pixel format and of same width.
20810 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20811 to create same output.
20813 The filter accepts the following options:
20817 Set number of input streams. Default is 2.
20820 If set to 1, force the output to terminate when the shortest input
20821 terminates. Default value is 0.
20826 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20827 Deinterlacing Filter").
20829 Based on the process described by Martin Weston for BBC R&D, and
20830 implemented based on the de-interlace algorithm written by Jim
20831 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20832 uses filter coefficients calculated by BBC R&D.
20834 This filter uses field-dominance information in frame to decide which
20835 of each pair of fields to place first in the output.
20836 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20838 There are two sets of filter coefficients, so called "simple"
20839 and "complex". Which set of filter coefficients is used can
20840 be set by passing an optional parameter:
20844 Set the interlacing filter coefficients. Accepts one of the following values:
20848 Simple filter coefficient set.
20850 More-complex filter coefficient set.
20852 Default value is @samp{complex}.
20855 Specify which frames to deinterlace. Accepts one of the following values:
20859 Deinterlace all frames,
20861 Only deinterlace frames marked as interlaced.
20864 Default value is @samp{all}.
20868 Video waveform monitor.
20870 The waveform monitor plots color component intensity. By default luminance
20871 only. Each column of the waveform corresponds to a column of pixels in the
20874 It accepts the following options:
20878 Can be either @code{row}, or @code{column}. Default is @code{column}.
20879 In row mode, the graph on the left side represents color component value 0 and
20880 the right side represents value = 255. In column mode, the top side represents
20881 color component value = 0 and bottom side represents value = 255.
20884 Set intensity. Smaller values are useful to find out how many values of the same
20885 luminance are distributed across input rows/columns.
20886 Default value is @code{0.04}. Allowed range is [0, 1].
20889 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20890 In mirrored mode, higher values will be represented on the left
20891 side for @code{row} mode and at the top for @code{column} mode. Default is
20892 @code{1} (mirrored).
20896 It accepts the following values:
20899 Presents information identical to that in the @code{parade}, except
20900 that the graphs representing color components are superimposed directly
20903 This display mode makes it easier to spot relative differences or similarities
20904 in overlapping areas of the color components that are supposed to be identical,
20905 such as neutral whites, grays, or blacks.
20908 Display separate graph for the color components side by side in
20909 @code{row} mode or one below the other in @code{column} mode.
20912 Display separate graph for the color components side by side in
20913 @code{column} mode or one below the other in @code{row} mode.
20915 Using this display mode makes it easy to spot color casts in the highlights
20916 and shadows of an image, by comparing the contours of the top and the bottom
20917 graphs of each waveform. Since whites, grays, and blacks are characterized
20918 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20919 should display three waveforms of roughly equal width/height. If not, the
20920 correction is easy to perform by making level adjustments the three waveforms.
20922 Default is @code{stack}.
20924 @item components, c
20925 Set which color components to display. Default is 1, which means only luminance
20926 or red color component if input is in RGB colorspace. If is set for example to
20927 7 it will display all 3 (if) available color components.
20932 No envelope, this is default.
20935 Instant envelope, minimum and maximum values presented in graph will be easily
20936 visible even with small @code{step} value.
20939 Hold minimum and maximum values presented in graph across time. This way you
20940 can still spot out of range values without constantly looking at waveforms.
20943 Peak and instant envelope combined together.
20949 No filtering, this is default.
20952 Luma and chroma combined together.
20955 Similar as above, but shows difference between blue and red chroma.
20958 Similar as above, but use different colors.
20961 Similar as above, but again with different colors.
20964 Displays only chroma.
20967 Displays actual color value on waveform.
20970 Similar as above, but with luma showing frequency of chroma values.
20974 Set which graticule to display.
20978 Do not display graticule.
20981 Display green graticule showing legal broadcast ranges.
20984 Display orange graticule showing legal broadcast ranges.
20987 Display invert graticule showing legal broadcast ranges.
20991 Set graticule opacity.
20994 Set graticule flags.
20998 Draw numbers above lines. By default enabled.
21001 Draw dots instead of lines.
21005 Set scale used for displaying graticule.
21012 Default is digital.
21015 Set background opacity.
21019 Set tint for output.
21020 Only used with lowpass filter and when display is not overlay and input
21021 pixel formats are not RGB.
21024 @section weave, doubleweave
21026 The @code{weave} takes a field-based video input and join
21027 each two sequential fields into single frame, producing a new double
21028 height clip with half the frame rate and half the frame count.
21030 The @code{doubleweave} works same as @code{weave} but without
21031 halving frame rate and frame count.
21033 It accepts the following option:
21037 Set first field. Available values are:
21041 Set the frame as top-field-first.
21044 Set the frame as bottom-field-first.
21048 @subsection Examples
21052 Interlace video using @ref{select} and @ref{separatefields} filter:
21054 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21059 Apply the xBR high-quality magnification filter which is designed for pixel
21060 art. It follows a set of edge-detection rules, see
21061 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
21063 It accepts the following option:
21067 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
21068 @code{3xBR} and @code{4} for @code{4xBR}.
21069 Default is @code{3}.
21074 Apply cross fade from one input video stream to another input video stream.
21075 The cross fade is applied for specified duration.
21077 The filter accepts the following options:
21081 Set one of available transition effects:
21129 Default transition effect is fade.
21132 Set cross fade duration in seconds.
21133 Default duration is 1 second.
21136 Set cross fade start relative to first input stream in seconds.
21137 Default offset is 0.
21140 Set expression for custom transition effect.
21142 The expressions can use the following variables and functions:
21147 The coordinates of the current sample.
21151 The width and height of the image.
21154 Progress of transition effect.
21157 Currently processed plane.
21160 Return value of first input at current location and plane.
21163 Return value of second input at current location and plane.
21169 Return the value of the pixel at location (@var{x},@var{y}) of the
21170 first/second/third/fourth component of first input.
21176 Return the value of the pixel at location (@var{x},@var{y}) of the
21177 first/second/third/fourth component of second input.
21181 @subsection Examples
21185 Cross fade from one input video to another input video, with fade transition and duration of transition
21186 of 2 seconds starting at offset of 5 seconds:
21188 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21193 Pick median pixels from several input videos.
21195 The filter accepts the following options:
21199 Set number of inputs.
21200 Default is 3. Allowed range is from 3 to 255.
21201 If number of inputs is even number, than result will be mean value between two median values.
21204 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
21207 Set median percentile. Default value is @code{0.5}.
21208 Default value of @code{0.5} will pick always median values, while @code{0} will pick
21209 minimum values, and @code{1} maximum values.
21213 Stack video inputs into custom layout.
21215 All streams must be of same pixel format.
21217 The filter accepts the following options:
21221 Set number of input streams. Default is 2.
21224 Specify layout of inputs.
21225 This option requires the desired layout configuration to be explicitly set by the user.
21226 This sets position of each video input in output. Each input
21227 is separated by '|'.
21228 The first number represents the column, and the second number represents the row.
21229 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
21230 where X is video input from which to take width or height.
21231 Multiple values can be used when separated by '+'. In such
21232 case values are summed together.
21234 Note that if inputs are of different sizes gaps may appear, as not all of
21235 the output video frame will be filled. Similarly, videos can overlap each
21236 other if their position doesn't leave enough space for the full frame of
21239 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
21240 a layout must be set by the user.
21243 If set to 1, force the output to terminate when the shortest input
21244 terminates. Default value is 0.
21247 If set to valid color, all unused pixels will be filled with that color.
21248 By default fill is set to none, so it is disabled.
21251 @subsection Examples
21255 Display 4 inputs into 2x2 grid.
21259 input1(0, 0) | input3(w0, 0)
21260 input2(0, h0) | input4(w0, h0)
21264 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21267 Note that if inputs are of different sizes, gaps or overlaps may occur.
21270 Display 4 inputs into 1x4 grid.
21277 input4(0, h0+h1+h2)
21281 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21284 Note that if inputs are of different widths, unused space will appear.
21287 Display 9 inputs into 3x3 grid.
21291 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21292 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21293 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21297 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
21300 Note that if inputs are of different sizes, gaps or overlaps may occur.
21303 Display 16 inputs into 4x4 grid.
21307 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21308 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21309 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21310 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21314 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|
21315 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
21318 Note that if inputs are of different sizes, gaps or overlaps may occur.
21325 Deinterlace the input video ("yadif" means "yet another deinterlacing
21328 It accepts the following parameters:
21334 The interlacing mode to adopt. It accepts one of the following values:
21337 @item 0, send_frame
21338 Output one frame for each frame.
21339 @item 1, send_field
21340 Output one frame for each field.
21341 @item 2, send_frame_nospatial
21342 Like @code{send_frame}, but it skips the spatial interlacing check.
21343 @item 3, send_field_nospatial
21344 Like @code{send_field}, but it skips the spatial interlacing check.
21347 The default value is @code{send_frame}.
21350 The picture field parity assumed for the input interlaced video. It accepts one
21351 of the following values:
21355 Assume the top field is first.
21357 Assume the bottom field is first.
21359 Enable automatic detection of field parity.
21362 The default value is @code{auto}.
21363 If the interlacing is unknown or the decoder does not export this information,
21364 top field first will be assumed.
21367 Specify which frames to deinterlace. Accepts one of the following
21372 Deinterlace all frames.
21373 @item 1, interlaced
21374 Only deinterlace frames marked as interlaced.
21377 The default value is @code{all}.
21380 @section yadif_cuda
21382 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21383 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21386 It accepts the following parameters:
21392 The interlacing mode to adopt. It accepts one of the following values:
21395 @item 0, send_frame
21396 Output one frame for each frame.
21397 @item 1, send_field
21398 Output one frame for each field.
21399 @item 2, send_frame_nospatial
21400 Like @code{send_frame}, but it skips the spatial interlacing check.
21401 @item 3, send_field_nospatial
21402 Like @code{send_field}, but it skips the spatial interlacing check.
21405 The default value is @code{send_frame}.
21408 The picture field parity assumed for the input interlaced video. It accepts one
21409 of the following values:
21413 Assume the top field is first.
21415 Assume the bottom field is first.
21417 Enable automatic detection of field parity.
21420 The default value is @code{auto}.
21421 If the interlacing is unknown or the decoder does not export this information,
21422 top field first will be assumed.
21425 Specify which frames to deinterlace. Accepts one of the following
21430 Deinterlace all frames.
21431 @item 1, interlaced
21432 Only deinterlace frames marked as interlaced.
21435 The default value is @code{all}.
21440 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21441 The algorithm is described in
21442 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21444 It accepts the following parameters:
21448 Set the window radius. Default value is 3.
21451 Set which planes to filter. Default is only the first plane.
21454 Set blur strength. Default value is 128.
21457 @subsection Commands
21458 This filter supports same @ref{commands} as options.
21462 Apply Zoom & Pan effect.
21464 This filter accepts the following options:
21468 Set the zoom expression. Range is 1-10. Default is 1.
21472 Set the x and y expression. Default is 0.
21475 Set the duration expression in number of frames.
21476 This sets for how many number of frames effect will last for
21477 single input image.
21480 Set the output image size, default is 'hd720'.
21483 Set the output frame rate, default is '25'.
21486 Each expression can contain the following constants:
21505 Output frame count.
21508 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21510 @item out_time, time, ot
21511 The output timestamp expressed in seconds.
21515 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21516 for current input frame.
21520 'x' and 'y' of last output frame of previous input frame or 0 when there was
21521 not yet such frame (first input frame).
21524 Last calculated zoom from 'z' expression for current input frame.
21527 Last calculated zoom of last output frame of previous input frame.
21530 Number of output frames for current input frame. Calculated from 'd' expression
21531 for each input frame.
21534 number of output frames created for previous input frame
21537 Rational number: input width / input height
21540 sample aspect ratio
21543 display aspect ratio
21547 @subsection Examples
21551 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21553 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
21557 Zoom in up to 1.5x and pan always at center of picture:
21559 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21563 Same as above but without pausing:
21565 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21569 Zoom in 2x into center of picture only for the first second of the input video:
21571 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21578 Scale (resize) the input video, using the z.lib library:
21579 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21580 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21582 The zscale filter forces the output display aspect ratio to be the same
21583 as the input, by changing the output sample aspect ratio.
21585 If the input image format is different from the format requested by
21586 the next filter, the zscale filter will convert the input to the
21589 @subsection Options
21590 The filter accepts the following options.
21595 Set the output video dimension expression. Default value is the input
21598 If the @var{width} or @var{w} value is 0, the input width is used for
21599 the output. If the @var{height} or @var{h} value is 0, the input height
21600 is used for the output.
