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, can be @var{2nd}, @var{4th} or @var{8th}.
527 Default is @var{4th}.
532 Reduce audio bit resolution.
534 This filter is bit crusher with enhanced functionality. A bit crusher
535 is used to audibly reduce number of bits an audio signal is sampled
536 with. This doesn't change the bit depth at all, it just produces the
537 effect. Material reduced in bit depth sounds more harsh and "digital".
538 This filter is able to even round to continuous values instead of discrete
540 Additionally it has a D/C offset which results in different crushing of
541 the lower and the upper half of the signal.
542 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
544 Another feature of this filter is the logarithmic mode.
545 This setting switches from linear distances between bits to logarithmic ones.
546 The result is a much more "natural" sounding crusher which doesn't gate low
547 signals for example. The human ear has a logarithmic perception,
548 so this kind of crushing is much more pleasant.
549 Logarithmic crushing is also able to get anti-aliased.
551 The filter accepts the following options:
567 Can be linear: @code{lin} or logarithmic: @code{log}.
576 Set sample reduction.
579 Enable LFO. By default disabled.
590 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
594 Remove impulsive noise from input audio.
596 Samples detected as impulsive noise are replaced by interpolated samples using
597 autoregressive modelling.
601 Set window size, in milliseconds. Allowed range is from @code{10} to
602 @code{100}. Default value is @code{55} milliseconds.
603 This sets size of window which will be processed at once.
606 Set window overlap, in percentage of window size. Allowed range is from
607 @code{50} to @code{95}. Default value is @code{75} percent.
608 Setting this to a very high value increases impulsive noise removal but makes
609 whole process much slower.
612 Set autoregression order, in percentage of window size. Allowed range is from
613 @code{0} to @code{25}. Default value is @code{2} percent. This option also
614 controls quality of interpolated samples using neighbour good samples.
617 Set threshold value. Allowed range is from @code{1} to @code{100}.
618 Default value is @code{2}.
619 This controls the strength of impulsive noise which is going to be removed.
620 The lower value, the more samples will be detected as impulsive noise.
623 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
624 @code{10}. Default value is @code{2}.
625 If any two samples detected as noise are spaced less than this value then any
626 sample between those two samples will be also detected as noise.
631 It accepts the following values:
634 Select overlap-add method. Even not interpolated samples are slightly
635 changed with this method.
638 Select overlap-save method. Not interpolated samples remain unchanged.
641 Default value is @code{a}.
645 Remove clipped samples from input audio.
647 Samples detected as clipped are replaced by interpolated samples using
648 autoregressive modelling.
652 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
653 Default value is @code{55} milliseconds.
654 This sets size of window which will be processed at once.
657 Set window overlap, in percentage of window size. Allowed range is from @code{50}
658 to @code{95}. Default value is @code{75} percent.
661 Set autoregression order, in percentage of window size. Allowed range is from
662 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
663 quality of interpolated samples using neighbour good samples.
666 Set threshold value. Allowed range is from @code{1} to @code{100}.
667 Default value is @code{10}. Higher values make clip detection less aggressive.
670 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
671 Default value is @code{1000}. Higher values make clip detection less aggressive.
676 It accepts the following values:
679 Select overlap-add method. Even not interpolated samples are slightly changed
683 Select overlap-save method. Not interpolated samples remain unchanged.
686 Default value is @code{a}.
691 Delay one or more audio channels.
693 Samples in delayed channel are filled with silence.
695 The filter accepts the following option:
699 Set list of delays in milliseconds for each channel separated by '|'.
700 Unused delays will be silently ignored. If number of given delays is
701 smaller than number of channels all remaining channels will not be delayed.
702 If you want to delay exact number of samples, append 'S' to number.
703 If you want instead to delay in seconds, append 's' to number.
706 Use last set delay for all remaining channels. By default is disabled.
707 This option if enabled changes how option @code{delays} is interpreted.
714 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
715 the second channel (and any other channels that may be present) unchanged.
721 Delay second channel by 500 samples, the third channel by 700 samples and leave
722 the first channel (and any other channels that may be present) unchanged.
728 Delay all channels by same number of samples:
730 adelay=delays=64S:all=1
734 @section aderivative, aintegral
736 Compute derivative/integral of audio stream.
738 Applying both filters one after another produces original audio.
742 Apply echoing to the input audio.
744 Echoes are reflected sound and can occur naturally amongst mountains
745 (and sometimes large buildings) when talking or shouting; digital echo
746 effects emulate this behaviour and are often used to help fill out the
747 sound of a single instrument or vocal. The time difference between the
748 original signal and the reflection is the @code{delay}, and the
749 loudness of the reflected signal is the @code{decay}.
750 Multiple echoes can have different delays and decays.
752 A description of the accepted parameters follows.
756 Set input gain of reflected signal. Default is @code{0.6}.
759 Set output gain of reflected signal. Default is @code{0.3}.
762 Set list of time intervals in milliseconds between original signal and reflections
763 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
764 Default is @code{1000}.
767 Set list of loudness of reflected signals separated by '|'.
768 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
769 Default is @code{0.5}.
776 Make it sound as if there are twice as many instruments as are actually playing:
778 aecho=0.8:0.88:60:0.4
782 If delay is very short, then it sounds like a (metallic) robot playing music:
788 A longer delay will sound like an open air concert in the mountains:
790 aecho=0.8:0.9:1000:0.3
794 Same as above but with one more mountain:
796 aecho=0.8:0.9:1000|1800:0.3|0.25
801 Audio emphasis filter creates or restores material directly taken from LPs or
802 emphased CDs with different filter curves. E.g. to store music on vinyl the
803 signal has to be altered by a filter first to even out the disadvantages of
804 this recording medium.
805 Once the material is played back the inverse filter has to be applied to
806 restore the distortion of the frequency response.
808 The filter accepts the following options:
818 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
819 use @code{production} mode. Default is @code{reproduction} mode.
822 Set filter type. Selects medium. Can be one of the following:
834 select Compact Disc (CD).
840 select 50µs (FM-KF).
842 select 75µs (FM-KF).
848 Modify an audio signal according to the specified expressions.
850 This filter accepts one or more expressions (one for each channel),
851 which are evaluated and used to modify a corresponding audio signal.
853 It accepts the following parameters:
857 Set the '|'-separated expressions list for each separate channel. If
858 the number of input channels is greater than the number of
859 expressions, the last specified expression is used for the remaining
862 @item channel_layout, c
863 Set output channel layout. If not specified, the channel layout is
864 specified by the number of expressions. If set to @samp{same}, it will
865 use by default the same input channel layout.
868 Each expression in @var{exprs} can contain the following constants and functions:
872 channel number of the current expression
875 number of the evaluated sample, starting from 0
881 time of the evaluated sample expressed in seconds
884 @item nb_out_channels
885 input and output number of channels
888 the value of input channel with number @var{CH}
891 Note: this filter is slow. For faster processing you should use a
900 aeval=val(ch)/2:c=same
904 Invert phase of the second channel:
913 Apply fade-in/out effect to input audio.
915 A description of the accepted parameters follows.
919 Specify the effect type, can be either @code{in} for fade-in, or
920 @code{out} for a fade-out effect. Default is @code{in}.
922 @item start_sample, ss
923 Specify the number of the start sample for starting to apply the fade
924 effect. Default is 0.
927 Specify the number of samples for which the fade effect has to last. At
928 the end of the fade-in effect the output audio will have the same
929 volume as the input audio, at the end of the fade-out transition
930 the output audio will be silence. Default is 44100.
933 Specify the start time of the fade effect. Default is 0.
934 The value must be specified as a time duration; see
935 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
936 for the accepted syntax.
937 If set this option is used instead of @var{start_sample}.
940 Specify the duration of the fade effect. See
941 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
942 for the accepted syntax.
943 At the end of the fade-in effect the output audio will have the same
944 volume as the input audio, at the end of the fade-out transition
945 the output audio will be silence.
946 By default the duration is determined by @var{nb_samples}.
947 If set this option is used instead of @var{nb_samples}.
950 Set curve for fade transition.
952 It accepts the following values:
955 select triangular, linear slope (default)
957 select quarter of sine wave
959 select half of sine wave
961 select exponential sine wave
965 select inverted parabola
979 select inverted quarter of sine wave
981 select inverted half of sine wave
983 select double-exponential seat
985 select double-exponential sigmoid
987 select logistic sigmoid
997 Fade in first 15 seconds of audio:
1003 Fade out last 25 seconds of a 900 seconds audio:
1005 afade=t=out:st=875:d=25
1010 Denoise audio samples with FFT.
1012 A description of the accepted parameters follows.
1016 Set the noise reduction in dB, allowed range is 0.01 to 97.
1017 Default value is 12 dB.
1020 Set the noise floor in dB, allowed range is -80 to -20.
1021 Default value is -50 dB.
1026 It accepts the following values:
1035 Select shellac noise.
1038 Select custom noise, defined in @code{bn} option.
1040 Default value is white noise.
1044 Set custom band noise for every one of 15 bands.
1045 Bands are separated by ' ' or '|'.
1048 Set the residual floor in dB, allowed range is -80 to -20.
1049 Default value is -38 dB.
1052 Enable noise tracking. By default is disabled.
1053 With this enabled, noise floor is automatically adjusted.
1056 Enable residual tracking. By default is disabled.
1059 Set the output mode.
1061 It accepts the following values:
1064 Pass input unchanged.
1067 Pass noise filtered out.
1072 Default value is @var{o}.
1076 @subsection Commands
1078 This filter supports the following commands:
1080 @item sample_noise, sn
1081 Start or stop measuring noise profile.
1082 Syntax for the command is : "start" or "stop" string.
1083 After measuring noise profile is stopped it will be
1084 automatically applied in filtering.
1086 @item noise_reduction, nr
1087 Change noise reduction. Argument is single float number.
1088 Syntax for the command is : "@var{noise_reduction}"
1090 @item noise_floor, nf
1091 Change noise floor. Argument is single float number.
1092 Syntax for the command is : "@var{noise_floor}"
1094 @item output_mode, om
1095 Change output mode operation.
1096 Syntax for the command is : "i", "o" or "n" string.
1100 Apply arbitrary expressions to samples in frequency domain.
1104 Set frequency domain real expression for each separate channel separated
1105 by '|'. Default is "re".
1106 If the number of input channels is greater than the number of
1107 expressions, the last specified expression is used for the remaining
1111 Set frequency domain imaginary expression for each separate channel
1112 separated by '|'. Default is "im".
1114 Each expression in @var{real} and @var{imag} can contain the following
1115 constants and functions:
1122 current frequency bin number
1125 number of available bins
1128 channel number of the current expression
1137 current real part of frequency bin of current channel
1140 current imaginary part of frequency bin of current channel
1143 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1146 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1150 Set window size. Allowed range is from 16 to 131072.
1151 Default is @code{4096}
1154 Set window function. Default is @code{hann}.
1157 Set window overlap. If set to 1, the recommended overlap for selected
1158 window function will be picked. Default is @code{0.75}.
1161 @subsection Examples
1165 Leave almost only low frequencies in audio:
1167 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1171 Apply robotize effect:
1173 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1177 Apply whisper effect:
1179 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"
1186 Apply an arbitrary Finite Impulse Response filter.
1188 This filter is designed for applying long FIR filters,
1189 up to 60 seconds long.
1191 It can be used as component for digital crossover filters,
1192 room equalization, cross talk cancellation, wavefield synthesis,
1193 auralization, ambiophonics, ambisonics and spatialization.
1195 This filter uses the streams higher than first one as FIR coefficients.
1196 If the non-first stream holds a single channel, it will be used
1197 for all input channels in the first stream, otherwise
1198 the number of channels in the non-first stream must be same as
1199 the number of channels in the first stream.
1201 It accepts the following parameters:
1205 Set dry gain. This sets input gain.
1208 Set wet gain. This sets final output gain.
1211 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1214 Enable applying gain measured from power of IR.
1216 Set which approach to use for auto gain measurement.
1220 Do not apply any gain.
1223 select peak gain, very conservative approach. This is default value.
1226 select DC gain, limited application.
1229 select gain to noise approach, this is most popular one.
1233 Set gain to be applied to IR coefficients before filtering.
1234 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1237 Set format of IR stream. Can be @code{mono} or @code{input}.
1238 Default is @code{input}.
1241 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1242 Allowed range is 0.1 to 60 seconds.
1245 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1246 By default it is disabled.
1249 Set for which IR channel to display frequency response. By default is first channel
1250 displayed. This option is used only when @var{response} is enabled.