21602 If one and only one of the values is -n with n >= 1, the zscale filter
21603 will use a value that maintains the aspect ratio of the input image,
21604 calculated from the other specified dimension. After that it will,
21605 however, make sure that the calculated dimension is divisible by n and
21606 adjust the value if necessary.
21608 If both values are -n with n >= 1, the behavior will be identical to
21609 both values being set to 0 as previously detailed.
21611 See below for the list of accepted constants for use in the dimension
21615 Set the video size. For the syntax of this option, check the
21616 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21619 Set the dither type.
21621 Possible values are:
21626 @item error_diffusion
21632 Set the resize filter type.
21634 Possible values are:
21644 Default is bilinear.
21647 Set the color range.
21649 Possible values are:
21656 Default is same as input.
21659 Set the color primaries.
21661 Possible values are:
21671 Default is same as input.
21674 Set the transfer characteristics.
21676 Possible values are:
21690 Default is same as input.
21693 Set the colorspace matrix.
21695 Possible value are:
21706 Default is same as input.
21709 Set the input color range.
21711 Possible values are:
21718 Default is same as input.
21720 @item primariesin, pin
21721 Set the input color primaries.
21723 Possible values are:
21733 Default is same as input.
21735 @item transferin, tin
21736 Set the input transfer characteristics.
21738 Possible values are:
21749 Default is same as input.
21751 @item matrixin, min
21752 Set the input colorspace matrix.
21754 Possible value are:
21766 Set the output chroma location.
21768 Possible values are:
21779 @item chromalin, cin
21780 Set the input chroma location.
21782 Possible values are:
21794 Set the nominal peak luminance.
21797 The values of the @option{w} and @option{h} options are expressions
21798 containing the following constants:
21803 The input width and height
21807 These are the same as @var{in_w} and @var{in_h}.
21811 The output (scaled) width and height
21815 These are the same as @var{out_w} and @var{out_h}
21818 The same as @var{iw} / @var{ih}
21821 input sample aspect ratio
21824 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21828 horizontal and vertical input chroma subsample values. For example for the
21829 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21833 horizontal and vertical output chroma subsample values. For example for the
21834 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21837 @subsection Commands
21839 This filter supports the following commands:
21843 Set the output video dimension expression.
21844 The command accepts the same syntax of the corresponding option.
21846 If the specified expression is not valid, it is kept at its current
21850 @c man end VIDEO FILTERS
21852 @chapter OpenCL Video Filters
21853 @c man begin OPENCL VIDEO FILTERS
21855 Below is a description of the currently available OpenCL video filters.
21857 To enable compilation of these filters you need to configure FFmpeg with
21858 @code{--enable-opencl}.
21860 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21863 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21864 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21865 given device parameters.
21867 @item -filter_hw_device @var{name}
21868 Pass the hardware device called @var{name} to all filters in any filter graph.
21872 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21876 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21878 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21882 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.
21884 @section avgblur_opencl
21886 Apply average blur filter.
21888 The filter accepts the following options:
21892 Set horizontal radius size.
21893 Range is @code{[1, 1024]} and default value is @code{1}.
21896 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21899 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21902 @subsection Example
21906 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.
21908 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21912 @section boxblur_opencl
21914 Apply a boxblur algorithm to the input video.
21916 It accepts the following parameters:
21920 @item luma_radius, lr
21921 @item luma_power, lp
21922 @item chroma_radius, cr
21923 @item chroma_power, cp
21924 @item alpha_radius, ar
21925 @item alpha_power, ap
21929 A description of the accepted options follows.
21932 @item luma_radius, lr
21933 @item chroma_radius, cr
21934 @item alpha_radius, ar
21935 Set an expression for the box radius in pixels used for blurring the
21936 corresponding input plane.
21938 The radius value must be a non-negative number, and must not be
21939 greater than the value of the expression @code{min(w,h)/2} for the
21940 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21943 Default value for @option{luma_radius} is "2". If not specified,
21944 @option{chroma_radius} and @option{alpha_radius} default to the
21945 corresponding value set for @option{luma_radius}.
21947 The expressions can contain the following constants:
21951 The input width and height in pixels.
21955 The input chroma image width and height in pixels.
21959 The horizontal and vertical chroma subsample values. For example, for the
21960 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21963 @item luma_power, lp
21964 @item chroma_power, cp
21965 @item alpha_power, ap
21966 Specify how many times the boxblur filter is applied to the
21967 corresponding plane.
21969 Default value for @option{luma_power} is 2. If not specified,
21970 @option{chroma_power} and @option{alpha_power} default to the
21971 corresponding value set for @option{luma_power}.
21973 A value of 0 will disable the effect.
21976 @subsection Examples
21978 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.
21982 Apply a boxblur filter with the luma, chroma, and alpha radius
21983 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.
21985 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21986 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21990 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.
21992 For the luma plane, a 2x2 box radius will be run once.
21994 For the chroma plane, a 4x4 box radius will be run 5 times.
21996 For the alpha plane, a 3x3 box radius will be run 7 times.
21998 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
22002 @section colorkey_opencl
22003 RGB colorspace color keying.
22005 The filter accepts the following options:
22009 The color which will be replaced with transparency.
22012 Similarity percentage with the key color.
22014 0.01 matches only the exact key color, while 1.0 matches everything.
22019 0.0 makes pixels either fully transparent, or not transparent at all.
22021 Higher values result in semi-transparent pixels, with a higher transparency
22022 the more similar the pixels color is to the key color.
22025 @subsection Examples
22029 Make every semi-green pixel in the input transparent with some slight blending:
22031 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22035 @section convolution_opencl
22037 Apply convolution of 3x3, 5x5, 7x7 matrix.
22039 The filter accepts the following options:
22046 Set matrix for each plane.
22047 Matrix is sequence of 9, 25 or 49 signed numbers.
22048 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
22054 Set multiplier for calculated value for each plane.
22055 If unset or 0, it will be sum of all matrix elements.
22056 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
22062 Set bias for each plane. This value is added to the result of the multiplication.
22063 Useful for making the overall image brighter or darker.
22064 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
22068 @subsection Examples
22074 -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
22080 -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
22084 Apply edge enhance:
22086 -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
22092 -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
22096 Apply laplacian edge detector which includes diagonals:
22098 -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
22104 -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
22108 @section erosion_opencl
22110 Apply erosion effect to the video.
22112 This filter replaces the pixel by the local(3x3) minimum.
22114 It accepts the following options:
22121 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22122 If @code{0}, plane will remain unchanged.
22125 Flag which specifies the pixel to refer to.
22126 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22128 Flags to local 3x3 coordinates region centered on @code{x}:
22137 @subsection Example
22141 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.
22143 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22147 @section deshake_opencl
22148 Feature-point based video stabilization filter.
22150 The filter accepts the following options:
22154 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
22157 Whether or not additional debug info should be displayed, both in the processed output and in the console.
22159 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
22161 Viewing point matches in the output video is only supported for RGB input.
22163 Defaults to @code{0}.
22165 @item adaptive_crop
22166 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
22168 Defaults to @code{1}.
22170 @item refine_features
22171 Whether or not feature points should be refined at a sub-pixel level.
22173 This can be turned off for a slight performance gain at the cost of precision.
22175 Defaults to @code{1}.
22177 @item smooth_strength
22178 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
22180 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
22182 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
22184 Defaults to @code{0.0}.
22186 @item smooth_window_multiplier
22187 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
22189 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
22191 Acceptable values range from @code{0.1} to @code{10.0}.
22193 Larger values increase the amount of motion data available for determining how to smooth the camera path,
22194 potentially improving smoothness, but also increase latency and memory usage.
22196 Defaults to @code{2.0}.
22200 @subsection Examples
22204 Stabilize a video with a fixed, medium smoothing strength:
22206 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22210 Stabilize a video with debugging (both in console and in rendered video):
22212 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22216 @section dilation_opencl
22218 Apply dilation effect to the video.
22220 This filter replaces the pixel by the local(3x3) maximum.
22222 It accepts the following options:
22229 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22230 If @code{0}, plane will remain unchanged.
22233 Flag which specifies the pixel to refer to.
22234 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22236 Flags to local 3x3 coordinates region centered on @code{x}:
22245 @subsection Example
22249 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.
22251 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22255 @section nlmeans_opencl
22257 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
22259 @section overlay_opencl
22261 Overlay one video on top of another.
22263 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
22264 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
22266 The filter accepts the following options:
22271 Set the x coordinate of the overlaid video on the main video.
22272 Default value is @code{0}.
22275 Set the y coordinate of the overlaid video on the main video.
22276 Default value is @code{0}.
22280 @subsection Examples
22284 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
22286 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22289 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
22291 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22296 @section pad_opencl
22298 Add paddings to the input image, and place the original input at the
22299 provided @var{x}, @var{y} coordinates.
22301 It accepts the following options:
22306 Specify an expression for the size of the output image with the
22307 paddings added. If the value for @var{width} or @var{height} is 0, the
22308 corresponding input size is used for the output.
22310 The @var{width} expression can reference the value set by the
22311 @var{height} expression, and vice versa.
22313 The default value of @var{width} and @var{height} is 0.
22317 Specify the offsets to place the input image at within the padded area,
22318 with respect to the top/left border of the output image.
22320 The @var{x} expression can reference the value set by the @var{y}
22321 expression, and vice versa.
22323 The default value of @var{x} and @var{y} is 0.
22325 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22326 so the input image is centered on the padded area.
22329 Specify the color of the padded area. For the syntax of this option,
22330 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22331 manual,ffmpeg-utils}.
22334 Pad to an aspect instead to a resolution.
22337 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22338 options are expressions containing the following constants:
22343 The input video width and height.
22347 These are the same as @var{in_w} and @var{in_h}.
22351 The output width and height (the size of the padded area), as
22352 specified by the @var{width} and @var{height} expressions.
22356 These are the same as @var{out_w} and @var{out_h}.
22360 The x and y offsets as specified by the @var{x} and @var{y}
22361 expressions, or NAN if not yet specified.
22364 same as @var{iw} / @var{ih}
22367 input sample aspect ratio
22370 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22373 @section prewitt_opencl
22375 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22377 The filter accepts the following option:
22381 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22384 Set value which will be multiplied with filtered result.
22385 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22388 Set value which will be added to filtered result.
22389 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22392 @subsection Example
22396 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22398 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22402 @anchor{program_opencl}
22403 @section program_opencl
22405 Filter video using an OpenCL program.
22410 OpenCL program source file.
22413 Kernel name in program.
22416 Number of inputs to the filter. Defaults to 1.
22419 Size of output frames. Defaults to the same as the first input.
22423 The @code{program_opencl} filter also supports the @ref{framesync} options.
22425 The program source file must contain a kernel function with the given name,
22426 which will be run once for each plane of the output. Each run on a plane
22427 gets enqueued as a separate 2D global NDRange with one work-item for each
22428 pixel to be generated. The global ID offset for each work-item is therefore
22429 the coordinates of a pixel in the destination image.
22431 The kernel function needs to take the following arguments:
22434 Destination image, @var{__write_only image2d_t}.
22436 This image will become the output; the kernel should write all of it.
22438 Frame index, @var{unsigned int}.
22440 This is a counter starting from zero and increasing by one for each frame.
22442 Source images, @var{__read_only image2d_t}.
22444 These are the most recent images on each input. The kernel may read from
22445 them to generate the output, but they can't be written to.
22452 Copy the input to the output (output must be the same size as the input).
22454 __kernel void copy(__write_only image2d_t destination,
22455 unsigned int index,
22456 __read_only image2d_t source)
22458 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22460 int2 location = (int2)(get_global_id(0), get_global_id(1));
22462 float4 value = read_imagef(source, sampler, location);
22464 write_imagef(destination, location, value);
22469 Apply a simple transformation, rotating the input by an amount increasing
22470 with the index counter. Pixel values are linearly interpolated by the
22471 sampler, and the output need not have the same dimensions as the input.
22473 __kernel void rotate_image(__write_only image2d_t dst,
22474 unsigned int index,
22475 __read_only image2d_t src)
22477 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22478 CLK_FILTER_LINEAR);
22480 float angle = (float)index / 100.0f;
22482 float2 dst_dim = convert_float2(get_image_dim(dst));
22483 float2 src_dim = convert_float2(get_image_dim(src));
22485 float2 dst_cen = dst_dim / 2.0f;
22486 float2 src_cen = src_dim / 2.0f;
22488 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22490 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22492 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22493 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22495 src_pos = src_pos * src_dim / dst_dim;
22497 float2 src_loc = src_pos + src_cen;
22499 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22500 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22501 write_imagef(dst, dst_loc, 0.5f);
22503 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22508 Blend two inputs together, with the amount of each input used varying
22509 with the index counter.
22511 __kernel void blend_images(__write_only image2d_t dst,
22512 unsigned int index,
22513 __read_only image2d_t src1,
22514 __read_only image2d_t src2)
22516 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22517 CLK_FILTER_LINEAR);
22519 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22521 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22522 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22523 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22525 float4 val1 = read_imagef(src1, sampler, src1_loc);
22526 float4 val2 = read_imagef(src2, sampler, src2_loc);
22528 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22534 @section roberts_opencl
22535 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22537 The filter accepts the following option:
22541 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22544 Set value which will be multiplied with filtered result.
22545 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22548 Set value which will be added to filtered result.
22549 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22552 @subsection Example
22556 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22558 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22562 @section sobel_opencl
22564 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22566 The filter accepts the following option:
22570 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22573 Set value which will be multiplied with filtered result.
22574 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22577 Set value which will be added to filtered result.
22578 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22581 @subsection Example
22585 Apply sobel operator with scale set to 2 and delta set to 10
22587 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22591 @section tonemap_opencl
22593 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22595 It accepts the following parameters:
22599 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22602 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22605 Apply desaturation for highlights that exceed this level of brightness. The
22606 higher the parameter, the more color information will be preserved. This
22607 setting helps prevent unnaturally blown-out colors for super-highlights, by
22608 (smoothly) turning into white instead. This makes images feel more natural,
22609 at the cost of reducing information about out-of-range colors.
22611 The default value is 0.5, and the algorithm here is a little different from
22612 the cpu version tonemap currently. A setting of 0.0 disables this option.
22615 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22616 is used to detect whether the scene has changed or not. If the distance between
22617 the current frame average brightness and the current running average exceeds
22618 a threshold value, we would re-calculate scene average and peak brightness.
22619 The default value is 0.2.
22622 Specify the output pixel format.
22624 Currently supported formats are:
22631 Set the output color range.
22633 Possible values are:
22639 Default is same as input.
22642 Set the output color primaries.
22644 Possible values are:
22650 Default is same as input.
22653 Set the output transfer characteristics.
22655 Possible values are:
22664 Set the output colorspace matrix.
22666 Possible value are:
22672 Default is same as input.
22676 @subsection Example
22680 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22682 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22686 @section unsharp_opencl
22688 Sharpen or blur the input video.
22690 It accepts the following parameters:
22693 @item luma_msize_x, lx
22694 Set the luma matrix horizontal size.
22695 Range is @code{[1, 23]} and default value is @code{5}.
22697 @item luma_msize_y, ly
22698 Set the luma matrix vertical size.
22699 Range is @code{[1, 23]} and default value is @code{5}.
22701 @item luma_amount, la
22702 Set the luma effect strength.
22703 Range is @code{[-10, 10]} and default value is @code{1.0}.
22705 Negative values will blur the input video, while positive values will
22706 sharpen it, a value of zero will disable the effect.
22708 @item chroma_msize_x, cx
22709 Set the chroma matrix horizontal size.
22710 Range is @code{[1, 23]} and default value is @code{5}.
22712 @item chroma_msize_y, cy
22713 Set the chroma matrix vertical size.
22714 Range is @code{[1, 23]} and default value is @code{5}.
22716 @item chroma_amount, ca
22717 Set the chroma effect strength.
22718 Range is @code{[-10, 10]} and default value is @code{0.0}.
22720 Negative values will blur the input video, while positive values will
22721 sharpen it, a value of zero will disable the effect.
22725 All parameters are optional and default to the equivalent of the
22726 string '5:5:1.0:5:5:0.0'.
22728 @subsection Examples
22732 Apply strong luma sharpen effect:
22734 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22738 Apply a strong blur of both luma and chroma parameters:
22740 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22744 @section xfade_opencl
22746 Cross fade two videos with custom transition effect by using OpenCL.
22748 It accepts the following options:
22752 Set one of possible transition effects.
22756 Select custom transition effect, the actual transition description
22757 will be picked from source and kernel options.
22769 Default transition is fade.
22773 OpenCL program source file for custom transition.
22776 Set name of kernel to use for custom transition from program source file.
22779 Set duration of video transition.
22782 Set time of start of transition relative to first video.
22785 The program source file must contain a kernel function with the given name,
22786 which will be run once for each plane of the output. Each run on a plane
22787 gets enqueued as a separate 2D global NDRange with one work-item for each
22788 pixel to be generated. The global ID offset for each work-item is therefore
22789 the coordinates of a pixel in the destination image.
22791 The kernel function needs to take the following arguments:
22794 Destination image, @var{__write_only image2d_t}.
22796 This image will become the output; the kernel should write all of it.
22799 First Source image, @var{__read_only image2d_t}.
22800 Second Source image, @var{__read_only image2d_t}.
22802 These are the most recent images on each input. The kernel may read from
22803 them to generate the output, but they can't be written to.
22806 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22813 Apply dots curtain transition effect:
22815 __kernel void blend_images(__write_only image2d_t dst,
22816 __read_only image2d_t src1,
22817 __read_only image2d_t src2,
22820 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22821 CLK_FILTER_LINEAR);
22822 int2 p = (int2)(get_global_id(0), get_global_id(1));
22823 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22824 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22827 float2 dots = (float2)(20.0, 20.0);
22828 float2 center = (float2)(0,0);
22831 float4 val1 = read_imagef(src1, sampler, p);
22832 float4 val2 = read_imagef(src2, sampler, p);
22833 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22835 write_imagef(dst, p, next ? val1 : val2);
22841 @c man end OPENCL VIDEO FILTERS
22843 @chapter VAAPI Video Filters
22844 @c man begin VAAPI VIDEO FILTERS
22846 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22848 To enable compilation of these filters you need to configure FFmpeg with
22849 @code{--enable-vaapi}.
22851 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}
22853 @section tonemap_vaapi
22855 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22856 It maps the dynamic range of HDR10 content to the SDR content.
22857 It currently only accepts HDR10 as input.
22859 It accepts the following parameters:
22863 Specify the output pixel format.
22865 Currently supported formats are:
22874 Set the output color primaries.
22876 Default is same as input.
22879 Set the output transfer characteristics.
22884 Set the output colorspace matrix.
22886 Default is same as input.
22890 @subsection Example
22894 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22896 tonemap_vaapi=format=p010:t=bt2020-10
22900 @c man end VAAPI VIDEO FILTERS
22902 @chapter Video Sources
22903 @c man begin VIDEO SOURCES
22905 Below is a description of the currently available video sources.
22909 Buffer video frames, and make them available to the filter chain.
22911 This source is mainly intended for a programmatic use, in particular
22912 through the interface defined in @file{libavfilter/buffersrc.h}.
22914 It accepts the following parameters:
22919 Specify the size (width and height) of the buffered video frames. For the
22920 syntax of this option, check the
22921 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22924 The input video width.
22927 The input video height.
22930 A string representing the pixel format of the buffered video frames.
22931 It may be a number corresponding to a pixel format, or a pixel format
22935 Specify the timebase assumed by the timestamps of the buffered frames.
22938 Specify the frame rate expected for the video stream.
22940 @item pixel_aspect, sar
22941 The sample (pixel) aspect ratio of the input video.
22944 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22945 to the filtergraph description to specify swscale flags for automatically
22946 inserted scalers. See @ref{Filtergraph syntax}.
22948 @item hw_frames_ctx
22949 When using a hardware pixel format, this should be a reference to an
22950 AVHWFramesContext describing input frames.
22955 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22958 will instruct the source to accept video frames with size 320x240 and
22959 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22960 square pixels (1:1 sample aspect ratio).
22961 Since the pixel format with name "yuv410p" corresponds to the number 6
22962 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22963 this example corresponds to:
22965 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22968 Alternatively, the options can be specified as a flat string, but this
22969 syntax is deprecated:
22971 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22975 Create a pattern generated by an elementary cellular automaton.
22977 The initial state of the cellular automaton can be defined through the
22978 @option{filename} and @option{pattern} options. If such options are
22979 not specified an initial state is created randomly.
22981 At each new frame a new row in the video is filled with the result of
22982 the cellular automaton next generation. The behavior when the whole
22983 frame is filled is defined by the @option{scroll} option.
22985 This source accepts the following options:
22989 Read the initial cellular automaton state, i.e. the starting row, from
22990 the specified file.
22991 In the file, each non-whitespace character is considered an alive
22992 cell, a newline will terminate the row, and further characters in the
22993 file will be ignored.
22996 Read the initial cellular automaton state, i.e. the starting row, from
22997 the specified string.
22999 Each non-whitespace character in the string is considered an alive
23000 cell, a newline will terminate the row, and further characters in the
23001 string will be ignored.
23004 Set the video rate, that is the number of frames generated per second.
23007 @item random_fill_ratio, ratio
23008 Set the random fill ratio for the initial cellular automaton row. It
23009 is a floating point number value ranging from 0 to 1, defaults to
23012 This option is ignored when a file or a pattern is specified.
23014 @item random_seed, seed
23015 Set the seed for filling randomly the initial row, must be an integer
23016 included between 0 and UINT32_MAX. If not specified, or if explicitly
23017 set to -1, the filter will try to use a good random seed on a best
23021 Set the cellular automaton rule, it is a number ranging from 0 to 255.
23022 Default value is 110.
23025 Set the size of the output video. For the syntax of this option, check the
23026 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23028 If @option{filename} or @option{pattern} is specified, the size is set
23029 by default to the width of the specified initial state row, and the
23030 height is set to @var{width} * PHI.
23032 If @option{size} is set, it must contain the width of the specified
23033 pattern string, and the specified pattern will be centered in the
23036 If a filename or a pattern string is not specified, the size value
23037 defaults to "320x518" (used for a randomly generated initial state).
23040 If set to 1, scroll the output upward when all the rows in the output
23041 have been already filled. If set to 0, the new generated row will be
23042 written over the top row just after the bottom row is filled.
23045 @item start_full, full
23046 If set to 1, completely fill the output with generated rows before
23047 outputting the first frame.
23048 This is the default behavior, for disabling set the value to 0.
23051 If set to 1, stitch the left and right row edges together.
23052 This is the default behavior, for disabling set the value to 0.
23055 @subsection Examples
23059 Read the initial state from @file{pattern}, and specify an output of
23062 cellauto=f=pattern:s=200x400
23066 Generate a random initial row with a width of 200 cells, with a fill
23069 cellauto=ratio=2/3:s=200x200
23073 Create a pattern generated by rule 18 starting by a single alive cell
23074 centered on an initial row with width 100:
23076 cellauto=p=@@:s=100x400:full=0:rule=18
23080 Specify a more elaborated initial pattern:
23082 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
23087 @anchor{coreimagesrc}
23088 @section coreimagesrc
23089 Video source generated on GPU using Apple's CoreImage API on OSX.
23091 This video source is a specialized version of the @ref{coreimage} video filter.
23092 Use a core image generator at the beginning of the applied filterchain to
23093 generate the content.
23095 The coreimagesrc video source accepts the following options:
23097 @item list_generators
23098 List all available generators along with all their respective options as well as
23099 possible minimum and maximum values along with the default values.
23101 list_generators=true
23105 Specify the size of the sourced video. For the syntax of this option, check the
23106 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23107 The default value is @code{320x240}.
23110 Specify the frame rate of the sourced video, as the number of frames
23111 generated per second. It has to be a string in the format
23112 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23113 number or a valid video frame rate abbreviation. The default value is
23117 Set the sample aspect ratio of the sourced video.
23120 Set the duration of the sourced video. See
23121 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23122 for the accepted syntax.
23124 If not specified, or the expressed duration is negative, the video is
23125 supposed to be generated forever.
23128 Additionally, all options of the @ref{coreimage} video filter are accepted.
23129 A complete filterchain can be used for further processing of the
23130 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
23131 and examples for details.
23133 @subsection Examples
23138 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23139 given as complete and escaped command-line for Apple's standard bash shell:
23141 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
23143 This example is equivalent to the QRCode example of @ref{coreimage} without the
23144 need for a nullsrc video source.