1253 Set video stream size. This option is used only when @var{response} is enabled.
1256 Set video stream frame rate. This option is used only when @var{response} is enabled.
1259 Set minimal partition size used for convolution. Default is @var{8192}.
1260 Allowed range is from @var{1} to @var{32768}.
1261 Lower values decreases latency at cost of higher CPU usage.
1264 Set maximal partition size used for convolution. Default is @var{8192}.
1265 Allowed range is from @var{8} to @var{32768}.
1266 Lower values may increase CPU usage.
1269 Set number of input impulse responses streams which will be switchable at runtime.
1270 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1273 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1274 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1275 This option can be changed at runtime via @ref{commands}.
1278 @subsection Examples
1282 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1284 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1291 Set output format constraints for the input audio. The framework will
1292 negotiate the most appropriate format to minimize conversions.
1294 It accepts the following parameters:
1297 @item sample_fmts, f
1298 A '|'-separated list of requested sample formats.
1300 @item sample_rates, r
1301 A '|'-separated list of requested sample rates.
1303 @item channel_layouts, cl
1304 A '|'-separated list of requested channel layouts.
1306 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1307 for the required syntax.
1310 If a parameter is omitted, all values are allowed.
1312 Force the output to either unsigned 8-bit or signed 16-bit stereo
1314 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1319 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1320 processing reduces disturbing noise between useful signals.
1322 Gating is done by detecting the volume below a chosen level @var{threshold}
1323 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1324 floor is set via @var{range}. Because an exact manipulation of the signal
1325 would cause distortion of the waveform the reduction can be levelled over
1326 time. This is done by setting @var{attack} and @var{release}.
1328 @var{attack} determines how long the signal has to fall below the threshold
1329 before any reduction will occur and @var{release} sets the time the signal
1330 has to rise above the threshold to reduce the reduction again.
1331 Shorter signals than the chosen attack time will be left untouched.
1335 Set input level before filtering.
1336 Default is 1. Allowed range is from 0.015625 to 64.
1339 Set the mode of operation. Can be @code{upward} or @code{downward}.
1340 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1341 will be amplified, expanding dynamic range in upward direction.
1342 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1345 Set the level of gain reduction when the signal is below the threshold.
1346 Default is 0.06125. Allowed range is from 0 to 1.
1347 Setting this to 0 disables reduction and then filter behaves like expander.
1350 If a signal rises above this level the gain reduction is released.
1351 Default is 0.125. Allowed range is from 0 to 1.
1354 Set a ratio by which the signal is reduced.
1355 Default is 2. Allowed range is from 1 to 9000.
1358 Amount of milliseconds the signal has to rise above the threshold before gain
1360 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1363 Amount of milliseconds the signal has to fall below the threshold before the
1364 reduction is increased again. Default is 250 milliseconds.
1365 Allowed range is from 0.01 to 9000.
1368 Set amount of amplification of signal after processing.
1369 Default is 1. Allowed range is from 1 to 64.
1372 Curve the sharp knee around the threshold to enter gain reduction more softly.
1373 Default is 2.828427125. Allowed range is from 1 to 8.
1376 Choose if exact signal should be taken for detection or an RMS like one.
1377 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1380 Choose if the average level between all channels or the louder channel affects
1382 Default is @code{average}. Can be @code{average} or @code{maximum}.
1387 Apply an arbitrary Infinite Impulse Response filter.
1389 It accepts the following parameters:
1393 Set numerator/zeros coefficients.
1396 Set denominator/poles coefficients.
1408 Set coefficients format.
1414 Z-plane zeros/poles, cartesian (default)
1416 Z-plane zeros/poles, polar radians
1418 Z-plane zeros/poles, polar degrees
1422 Set kind of processing.
1423 Can be @code{d} - direct or @code{s} - serial cascading. Default is @code{s}.
1426 Set filtering precision.
1430 double-precision floating-point (default)
1432 single-precision floating-point
1440 How much to use filtered signal in output. Default is 1.
1441 Range is between 0 and 1.
1444 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1445 By default it is disabled.
1448 Set for which IR channel to display frequency response. By default is first channel
1449 displayed. This option is used only when @var{response} is enabled.
1452 Set video stream size. This option is used only when @var{response} is enabled.
1455 Coefficients in @code{tf} format are separated by spaces and are in ascending
1458 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1459 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1462 Different coefficients and gains can be provided for every channel, in such case
1463 use '|' to separate coefficients or gains. Last provided coefficients will be
1464 used for all remaining channels.
1466 @subsection Examples
1470 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1472 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
1476 Same as above but in @code{zp} format:
1478 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
1484 The limiter prevents an input signal from rising over a desired threshold.
1485 This limiter uses lookahead technology to prevent your signal from distorting.
1486 It means that there is a small delay after the signal is processed. Keep in mind
1487 that the delay it produces is the attack time you set.
1489 The filter accepts the following options:
1493 Set input gain. Default is 1.
1496 Set output gain. Default is 1.
1499 Don't let signals above this level pass the limiter. Default is 1.
1502 The limiter will reach its attenuation level in this amount of time in
1503 milliseconds. Default is 5 milliseconds.
1506 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1507 Default is 50 milliseconds.
1510 When gain reduction is always needed ASC takes care of releasing to an
1511 average reduction level rather than reaching a reduction of 0 in the release
1515 Select how much the release time is affected by ASC, 0 means nearly no changes
1516 in release time while 1 produces higher release times.
1519 Auto level output signal. Default is enabled.
1520 This normalizes audio back to 0dB if enabled.
1523 Depending on picked setting it is recommended to upsample input 2x or 4x times
1524 with @ref{aresample} before applying this filter.
1528 Apply a two-pole all-pass filter with central frequency (in Hz)
1529 @var{frequency}, and filter-width @var{width}.
1530 An all-pass filter changes the audio's frequency to phase relationship
1531 without changing its frequency to amplitude relationship.
1533 The filter accepts the following options:
1537 Set frequency in Hz.
1540 Set method to specify band-width of filter.
1555 Specify the band-width of a filter in width_type units.
1558 How much to use filtered signal in output. Default is 1.
1559 Range is between 0 and 1.
1562 Specify which channels to filter, by default all available are filtered.
1565 Normalize biquad coefficients, by default is disabled.
1566 Enabling it will normalize magnitude response at DC to 0dB.
1569 @subsection Commands
1571 This filter supports the following commands:
1574 Change allpass frequency.
1575 Syntax for the command is : "@var{frequency}"
1578 Change allpass width_type.
1579 Syntax for the command is : "@var{width_type}"
1582 Change allpass width.
1583 Syntax for the command is : "@var{width}"
1587 Syntax for the command is : "@var{mix}"
1594 The filter accepts the following options:
1598 Set the number of loops. Setting this value to -1 will result in infinite loops.
1602 Set maximal number of samples. Default is 0.
1605 Set first sample of loop. Default is 0.
1611 Merge two or more audio streams into a single multi-channel stream.
1613 The filter accepts the following options:
1618 Set the number of inputs. Default is 2.
1622 If the channel layouts of the inputs are disjoint, and therefore compatible,
1623 the channel layout of the output will be set accordingly and the channels
1624 will be reordered as necessary. If the channel layouts of the inputs are not
1625 disjoint, the output will have all the channels of the first input then all
1626 the channels of the second input, in that order, and the channel layout of
1627 the output will be the default value corresponding to the total number of
1630 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1631 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1632 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1633 first input, b1 is the first channel of the second input).
1635 On the other hand, if both input are in stereo, the output channels will be
1636 in the default order: a1, a2, b1, b2, and the channel layout will be
1637 arbitrarily set to 4.0, which may or may not be the expected value.
1639 All inputs must have the same sample rate, and format.
1641 If inputs do not have the same duration, the output will stop with the
1644 @subsection Examples
1648 Merge two mono files into a stereo stream:
1650 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1654 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1656 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
1662 Mixes multiple audio inputs into a single output.
1664 Note that this filter only supports float samples (the @var{amerge}
1665 and @var{pan} audio filters support many formats). If the @var{amix}
1666 input has integer samples then @ref{aresample} will be automatically
1667 inserted to perform the conversion to float samples.
1671 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1673 will mix 3 input audio streams to a single output with the same duration as the
1674 first input and a dropout transition time of 3 seconds.
1676 It accepts the following parameters:
1680 The number of inputs. If unspecified, it defaults to 2.
1683 How to determine the end-of-stream.
1687 The duration of the longest input. (default)
1690 The duration of the shortest input.
1693 The duration of the first input.
1697 @item dropout_transition
1698 The transition time, in seconds, for volume renormalization when an input
1699 stream ends. The default value is 2 seconds.
1702 Specify weight of each input audio stream as sequence.
1703 Each weight is separated by space. By default all inputs have same weight.
1708 Multiply first audio stream with second audio stream and store result
1709 in output audio stream. Multiplication is done by multiplying each
1710 sample from first stream with sample at same position from second stream.
1712 With this element-wise multiplication one can create amplitude fades and
1713 amplitude modulations.
1715 @section anequalizer
1717 High-order parametric multiband equalizer for each channel.
1719 It accepts the following parameters:
1723 This option string is in format:
1724 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1725 Each equalizer band is separated by '|'.
1729 Set channel number to which equalization will be applied.
1730 If input doesn't have that channel the entry is ignored.
1733 Set central frequency for band.
1734 If input doesn't have that frequency the entry is ignored.
1737 Set band width in hertz.
1740 Set band gain in dB.
1743 Set filter type for band, optional, can be:
1747 Butterworth, this is default.
1758 With this option activated frequency response of anequalizer is displayed
1762 Set video stream size. Only useful if curves option is activated.
1765 Set max gain that will be displayed. Only useful if curves option is activated.
1766 Setting this to a reasonable value makes it possible to display gain which is derived from
1767 neighbour bands which are too close to each other and thus produce higher gain
1768 when both are activated.
1771 Set frequency scale used to draw frequency response in video output.
1772 Can be linear or logarithmic. Default is logarithmic.
1775 Set color for each channel curve which is going to be displayed in video stream.
1776 This is list of color names separated by space or by '|'.
1777 Unrecognised or missing colors will be replaced by white color.
1780 @subsection Examples
1784 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1785 for first 2 channels using Chebyshev type 1 filter:
1787 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1791 @subsection Commands
1793 This filter supports the following commands:
1796 Alter existing filter parameters.
1797 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1799 @var{fN} is existing filter number, starting from 0, if no such filter is available
1801 @var{freq} set new frequency parameter.
1802 @var{width} set new width parameter in herz.
1803 @var{gain} set new gain parameter in dB.
1805 Full filter invocation with asendcmd may look like this:
1806 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1811 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1813 Each sample is adjusted by looking for other samples with similar contexts. This
1814 context similarity is defined by comparing their surrounding patches of size
1815 @option{p}. Patches are searched in an area of @option{r} around the sample.
1817 The filter accepts the following options:
1821 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1824 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1825 Default value is 2 milliseconds.
1828 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1829 Default value is 6 milliseconds.
1832 Set the output mode.
1834 It accepts the following values:
1837 Pass input unchanged.
1840 Pass noise filtered out.
1845 Default value is @var{o}.
1849 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
1852 @subsection Commands
1854 This filter supports the following commands:
1857 Change denoise strength. Argument is single float number.
1858 Syntax for the command is : "@var{s}"
1862 Syntax for the command is : "i", "o" or "n" string.
1866 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
1868 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
1869 relate to producing the least mean square of the error signal (difference between the desired,
1870 2nd input audio stream and the actual signal, the 1st input audio stream).
1872 A description of the accepted options follows.
1885 Set the filter leakage.
1888 It accepts the following values:
1897 Pass filtered samples.
1900 Pass difference between desired and filtered samples.
1902 Default value is @var{o}.
1906 @subsection Examples
1910 One of many usages of this filter is noise reduction, input audio is filtered
1911 with same samples that are delayed by fixed amount, one such example for stereo audio is:
1913 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
1917 @subsection Commands
1919 This filter supports the same commands as options, excluding option @code{order}.
1923 Pass the audio source unchanged to the output.
1927 Pad the end of an audio stream with silence.
1929 This can be used together with @command{ffmpeg} @option{-shortest} to
1930 extend audio streams to the same length as the video stream.
1932 A description of the accepted options follows.
1936 Set silence packet size. Default value is 4096.
1939 Set the number of samples of silence to add to the end. After the
1940 value is reached, the stream is terminated. This option is mutually
1941 exclusive with @option{whole_len}.
1944 Set the minimum total number of samples in the output audio stream. If
1945 the value is longer than the input audio length, silence is added to
1946 the end, until the value is reached. This option is mutually exclusive
1947 with @option{pad_len}.
1950 Specify the duration of samples of silence to add. See
1951 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1952 for the accepted syntax. Used only if set to non-zero value.
1955 Specify the minimum total duration in the output audio stream. See
1956 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1957 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
1958 the input audio length, silence is added to the end, until the value is reached.
1959 This option is mutually exclusive with @option{pad_dur}
1962 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
1963 nor @option{whole_dur} option is set, the filter will add silence to the end of
1964 the input stream indefinitely.
1966 @subsection Examples
1970 Add 1024 samples of silence to the end of the input:
1976 Make sure the audio output will contain at least 10000 samples, pad
1977 the input with silence if required:
1979 apad=whole_len=10000
1983 Use @command{ffmpeg} to pad the audio input with silence, so that the
1984 video stream will always result the shortest and will be converted
1985 until the end in the output file when using the @option{shortest}
1988 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
1993 Add a phasing effect to the input audio.
1995 A phaser filter creates series of peaks and troughs in the frequency spectrum.
1996 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
1998 A description of the accepted parameters follows.
2002 Set input gain. Default is 0.4.
2005 Set output gain. Default is 0.74
2008 Set delay in milliseconds. Default is 3.0.
2011 Set decay. Default is 0.4.
2014 Set modulation speed in Hz. Default is 0.5.
2017 Set modulation type. Default is triangular.
2019 It accepts the following values:
2028 Audio pulsator is something between an autopanner and a tremolo.
2029 But it can produce funny stereo effects as well. Pulsator changes the volume
2030 of the left and right channel based on a LFO (low frequency oscillator) with
2031 different waveforms and shifted phases.
2032 This filter have the ability to define an offset between left and right
2033 channel. An offset of 0 means that both LFO shapes match each other.
2034 The left and right channel are altered equally - a conventional tremolo.
2035 An offset of 50% means that the shape of the right channel is exactly shifted
2036 in phase (or moved backwards about half of the frequency) - pulsator acts as
2037 an autopanner. At 1 both curves match again. Every setting in between moves the
2038 phase shift gapless between all stages and produces some "bypassing" sounds with
2039 sine and triangle waveforms. The more you set the offset near 1 (starting from
2040 the 0.5) the faster the signal passes from the left to the right speaker.
2042 The filter accepts the following options:
2046 Set input gain. By default it is 1. Range is [0.015625 - 64].
2049 Set output gain. By default it is 1. Range is [0.015625 - 64].
2052 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2053 sawup or sawdown. Default is sine.
2056 Set modulation. Define how much of original signal is affected by the LFO.
2059 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2062 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2065 Set pulse width. Default is 1. Allowed range is [0 - 2].
2068 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2071 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2075 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2079 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2080 if timing is set to hz.
2086 Resample the input audio to the specified parameters, using the
2087 libswresample library. If none are specified then the filter will
2088 automatically convert between its input and output.
2090 This filter is also able to stretch/squeeze the audio data to make it match
2091 the timestamps or to inject silence / cut out audio to make it match the
2092 timestamps, do a combination of both or do neither.
2094 The filter accepts the syntax
2095 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2096 expresses a sample rate and @var{resampler_options} is a list of
2097 @var{key}=@var{value} pairs, separated by ":". See the
2098 @ref{Resampler Options,,"Resampler Options" section in the
2099 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2100 for the complete list of supported options.
2102 @subsection Examples
2106 Resample the input audio to 44100Hz:
2112 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2113 samples per second compensation:
2115 aresample=async=1000
2121 Reverse an audio clip.
2123 Warning: This filter requires memory to buffer the entire clip, so trimming
2126 @subsection Examples
2130 Take the first 5 seconds of a clip, and reverse it.
2132 atrim=end=5,areverse
2138 Reduce noise from speech using Recurrent Neural Networks.
2140 This filter accepts the following options:
2144 Set train model file to load. This option is always required.
2147 @section asetnsamples
2149 Set the number of samples per each output audio frame.
2151 The last output packet may contain a different number of samples, as
2152 the filter will flush all the remaining samples when the input audio
2155 The filter accepts the following options:
2159 @item nb_out_samples, n
2160 Set the number of frames per each output audio frame. The number is
2161 intended as the number of samples @emph{per each channel}.
2162 Default value is 1024.
2165 If set to 1, the filter will pad the last audio frame with zeroes, so
2166 that the last frame will contain the same number of samples as the
2167 previous ones. Default value is 1.
2170 For example, to set the number of per-frame samples to 1234 and
2171 disable padding for the last frame, use:
2173 asetnsamples=n=1234:p=0
2178 Set the sample rate without altering the PCM data.
2179 This will result in a change of speed and pitch.
2181 The filter accepts the following options:
2184 @item sample_rate, r
2185 Set the output sample rate. Default is 44100 Hz.
2190 Show a line containing various information for each input audio frame.
2191 The input audio is not modified.
2193 The shown line contains a sequence of key/value pairs of the form
2194 @var{key}:@var{value}.
2196 The following values are shown in the output:
2200 The (sequential) number of the input frame, starting from 0.
2203 The presentation timestamp of the input frame, in time base units; the time base
2204 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2207 The presentation timestamp of the input frame in seconds.
2210 position of the frame in the input stream, -1 if this information in
2211 unavailable and/or meaningless (for example in case of synthetic audio)
2220 The sample rate for the audio frame.
2223 The number of samples (per channel) in the frame.
2226 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2227 audio, the data is treated as if all the planes were concatenated.
2229 @item plane_checksums
2230 A list of Adler-32 checksums for each data plane.
2234 Apply audio soft clipping.
2236 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2237 along a smooth curve, rather than the abrupt shape of hard-clipping.
2239 This filter accepts the following options:
2243 Set type of soft-clipping.
2245 It accepts the following values:
2257 Set additional parameter which controls sigmoid function.
2260 @subsection Commands
2262 This filter supports the all above options as @ref{commands}.
2265 Automatic Speech Recognition
2267 This filter uses PocketSphinx for speech recognition. To enable
2268 compilation of this filter, you need to configure FFmpeg with
2269 @code{--enable-pocketsphinx}.
2271 It accepts the following options:
2275 Set sampling rate of input audio. Defaults is @code{16000}.
2276 This need to match speech models, otherwise one will get poor results.
2279 Set dictionary containing acoustic model files.
2282 Set pronunciation dictionary.
2285 Set language model file.
2288 Set language model set.
2291 Set which language model to use.
2294 Set output for log messages.
2297 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2302 Display time domain statistical information about the audio channels.
2303 Statistics are calculated and displayed for each audio channel and,
2304 where applicable, an overall figure is also given.
2306 It accepts the following option:
2309 Short window length in seconds, used for peak and trough RMS measurement.
2310 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2314 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2315 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2318 Available keys for each channel are:
2360 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2361 this @code{lavfi.astats.Overall.Peak_count}.
2363 For description what each key means read below.
2366 Set number of frame after which stats are going to be recalculated.
2367 Default is disabled.
2369 @item measure_perchannel
2370 Select the entries which need to be measured per channel. The metadata keys can
2371 be used as flags, default is @option{all} which measures everything.
2372 @option{none} disables all per channel measurement.
2374 @item measure_overall
2375 Select the entries which need to be measured overall. The metadata keys can
2376 be used as flags, default is @option{all} which measures everything.
2377 @option{none} disables all overall measurement.
2381 A description of each shown parameter follows:
2385 Mean amplitude displacement from zero.
2388 Minimal sample level.
2391 Maximal sample level.
2393 @item Min difference
2394 Minimal difference between two consecutive samples.
2396 @item Max difference
2397 Maximal difference between two consecutive samples.
2399 @item Mean difference
2400 Mean difference between two consecutive samples.
2401 The average of each difference between two consecutive samples.
2403 @item RMS difference
2404 Root Mean Square difference between two consecutive samples.
2408 Standard peak and RMS level measured in dBFS.
2412 Peak and trough values for RMS level measured over a short window.
2415 Standard ratio of peak to RMS level (note: not in dB).
2418 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2419 (i.e. either @var{Min level} or @var{Max level}).
2422 Number of occasions (not the number of samples) that the signal attained either
2423 @var{Min level} or @var{Max level}.
2426 Overall bit depth of audio. Number of bits used for each sample.
2429 Measured dynamic range of audio in dB.
2431 @item Zero crossings
2432 Number of points where the waveform crosses the zero level axis.
2434 @item Zero crossings rate
2435 Rate of Zero crossings and number of audio samples.
2442 The filter accepts exactly one parameter, the audio tempo. If not
2443 specified then the filter will assume nominal 1.0 tempo. Tempo must
2444 be in the [0.5, 100.0] range.
2446 Note that tempo greater than 2 will skip some samples rather than
2447 blend them in. If for any reason this is a concern it is always
2448 possible to daisy-chain several instances of atempo to achieve the
2449 desired product tempo.
2451 @subsection Examples
2455 Slow down audio to 80% tempo:
2461 To speed up audio to 300% tempo:
2467 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2469 atempo=sqrt(3),atempo=sqrt(3)
2473 @subsection Commands
2475 This filter supports the following commands:
2478 Change filter tempo scale factor.
2479 Syntax for the command is : "@var{tempo}"
2484 Trim the input so that the output contains one continuous subpart of the input.
2486 It accepts the following parameters:
2489 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2490 sample with the timestamp @var{start} will be the first sample in the output.
2493 Specify time of the first audio sample that will be dropped, i.e. the
2494 audio sample immediately preceding the one with the timestamp @var{end} will be
2495 the last sample in the output.
2498 Same as @var{start}, except this option sets the start timestamp in samples
2502 Same as @var{end}, except this option sets the end timestamp in samples instead
2506 The maximum duration of the output in seconds.
2509 The number of the first sample that should be output.
2512 The number of the first sample that should be dropped.
2515 @option{start}, @option{end}, and @option{duration} are expressed as time
2516 duration specifications; see
2517 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2519 Note that the first two sets of the start/end options and the @option{duration}
2520 option look at the frame timestamp, while the _sample options simply count the
2521 samples that pass through the filter. So start/end_pts and start/end_sample will
2522 give different results when the timestamps are wrong, inexact or do not start at
2523 zero. Also note that this filter does not modify the timestamps. If you wish
2524 to have the output timestamps start at zero, insert the asetpts filter after the
2527 If multiple start or end options are set, this filter tries to be greedy and
2528 keep all samples that match at least one of the specified constraints. To keep
2529 only the part that matches all the constraints at once, chain multiple atrim
2532 The defaults are such that all the input is kept. So it is possible to set e.g.
2533 just the end values to keep everything before the specified time.
2538 Drop everything except the second minute of input:
2540 ffmpeg -i INPUT -af atrim=60:120
2544 Keep only the first 1000 samples:
2546 ffmpeg -i INPUT -af atrim=end_sample=1000
2551 @section axcorrelate
2552 Calculate normalized cross-correlation between two input audio streams.
2554 Resulted samples are always between -1 and 1 inclusive.
2555 If result is 1 it means two input samples are highly correlated in that selected segment.
2556 Result 0 means they are not correlated at all.
2557 If result is -1 it means two input samples are out of phase, which means they cancel each
2560 The filter accepts the following options:
2564 Set size of segment over which cross-correlation is calculated.
2565 Default is 256. Allowed range is from 2 to 131072.
2568 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2569 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2570 are always zero and thus need much less calculations to make.
2571 This is generally not true, but is valid for typical audio streams.
2574 @subsection Examples
2578 Calculate correlation between channels in stereo audio stream:
2580 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2586 Apply a two-pole Butterworth band-pass filter with central
2587 frequency @var{frequency}, and (3dB-point) band-width width.
2588 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2589 instead of the default: constant 0dB peak gain.
2590 The filter roll off at 6dB per octave (20dB per decade).
2592 The filter accepts the following options:
2596 Set the filter's central frequency. Default is @code{3000}.
2599 Constant skirt gain if set to 1. Defaults to 0.
2602 Set method to specify band-width of filter.
2617 Specify the band-width of a filter in width_type units.
2620 How much to use filtered signal in output. Default is 1.
2621 Range is between 0 and 1.
2624 Specify which channels to filter, by default all available are filtered.
2627 Normalize biquad coefficients, by default is disabled.
2628 Enabling it will normalize magnitude response at DC to 0dB.
2631 @subsection Commands
2633 This filter supports the following commands:
2636 Change bandpass frequency.
2637 Syntax for the command is : "@var{frequency}"
2640 Change bandpass width_type.
2641 Syntax for the command is : "@var{width_type}"
2644 Change bandpass width.
2645 Syntax for the command is : "@var{width}"
2648 Change bandpass mix.
2649 Syntax for the command is : "@var{mix}"
2654 Apply a two-pole Butterworth band-reject filter with central
2655 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2656 The filter roll off at 6dB per octave (20dB per decade).