23149 Generate several gradients.
23153 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23154 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23157 Set frame rate, expressed as number of frames per second. Default
23160 @item c0, c1, c2, c3, c4, c5, c6, c7
23161 Set 8 colors. Default values for colors is to pick random one.
23163 @item x0, y0, y0, y1
23164 Set gradient line source and destination points. If negative or out of range, random ones
23168 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
23171 Set seed for picking gradient line points.
23174 Set the duration of the sourced video. See
23175 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23176 for the accepted syntax.
23178 If not specified, or the expressed duration is negative, the video is
23179 supposed to be generated forever.
23182 Set speed of gradients rotation.
23186 @section mandelbrot
23188 Generate a Mandelbrot set fractal, and progressively zoom towards the
23189 point specified with @var{start_x} and @var{start_y}.
23191 This source accepts the following options:
23196 Set the terminal pts value. Default value is 400.
23199 Set the terminal scale value.
23200 Must be a floating point value. Default value is 0.3.
23203 Set the inner coloring mode, that is the algorithm used to draw the
23204 Mandelbrot fractal internal region.
23206 It shall assume one of the following values:
23211 Show time until convergence.
23213 Set color based on point closest to the origin of the iterations.
23218 Default value is @var{mincol}.
23221 Set the bailout value. Default value is 10.0.
23224 Set the maximum of iterations performed by the rendering
23225 algorithm. Default value is 7189.
23228 Set outer coloring mode.
23229 It shall assume one of following values:
23231 @item iteration_count
23232 Set iteration count mode.
23233 @item normalized_iteration_count
23234 set normalized iteration count mode.
23236 Default value is @var{normalized_iteration_count}.
23239 Set frame rate, expressed as number of frames per second. Default
23243 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23244 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23247 Set the initial scale value. Default value is 3.0.
23250 Set the initial x position. Must be a floating point value between
23251 -100 and 100. Default value is -0.743643887037158704752191506114774.
23254 Set the initial y position. Must be a floating point value between
23255 -100 and 100. Default value is -0.131825904205311970493132056385139.
23260 Generate various test patterns, as generated by the MPlayer test filter.
23262 The size of the generated video is fixed, and is 256x256.
23263 This source is useful in particular for testing encoding features.
23265 This source accepts the following options:
23270 Specify the frame rate of the sourced video, as the number of frames
23271 generated per second. It has to be a string in the format
23272 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23273 number or a valid video frame rate abbreviation. The default value is
23277 Set the duration of the sourced video. See
23278 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23279 for the accepted syntax.
23281 If not specified, or the expressed duration is negative, the video is
23282 supposed to be generated forever.
23286 Set the number or the name of the test to perform. Supported tests are:
23300 @item max_frames, m
23301 Set the maximum number of frames generated for each test, default value is 30.
23305 Default value is "all", which will cycle through the list of all tests.
23310 mptestsrc=t=dc_luma
23313 will generate a "dc_luma" test pattern.
23315 @section frei0r_src
23317 Provide a frei0r source.
23319 To enable compilation of this filter you need to install the frei0r
23320 header and configure FFmpeg with @code{--enable-frei0r}.
23322 This source accepts the following parameters:
23327 The size of the video to generate. For the syntax of this option, check the
23328 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23331 The framerate of the generated video. It may be a string of the form
23332 @var{num}/@var{den} or a frame rate abbreviation.
23335 The name to the frei0r source to load. For more information regarding frei0r and
23336 how to set the parameters, read the @ref{frei0r} section in the video filters
23339 @item filter_params
23340 A '|'-separated list of parameters to pass to the frei0r source.
23344 For example, to generate a frei0r partik0l source with size 200x200
23345 and frame rate 10 which is overlaid on the overlay filter main input:
23347 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23352 Generate a life pattern.
23354 This source is based on a generalization of John Conway's life game.
23356 The sourced input represents a life grid, each pixel represents a cell
23357 which can be in one of two possible states, alive or dead. Every cell
23358 interacts with its eight neighbours, which are the cells that are
23359 horizontally, vertically, or diagonally adjacent.
23361 At each interaction the grid evolves according to the adopted rule,
23362 which specifies the number of neighbor alive cells which will make a
23363 cell stay alive or born. The @option{rule} option allows one to specify
23366 This source accepts the following options:
23370 Set the file from which to read the initial grid state. In the file,
23371 each non-whitespace character is considered an alive cell, and newline
23372 is used to delimit the end of each row.
23374 If this option is not specified, the initial grid is generated
23378 Set the video rate, that is the number of frames generated per second.
23381 @item random_fill_ratio, ratio
23382 Set the random fill ratio for the initial random grid. It is a
23383 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23384 It is ignored when a file is specified.
23386 @item random_seed, seed
23387 Set the seed for filling the initial random grid, must be an integer
23388 included between 0 and UINT32_MAX. If not specified, or if explicitly
23389 set to -1, the filter will try to use a good random seed on a best
23395 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23396 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23397 @var{NS} specifies the number of alive neighbor cells which make a
23398 live cell stay alive, and @var{NB} the number of alive neighbor cells
23399 which make a dead cell to become alive (i.e. to "born").
23400 "s" and "b" can be used in place of "S" and "B", respectively.
23402 Alternatively a rule can be specified by an 18-bits integer. The 9
23403 high order bits are used to encode the next cell state if it is alive
23404 for each number of neighbor alive cells, the low order bits specify
23405 the rule for "borning" new cells. Higher order bits encode for an
23406 higher number of neighbor cells.
23407 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23408 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23410 Default value is "S23/B3", which is the original Conway's game of life
23411 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23412 cells, and will born a new cell if there are three alive cells around
23416 Set the size of the output video. For the syntax of this option, check the
23417 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23419 If @option{filename} is specified, the size is set by default to the
23420 same size of the input file. If @option{size} is set, it must contain
23421 the size specified in the input file, and the initial grid defined in
23422 that file is centered in the larger resulting area.
23424 If a filename is not specified, the size value defaults to "320x240"
23425 (used for a randomly generated initial grid).
23428 If set to 1, stitch the left and right grid edges together, and the
23429 top and bottom edges also. Defaults to 1.
23432 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23433 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23434 value from 0 to 255.
23437 Set the color of living (or new born) cells.
23440 Set the color of dead cells. If @option{mold} is set, this is the first color
23441 used to represent a dead cell.
23444 Set mold color, for definitely dead and moldy cells.
23446 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23447 ffmpeg-utils manual,ffmpeg-utils}.
23450 @subsection Examples
23454 Read a grid from @file{pattern}, and center it on a grid of size
23457 life=f=pattern:s=300x300
23461 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23463 life=ratio=2/3:s=200x200
23467 Specify a custom rule for evolving a randomly generated grid:
23473 Full example with slow death effect (mold) using @command{ffplay}:
23475 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23482 @anchor{haldclutsrc}
23485 @anchor{pal100bars}
23486 @anchor{rgbtestsrc}
23488 @anchor{smptehdbars}
23491 @anchor{yuvtestsrc}
23492 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23494 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23496 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23498 The @code{color} source provides an uniformly colored input.
23500 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23501 @ref{haldclut} filter.
23503 The @code{nullsrc} source returns unprocessed video frames. It is
23504 mainly useful to be employed in analysis / debugging tools, or as the
23505 source for filters which ignore the input data.
23507 The @code{pal75bars} source generates a color bars pattern, based on
23508 EBU PAL recommendations with 75% color levels.
23510 The @code{pal100bars} source generates a color bars pattern, based on
23511 EBU PAL recommendations with 100% color levels.
23513 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23514 detecting RGB vs BGR issues. You should see a red, green and blue
23515 stripe from top to bottom.
23517 The @code{smptebars} source generates a color bars pattern, based on
23518 the SMPTE Engineering Guideline EG 1-1990.
23520 The @code{smptehdbars} source generates a color bars pattern, based on
23521 the SMPTE RP 219-2002.
23523 The @code{testsrc} source generates a test video pattern, showing a
23524 color pattern, a scrolling gradient and a timestamp. This is mainly
23525 intended for testing purposes.
23527 The @code{testsrc2} source is similar to testsrc, but supports more
23528 pixel formats instead of just @code{rgb24}. This allows using it as an
23529 input for other tests without requiring a format conversion.
23531 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23532 see a y, cb and cr stripe from top to bottom.
23534 The sources accept the following parameters:
23539 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23540 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23541 pixels to be used as identity matrix for 3D lookup tables. Each component is
23542 coded on a @code{1/(N*N)} scale.
23545 Specify the color of the source, only available in the @code{color}
23546 source. For the syntax of this option, check the
23547 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23550 Specify the size of the sourced video. For the syntax of this option, check the
23551 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23552 The default value is @code{320x240}.
23554 This option is not available with the @code{allrgb}, @code{allyuv}, and
23555 @code{haldclutsrc} filters.
23558 Specify the frame rate of the sourced video, as the number of frames
23559 generated per second. It has to be a string in the format
23560 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23561 number or a valid video frame rate abbreviation. The default value is
23565 Set the duration of the sourced video. See
23566 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23567 for the accepted syntax.
23569 If not specified, or the expressed duration is negative, the video is
23570 supposed to be generated forever.
23572 Since the frame rate is used as time base, all frames including the last one
23573 will have their full duration. If the specified duration is not a multiple
23574 of the frame duration, it will be rounded up.
23577 Set the sample aspect ratio of the sourced video.
23580 Specify the alpha (opacity) of the background, only available in the
23581 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23582 255 (fully opaque, the default).
23585 Set the number of decimals to show in the timestamp, only available in the
23586 @code{testsrc} source.
23588 The displayed timestamp value will correspond to the original
23589 timestamp value multiplied by the power of 10 of the specified
23590 value. Default value is 0.
23593 @subsection Examples
23597 Generate a video with a duration of 5.3 seconds, with size
23598 176x144 and a frame rate of 10 frames per second:
23600 testsrc=duration=5.3:size=qcif:rate=10
23604 The following graph description will generate a red source
23605 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23608 color=c=red@@0.2:s=qcif:r=10
23612 If the input content is to be ignored, @code{nullsrc} can be used. The
23613 following command generates noise in the luminance plane by employing
23614 the @code{geq} filter:
23616 nullsrc=s=256x256, geq=random(1)*255:128:128
23620 @subsection Commands
23622 The @code{color} source supports the following commands:
23626 Set the color of the created image. Accepts the same syntax of the
23627 corresponding @option{color} option.
23632 Generate video using an OpenCL program.
23637 OpenCL program source file.
23640 Kernel name in program.
23643 Size of frames to generate. This must be set.
23646 Pixel format to use for the generated frames. This must be set.
23649 Number of frames generated every second. Default value is '25'.
23653 For details of how the program loading works, see the @ref{program_opencl}
23660 Generate a colour ramp by setting pixel values from the position of the pixel
23661 in the output image. (Note that this will work with all pixel formats, but
23662 the generated output will not be the same.)
23664 __kernel void ramp(__write_only image2d_t dst,
23665 unsigned int index)
23667 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23670 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23672 write_imagef(dst, loc, val);
23677 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23679 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23680 unsigned int index)
23682 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23684 float4 value = 0.0f;
23685 int x = loc.x + index;
23686 int y = loc.y + index;
23687 while (x > 0 || y > 0) {
23688 if (x % 3 == 1 && y % 3 == 1) {
23696 write_imagef(dst, loc, value);
23702 @section sierpinski
23704 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23706 This source accepts the following options:
23710 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23711 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23714 Set frame rate, expressed as number of frames per second. Default
23718 Set seed which is used for random panning.
23721 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23724 Set fractal type, can be default @code{carpet} or @code{triangle}.
23727 @c man end VIDEO SOURCES
23729 @chapter Video Sinks
23730 @c man begin VIDEO SINKS
23732 Below is a description of the currently available video sinks.
23734 @section buffersink
23736 Buffer video frames, and make them available to the end of the filter
23739 This sink is mainly intended for programmatic use, in particular
23740 through the interface defined in @file{libavfilter/buffersink.h}
23741 or the options system.
23743 It accepts a pointer to an AVBufferSinkContext structure, which
23744 defines the incoming buffers' formats, to be passed as the opaque
23745 parameter to @code{avfilter_init_filter} for initialization.