2658 The filter accepts the following options:
2662 Set the filter's central frequency. Default is @code{3000}.
2665 Set method to specify band-width of filter.
2680 Specify the band-width of a filter in width_type units.
2683 How much to use filtered signal in output. Default is 1.
2684 Range is between 0 and 1.
2687 Specify which channels to filter, by default all available are filtered.
2690 Normalize biquad coefficients, by default is disabled.
2691 Enabling it will normalize magnitude response at DC to 0dB.
2694 @subsection Commands
2696 This filter supports the following commands:
2699 Change bandreject frequency.
2700 Syntax for the command is : "@var{frequency}"
2703 Change bandreject width_type.
2704 Syntax for the command is : "@var{width_type}"
2707 Change bandreject width.
2708 Syntax for the command is : "@var{width}"
2711 Change bandreject mix.
2712 Syntax for the command is : "@var{mix}"
2715 @section bass, lowshelf
2717 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2718 shelving filter with a response similar to that of a standard
2719 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2721 The filter accepts the following options:
2725 Give the gain at 0 Hz. Its useful range is about -20
2726 (for a large cut) to +20 (for a large boost).
2727 Beware of clipping when using a positive gain.
2730 Set the filter's central frequency and so can be used
2731 to extend or reduce the frequency range to be boosted or cut.
2732 The default value is @code{100} Hz.
2735 Set method to specify band-width of filter.
2750 Determine how steep is the filter's shelf transition.
2753 How much to use filtered signal in output. Default is 1.
2754 Range is between 0 and 1.
2757 Specify which channels to filter, by default all available are filtered.
2760 Normalize biquad coefficients, by default is disabled.
2761 Enabling it will normalize magnitude response at DC to 0dB.
2764 @subsection Commands
2766 This filter supports the following commands:
2769 Change bass frequency.
2770 Syntax for the command is : "@var{frequency}"
2773 Change bass width_type.
2774 Syntax for the command is : "@var{width_type}"
2778 Syntax for the command is : "@var{width}"
2782 Syntax for the command is : "@var{gain}"
2786 Syntax for the command is : "@var{mix}"
2791 Apply a biquad IIR filter with the given coefficients.
2792 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
2793 are the numerator and denominator coefficients respectively.
2794 and @var{channels}, @var{c} specify which channels to filter, by default all
2795 available are filtered.
2797 @subsection Commands
2799 This filter supports the following commands:
2807 Change biquad parameter.
2808 Syntax for the command is : "@var{value}"
2811 How much to use filtered signal in output. Default is 1.
2812 Range is between 0 and 1.
2815 Specify which channels to filter, by default all available are filtered.
2818 Normalize biquad coefficients, by default is disabled.
2819 Enabling it will normalize magnitude response at DC to 0dB.
2823 Bauer stereo to binaural transformation, which improves headphone listening of
2824 stereo audio records.
2826 To enable compilation of this filter you need to configure FFmpeg with
2827 @code{--enable-libbs2b}.
2829 It accepts the following parameters:
2833 Pre-defined crossfeed level.
2837 Default level (fcut=700, feed=50).
2840 Chu Moy circuit (fcut=700, feed=60).
2843 Jan Meier circuit (fcut=650, feed=95).
2848 Cut frequency (in Hz).
2857 Remap input channels to new locations.
2859 It accepts the following parameters:
2862 Map channels from input to output. The argument is a '|'-separated list of
2863 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
2864 @var{in_channel} form. @var{in_channel} can be either the name of the input
2865 channel (e.g. FL for front left) or its index in the input channel layout.
2866 @var{out_channel} is the name of the output channel or its index in the output
2867 channel layout. If @var{out_channel} is not given then it is implicitly an
2868 index, starting with zero and increasing by one for each mapping.
2870 @item channel_layout
2871 The channel layout of the output stream.
2874 If no mapping is present, the filter will implicitly map input channels to
2875 output channels, preserving indices.
2877 @subsection Examples
2881 For example, assuming a 5.1+downmix input MOV file,
2883 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
2885 will create an output WAV file tagged as stereo from the downmix channels of
2889 To fix a 5.1 WAV improperly encoded in AAC's native channel order
2891 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
2895 @section channelsplit
2897 Split each channel from an input audio stream into a separate output stream.
2899 It accepts the following parameters:
2901 @item channel_layout
2902 The channel layout of the input stream. The default is "stereo".
2904 A channel layout describing the channels to be extracted as separate output streams
2905 or "all" to extract each input channel as a separate stream. The default is "all".
2907 Choosing channels not present in channel layout in the input will result in an error.
2910 @subsection Examples
2914 For example, assuming a stereo input MP3 file,
2916 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
2918 will create an output Matroska file with two audio streams, one containing only
2919 the left channel and the other the right channel.
2922 Split a 5.1 WAV file into per-channel files:
2924 ffmpeg -i in.wav -filter_complex
2925 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
2926 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
2927 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
2932 Extract only LFE from a 5.1 WAV file:
2934 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
2935 -map '[LFE]' lfe.wav
2940 Add a chorus effect to the audio.
2942 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
2944 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
2945 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
2946 The modulation depth defines the range the modulated delay is played before or after
2947 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
2948 sound tuned around the original one, like in a chorus where some vocals are slightly
2951 It accepts the following parameters:
2954 Set input gain. Default is 0.4.
2957 Set output gain. Default is 0.4.
2960 Set delays. A typical delay is around 40ms to 60ms.
2972 @subsection Examples
2978 chorus=0.7:0.9:55:0.4:0.25:2
2984 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
2988 Fuller sounding chorus with three delays:
2990 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
2995 Compress or expand the audio's dynamic range.
2997 It accepts the following parameters:
3003 A list of times in seconds for each channel over which the instantaneous level
3004 of the input signal is averaged to determine its volume. @var{attacks} refers to
3005 increase of volume and @var{decays} refers to decrease of volume. For most
3006 situations, the attack time (response to the audio getting louder) should be
3007 shorter than the decay time, because the human ear is more sensitive to sudden
3008 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3009 a typical value for decay is 0.8 seconds.
3010 If specified number of attacks & decays is lower than number of channels, the last
3011 set attack/decay will be used for all remaining channels.
3014 A list of points for the transfer function, specified in dB relative to the
3015 maximum possible signal amplitude. Each key points list must be defined using
3016 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3017 @code{x0/y0 x1/y1 x2/y2 ....}
3019 The input values must be in strictly increasing order but the transfer function
3020 does not have to be monotonically rising. The point @code{0/0} is assumed but
3021 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3022 function are @code{-70/-70|-60/-20|1/0}.
3025 Set the curve radius in dB for all joints. It defaults to 0.01.
3028 Set the additional gain in dB to be applied at all points on the transfer
3029 function. This allows for easy adjustment of the overall gain.
3033 Set an initial volume, in dB, to be assumed for each channel when filtering
3034 starts. This permits the user to supply a nominal level initially, so that, for
3035 example, a very large gain is not applied to initial signal levels before the
3036 companding has begun to operate. A typical value for audio which is initially
3037 quiet is -90 dB. It defaults to 0.
3040 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3041 delayed before being fed to the volume adjuster. Specifying a delay
3042 approximately equal to the attack/decay times allows the filter to effectively
3043 operate in predictive rather than reactive mode. It defaults to 0.
3047 @subsection Examples
3051 Make music with both quiet and loud passages suitable for listening to in a
3054 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3057 Another example for audio with whisper and explosion parts:
3059 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3063 A noise gate for when the noise is at a lower level than the signal:
3065 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3069 Here is another noise gate, this time for when the noise is at a higher level
3070 than the signal (making it, in some ways, similar to squelch):
3072 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3076 2:1 compression starting at -6dB:
3078 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3082 2:1 compression starting at -9dB:
3084 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3088 2:1 compression starting at -12dB:
3090 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3094 2:1 compression starting at -18dB:
3096 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3100 3:1 compression starting at -15dB:
3102 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3108 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3114 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
3118 Hard limiter at -6dB:
3120 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3124 Hard limiter at -12dB:
3126 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3130 Hard noise gate at -35 dB:
3132 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3138 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3142 @section compensationdelay
3144 Compensation Delay Line is a metric based delay to compensate differing
3145 positions of microphones or speakers.
3147 For example, you have recorded guitar with two microphones placed in
3148 different locations. Because the front of sound wave has fixed speed in
3149 normal conditions, the phasing of microphones can vary and depends on
3150 their location and interposition. The best sound mix can be achieved when
3151 these microphones are in phase (synchronized). Note that a distance of
3152 ~30 cm between microphones makes one microphone capture the signal in
3153 antiphase to the other microphone. That makes the final mix sound moody.
3154 This filter helps to solve phasing problems by adding different delays
3155 to each microphone track and make them synchronized.
3157 The best result can be reached when you take one track as base and
3158 synchronize other tracks one by one with it.
3159 Remember that synchronization/delay tolerance depends on sample rate, too.
3160 Higher sample rates will give more tolerance.
3162 The filter accepts the following parameters:
3166 Set millimeters distance. This is compensation distance for fine tuning.
3170 Set cm distance. This is compensation distance for tightening distance setup.
3174 Set meters distance. This is compensation distance for hard distance setup.
3178 Set dry amount. Amount of unprocessed (dry) signal.
3182 Set wet amount. Amount of processed (wet) signal.
3186 Set temperature in degrees Celsius. This is the temperature of the environment.
3191 Apply headphone crossfeed filter.
3193 Crossfeed is the process of blending the left and right channels of stereo
3195 It is mainly used to reduce extreme stereo separation of low frequencies.
3197 The intent is to produce more speaker like sound to the listener.
3199 The filter accepts the following options:
3203 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3204 This sets gain of low shelf filter for side part of stereo image.
3205 Default is -6dB. Max allowed is -30db when strength is set to 1.
3208 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3209 This sets cut off frequency of low shelf filter. Default is cut off near
3210 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3213 Set input gain. Default is 0.9.
3216 Set output gain. Default is 1.
3219 @section crystalizer
3220 Simple algorithm to expand audio dynamic range.
3222 The filter accepts the following options:
3226 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3227 (unchanged sound) to 10.0 (maximum effect).
3230 Enable clipping. By default is enabled.
3233 @subsection Commands
3235 This filter supports the all above options as @ref{commands}.
3238 Apply a DC shift to the audio.
3240 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3241 in the recording chain) from the audio. The effect of a DC offset is reduced
3242 headroom and hence volume. The @ref{astats} filter can be used to determine if
3243 a signal has a DC offset.
3247 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3251 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3252 used to prevent clipping.
3257 Apply de-essing to the audio samples.
3261 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3265 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3269 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3273 Set the output mode.
3275 It accepts the following values:
3278 Pass input unchanged.
3281 Pass ess filtered out.
3286 Default value is @var{o}.
3292 Measure audio dynamic range.
3294 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3295 is found in transition material. And anything less that 8 have very poor dynamics
3296 and is very compressed.
3298 The filter accepts the following options:
3302 Set window length in seconds used to split audio into segments of equal length.
3303 Default is 3 seconds.
3307 Dynamic Audio Normalizer.
3309 This filter applies a certain amount of gain to the input audio in order
3310 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3311 contrast to more "simple" normalization algorithms, the Dynamic Audio
3312 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3313 This allows for applying extra gain to the "quiet" sections of the audio
3314 while avoiding distortions or clipping the "loud" sections. In other words:
3315 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3316 sections, in the sense that the volume of each section is brought to the
3317 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3318 this goal *without* applying "dynamic range compressing". It will retain 100%
3319 of the dynamic range *within* each section of the audio file.
3323 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3324 Default is 500 milliseconds.
3325 The Dynamic Audio Normalizer processes the input audio in small chunks,
3326 referred to as frames. This is required, because a peak magnitude has no
3327 meaning for just a single sample value. Instead, we need to determine the
3328 peak magnitude for a contiguous sequence of sample values. While a "standard"
3329 normalizer would simply use the peak magnitude of the complete file, the
3330 Dynamic Audio Normalizer determines the peak magnitude individually for each
3331 frame. The length of a frame is specified in milliseconds. By default, the
3332 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3333 been found to give good results with most files.
3334 Note that the exact frame length, in number of samples, will be determined
3335 automatically, based on the sampling rate of the individual input audio file.
3338 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3339 number. Default is 31.
3340 Probably the most important parameter of the Dynamic Audio Normalizer is the
3341 @code{window size} of the Gaussian smoothing filter. The filter's window size
3342 is specified in frames, centered around the current frame. For the sake of
3343 simplicity, this must be an odd number. Consequently, the default value of 31
3344 takes into account the current frame, as well as the 15 preceding frames and
3345 the 15 subsequent frames. Using a larger window results in a stronger
3346 smoothing effect and thus in less gain variation, i.e. slower gain
3347 adaptation. Conversely, using a smaller window results in a weaker smoothing
3348 effect and thus in more gain variation, i.e. faster gain adaptation.