23749 Null video sink: do absolutely nothing with the input video. It is
23750 mainly useful as a template and for use in analysis / debugging
23753 @c man end VIDEO SINKS
23755 @chapter Multimedia Filters
23756 @c man begin MULTIMEDIA FILTERS
23758 Below is a description of the currently available multimedia filters.
23762 Convert input audio to a video output, displaying the audio bit scope.
23764 The filter accepts the following options:
23768 Set frame rate, expressed as number of frames per second. Default
23772 Specify the video size for the output. For the syntax of this option, check the
23773 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23774 Default value is @code{1024x256}.
23777 Specify list of colors separated by space or by '|' which will be used to
23778 draw channels. Unrecognized or missing colors will be replaced
23782 @section adrawgraph
23783 Draw a graph using input audio metadata.
23785 See @ref{drawgraph}
23787 @section agraphmonitor
23789 See @ref{graphmonitor}.
23791 @section ahistogram
23793 Convert input audio to a video output, displaying the volume histogram.
23795 The filter accepts the following options:
23799 Specify how histogram is calculated.
23801 It accepts the following values:
23804 Use single histogram for all channels.
23806 Use separate histogram for each channel.
23808 Default is @code{single}.
23811 Set frame rate, expressed as number of frames per second. Default
23815 Specify the video size for the output. For the syntax of this option, check the
23816 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23817 Default value is @code{hd720}.
23822 It accepts the following values:
23833 reverse logarithmic
23835 Default is @code{log}.
23838 Set amplitude scale.
23840 It accepts the following values:
23847 Default is @code{log}.
23850 Set how much frames to accumulate in histogram.
23851 Default is 1. Setting this to -1 accumulates all frames.
23854 Set histogram ratio of window height.
23857 Set sonogram sliding.
23859 It accepts the following values:
23862 replace old rows with new ones.
23864 scroll from top to bottom.
23866 Default is @code{replace}.
23869 @section aphasemeter
23871 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23872 representing mean phase of current audio frame. A video output can also be produced and is
23873 enabled by default. The audio is passed through as first output.
23875 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23876 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23877 and @code{1} means channels are in phase.
23879 The filter accepts the following options, all related to its video output:
23883 Set the output frame rate. Default value is @code{25}.
23886 Set the video size for the output. For the syntax of this option, check the
23887 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23888 Default value is @code{800x400}.
23893 Specify the red, green, blue contrast. Default values are @code{2},
23894 @code{7} and @code{1}.
23895 Allowed range is @code{[0, 255]}.
23898 Set color which will be used for drawing median phase. If color is
23899 @code{none} which is default, no median phase value will be drawn.
23902 Enable video output. Default is enabled.
23905 @subsection phasing detection
23907 The filter also detects out of phase and mono sequences in stereo streams.
23908 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23910 The filter accepts the following options for this detection:
23914 Enable mono and out of phase detection. Default is disabled.
23917 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23918 Allowed range is @code{[0, 1]}.
23921 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23922 Allowed range is @code{[90, 180]}.
23925 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23928 @subsection Examples
23932 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23934 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23938 @section avectorscope
23940 Convert input audio to a video output, representing the audio vector
23943 The filter is used to measure the difference between channels of stereo
23944 audio stream. A monaural signal, consisting of identical left and right
23945 signal, results in straight vertical line. Any stereo separation is visible
23946 as a deviation from this line, creating a Lissajous figure.
23947 If the straight (or deviation from it) but horizontal line appears this
23948 indicates that the left and right channels are out of phase.
23950 The filter accepts the following options:
23954 Set the vectorscope mode.
23956 Available values are:
23959 Lissajous rotated by 45 degrees.
23962 Same as above but not rotated.
23965 Shape resembling half of circle.
23968 Default value is @samp{lissajous}.
23971 Set the video size for the output. For the syntax of this option, check the
23972 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23973 Default value is @code{400x400}.
23976 Set the output frame rate. Default value is @code{25}.
23982 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23983 @code{160}, @code{80} and @code{255}.
23984 Allowed range is @code{[0, 255]}.
23990 Specify the red, green, blue and alpha fade. Default values are @code{15},
23991 @code{10}, @code{5} and @code{5}.
23992 Allowed range is @code{[0, 255]}.
23995 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
23996 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
23999 Set the vectorscope drawing mode.
24001 Available values are:
24004 Draw dot for each sample.
24007 Draw line between previous and current sample.
24010 Default value is @samp{dot}.
24013 Specify amplitude scale of audio samples.
24015 Available values are:
24031 Swap left channel axis with right channel axis.
24041 Mirror only x axis.
24044 Mirror only y axis.
24052 @subsection Examples
24056 Complete example using @command{ffplay}:
24058 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24059 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24063 @section bench, abench
24065 Benchmark part of a filtergraph.
24067 The filter accepts the following options:
24071 Start or stop a timer.
24073 Available values are:
24076 Get the current time, set it as frame metadata (using the key
24077 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
24080 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
24081 the input frame metadata to get the time difference. Time difference, average,
24082 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
24083 @code{min}) are then printed. The timestamps are expressed in seconds.
24087 @subsection Examples
24091 Benchmark @ref{selectivecolor} filter:
24093 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24099 Concatenate audio and video streams, joining them together one after the
24102 The filter works on segments of synchronized video and audio streams. All
24103 segments must have the same number of streams of each type, and that will
24104 also be the number of streams at output.
24106 The filter accepts the following options:
24111 Set the number of segments. Default is 2.
24114 Set the number of output video streams, that is also the number of video
24115 streams in each segment. Default is 1.
24118 Set the number of output audio streams, that is also the number of audio
24119 streams in each segment. Default is 0.
24122 Activate unsafe mode: do not fail if segments have a different format.
24126 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
24127 @var{a} audio outputs.
24129 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
24130 segment, in the same order as the outputs, then the inputs for the second
24133 Related streams do not always have exactly the same duration, for various
24134 reasons including codec frame size or sloppy authoring. For that reason,
24135 related synchronized streams (e.g. a video and its audio track) should be
24136 concatenated at once. The concat filter will use the duration of the longest
24137 stream in each segment (except the last one), and if necessary pad shorter
24138 audio streams with silence.
24140 For this filter to work correctly, all segments must start at timestamp 0.
24142 All corresponding streams must have the same parameters in all segments; the
24143 filtering system will automatically select a common pixel format for video
24144 streams, and a common sample format, sample rate and channel layout for
24145 audio streams, but other settings, such as resolution, must be converted
24146 explicitly by the user.
24148 Different frame rates are acceptable but will result in variable frame rate
24149 at output; be sure to configure the output file to handle it.
24151 @subsection Examples
24155 Concatenate an opening, an episode and an ending, all in bilingual version
24156 (video in stream 0, audio in streams 1 and 2):
24158 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24159 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24160 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24161 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24165 Concatenate two parts, handling audio and video separately, using the
24166 (a)movie sources, and adjusting the resolution:
24168 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24169 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24170 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24172 Note that a desync will happen at the stitch if the audio and video streams
24173 do not have exactly the same duration in the first file.
24177 @subsection Commands
24179 This filter supports the following commands:
24182 Close the current segment and step to the next one
24188 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
24189 level. By default, it logs a message at a frequency of 10Hz with the
24190 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
24191 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
24193 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
24194 sample format is double-precision floating point. The input stream will be converted to
24195 this specification, if needed. Users may need to insert aformat and/or aresample filters
24196 after this filter to obtain the original parameters.
24198 The filter also has a video output (see the @var{video} option) with a real
24199 time graph to observe the loudness evolution. The graphic contains the logged
24200 message mentioned above, so it is not printed anymore when this option is set,
24201 unless the verbose logging is set. The main graphing area contains the
24202 short-term loudness (3 seconds of analysis), and the gauge on the right is for
24203 the momentary loudness (400 milliseconds), but can optionally be configured
24204 to instead display short-term loudness (see @var{gauge}).
24206 The green area marks a +/- 1LU target range around the target loudness
24207 (-23LUFS by default, unless modified through @var{target}).
24209 More information about the Loudness Recommendation EBU R128 on
24210 @url{http://tech.ebu.ch/loudness}.
24212 The filter accepts the following options:
24217 Activate the video output. The audio stream is passed unchanged whether this
24218 option is set or no. The video stream will be the first output stream if
24219 activated. Default is @code{0}.
24222 Set the video size. This option is for video only. For the syntax of this
24224 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24225 Default and minimum resolution is @code{640x480}.
24228 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
24229 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
24230 other integer value between this range is allowed.
24233 Set metadata injection. If set to @code{1}, the audio input will be segmented
24234 into 100ms output frames, each of them containing various loudness information
24235 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
24237 Default is @code{0}.
24240 Force the frame logging level.
24242 Available values are:
24245 information logging level
24247 verbose logging level
24250 By default, the logging level is set to @var{info}. If the @option{video} or
24251 the @option{metadata} options are set, it switches to @var{verbose}.
24256 Available modes can be cumulated (the option is a @code{flag} type). Possible
24260 Disable any peak mode (default).
24262 Enable sample-peak mode.
24264 Simple peak mode looking for the higher sample value. It logs a message
24265 for sample-peak (identified by @code{SPK}).
24267 Enable true-peak mode.
24269 If enabled, the peak lookup is done on an over-sampled version of the input
24270 stream for better peak accuracy. It logs a message for true-peak.
24271 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
24272 This mode requires a build with @code{libswresample}.
24276 Treat mono input files as "dual mono". If a mono file is intended for playback
24277 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
24278 If set to @code{true}, this option will compensate for this effect.
24279 Multi-channel input files are not affected by this option.
24282 Set a specific pan law to be used for the measurement of dual mono files.
24283 This parameter is optional, and has a default value of -3.01dB.
24286 Set a specific target level (in LUFS) used as relative zero in the visualization.
24287 This parameter is optional and has a default value of -23LUFS as specified
24288 by EBU R128. However, material published online may prefer a level of -16LUFS
24289 (e.g. for use with podcasts or video platforms).
24292 Set the value displayed by the gauge. Valid values are @code{momentary} and s
24293 @code{shortterm}. By default the momentary value will be used, but in certain
24294 scenarios it may be more useful to observe the short term value instead (e.g.
24298 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24299 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24300 video output, not the summary or continuous log output.
24303 @subsection Examples
24307 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24309 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24313 Run an analysis with @command{ffmpeg}:
24315 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24319 @section interleave, ainterleave
24321 Temporally interleave frames from several inputs.
24323 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24325 These filters read frames from several inputs and send the oldest
24326 queued frame to the output.
24328 Input streams must have well defined, monotonically increasing frame
24331 In order to submit one frame to output, these filters need to enqueue
24332 at least one frame for each input, so they cannot work in case one
24333 input is not yet terminated and will not receive incoming frames.
24335 For example consider the case when one input is a @code{select} filter
24336 which always drops input frames. The @code{interleave} filter will keep
24337 reading from that input, but it will never be able to send new frames
24338 to output until the input sends an end-of-stream signal.
24340 Also, depending on inputs synchronization, the filters will drop
24341 frames in case one input receives more frames than the other ones, and
24342 the queue is already filled.
24344 These filters accept the following options:
24348 Set the number of different inputs, it is 2 by default.
24351 How to determine the end-of-stream.
24355 The duration of the longest input. (default)
24358 The duration of the shortest input.
24361 The duration of the first input.
24366 @subsection Examples
24370 Interleave frames belonging to different streams using @command{ffmpeg}:
24372 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24376 Add flickering blur effect:
24378 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24382 @section metadata, ametadata
24384 Manipulate frame metadata.
24386 This filter accepts the following options:
24390 Set mode of operation of the filter.
24392 Can be one of the following:
24396 If both @code{value} and @code{key} is set, select frames
24397 which have such metadata. If only @code{key} is set, select
24398 every frame that has such key in metadata.
24401 Add new metadata @code{key} and @code{value}. If key is already available
24405 Modify value of already present key.
24408 If @code{value} is set, delete only keys that have such value.
24409 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24413 Print key and its value if metadata was found. If @code{key} is not set print all
24414 metadata values available in frame.
24418 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24421 Set metadata value which will be used. This option is mandatory for
24422 @code{modify} and @code{add} mode.
24425 Which function to use when comparing metadata value and @code{value}.
24427 Can be one of following:
24431 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24434 Values are interpreted as strings, returns true if metadata value starts with
24435 the @code{value} option string.
24438 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24441 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24444 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24447 Values are interpreted as floats, returns true if expression from option @code{expr}
24451 Values are interpreted as strings, returns true if metadata value ends with
24452 the @code{value} option string.