3349 In other words, the more you increase this value, the more the Dynamic Audio
3350 Normalizer will behave like a "traditional" normalization filter. On the
3351 contrary, the more you decrease this value, the more the Dynamic Audio
3352 Normalizer will behave like a dynamic range compressor.
3355 Set the target peak value. This specifies the highest permissible magnitude
3356 level for the normalized audio input. This filter will try to approach the
3357 target peak magnitude as closely as possible, but at the same time it also
3358 makes sure that the normalized signal will never exceed the peak magnitude.
3359 A frame's maximum local gain factor is imposed directly by the target peak
3360 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3361 It is not recommended to go above this value.
3364 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3365 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3366 factor for each input frame, i.e. the maximum gain factor that does not
3367 result in clipping or distortion. The maximum gain factor is determined by
3368 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3369 additionally bounds the frame's maximum gain factor by a predetermined
3370 (global) maximum gain factor. This is done in order to avoid excessive gain
3371 factors in "silent" or almost silent frames. By default, the maximum gain
3372 factor is 10.0, For most inputs the default value should be sufficient and
3373 it usually is not recommended to increase this value. Though, for input
3374 with an extremely low overall volume level, it may be necessary to allow even
3375 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3376 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3377 Instead, a "sigmoid" threshold function will be applied. This way, the
3378 gain factors will smoothly approach the threshold value, but never exceed that
3382 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3383 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3384 This means that the maximum local gain factor for each frame is defined
3385 (only) by the frame's highest magnitude sample. This way, the samples can
3386 be amplified as much as possible without exceeding the maximum signal
3387 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3388 Normalizer can also take into account the frame's root mean square,
3389 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3390 determine the power of a time-varying signal. It is therefore considered
3391 that the RMS is a better approximation of the "perceived loudness" than
3392 just looking at the signal's peak magnitude. Consequently, by adjusting all
3393 frames to a constant RMS value, a uniform "perceived loudness" can be
3394 established. If a target RMS value has been specified, a frame's local gain
3395 factor is defined as the factor that would result in exactly that RMS value.
3396 Note, however, that the maximum local gain factor is still restricted by the
3397 frame's highest magnitude sample, in order to prevent clipping.
3400 Enable channels coupling. By default is enabled.
3401 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3402 amount. This means the same gain factor will be applied to all channels, i.e.
3403 the maximum possible gain factor is determined by the "loudest" channel.
3404 However, in some recordings, it may happen that the volume of the different
3405 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3406 In this case, this option can be used to disable the channel coupling. This way,
3407 the gain factor will be determined independently for each channel, depending
3408 only on the individual channel's highest magnitude sample. This allows for
3409 harmonizing the volume of the different channels.
3412 Enable DC bias correction. By default is disabled.
3413 An audio signal (in the time domain) is a sequence of sample values.
3414 In the Dynamic Audio Normalizer these sample values are represented in the
3415 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3416 audio signal, or "waveform", should be centered around the zero point.
3417 That means if we calculate the mean value of all samples in a file, or in a
3418 single frame, then the result should be 0.0 or at least very close to that
3419 value. If, however, there is a significant deviation of the mean value from
3420 0.0, in either positive or negative direction, this is referred to as a
3421 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3422 Audio Normalizer provides optional DC bias correction.
3423 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3424 the mean value, or "DC correction" offset, of each input frame and subtract
3425 that value from all of the frame's sample values which ensures those samples
3426 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3427 boundaries, the DC correction offset values will be interpolated smoothly
3428 between neighbouring frames.
3430 @item altboundary, b
3431 Enable alternative boundary mode. By default is disabled.
3432 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3433 around each frame. This includes the preceding frames as well as the
3434 subsequent frames. However, for the "boundary" frames, located at the very
3435 beginning and at the very end of the audio file, not all neighbouring
3436 frames are available. In particular, for the first few frames in the audio
3437 file, the preceding frames are not known. And, similarly, for the last few
3438 frames in the audio file, the subsequent frames are not known. Thus, the
3439 question arises which gain factors should be assumed for the missing frames
3440 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3441 to deal with this situation. The default boundary mode assumes a gain factor
3442 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3443 "fade out" at the beginning and at the end of the input, respectively.
3446 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3447 By default, the Dynamic Audio Normalizer does not apply "traditional"
3448 compression. This means that signal peaks will not be pruned and thus the
3449 full dynamic range will be retained within each local neighbourhood. However,
3450 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3451 normalization algorithm with a more "traditional" compression.
3452 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3453 (thresholding) function. If (and only if) the compression feature is enabled,
3454 all input frames will be processed by a soft knee thresholding function prior
3455 to the actual normalization process. Put simply, the thresholding function is
3456 going to prune all samples whose magnitude exceeds a certain threshold value.
3457 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3458 value. Instead, the threshold value will be adjusted for each individual
3460 In general, smaller parameters result in stronger compression, and vice versa.
3461 Values below 3.0 are not recommended, because audible distortion may appear.
3464 Set the target threshold value. This specifies the lowest permissible
3465 magnitude level for the audio input which will be normalized.
3466 If input frame volume is above this value frame will be normalized.
3467 Otherwise frame may not be normalized at all. The default value is set
3468 to 0, which means all input frames will be normalized.
3469 This option is mostly useful if digital noise is not wanted to be amplified.
3472 @subsection Commands
3474 This filter supports the all above options as @ref{commands}.
3478 Make audio easier to listen to on headphones.
3480 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3481 so that when listened to on headphones the stereo image is moved from
3482 inside your head (standard for headphones) to outside and in front of
3483 the listener (standard for speakers).
3489 Apply a two-pole peaking equalisation (EQ) filter. With this
3490 filter, the signal-level at and around a selected frequency can
3491 be increased or decreased, whilst (unlike bandpass and bandreject
3492 filters) that at all other frequencies is unchanged.
3494 In order to produce complex equalisation curves, this filter can
3495 be given several times, each with a different central frequency.
3497 The filter accepts the following options:
3501 Set the filter's central frequency in Hz.
3504 Set method to specify band-width of filter.
3519 Specify the band-width of a filter in width_type units.
3522 Set the required gain or attenuation in dB.
3523 Beware of clipping when using a positive gain.
3526 How much to use filtered signal in output. Default is 1.
3527 Range is between 0 and 1.
3530 Specify which channels to filter, by default all available are filtered.
3533 Normalize biquad coefficients, by default is disabled.
3534 Enabling it will normalize magnitude response at DC to 0dB.
3537 @subsection Examples
3540 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3542 equalizer=f=1000:t=h:width=200:g=-10
3546 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3548 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3552 @subsection Commands
3554 This filter supports the following commands:
3557 Change equalizer frequency.
3558 Syntax for the command is : "@var{frequency}"
3561 Change equalizer width_type.
3562 Syntax for the command is : "@var{width_type}"
3565 Change equalizer width.
3566 Syntax for the command is : "@var{width}"
3569 Change equalizer gain.
3570 Syntax for the command is : "@var{gain}"
3573 Change equalizer mix.
3574 Syntax for the command is : "@var{mix}"
3577 @section extrastereo
3579 Linearly increases the difference between left and right channels which
3580 adds some sort of "live" effect to playback.
3582 The filter accepts the following options:
3586 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3587 (average of both channels), with 1.0 sound will be unchanged, with
3588 -1.0 left and right channels will be swapped.
3591 Enable clipping. By default is enabled.
3594 @subsection Commands
3596 This filter supports the all above options as @ref{commands}.
3598 @section firequalizer
3599 Apply FIR Equalization using arbitrary frequency response.
3601 The filter accepts the following option:
3605 Set gain curve equation (in dB). The expression can contain variables:
3608 the evaluated frequency
3612 channel number, set to 0 when multichannels evaluation is disabled
3614 channel id, see libavutil/channel_layout.h, set to the first channel id when
3615 multichannels evaluation is disabled
3619 channel_layout, see libavutil/channel_layout.h
3624 @item gain_interpolate(f)
3625 interpolate gain on frequency f based on gain_entry
3626 @item cubic_interpolate(f)
3627 same as gain_interpolate, but smoother
3629 This option is also available as command. Default is @code{gain_interpolate(f)}.
3632 Set gain entry for gain_interpolate function. The expression can
3636 store gain entry at frequency f with value g
3638 This option is also available as command.
3641 Set filter delay in seconds. Higher value means more accurate.
3642 Default is @code{0.01}.
3645 Set filter accuracy in Hz. Lower value means more accurate.
3646 Default is @code{5}.
3649 Set window function. Acceptable values are:
3652 rectangular window, useful when gain curve is already smooth
3654 hann window (default)
3660 3-terms continuous 1st derivative nuttall window
3662 minimum 3-terms discontinuous nuttall window
3664 4-terms continuous 1st derivative nuttall window
3666 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3668 blackman-harris window
3674 If enabled, use fixed number of audio samples. This improves speed when
3675 filtering with large delay. Default is disabled.
3678 Enable multichannels evaluation on gain. Default is disabled.
3681 Enable zero phase mode by subtracting timestamp to compensate delay.
3682 Default is disabled.
3685 Set scale used by gain. Acceptable values are:
3688 linear frequency, linear gain
3690 linear frequency, logarithmic (in dB) gain (default)
3692 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3694 logarithmic frequency, logarithmic gain
3698 Set file for dumping, suitable for gnuplot.
3701 Set scale for dumpfile. Acceptable values are same with scale option.
3705 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3706 Default is disabled.
3709 Enable minimum phase impulse response. Default is disabled.
3712 @subsection Examples
3717 firequalizer=gain='if(lt(f,1000), 0, -INF)'
3720 lowpass at 1000 Hz with gain_entry:
3722 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
3725 custom equalization:
3727 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
3730 higher delay with zero phase to compensate delay:
3732 firequalizer=delay=0.1:fixed=on:zero_phase=on
3735 lowpass on left channel, highpass on right channel:
3737 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
3738 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
3743 Apply a flanging effect to the audio.
3745 The filter accepts the following options:
3749 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
3752 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
3755 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
3759 Set percentage of delayed signal mixed with original. Range from 0 to 100.
3760 Default value is 71.
3763 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
3766 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
3767 Default value is @var{sinusoidal}.
3770 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
3771 Default value is 25.
3774 Set delay-line interpolation, @var{linear} or @var{quadratic}.
3775 Default is @var{linear}.
3779 Apply Haas effect to audio.
3781 Note that this makes most sense to apply on mono signals.
3782 With this filter applied to mono signals it give some directionality and
3783 stretches its stereo image.
3785 The filter accepts the following options:
3789 Set input level. By default is @var{1}, or 0dB
3792 Set output level. By default is @var{1}, or 0dB.
3795 Set gain applied to side part of signal. By default is @var{1}.
3798 Set kind of middle source. Can be one of the following:
3808 Pick middle part signal of stereo image.
3811 Pick side part signal of stereo image.
3815 Change middle phase. By default is disabled.
3818 Set left channel delay. By default is @var{2.05} milliseconds.
3821 Set left channel balance. By default is @var{-1}.
3824 Set left channel gain. By default is @var{1}.
3827 Change left phase. By default is disabled.
3830 Set right channel delay. By defaults is @var{2.12} milliseconds.
3833 Set right channel balance. By default is @var{1}.
3836 Set right channel gain. By default is @var{1}.
3839 Change right phase. By default is enabled.
3844 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
3845 embedded HDCD codes is expanded into a 20-bit PCM stream.
3847 The filter supports the Peak Extend and Low-level Gain Adjustment features
3848 of HDCD, and detects the Transient Filter flag.
3851 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
3854 When using the filter with wav, note the default encoding for wav is 16-bit,
3855 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
3856 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
3858 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
3859 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
3862 The filter accepts the following options:
3865 @item disable_autoconvert
3866 Disable any automatic format conversion or resampling in the filter graph.
3868 @item process_stereo
3869 Process the stereo channels together. If target_gain does not match between
3870 channels, consider it invalid and use the last valid target_gain.
3873 Set the code detect timer period in ms.
3876 Always extend peaks above -3dBFS even if PE isn't signaled.
3879 Replace audio with a solid tone and adjust the amplitude to signal some
3880 specific aspect of the decoding process. The output file can be loaded in
3881 an audio editor alongside the original to aid analysis.
3883 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
3890 Gain adjustment level at each sample
3892 Samples where peak extend occurs
3894 Samples where the code detect timer is active
3896 Samples where the target gain does not match between channels
3902 Apply head-related transfer functions (HRTFs) to create virtual
3903 loudspeakers around the user for binaural listening via headphones.
3904 The HRIRs are provided via additional streams, for each channel
3905 one stereo input stream is needed.
3907 The filter accepts the following options:
3911 Set mapping of input streams for convolution.
3912 The argument is a '|'-separated list of channel names in order as they
3913 are given as additional stream inputs for filter.
3914 This also specify number of input streams. Number of input streams
3915 must be not less than number of channels in first stream plus one.
3918 Set gain applied to audio. Value is in dB. Default is 0.
3921 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
3922 processing audio in time domain which is slow.
3923 @var{freq} is processing audio in frequency domain which is fast.
3924 Default is @var{freq}.
3927 Set custom gain for LFE channels. Value is in dB. Default is 0.
3930 Set size of frame in number of samples which will be processed at once.
3931 Default value is @var{1024}. Allowed range is from 1024 to 96000.
3934 Set format of hrir stream.
3935 Default value is @var{stereo}. Alternative value is @var{multich}.
3936 If value is set to @var{stereo}, number of additional streams should
3937 be greater or equal to number of input channels in first input stream.
3938 Also each additional stream should have stereo number of channels.
3939 If value is set to @var{multich}, number of additional streams should
3940 be exactly one. Also number of input channels of additional stream
3941 should be equal or greater than twice number of channels of first input
3945 @subsection Examples
3949 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3950 each amovie filter use stereo file with IR coefficients as input.
3951 The files give coefficients for each position of virtual loudspeaker:
3954 -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"
3959 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3960 but now in @var{multich} @var{hrir} format.
3962 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"
3969 Apply a high-pass filter with 3dB point frequency.
3970 The filter can be either single-pole, or double-pole (the default).
3971 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
3973 The filter accepts the following options:
3977 Set frequency in Hz. Default is 3000.
3980 Set number of poles. Default is 2.
3983 Set method to specify band-width of filter.
3998 Specify the band-width of a filter in width_type units.
3999 Applies only to double-pole filter.
4000 The default is 0.707q and gives a Butterworth response.
4003 How much to use filtered signal in output. Default is 1.
4004 Range is between 0 and 1.
4007 Specify which channels to filter, by default all available are filtered.
4010 Normalize biquad coefficients, by default is disabled.
4011 Enabling it will normalize magnitude response at DC to 0dB.
4014 @subsection Commands
4016 This filter supports the following commands:
4019 Change highpass frequency.
4020 Syntax for the command is : "@var{frequency}"
4023 Change highpass width_type.
4024 Syntax for the command is : "@var{width_type}"
4027 Change highpass width.
4028 Syntax for the command is : "@var{width}"
4031 Change highpass mix.
4032 Syntax for the command is : "@var{mix}"
4037 Join multiple input streams into one multi-channel stream.
4039 It accepts the following parameters:
4043 The number of input streams. It defaults to 2.
4045 @item channel_layout
4046 The desired output channel layout. It defaults to stereo.
4049 Map channels from inputs to output. The argument is a '|'-separated list of
4050 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4051 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4052 can be either the name of the input channel (e.g. FL for front left) or its
4053 index in the specified input stream. @var{out_channel} is the name of the output
4057 The filter will attempt to guess the mappings when they are not specified
4058 explicitly. It does so by first trying to find an unused matching input channel
4059 and if that fails it picks the first unused input channel.
4061 Join 3 inputs (with properly set channel layouts):
4063 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4066 Build a 5.1 output from 6 single-channel streams:
4068 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4069 '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'
4075 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4077 To enable compilation of this filter you need to configure FFmpeg with
4078 @code{--enable-ladspa}.
4082 Specifies the name of LADSPA plugin library to load. If the environment
4083 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4084 each one of the directories specified by the colon separated list in
4085 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4086 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4087 @file{/usr/lib/ladspa/}.
4090 Specifies the plugin within the library. Some libraries contain only
4091 one plugin, but others contain many of them. If this is not set filter
4092 will list all available plugins within the specified library.
4095 Set the '|' separated list of controls which are zero or more floating point
4096 values that determine the behavior of the loaded plugin (for example delay,
4098 Controls need to be defined using the following syntax:
4099 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4100 @var{valuei} is the value set on the @var{i}-th control.
4101 Alternatively they can be also defined using the following syntax:
4102 @var{value0}|@var{value1}|@var{value2}|..., where
4103 @var{valuei} is the value set on the @var{i}-th control.
4104 If @option{controls} is set to @code{help}, all available controls and
4105 their valid ranges are printed.
4107 @item sample_rate, s
4108 Specify the sample rate, default to 44100. Only used if plugin have
4112 Set the number of samples per channel per each output frame, default
4113 is 1024. Only used if plugin have zero inputs.
4116 Set the minimum duration of the sourced audio. See
4117 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4118 for the accepted syntax.
4119 Note that the resulting duration may be greater than the specified duration,
4120 as the generated audio is always cut at the end of a complete frame.
4121 If not specified, or the expressed duration is negative, the audio is
4122 supposed to be generated forever.
4123 Only used if plugin have zero inputs.
4127 @subsection Examples
4131 List all available plugins within amp (LADSPA example plugin) library:
4137 List all available controls and their valid ranges for @code{vcf_notch}
4138 plugin from @code{VCF} library:
4140 ladspa=f=vcf:p=vcf_notch:c=help
4144 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4147 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4151 Add reverberation to the audio using TAP-plugins
4152 (Tom's Audio Processing plugins):
4154 ladspa=file=tap_reverb:tap_reverb
4158 Generate white noise, with 0.2 amplitude:
4160 ladspa=file=cmt:noise_source_white:c=c0=.2
4164 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4165 @code{C* Audio Plugin Suite} (CAPS) library:
4167 ladspa=file=caps:Click:c=c1=20'
4171 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4173 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4177 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4178 @code{SWH Plugins} collection:
4180 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4184 Attenuate low frequencies using Multiband EQ from Steve Harris
4185 @code{SWH Plugins} collection:
4187 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4191 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4194 ladspa=caps:Narrower
4198 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4200 ladspa=caps:White:.2
4204 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4206 ladspa=caps:Fractal:c=c1=1
4210 Dynamic volume normalization using @code{VLevel} plugin:
4212 ladspa=vlevel-ladspa:vlevel_mono
4216 @subsection Commands
4218 This filter supports the following commands:
4221 Modify the @var{N}-th control value.
4223 If the specified value is not valid, it is ignored and prior one is kept.
4228 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4229 Support for both single pass (livestreams, files) and double pass (files) modes.
4230 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4231 detect true peaks, the audio stream will be upsampled to 192 kHz.
4232 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4234 The filter accepts the following options:
4238 Set integrated loudness target.
4239 Range is -70.0 - -5.0. Default value is -24.0.
4242 Set loudness range target.
4243 Range is 1.0 - 20.0. Default value is 7.0.
4246 Set maximum true peak.
4247 Range is -9.0 - +0.0. Default value is -2.0.
4249 @item measured_I, measured_i
4250 Measured IL of input file.
4251 Range is -99.0 - +0.0.
4253 @item measured_LRA, measured_lra
4254 Measured LRA of input file.
4255 Range is 0.0 - 99.0.
4257 @item measured_TP, measured_tp
4258 Measured true peak of input file.
4259 Range is -99.0 - +99.0.
4261 @item measured_thresh
4262 Measured threshold of input file.
4263 Range is -99.0 - +0.0.
4266 Set offset gain. Gain is applied before the true-peak limiter.
4267 Range is -99.0 - +99.0. Default is +0.0.
4270 Normalize by linearly scaling the source audio.
4271 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4272 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4273 be lower than source LRA and the change in integrated loudness shouldn't
4274 result in a true peak which exceeds the target TP. If any of these
4275 conditions aren't met, normalization mode will revert to @var{dynamic}.
4276 Options are @code{true} or @code{false}. Default is @code{true}.
4279 Treat mono input files as "dual-mono". If a mono file is intended for playback
4280 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4281 If set to @code{true}, this option will compensate for this effect.
4282 Multi-channel input files are not affected by this option.
4283 Options are true or false. Default is false.
4286 Set print format for stats. Options are summary, json, or none.
4287 Default value is none.
4292 Apply a low-pass filter with 3dB point frequency.
4293 The filter can be either single-pole or double-pole (the default).
4294 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4296 The filter accepts the following options:
4300 Set frequency in Hz. Default is 500.
4303 Set number of poles. Default is 2.
4306 Set method to specify band-width of filter.
4321 Specify the band-width of a filter in width_type units.
4322 Applies only to double-pole filter.
4323 The default is 0.707q and gives a Butterworth response.
4326 How much to use filtered signal in output. Default is 1.
4327 Range is between 0 and 1.
4330 Specify which channels to filter, by default all available are filtered.
4333 Normalize biquad coefficients, by default is disabled.
4334 Enabling it will normalize magnitude response at DC to 0dB.
4337 @subsection Examples
4340 Lowpass only LFE channel, it LFE is not present it does nothing:
4346 @subsection Commands
4348 This filter supports the following commands:
4351 Change lowpass frequency.
4352 Syntax for the command is : "@var{frequency}"
4355 Change lowpass width_type.
4356 Syntax for the command is : "@var{width_type}"
4359 Change lowpass width.
4360 Syntax for the command is : "@var{width}"
4364 Syntax for the command is : "@var{mix}"
4369 Load a LV2 (LADSPA Version 2) plugin.
4371 To enable compilation of this filter you need to configure FFmpeg with
4372 @code{--enable-lv2}.
4376 Specifies the plugin URI. You may need to escape ':'.
4379 Set the '|' separated list of controls which are zero or more floating point
4380 values that determine the behavior of the loaded plugin (for example delay,
4382 If @option{controls} is set to @code{help}, all available controls and
4383 their valid ranges are printed.
4385 @item sample_rate, s
4386 Specify the sample rate, default to 44100. Only used if plugin have
4390 Set the number of samples per channel per each output frame, default
4391 is 1024. Only used if plugin have zero inputs.
4394 Set the minimum duration of the sourced audio. See
4395 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4396 for the accepted syntax.
4397 Note that the resulting duration may be greater than the specified duration,
4398 as the generated audio is always cut at the end of a complete frame.
4399 If not specified, or the expressed duration is negative, the audio is
4400 supposed to be generated forever.
4401 Only used if plugin have zero inputs.
4404 @subsection Examples
4408 Apply bass enhancer plugin from Calf:
4410 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4414 Apply vinyl plugin from Calf:
4416 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4420 Apply bit crusher plugin from ArtyFX:
4422 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4427 Multiband Compress or expand the audio's dynamic range.
4429 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4430 This is akin to the crossover of a loudspeaker, and results in flat frequency
4431 response when absent compander action.
4433 It accepts the following parameters:
4437 This option syntax is:
4438 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4439 For explanation of each item refer to compand filter documentation.
4445 Mix channels with specific gain levels. The filter accepts the output
4446 channel layout followed by a set of channels definitions.
4448 This filter is also designed to efficiently remap the channels of an audio
4451 The filter accepts parameters of the form:
4452 "@var{l}|@var{outdef}|@var{outdef}|..."
4456 output channel layout or number of channels
4459 output channel specification, of the form:
4460 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4463 output channel to define, either a channel name (FL, FR, etc.) or a channel
4464 number (c0, c1, etc.)
4467 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4470 input channel to use, see out_name for details; it is not possible to mix
4471 named and numbered input channels
4474 If the `=' in a channel specification is replaced by `<', then the gains for
4475 that specification will be renormalized so that the total is 1, thus
4476 avoiding clipping noise.
4478 @subsection Mixing examples
4480 For example, if you want to down-mix from stereo to mono, but with a bigger
4481 factor for the left channel:
4483 pan=1c|c0=0.9*c0+0.1*c1
4486 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4487 7-channels surround:
4489 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4492 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4493 that should be preferred (see "-ac" option) unless you have very specific
4496 @subsection Remapping examples
4498 The channel remapping will be effective if, and only if:
4501 @item gain coefficients are zeroes or ones,
4502 @item only one input per channel output,
4505 If all these conditions are satisfied, the filter will notify the user ("Pure
4506 channel mapping detected"), and use an optimized and lossless method to do the
4509 For example, if you have a 5.1 source and want a stereo audio stream by
4510 dropping the extra channels:
4512 pan="stereo| c0=FL | c1=FR"
4515 Given the same source, you can also switch front left and front right channels
4516 and keep the input channel layout:
4518 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4521 If the input is a stereo audio stream, you can mute the front left channel (and
4522 still keep the stereo channel layout) with:
4527 Still with a stereo audio stream input, you can copy the right channel in both
4528 front left and right:
4530 pan="stereo| c0=FR | c1=FR"
4535 ReplayGain scanner filter. This filter takes an audio stream as an input and
4536 outputs it unchanged.
4537 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4541 Convert the audio sample format, sample rate and channel layout. It is
4542 not meant to be used directly.