24456 Set expression which is used when @code{function} is set to @code{expr}.
24457 The expression is evaluated through the eval API and can contain the following
24462 Float representation of @code{value} from metadata key.
24465 Float representation of @code{value} as supplied by user in @code{value} option.
24469 If specified in @code{print} mode, output is written to the named file. Instead of
24470 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24471 for standard output. If @code{file} option is not set, output is written to the log
24472 with AV_LOG_INFO loglevel.
24475 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24479 @subsection Examples
24483 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24486 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24489 Print silencedetect output to file @file{metadata.txt}.
24491 silencedetect,ametadata=mode=print:file=metadata.txt
24494 Direct all metadata to a pipe with file descriptor 4.
24496 metadata=mode=print:file='pipe\:4'
24500 @section perms, aperms
24502 Set read/write permissions for the output frames.
24504 These filters are mainly aimed at developers to test direct path in the
24505 following filter in the filtergraph.
24507 The filters accept the following options:
24511 Select the permissions mode.
24513 It accepts the following values:
24516 Do nothing. This is the default.
24518 Set all the output frames read-only.
24520 Set all the output frames directly writable.
24522 Make the frame read-only if writable, and writable if read-only.
24524 Set each output frame read-only or writable randomly.
24528 Set the seed for the @var{random} mode, must be an integer included between
24529 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24530 @code{-1}, the filter will try to use a good random seed on a best effort
24534 Note: in case of auto-inserted filter between the permission filter and the
24535 following one, the permission might not be received as expected in that
24536 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24537 perms/aperms filter can avoid this problem.
24539 @section realtime, arealtime
24541 Slow down filtering to match real time approximately.
24543 These filters will pause the filtering for a variable amount of time to
24544 match the output rate with the input timestamps.
24545 They are similar to the @option{re} option to @code{ffmpeg}.
24547 They accept the following options:
24551 Time limit for the pauses. Any pause longer than that will be considered
24552 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24554 Speed factor for processing. The value must be a float larger than zero.
24555 Values larger than 1.0 will result in faster than realtime processing,
24556 smaller will slow processing down. The @var{limit} is automatically adapted
24557 accordingly. Default is 1.0.
24559 A processing speed faster than what is possible without these filters cannot
24564 @section select, aselect
24566 Select frames to pass in output.
24568 This filter accepts the following options:
24573 Set expression, which is evaluated for each input frame.
24575 If the expression is evaluated to zero, the frame is discarded.
24577 If the evaluation result is negative or NaN, the frame is sent to the
24578 first output; otherwise it is sent to the output with index
24579 @code{ceil(val)-1}, assuming that the input index starts from 0.
24581 For example a value of @code{1.2} corresponds to the output with index
24582 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24585 Set the number of outputs. The output to which to send the selected
24586 frame is based on the result of the evaluation. Default value is 1.
24589 The expression can contain the following constants:
24593 The (sequential) number of the filtered frame, starting from 0.
24596 The (sequential) number of the selected frame, starting from 0.
24598 @item prev_selected_n
24599 The sequential number of the last selected frame. It's NAN if undefined.
24602 The timebase of the input timestamps.
24605 The PTS (Presentation TimeStamp) of the filtered video frame,
24606 expressed in @var{TB} units. It's NAN if undefined.
24609 The PTS of the filtered video frame,
24610 expressed in seconds. It's NAN if undefined.
24613 The PTS of the previously filtered video frame. It's NAN if undefined.
24615 @item prev_selected_pts
24616 The PTS of the last previously filtered video frame. It's NAN if undefined.
24618 @item prev_selected_t
24619 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24622 The PTS of the first video frame in the video. It's NAN if undefined.
24625 The time of the first video frame in the video. It's NAN if undefined.
24627 @item pict_type @emph{(video only)}
24628 The type of the filtered frame. It can assume one of the following
24640 @item interlace_type @emph{(video only)}
24641 The frame interlace type. It can assume one of the following values:
24644 The frame is progressive (not interlaced).
24646 The frame is top-field-first.
24648 The frame is bottom-field-first.
24651 @item consumed_sample_n @emph{(audio only)}
24652 the number of selected samples before the current frame
24654 @item samples_n @emph{(audio only)}
24655 the number of samples in the current frame
24657 @item sample_rate @emph{(audio only)}
24658 the input sample rate
24661 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24664 the position in the file of the filtered frame, -1 if the information
24665 is not available (e.g. for synthetic video)
24667 @item scene @emph{(video only)}
24668 value between 0 and 1 to indicate a new scene; a low value reflects a low
24669 probability for the current frame to introduce a new scene, while a higher
24670 value means the current frame is more likely to be one (see the example below)
24672 @item concatdec_select
24673 The concat demuxer can select only part of a concat input file by setting an
24674 inpoint and an outpoint, but the output packets may not be entirely contained
24675 in the selected interval. By using this variable, it is possible to skip frames
24676 generated by the concat demuxer which are not exactly contained in the selected
24679 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24680 and the @var{lavf.concat.duration} packet metadata values which are also
24681 present in the decoded frames.
24683 The @var{concatdec_select} variable is -1 if the frame pts is at least
24684 start_time and either the duration metadata is missing or the frame pts is less
24685 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24688 That basically means that an input frame is selected if its pts is within the
24689 interval set by the concat demuxer.
24693 The default value of the select expression is "1".
24695 @subsection Examples
24699 Select all frames in input:
24704 The example above is the same as:
24716 Select only I-frames:
24718 select='eq(pict_type\,I)'
24722 Select one frame every 100:
24724 select='not(mod(n\,100))'
24728 Select only frames contained in the 10-20 time interval:
24730 select=between(t\,10\,20)
24734 Select only I-frames contained in the 10-20 time interval:
24736 select=between(t\,10\,20)*eq(pict_type\,I)
24740 Select frames with a minimum distance of 10 seconds:
24742 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24746 Use aselect to select only audio frames with samples number > 100:
24748 aselect='gt(samples_n\,100)'
24752 Create a mosaic of the first scenes:
24754 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24757 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24761 Send even and odd frames to separate outputs, and compose them:
24763 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24767 Select useful frames from an ffconcat file which is using inpoints and
24768 outpoints but where the source files are not intra frame only.
24770 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24774 @section sendcmd, asendcmd
24776 Send commands to filters in the filtergraph.
24778 These filters read commands to be sent to other filters in the
24781 @code{sendcmd} must be inserted between two video filters,
24782 @code{asendcmd} must be inserted between two audio filters, but apart
24783 from that they act the same way.
24785 The specification of commands can be provided in the filter arguments
24786 with the @var{commands} option, or in a file specified by the
24787 @var{filename} option.
24789 These filters accept the following options:
24792 Set the commands to be read and sent to the other filters.
24794 Set the filename of the commands to be read and sent to the other
24798 @subsection Commands syntax
24800 A commands description consists of a sequence of interval
24801 specifications, comprising a list of commands to be executed when a
24802 particular event related to that interval occurs. The occurring event
24803 is typically the current frame time entering or leaving a given time
24806 An interval is specified by the following syntax:
24808 @var{START}[-@var{END}] @var{COMMANDS};
24811 The time interval is specified by the @var{START} and @var{END} times.
24812 @var{END} is optional and defaults to the maximum time.
24814 The current frame time is considered within the specified interval if
24815 it is included in the interval [@var{START}, @var{END}), that is when
24816 the time is greater or equal to @var{START} and is lesser than
24819 @var{COMMANDS} consists of a sequence of one or more command
24820 specifications, separated by ",", relating to that interval. The
24821 syntax of a command specification is given by:
24823 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24826 @var{FLAGS} is optional and specifies the type of events relating to
24827 the time interval which enable sending the specified command, and must
24828 be a non-null sequence of identifier flags separated by "+" or "|" and
24829 enclosed between "[" and "]".
24831 The following flags are recognized:
24834 The command is sent when the current frame timestamp enters the
24835 specified interval. In other words, the command is sent when the
24836 previous frame timestamp was not in the given interval, and the
24840 The command is sent when the current frame timestamp leaves the
24841 specified interval. In other words, the command is sent when the
24842 previous frame timestamp was in the given interval, and the
24846 The command @var{ARG} is interpreted as expression and result of
24847 expression is passed as @var{ARG}.
24849 The expression is evaluated through the eval API and can contain the following
24854 Original position in the file of the frame, or undefined if undefined
24855 for the current frame.
24858 The presentation timestamp in input.
24861 The count of the input frame for video or audio, starting from 0.
24864 The time in seconds of the current frame.
24867 The start time in seconds of the current command interval.
24870 The end time in seconds of the current command interval.
24873 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24878 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24881 @var{TARGET} specifies the target of the command, usually the name of
24882 the filter class or a specific filter instance name.
24884 @var{COMMAND} specifies the name of the command for the target filter.
24886 @var{ARG} is optional and specifies the optional list of argument for
24887 the given @var{COMMAND}.
24889 Between one interval specification and another, whitespaces, or
24890 sequences of characters starting with @code{#} until the end of line,
24891 are ignored and can be used to annotate comments.
24893 A simplified BNF description of the commands specification syntax
24896 @var{COMMAND_FLAG} ::= "enter" | "leave"
24897 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24898 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24899 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24900 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24901 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24904 @subsection Examples
24908 Specify audio tempo change at second 4:
24910 asendcmd=c='4.0 atempo tempo 1.5',atempo
24914 Target a specific filter instance:
24916 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24920 Specify a list of drawtext and hue commands in a file.
24922 # show text in the interval 5-10
24923 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24924 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24926 # desaturate the image in the interval 15-20
24927 15.0-20.0 [enter] hue s 0,
24928 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24930 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24932 # apply an exponential saturation fade-out effect, starting from time 25
24933 25 [enter] hue s exp(25-t)
24936 A filtergraph allowing to read and process the above command list
24937 stored in a file @file{test.cmd}, can be specified with:
24939 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24944 @section setpts, asetpts
24946 Change the PTS (presentation timestamp) of the input frames.
24948 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24950 This filter accepts the following options:
24955 The expression which is evaluated for each frame to construct its timestamp.
24959 The expression is evaluated through the eval API and can contain the following
24963 @item FRAME_RATE, FR
24964 frame rate, only defined for constant frame-rate video
24967 The presentation timestamp in input
24970 The count of the input frame for video or the number of consumed samples,
24971 not including the current frame for audio, starting from 0.
24973 @item NB_CONSUMED_SAMPLES
24974 The number of consumed samples, not including the current frame (only
24977 @item NB_SAMPLES, S
24978 The number of samples in the current frame (only audio)
24980 @item SAMPLE_RATE, SR
24981 The audio sample rate.
24984 The PTS of the first frame.
24987 the time in seconds of the first frame
24990 State whether the current frame is interlaced.
24993 the time in seconds of the current frame
24996 original position in the file of the frame, or undefined if undefined
24997 for the current frame
25000 The previous input PTS.
25003 previous input time in seconds
25006 The previous output PTS.
25009 previous output time in seconds
25012 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
25016 The wallclock (RTC) time at the start of the movie in microseconds.
25019 The timebase of the input timestamps.
25023 @subsection Examples
25027 Start counting PTS from zero
25029 setpts=PTS-STARTPTS
25033 Apply fast motion effect:
25039 Apply slow motion effect:
25045 Set fixed rate of 25 frames per second:
25051 Set fixed rate 25 fps with some jitter:
25053 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25057 Apply an offset of 10 seconds to the input PTS:
25063 Generate timestamps from a "live source" and rebase onto the current timebase:
25065 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25069 Generate timestamps by counting samples:
25078 Force color range for the output video frame.
25080 The @code{setrange} filter marks the color range property for the
25081 output frames. It does not change the input frame, but only sets the
25082 corresponding property, which affects how the frame is treated by
25085 The filter accepts the following options:
25090 Available values are:
25094 Keep the same color range property.
25096 @item unspecified, unknown
25097 Set the color range as unspecified.
25099 @item limited, tv, mpeg
25100 Set the color range as limited.
25102 @item full, pc, jpeg
25103 Set the color range as full.
25107 @section settb, asettb
25109 Set the timebase to use for the output frames timestamps.
25110 It is mainly useful for testing timebase configuration.
25112 It accepts the following parameters:
25117 The expression which is evaluated into the output timebase.
25121 The value for @option{tb} is an arithmetic expression representing a
25122 rational. The expression can contain the constants "AVTB" (the default
25123 timebase), "intb" (the input timebase) and "sr" (the sample rate,
25124 audio only). Default value is "intb".