4545 Apply time-stretching and pitch-shifting with librubberband.
4547 To enable compilation of this filter, you need to configure FFmpeg with
4548 @code{--enable-librubberband}.
4550 The filter accepts the following options:
4554 Set tempo scale factor.
4557 Set pitch scale factor.
4560 Set transients detector.
4561 Possible values are:
4570 Possible values are:
4579 Possible values are:
4586 Set processing window size.
4587 Possible values are:
4596 Possible values are:
4603 Enable formant preservation when shift pitching.
4604 Possible values are:
4612 Possible values are:
4621 Possible values are:
4628 @subsection Commands
4630 This filter supports the following commands:
4633 Change filter tempo scale factor.
4634 Syntax for the command is : "@var{tempo}"
4637 Change filter pitch scale factor.
4638 Syntax for the command is : "@var{pitch}"
4641 @section sidechaincompress
4643 This filter acts like normal compressor but has the ability to compress
4644 detected signal using second input signal.
4645 It needs two input streams and returns one output stream.
4646 First input stream will be processed depending on second stream signal.
4647 The filtered signal then can be filtered with other filters in later stages of
4648 processing. See @ref{pan} and @ref{amerge} filter.
4650 The filter accepts the following options:
4654 Set input gain. Default is 1. Range is between 0.015625 and 64.
4657 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4658 Default is @code{downward}.
4661 If a signal of second stream raises above this level it will affect the gain
4662 reduction of first stream.
4663 By default is 0.125. Range is between 0.00097563 and 1.
4666 Set a ratio about which the signal is reduced. 1:2 means that if the level
4667 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4668 Default is 2. Range is between 1 and 20.
4671 Amount of milliseconds the signal has to rise above the threshold before gain
4672 reduction starts. Default is 20. Range is between 0.01 and 2000.
4675 Amount of milliseconds the signal has to fall below the threshold before
4676 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4679 Set the amount by how much signal will be amplified after processing.
4680 Default is 1. Range is from 1 to 64.
4683 Curve the sharp knee around the threshold to enter gain reduction more softly.
4684 Default is 2.82843. Range is between 1 and 8.
4687 Choose if the @code{average} level between all channels of side-chain stream
4688 or the louder(@code{maximum}) channel of side-chain stream affects the
4689 reduction. Default is @code{average}.
4692 Should the exact signal be taken in case of @code{peak} or an RMS one in case
4693 of @code{rms}. Default is @code{rms} which is mainly smoother.
4696 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
4699 How much to use compressed signal in output. Default is 1.
4700 Range is between 0 and 1.
4703 @subsection Commands
4705 This filter supports the all above options as @ref{commands}.
4707 @subsection Examples
4711 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
4712 depending on the signal of 2nd input and later compressed signal to be
4713 merged with 2nd input:
4715 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
4719 @section sidechaingate
4721 A sidechain gate acts like a normal (wideband) gate but has the ability to
4722 filter the detected signal before sending it to the gain reduction stage.
4723 Normally a gate uses the full range signal to detect a level above the
4725 For example: If you cut all lower frequencies from your sidechain signal
4726 the gate will decrease the volume of your track only if not enough highs
4727 appear. With this technique you are able to reduce the resonation of a
4728 natural drum or remove "rumbling" of muted strokes from a heavily distorted
4730 It needs two input streams and returns one output stream.
4731 First input stream will be processed depending on second stream signal.
4733 The filter accepts the following options:
4737 Set input level before filtering.
4738 Default is 1. Allowed range is from 0.015625 to 64.
4741 Set the mode of operation. Can be @code{upward} or @code{downward}.
4742 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
4743 will be amplified, expanding dynamic range in upward direction.
4744 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
4747 Set the level of gain reduction when the signal is below the threshold.
4748 Default is 0.06125. Allowed range is from 0 to 1.
4749 Setting this to 0 disables reduction and then filter behaves like expander.
4752 If a signal rises above this level the gain reduction is released.
4753 Default is 0.125. Allowed range is from 0 to 1.
4756 Set a ratio about which the signal is reduced.
4757 Default is 2. Allowed range is from 1 to 9000.
4760 Amount of milliseconds the signal has to rise above the threshold before gain
4762 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
4765 Amount of milliseconds the signal has to fall below the threshold before the
4766 reduction is increased again. Default is 250 milliseconds.
4767 Allowed range is from 0.01 to 9000.
4770 Set amount of amplification of signal after processing.
4771 Default is 1. Allowed range is from 1 to 64.
4774 Curve the sharp knee around the threshold to enter gain reduction more softly.
4775 Default is 2.828427125. Allowed range is from 1 to 8.
4778 Choose if exact signal should be taken for detection or an RMS like one.
4779 Default is rms. Can be peak or rms.
4782 Choose if the average level between all channels or the louder channel affects
4784 Default is average. Can be average or maximum.
4787 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
4790 @section silencedetect
4792 Detect silence in an audio stream.
4794 This filter logs a message when it detects that the input audio volume is less
4795 or equal to a noise tolerance value for a duration greater or equal to the
4796 minimum detected noise duration.
4798 The printed times and duration are expressed in seconds. The
4799 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
4800 is set on the first frame whose timestamp equals or exceeds the detection
4801 duration and it contains the timestamp of the first frame of the silence.
4803 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
4804 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
4805 keys are set on the first frame after the silence. If @option{mono} is
4806 enabled, and each channel is evaluated separately, the @code{.X}
4807 suffixed keys are used, and @code{X} corresponds to the channel number.
4809 The filter accepts the following options:
4813 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
4814 specified value) or amplitude ratio. Default is -60dB, or 0.001.
4817 Set silence duration until notification (default is 2 seconds). See
4818 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4819 for the accepted syntax.
4822 Process each channel separately, instead of combined. By default is disabled.
4825 @subsection Examples
4829 Detect 5 seconds of silence with -50dB noise tolerance:
4831 silencedetect=n=-50dB:d=5
4835 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
4836 tolerance in @file{silence.mp3}:
4838 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
4842 @section silenceremove
4844 Remove silence from the beginning, middle or end of the audio.
4846 The filter accepts the following options:
4850 This value is used to indicate if audio should be trimmed at beginning of
4851 the audio. A value of zero indicates no silence should be trimmed from the
4852 beginning. When specifying a non-zero value, it trims audio up until it
4853 finds non-silence. Normally, when trimming silence from beginning of audio
4854 the @var{start_periods} will be @code{1} but it can be increased to higher
4855 values to trim all audio up to specific count of non-silence periods.
4856 Default value is @code{0}.
4858 @item start_duration
4859 Specify the amount of time that non-silence must be detected before it stops
4860 trimming audio. By increasing the duration, bursts of noises can be treated
4861 as silence and trimmed off. Default value is @code{0}.
4863 @item start_threshold
4864 This indicates what sample value should be treated as silence. For digital
4865 audio, a value of @code{0} may be fine but for audio recorded from analog,
4866 you may wish to increase the value to account for background noise.
4867 Can be specified in dB (in case "dB" is appended to the specified value)
4868 or amplitude ratio. Default value is @code{0}.
4871 Specify max duration of silence at beginning that will be kept after
4872 trimming. Default is 0, which is equal to trimming all samples detected
4876 Specify mode of detection of silence end in start of multi-channel audio.
4877 Can be @var{any} or @var{all}. Default is @var{any}.
4878 With @var{any}, any sample that is detected as non-silence will cause
4879 stopped trimming of silence.
4880 With @var{all}, only if all channels are detected as non-silence will cause
4881 stopped trimming of silence.
4884 Set the count for trimming silence from the end of audio.
4885 To remove silence from the middle of a file, specify a @var{stop_periods}
4886 that is negative. This value is then treated as a positive value and is
4887 used to indicate the effect should restart processing as specified by
4888 @var{start_periods}, making it suitable for removing periods of silence
4889 in the middle of the audio.
4890 Default value is @code{0}.
4893 Specify a duration of silence that must exist before audio is not copied any
4894 more. By specifying a higher duration, silence that is wanted can be left in
4896 Default value is @code{0}.
4898 @item stop_threshold
4899 This is the same as @option{start_threshold} but for trimming silence from
4901 Can be specified in dB (in case "dB" is appended to the specified value)
4902 or amplitude ratio. Default value is @code{0}.
4905 Specify max duration of silence at end that will be kept after
4906 trimming. Default is 0, which is equal to trimming all samples detected
4910 Specify mode of detection of silence start in end of multi-channel audio.
4911 Can be @var{any} or @var{all}. Default is @var{any}.
4912 With @var{any}, any sample that is detected as non-silence will cause
4913 stopped trimming of silence.
4914 With @var{all}, only if all channels are detected as non-silence will cause
4915 stopped trimming of silence.
4918 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
4919 and works better with digital silence which is exactly 0.
4920 Default value is @code{rms}.
4923 Set duration in number of seconds used to calculate size of window in number
4924 of samples for detecting silence.
4925 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
4928 @subsection Examples
4932 The following example shows how this filter can be used to start a recording
4933 that does not contain the delay at the start which usually occurs between
4934 pressing the record button and the start of the performance:
4936 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
4940 Trim all silence encountered from beginning to end where there is more than 1
4941 second of silence in audio:
4943 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
4947 Trim all digital silence samples, using peak detection, from beginning to end
4948 where there is more than 0 samples of digital silence in audio and digital
4949 silence is detected in all channels at same positions in stream:
4951 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
4957 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
4958 loudspeakers around the user for binaural listening via headphones (audio
4959 formats up to 9 channels supported).
4960 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
4961 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
4962 Austrian Academy of Sciences.
4964 To enable compilation of this filter you need to configure FFmpeg with
4965 @code{--enable-libmysofa}.
4967 The filter accepts the following options:
4971 Set the SOFA file used for rendering.
4974 Set gain applied to audio. Value is in dB. Default is 0.
4977 Set rotation of virtual loudspeakers in deg. Default is 0.
4980 Set elevation of virtual speakers in deg. Default is 0.
4983 Set distance in meters between loudspeakers and the listener with near-field
4984 HRTFs. Default is 1.
4987 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4988 processing audio in time domain which is slow.
4989 @var{freq} is processing audio in frequency domain which is fast.
4990 Default is @var{freq}.
4993 Set custom positions of virtual loudspeakers. Syntax for this option is:
4994 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
4995 Each virtual loudspeaker is described with short channel name following with
4996 azimuth and elevation in degrees.
4997 Each virtual loudspeaker description is separated by '|'.
4998 For example to override front left and front right channel positions use:
4999 'speakers=FL 45 15|FR 345 15'.
5000 Descriptions with unrecognised channel names are ignored.
5003 Set custom gain for LFE channels. Value is in dB. Default is 0.
5006 Set custom frame size in number of samples. Default is 1024.
5007 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5008 is set to @var{freq}.
5011 Should all IRs be normalized upon importing SOFA file.
5012 By default is enabled.
5015 Should nearest IRs be interpolated with neighbor IRs if exact position
5016 does not match. By default is disabled.
5019 Minphase all IRs upon loading of SOFA file. By default is disabled.
5022 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5025 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5028 @subsection Examples
5032 Using ClubFritz6 sofa file:
5034 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5038 Using ClubFritz12 sofa file and bigger radius with small rotation:
5040 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5044 Similar as above but with custom speaker positions for front left, front right, back left and back right
5045 and also with custom gain:
5047 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5051 @section stereotools
5053 This filter has some handy utilities to manage stereo signals, for converting
5054 M/S stereo recordings to L/R signal while having control over the parameters
5055 or spreading the stereo image of master track.
5057 The filter accepts the following options:
5061 Set input level before filtering for both channels. Defaults is 1.
5062 Allowed range is from 0.015625 to 64.
5065 Set output level after filtering for both channels. Defaults is 1.
5066 Allowed range is from 0.015625 to 64.
5069 Set input balance between both channels. Default is 0.
5070 Allowed range is from -1 to 1.
5073 Set output balance between both channels. Default is 0.
5074 Allowed range is from -1 to 1.
5077 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5078 clipping. Disabled by default.
5081 Mute the left channel. Disabled by default.
5084 Mute the right channel. Disabled by default.
5087 Change the phase of the left channel. Disabled by default.
5090 Change the phase of the right channel. Disabled by default.
5093 Set stereo mode. Available values are:
5097 Left/Right to Left/Right, this is default.
5100 Left/Right to Mid/Side.
5103 Mid/Side to Left/Right.
5106 Left/Right to Left/Left.
5109 Left/Right to Right/Right.
5112 Left/Right to Left + Right.
5115 Left/Right to Right/Left.
5118 Mid/Side to Left/Left.
5121 Mid/Side to Right/Right.
5125 Set level of side signal. Default is 1.
5126 Allowed range is from 0.015625 to 64.
5129 Set balance of side signal. Default is 0.
5130 Allowed range is from -1 to 1.
5133 Set level of the middle signal. Default is 1.
5134 Allowed range is from 0.015625 to 64.
5137 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5140 Set stereo base between mono and inversed channels. Default is 0.