25126 @subsection Examples
25130 Set the timebase to 1/25:
25136 Set the timebase to 1/10:
25142 Set the timebase to 1001/1000:
25148 Set the timebase to 2*intb:
25154 Set the default timebase value:
25161 Convert input audio to a video output representing frequency spectrum
25162 logarithmically using Brown-Puckette constant Q transform algorithm with
25163 direct frequency domain coefficient calculation (but the transform itself
25164 is not really constant Q, instead the Q factor is actually variable/clamped),
25165 with musical tone scale, from E0 to D#10.
25167 The filter accepts the following options:
25171 Specify the video size for the output. It must be even. For the syntax of this option,
25172 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25173 Default value is @code{1920x1080}.
25176 Set the output frame rate. Default value is @code{25}.
25179 Set the bargraph height. It must be even. Default value is @code{-1} which
25180 computes the bargraph height automatically.
25183 Set the axis height. It must be even. Default value is @code{-1} which computes
25184 the axis height automatically.
25187 Set the sonogram height. It must be even. Default value is @code{-1} which
25188 computes the sonogram height automatically.
25191 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
25192 instead. Default value is @code{1}.
25194 @item sono_v, volume
25195 Specify the sonogram volume expression. It can contain variables:
25198 the @var{bar_v} evaluated expression
25199 @item frequency, freq, f
25200 the frequency where it is evaluated
25201 @item timeclamp, tc
25202 the value of @var{timeclamp} option
25206 @item a_weighting(f)
25207 A-weighting of equal loudness
25208 @item b_weighting(f)
25209 B-weighting of equal loudness
25210 @item c_weighting(f)
25211 C-weighting of equal loudness.
25213 Default value is @code{16}.
25215 @item bar_v, volume2
25216 Specify the bargraph volume expression. It can contain variables:
25219 the @var{sono_v} evaluated expression
25220 @item frequency, freq, f
25221 the frequency where it is evaluated
25222 @item timeclamp, tc
25223 the value of @var{timeclamp} option
25227 @item a_weighting(f)
25228 A-weighting of equal loudness
25229 @item b_weighting(f)
25230 B-weighting of equal loudness
25231 @item c_weighting(f)
25232 C-weighting of equal loudness.
25234 Default value is @code{sono_v}.
25236 @item sono_g, gamma
25237 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
25238 higher gamma makes the spectrum having more range. Default value is @code{3}.
25239 Acceptable range is @code{[1, 7]}.
25241 @item bar_g, gamma2
25242 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
25246 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
25247 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
25249 @item timeclamp, tc
25250 Specify the transform timeclamp. At low frequency, there is trade-off between
25251 accuracy in time domain and frequency domain. If timeclamp is lower,
25252 event in time domain is represented more accurately (such as fast bass drum),
25253 otherwise event in frequency domain is represented more accurately
25254 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
25257 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
25258 limits future samples by applying asymmetric windowing in time domain, useful
25259 when low latency is required. Accepted range is @code{[0, 1]}.
25262 Specify the transform base frequency. Default value is @code{20.01523126408007475},
25263 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
25266 Specify the transform end frequency. Default value is @code{20495.59681441799654},
25267 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
25270 This option is deprecated and ignored.
25273 Specify the transform length in time domain. Use this option to control accuracy
25274 trade-off between time domain and frequency domain at every frequency sample.
25275 It can contain variables:
25277 @item frequency, freq, f
25278 the frequency where it is evaluated
25279 @item timeclamp, tc
25280 the value of @var{timeclamp} option.
25282 Default value is @code{384*tc/(384+tc*f)}.
25285 Specify the transform count for every video frame. Default value is @code{6}.
25286 Acceptable range is @code{[1, 30]}.
25289 Specify the transform count for every single pixel. Default value is @code{0},
25290 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
25293 Specify font file for use with freetype to draw the axis. If not specified,
25294 use embedded font. Note that drawing with font file or embedded font is not
25295 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25299 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25300 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25304 Specify font color expression. This is arithmetic expression that should return
25305 integer value 0xRRGGBB. It can contain variables:
25307 @item frequency, freq, f
25308 the frequency where it is evaluated
25309 @item timeclamp, tc
25310 the value of @var{timeclamp} option
25315 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25316 @item r(x), g(x), b(x)
25317 red, green, and blue value of intensity x.
25319 Default value is @code{st(0, (midi(f)-59.5)/12);
25320 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25321 r(1-ld(1)) + b(ld(1))}.
25324 Specify image file to draw the axis. This option override @var{fontfile} and
25325 @var{fontcolor} option.
25328 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25329 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25330 Default value is @code{1}.
25333 Set colorspace. The accepted values are:
25336 Unspecified (default)
25345 BT.470BG or BT.601-6 625
25348 SMPTE-170M or BT.601-6 525
25354 BT.2020 with non-constant luminance
25359 Set spectrogram color scheme. This is list of floating point values with format
25360 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25361 The default is @code{1|0.5|0|0|0.5|1}.
25365 @subsection Examples
25369 Playing audio while showing the spectrum:
25371 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25375 Same as above, but with frame rate 30 fps:
25377 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25381 Playing at 1280x720:
25383 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25387 Disable sonogram display:
25393 A1 and its harmonics: A1, A2, (near)E3, A3:
25395 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),
25396 asplit[a][out1]; [a] showcqt [out0]'
25400 Same as above, but with more accuracy in frequency domain:
25402 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),
25403 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25409 bar_v=10:sono_v=bar_v*a_weighting(f)
25413 Custom gamma, now spectrum is linear to the amplitude.
25419 Custom tlength equation:
25421 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)))'
25425 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25427 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25431 Custom font using fontconfig:
25433 font='Courier New,Monospace,mono|bold'
25437 Custom frequency range with custom axis using image file:
25439 axisfile=myaxis.png:basefreq=40:endfreq=10000
25445 Convert input audio to video output representing the audio power spectrum.
25446 Audio amplitude is on Y-axis while frequency is on X-axis.
25448 The filter accepts the following options:
25452 Specify size of video. For the syntax of this option, check the
25453 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25454 Default is @code{1024x512}.
25458 This set how each frequency bin will be represented.
25460 It accepts the following values:
25466 Default is @code{bar}.
25469 Set amplitude scale.
25471 It accepts the following values:
25485 Default is @code{log}.
25488 Set frequency scale.
25490 It accepts the following values:
25499 Reverse logarithmic scale.
25501 Default is @code{lin}.
25504 Set window size. Allowed range is from 16 to 65536.
25506 Default is @code{2048}
25509 Set windowing function.
25511 It accepts the following values:
25534 Default is @code{hanning}.
25537 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25538 which means optimal overlap for selected window function will be picked.
25541 Set time averaging. Setting this to 0 will display current maximal peaks.
25542 Default is @code{1}, which means time averaging is disabled.
25545 Specify list of colors separated by space or by '|' which will be used to
25546 draw channel frequencies. Unrecognized or missing colors will be replaced
25550 Set channel display mode.
25552 It accepts the following values:
25557 Default is @code{combined}.
25560 Set minimum amplitude used in @code{log} amplitude scaler.
25563 Set data display mode.
25565 It accepts the following values:
25571 Default is @code{magnitude}.
25574 @section showspatial
25576 Convert stereo input audio to a video output, representing the spatial relationship
25577 between two channels.
25579 The filter accepts the following options:
25583 Specify the video size for the output. For the syntax of this option, check the
25584 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25585 Default value is @code{512x512}.
25588 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25591 Set window function.
25593 It accepts the following values:
25618 Default value is @code{hann}.
25621 Set ratio of overlap window. Default value is @code{0.5}.
25622 When value is @code{1} overlap is set to recommended size for specific
25623 window function currently used.
25626 @anchor{showspectrum}
25627 @section showspectrum
25629 Convert input audio to a video output, representing the audio frequency
25632 The filter accepts the following options:
25636 Specify the video size for the output. For the syntax of this option, check the
25637 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25638 Default value is @code{640x512}.
25641 Specify how the spectrum should slide along the window.
25643 It accepts the following values:
25646 the samples start again on the left when they reach the right
25648 the samples scroll from right to left
25650 frames are only produced when the samples reach the right
25652 the samples scroll from left to right
25655 Default value is @code{replace}.
25658 Specify display mode.
25660 It accepts the following values:
25663 all channels are displayed in the same row
25665 all channels are displayed in separate rows
25668 Default value is @samp{combined}.
25671 Specify display color mode.
25673 It accepts the following values:
25676 each channel is displayed in a separate color
25678 each channel is displayed using the same color scheme
25680 each channel is displayed using the rainbow color scheme
25682 each channel is displayed using the moreland color scheme
25684 each channel is displayed using the nebulae color scheme
25686 each channel is displayed using the fire color scheme
25688 each channel is displayed using the fiery color scheme
25690 each channel is displayed using the fruit color scheme
25692 each channel is displayed using the cool color scheme
25694 each channel is displayed using the magma color scheme
25696 each channel is displayed using the green color scheme
25698 each channel is displayed using the viridis color scheme
25700 each channel is displayed using the plasma color scheme
25702 each channel is displayed using the cividis color scheme
25704 each channel is displayed using the terrain color scheme
25707 Default value is @samp{channel}.
25710 Specify scale used for calculating intensity color values.
25712 It accepts the following values:
25717 square root, default
25728 Default value is @samp{sqrt}.
25731 Specify frequency scale.
25733 It accepts the following values:
25741 Default value is @samp{lin}.
25744 Set saturation modifier for displayed colors. Negative values provide
25745 alternative color scheme. @code{0} is no saturation at all.
25746 Saturation must be in [-10.0, 10.0] range.
25747 Default value is @code{1}.
25750 Set window function.
25752 It accepts the following values:
25777 Default value is @code{hann}.
25780 Set orientation of time vs frequency axis. Can be @code{vertical} or
25781 @code{horizontal}. Default is @code{vertical}.
25784 Set ratio of overlap window. Default value is @code{0}.
25785 When value is @code{1} overlap is set to recommended size for specific
25786 window function currently used.
25789 Set scale gain for calculating intensity color values.
25790 Default value is @code{1}.
25793 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25796 Set color rotation, must be in [-1.0, 1.0] range.
25797 Default value is @code{0}.
25800 Set start frequency from which to display spectrogram. Default is @code{0}.
25803 Set stop frequency to which to display spectrogram. Default is @code{0}.
25806 Set upper frame rate limit. Default is @code{auto}, unlimited.
25809 Draw time and frequency axes and legends. Default is disabled.
25812 The usage is very similar to the showwaves filter; see the examples in that
25815 @subsection Examples
25819 Large window with logarithmic color scaling:
25821 showspectrum=s=1280x480:scale=log
25825 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25827 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25828 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25832 @section showspectrumpic
25834 Convert input audio to a single video frame, representing the audio frequency
25837 The filter accepts the following options:
25841 Specify the video size for the output. For the syntax of this option, check the
25842 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25843 Default value is @code{4096x2048}.
25846 Specify display mode.
25848 It accepts the following values:
25851 all channels are displayed in the same row
25853 all channels are displayed in separate rows
25855 Default value is @samp{combined}.
25858 Specify display color mode.
25860 It accepts the following values:
25863 each channel is displayed in a separate color
25865 each channel is displayed using the same color scheme
25867 each channel is displayed using the rainbow color scheme
25869 each channel is displayed using the moreland color scheme
25871 each channel is displayed using the nebulae color scheme
25873 each channel is displayed using the fire color scheme
25875 each channel is displayed using the fiery color scheme
25877 each channel is displayed using the fruit color scheme
25879 each channel is displayed using the cool color scheme
25881 each channel is displayed using the magma color scheme
25883 each channel is displayed using the green color scheme
25885 each channel is displayed using the viridis color scheme
25887 each channel is displayed using the plasma color scheme
25889 each channel is displayed using the cividis color scheme
25891 each channel is displayed using the terrain color scheme
25893 Default value is @samp{intensity}.
25896 Specify scale used for calculating intensity color values.
25898 It accepts the following values:
25903 square root, default
25913 Default value is @samp{log}.
25916 Specify frequency scale.
25918 It accepts the following values:
25926 Default value is @samp{lin}.
25929 Set saturation modifier for displayed colors. Negative values provide
25930 alternative color scheme. @code{0} is no saturation at all.
25931 Saturation must be in [-10.0, 10.0] range.
25932 Default value is @code{1}.
25935 Set window function.
25937 It accepts the following values:
25961 Default value is @code{hann}.
25964 Set orientation of time vs frequency axis. Can be @code{vertical} or
25965 @code{horizontal}. Default is @code{vertical}.
25968 Set scale gain for calculating intensity color values.
25969 Default value is @code{1}.
25972 Draw time and frequency axes and legends. Default is enabled.
25975 Set color rotation, must be in [-1.0, 1.0] range.
25976 Default value is @code{0}.
25979 Set start frequency from which to display spectrogram. Default is @code{0}.
25982 Set stop frequency to which to display spectrogram. Default is @code{0}.
25985 @subsection Examples
25989 Extract an audio spectrogram of a whole audio track
25990 in a 1024x1024 picture using @command{ffmpeg}:
25992 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
25996 @section showvolume
25998 Convert input audio volume to a video output.
26000 The filter accepts the following options:
26007 Set border width, allowed range is [0, 5]. Default is 1.
26010 Set channel width, allowed range is [80, 8192]. Default is 400.
26013 Set channel height, allowed range is [1, 900]. Default is 20.
26016 Set fade, allowed range is [0, 1]. Default is 0.95.
26019 Set volume color expression.
26021 The expression can use the following variables:
26025 Current max volume of channel in dB.
26031 Current channel number, starting from 0.
26035 If set, displays channel names. Default is enabled.
26038 If set, displays volume values. Default is enabled.
26041 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
26042 default is @code{h}.
26045 Set step size, allowed range is [0, 5]. Default is 0, which means
26049 Set background opacity, allowed range is [0, 1]. Default is 0.
26052 Set metering mode, can be peak: @code{p} or rms: @code{r},
26053 default is @code{p}.
26056 Set display scale, can be linear: @code{lin} or log: @code{log},
26057 default is @code{lin}.
26061 If set to > 0., display a line for the max level
26062 in the previous seconds.
26063 default is disabled: @code{0.}
26066 The color of the max line. Use when @code{dm} option is set to > 0.
26067 default is: @code{orange}
26072 Convert input audio to a video output, representing the samples waves.
26074 The filter accepts the following options:
26078 Specify the video size for the output. For the syntax of this option, check the
26079 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26080 Default value is @code{600x240}.
26085 Available values are:
26088 Draw a point for each sample.
26091 Draw a vertical line for each sample.
26094 Draw a point for each sample and a line between them.
26097 Draw a centered vertical line for each sample.
26100 Default value is @code{point}.
26103 Set the number of samples which are printed on the same column. A
26104 larger value will decrease the frame rate. Must be a positive
26105 integer. This option can be set only if the value for @var{rate}
26106 is not explicitly specified.
26109 Set the (approximate) output frame rate. This is done by setting the
26110 option @var{n}. Default value is "25".
26112 @item split_channels
26113 Set if channels should be drawn separately or overlap. Default value is 0.
26116 Set colors separated by '|' which are going to be used for drawing of each channel.
26119 Set amplitude scale.
26121 Available values are:
26139 Set the draw mode. This is mostly useful to set for high @var{n}.
26141 Available values are:
26144 Scale pixel values for each drawn sample.
26147 Draw every sample directly.
26150 Default value is @code{scale}.
26153 @subsection Examples
26157 Output the input file audio and the corresponding video representation
26160 amovie=a.mp3,asplit[out0],showwaves[out1]
26164 Create a synthetic signal and show it with showwaves, forcing a
26165 frame rate of 30 frames per second:
26167 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26171 @section showwavespic
26173 Convert input audio to a single video frame, representing the samples waves.
26175 The filter accepts the following options:
26179 Specify the video size for the output. For the syntax of this option, check the
26180 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26181 Default value is @code{600x240}.
26183 @item split_channels
26184 Set if channels should be drawn separately or overlap. Default value is 0.
26187 Set colors separated by '|' which are going to be used for drawing of each channel.
26190 Set amplitude scale.
26192 Available values are:
26212 Available values are:
26215 Scale pixel values for each drawn sample.
26218 Draw every sample directly.
26221 Default value is @code{scale}.
26224 Set the filter mode.
26226 Available values are:
26229 Use average samples values for each drawn sample.
26232 Use peak samples values for each drawn sample.
26235 Default value is @code{average}.
26238 @subsection Examples
26242 Extract a channel split representation of the wave form of a whole audio track
26243 in a 1024x800 picture using @command{ffmpeg}:
26245 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26249 @section sidedata, asidedata
26251 Delete frame side data, or select frames based on it.
26253 This filter accepts the following options:
26257 Set mode of operation of the filter.
26259 Can be one of the following:
26263 Select every frame with side data of @code{type}.
26266 Delete side data of @code{type}. If @code{type} is not set, delete all side
26272 Set side data type used with all modes. Must be set for @code{select} mode. For
26273 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
26274 in @file{libavutil/frame.h}. For example, to choose
26275 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
26279 @section spectrumsynth
26281 Synthesize audio from 2 input video spectrums, first input stream represents
26282 magnitude across time and second represents phase across time.
26283 The filter will transform from frequency domain as displayed in videos back
26284 to time domain as presented in audio output.
26286 This filter is primarily created for reversing processed @ref{showspectrum}
26287 filter outputs, but can synthesize sound from other spectrograms too.
26288 But in such case results are going to be poor if the phase data is not
26289 available, because in such cases phase data need to be recreated, usually
26290 it's just recreated from random noise.
26291 For best results use gray only output (@code{channel} color mode in
26292 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
26293 @code{lin} scale for phase video. To produce phase, for 2nd video, use
26294 @code{data} option. Inputs videos should generally use @code{fullframe}
26295 slide mode as that saves resources needed for decoding video.
26297 The filter accepts the following options:
26301 Specify sample rate of output audio, the sample rate of audio from which
26302 spectrum was generated may differ.
26305 Set number of channels represented in input video spectrums.
26308 Set scale which was used when generating magnitude input spectrum.
26309 Can be @code{lin} or @code{log}. Default is @code{log}.
26312 Set slide which was used when generating inputs spectrums.
26313 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26314 Default is @code{fullframe}.
26317 Set window function used for resynthesis.
26320 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26321 which means optimal overlap for selected window function will be picked.
26324 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26325 Default is @code{vertical}.
26328 @subsection Examples
26332 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26333 then resynthesize videos back to audio with spectrumsynth:
26335 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
26336 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
26337 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26341 @section split, asplit
26343 Split input into several identical outputs.
26345 @code{asplit} works with audio input, @code{split} with video.
26347 The filter accepts a single parameter which specifies the number of outputs. If
26348 unspecified, it defaults to 2.
26350 @subsection Examples
26354 Create two separate outputs from the same input:
26356 [in] split [out0][out1]
26360 To create 3 or more outputs, you need to specify the number of
26363 [in] asplit=3 [out0][out1][out2]
26367 Create two separate outputs from the same input, one cropped and
26370 [in] split [splitout1][splitout2];
26371 [splitout1] crop=100:100:0:0 [cropout];
26372 [splitout2] pad=200:200:100:100 [padout];
26376 Create 5 copies of the input audio with @command{ffmpeg}:
26378 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26384 Receive commands sent through a libzmq client, and forward them to
26385 filters in the filtergraph.
26387 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26388 must be inserted between two video filters, @code{azmq} between two
26389 audio filters. Both are capable to send messages to any filter type.
26391 To enable these filters you need to install the libzmq library and
26392 headers and configure FFmpeg with @code{--enable-libzmq}.
26394 For more information about libzmq see:
26395 @url{http://www.zeromq.org/}
26397 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26398 receives messages sent through a network interface defined by the
26399 @option{bind_address} (or the abbreviation "@option{b}") option.
26400 Default value of this option is @file{tcp://localhost:5555}. You may
26401 want to alter this value to your needs, but do not forget to escape any
26402 ':' signs (see @ref{filtergraph escaping}).
26404 The received message must be in the form:
26406 @var{TARGET} @var{COMMAND} [@var{ARG}]
26409 @var{TARGET} specifies the target of the command, usually the name of
26410 the filter class or a specific filter instance name. The default
26411 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26412 but you can override this by using the @samp{filter_name@@id} syntax
26413 (see @ref{Filtergraph syntax}).
26415 @var{COMMAND} specifies the name of the command for the target filter.
26417 @var{ARG} is optional and specifies the optional argument list for the
26418 given @var{COMMAND}.
26420 Upon reception, the message is processed and the corresponding command
26421 is injected into the filtergraph. Depending on the result, the filter
26422 will send a reply to the client, adopting the format:
26424 @var{ERROR_CODE} @var{ERROR_REASON}
26428 @var{MESSAGE} is optional.
26430 @subsection Examples
26432 Look at @file{tools/zmqsend} for an example of a zmq client which can
26433 be used to send commands processed by these filters.
26435 Consider the following filtergraph generated by @command{ffplay}.
26436 In this example the last overlay filter has an instance name. All other
26437 filters will have default instance names.
26440 ffplay -dumpgraph 1 -f lavfi "
26441 color=s=100x100:c=red [l];
26442 color=s=100x100:c=blue [r];
26443 nullsrc=s=200x100, zmq [bg];
26444 [bg][l] overlay [bg+l];
26445 [bg+l][r] overlay@@my=x=100 "
26448 To change the color of the left side of the video, the following
26449 command can be used:
26451 echo Parsed_color_0 c yellow | tools/zmqsend
26454 To change the right side:
26456 echo Parsed_color_1 c pink | tools/zmqsend
26459 To change the position of the right side:
26461 echo overlay@@my x 150 | tools/zmqsend
26465 @c man end MULTIMEDIA FILTERS
26467 @chapter Multimedia Sources
26468 @c man begin MULTIMEDIA SOURCES
26470 Below is a description of the currently available multimedia sources.
26474 This is the same as @ref{movie} source, except it selects an audio
26480 Read audio and/or video stream(s) from a movie container.
26482 It accepts the following parameters:
26486 The name of the resource to read (not necessarily a file; it can also be a
26487 device or a stream accessed through some protocol).
26489 @item format_name, f
26490 Specifies the format assumed for the movie to read, and can be either
26491 the name of a container or an input device. If not specified, the
26492 format is guessed from @var{movie_name} or by probing.
26494 @item seek_point, sp
26495 Specifies the seek point in seconds. The frames will be output
26496 starting from this seek point. The parameter is evaluated with
26497 @code{av_strtod}, so the numerical value may be suffixed by an IS
26498 postfix. The default value is "0".
26501 Specifies the streams to read. Several streams can be specified,
26502 separated by "+". The source will then have as many outputs, in the
26503 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26504 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26505 respectively the default (best suited) video and audio stream. Default
26506 is "dv", or "da" if the filter is called as "amovie".
26508 @item stream_index, si
26509 Specifies the index of the video stream to read. If the value is -1,
26510 the most suitable video stream will be automatically selected. The default
26511 value is "-1". Deprecated. If the filter is called "amovie", it will select
26512 audio instead of video.
26515 Specifies how many times to read the stream in sequence.
26516 If the value is 0, the stream will be looped infinitely.
26517 Default value is "1".
26519 Note that when the movie is looped the source timestamps are not
26520 changed, so it will generate non monotonically increasing timestamps.
26522 @item discontinuity
26523 Specifies the time difference between frames above which the point is
26524 considered a timestamp discontinuity which is removed by adjusting the later
26528 It allows overlaying a second video on top of the main input of
26529 a filtergraph, as shown in this graph:
26531 input -----------> deltapts0 --> overlay --> output
26534 movie --> scale--> deltapts1 -------+
26536 @subsection Examples
26540 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26541 on top of the input labelled "in":
26543 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26544 [in] setpts=PTS-STARTPTS [main];
26545 [main][over] overlay=16:16 [out]
26549 Read from a video4linux2 device, and overlay it on top of the input
26552 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26553 [in] setpts=PTS-STARTPTS [main];
26554 [main][over] overlay=16:16 [out]
26558 Read the first video stream and the audio stream with id 0x81 from
26559 dvd.vob; the video is connected to the pad named "video" and the audio is
26560 connected to the pad named "audio":
26562 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26566 @subsection Commands
26568 Both movie and amovie support the following commands:
26571 Perform seek using "av_seek_frame".
26572 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26575 @var{stream_index}: If stream_index is -1, a default
26576 stream is selected, and @var{timestamp} is automatically converted
26577 from AV_TIME_BASE units to the stream specific time_base.
26579 @var{timestamp}: Timestamp in AVStream.time_base units
26580 or, if no stream is specified, in AV_TIME_BASE units.
26582 @var{flags}: Flags which select direction and seeking mode.
26586 Get movie duration in AV_TIME_BASE units.
26590 @c man end MULTIMEDIA SOURCES