5141 Allowed range is from -1 to 1.
5144 Set delay in milliseconds how much to delay left from right channel and
5145 vice versa. Default is 0. Allowed range is from -20 to 20.
5148 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5151 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5153 @item bmode_in, bmode_out
5154 Set balance mode for balance_in/balance_out option.
5156 Can be one of the following:
5160 Classic balance mode. Attenuate one channel at time.
5161 Gain is raised up to 1.
5164 Similar as classic mode above but gain is raised up to 2.
5167 Equal power distribution, from -6dB to +6dB range.
5171 @subsection Examples
5175 Apply karaoke like effect:
5177 stereotools=mlev=0.015625
5181 Convert M/S signal to L/R:
5183 "stereotools=mode=ms>lr"
5187 @section stereowiden
5189 This filter enhance the stereo effect by suppressing signal common to both
5190 channels and by delaying the signal of left into right and vice versa,
5191 thereby widening the stereo effect.
5193 The filter accepts the following options:
5197 Time in milliseconds of the delay of left signal into right and vice versa.
5198 Default is 20 milliseconds.
5201 Amount of gain in delayed signal into right and vice versa. Gives a delay
5202 effect of left signal in right output and vice versa which gives widening
5203 effect. Default is 0.3.
5206 Cross feed of left into right with inverted phase. This helps in suppressing
5207 the mono. If the value is 1 it will cancel all the signal common to both
5208 channels. Default is 0.3.
5211 Set level of input signal of original channel. Default is 0.8.
5214 @subsection Commands
5216 This filter supports the all above options except @code{delay} as @ref{commands}.
5218 @section superequalizer
5219 Apply 18 band equalizer.
5221 The filter accepts the following options:
5228 Set 131Hz band gain.
5230 Set 185Hz band gain.
5232 Set 262Hz band gain.
5234 Set 370Hz band gain.
5236 Set 523Hz band gain.
5238 Set 740Hz band gain.
5240 Set 1047Hz band gain.
5242 Set 1480Hz band gain.
5244 Set 2093Hz band gain.
5246 Set 2960Hz band gain.
5248 Set 4186Hz band gain.
5250 Set 5920Hz band gain.
5252 Set 8372Hz band gain.
5254 Set 11840Hz band gain.
5256 Set 16744Hz band gain.
5258 Set 20000Hz band gain.
5262 Apply audio surround upmix filter.
5264 This filter allows to produce multichannel output from audio stream.
5266 The filter accepts the following options:
5270 Set output channel layout. By default, this is @var{5.1}.
5272 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5273 for the required syntax.
5276 Set input channel layout. By default, this is @var{stereo}.
5278 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5279 for the required syntax.
5282 Set input volume level. By default, this is @var{1}.
5285 Set output volume level. By default, this is @var{1}.
5288 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5291 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5294 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5297 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5298 In @var{add} mode, LFE channel is created from input audio and added to output.
5299 In @var{sub} mode, LFE channel is created from input audio and added to output but
5300 also all non-LFE output channels are subtracted with output LFE channel.
5303 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5304 Default is @var{90}.
5307 Set front center input volume. By default, this is @var{1}.
5310 Set front center output volume. By default, this is @var{1}.
5313 Set front left input volume. By default, this is @var{1}.
5316 Set front left output volume. By default, this is @var{1}.
5319 Set front right input volume. By default, this is @var{1}.
5322 Set front right output volume. By default, this is @var{1}.
5325 Set side left input volume. By default, this is @var{1}.
5328 Set side left output volume. By default, this is @var{1}.
5331 Set side right input volume. By default, this is @var{1}.
5334 Set side right output volume. By default, this is @var{1}.
5337 Set back left input volume. By default, this is @var{1}.
5340 Set back left output volume. By default, this is @var{1}.
5343 Set back right input volume. By default, this is @var{1}.
5346 Set back right output volume. By default, this is @var{1}.
5349 Set back center input volume. By default, this is @var{1}.
5352 Set back center output volume. By default, this is @var{1}.
5355 Set LFE input volume. By default, this is @var{1}.
5358 Set LFE output volume. By default, this is @var{1}.
5361 Set spread usage of stereo image across X axis for all channels.
5364 Set spread usage of stereo image across Y axis for all channels.
5366 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5367 Set spread usage of stereo image across X axis for each channel.
5369 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5370 Set spread usage of stereo image across Y axis for each channel.
5373 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5376 Set window function.
5378 It accepts the following values:
5401 Default is @code{hann}.
5404 Set window overlap. If set to 1, the recommended overlap for selected
5405 window function will be picked. Default is @code{0.5}.
5408 @section treble, highshelf
5410 Boost or cut treble (upper) frequencies of the audio using a two-pole
5411 shelving filter with a response similar to that of a standard
5412 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5414 The filter accepts the following options:
5418 Give the gain at whichever is the lower of ~22 kHz and the
5419 Nyquist frequency. Its useful range is about -20 (for a large cut)
5420 to +20 (for a large boost). Beware of clipping when using a positive gain.
5423 Set the filter's central frequency and so can be used
5424 to extend or reduce the frequency range to be boosted or cut.
5425 The default value is @code{3000} Hz.
5428 Set method to specify band-width of filter.
5443 Determine how steep is the filter's shelf transition.
5446 How much to use filtered signal in output. Default is 1.
5447 Range is between 0 and 1.
5450 Specify which channels to filter, by default all available are filtered.
5453 Normalize biquad coefficients, by default is disabled.
5454 Enabling it will normalize magnitude response at DC to 0dB.
5457 @subsection Commands
5459 This filter supports the following commands:
5462 Change treble frequency.
5463 Syntax for the command is : "@var{frequency}"
5466 Change treble width_type.
5467 Syntax for the command is : "@var{width_type}"
5470 Change treble width.
5471 Syntax for the command is : "@var{width}"
5475 Syntax for the command is : "@var{gain}"
5479 Syntax for the command is : "@var{mix}"
5484 Sinusoidal amplitude modulation.
5486 The filter accepts the following options:
5490 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5491 (20 Hz or lower) will result in a tremolo effect.
5492 This filter may also be used as a ring modulator by specifying
5493 a modulation frequency higher than 20 Hz.
5494 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5497 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5498 Default value is 0.5.
5503 Sinusoidal phase modulation.
5505 The filter accepts the following options:
5509 Modulation frequency in Hertz.
5510 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5513 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5514 Default value is 0.5.
5519 Adjust the input audio volume.
5521 It accepts the following parameters:
5525 Set audio volume expression.
5527 Output values are clipped to the maximum value.
5529 The output audio volume is given by the relation:
5531 @var{output_volume} = @var{volume} * @var{input_volume}
5534 The default value for @var{volume} is "1.0".
5537 This parameter represents the mathematical precision.
5539 It determines which input sample formats will be allowed, which affects the
5540 precision of the volume scaling.
5544 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5546 32-bit floating-point; this limits input sample format to FLT. (default)
5548 64-bit floating-point; this limits input sample format to DBL.
5552 Choose the behaviour on encountering ReplayGain side data in input frames.
5556 Remove ReplayGain side data, ignoring its contents (the default).
5559 Ignore ReplayGain side data, but leave it in the frame.
5562 Prefer the track gain, if present.
5565 Prefer the album gain, if present.
5568 @item replaygain_preamp
5569 Pre-amplification gain in dB to apply to the selected replaygain gain.
5571 Default value for @var{replaygain_preamp} is 0.0.
5573 @item replaygain_noclip
5574 Prevent clipping by limiting the gain applied.
5576 Default value for @var{replaygain_noclip} is 1.
5579 Set when the volume expression is evaluated.
5581 It accepts the following values:
5584 only evaluate expression once during the filter initialization, or
5585 when the @samp{volume} command is sent
5588 evaluate expression for each incoming frame
5591 Default value is @samp{once}.
5594 The volume expression can contain the following parameters.
5598 frame number (starting at zero)
5601 @item nb_consumed_samples
5602 number of samples consumed by the filter
5604 number of samples in the current frame
5606 original frame position in the file
5612 PTS at start of stream
5614 time at start of stream
5620 last set volume value
5623 Note that when @option{eval} is set to @samp{once} only the
5624 @var{sample_rate} and @var{tb} variables are available, all other
5625 variables will evaluate to NAN.
5627 @subsection Commands
5629 This filter supports the following commands:
5632 Modify the volume expression.
5633 The command accepts the same syntax of the corresponding option.
5635 If the specified expression is not valid, it is kept at its current
5639 @subsection Examples
5643 Halve the input audio volume:
5647 volume=volume=-6.0206dB
5650 In all the above example the named key for @option{volume} can be
5651 omitted, for example like in:
5657 Increase input audio power by 6 decibels using fixed-point precision:
5659 volume=volume=6dB:precision=fixed
5663 Fade volume after time 10 with an annihilation period of 5 seconds:
5665 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
5669 @section volumedetect
5671 Detect the volume of the input video.
5673 The filter has no parameters. The input is not modified. Statistics about
5674 the volume will be printed in the log when the input stream end is reached.
5676 In particular it will show the mean volume (root mean square), maximum
5677 volume (on a per-sample basis), and the beginning of a histogram of the
5678 registered volume values (from the maximum value to a cumulated 1/1000 of
5681 All volumes are in decibels relative to the maximum PCM value.
5683 @subsection Examples
5685 Here is an excerpt of the output:
5687 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
5688 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
5689 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
5690 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
5691 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
5692 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
5693 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
5694 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
5695 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
5701 The mean square energy is approximately -27 dB, or 10^-2.7.
5703 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
5705 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
5708 In other words, raising the volume by +4 dB does not cause any clipping,
5709 raising it by +5 dB causes clipping for 6 samples, etc.
5711 @c man end AUDIO FILTERS
5713 @chapter Audio Sources
5714 @c man begin AUDIO SOURCES
5716 Below is a description of the currently available audio sources.
5720 Buffer audio frames, and make them available to the filter chain.
5722 This source is mainly intended for a programmatic use, in particular
5723 through the interface defined in @file{libavfilter/asrc_abuffer.h}.
5725 It accepts the following parameters:
5729 The timebase which will be used for timestamps of submitted frames. It must be
5730 either a floating-point number or in @var{numerator}/@var{denominator} form.
5733 The sample rate of the incoming audio buffers.
5736 The sample format of the incoming audio buffers.
5737 Either a sample format name or its corresponding integer representation from
5738 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
5740 @item channel_layout
5741 The channel layout of the incoming audio buffers.
5742 Either a channel layout name from channel_layout_map in
5743 @file{libavutil/channel_layout.c} or its corresponding integer representation
5744 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
5747 The number of channels of the incoming audio buffers.
5748 If both @var{channels} and @var{channel_layout} are specified, then they
5753 @subsection Examples
5756 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
5759 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
5760 Since the sample format with name "s16p" corresponds to the number
5761 6 and the "stereo" channel layout corresponds to the value 0x3, this is
5764 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
5769 Generate an audio signal specified by an expression.
5771 This source accepts in input one or more expressions (one for each
5772 channel), which are evaluated and used to generate a corresponding
5775 This source accepts the following options:
5779 Set the '|'-separated expressions list for each separate channel. In case the
5780 @option{channel_layout} option is not specified, the selected channel layout
5781 depends on the number of provided expressions. Otherwise the last
5782 specified expression is applied to the remaining output channels.
5784 @item channel_layout, c
5785 Set the channel layout. The number of channels in the specified layout
5786 must be equal to the number of specified expressions.
5789 Set the minimum duration of the sourced audio. See
5790 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5791 for the accepted syntax.
5792 Note that the resulting duration may be greater than the specified
5793 duration, as the generated audio is always cut at the end of a
5796 If not specified, or the expressed duration is negative, the audio is
5797 supposed to be generated forever.
5800 Set the number of samples per channel per each output frame,
5803 @item sample_rate, s
5804 Specify the sample rate, default to 44100.
5807 Each expression in @var{exprs} can contain the following constants:
5811 number of the evaluated sample, starting from 0
5814 time of the evaluated sample expressed in seconds, starting from 0
5821 @subsection Examples
5831 Generate a sin signal with frequency of 440 Hz, set sample rate to
5834 aevalsrc="sin(440*2*PI*t):s=8000"
5838 Generate a two channels signal, specify the channel layout (Front
5839 Center + Back Center) explicitly:
5841 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
5845 Generate white noise:
5847 aevalsrc="-2+random(0)"
5851 Generate an amplitude modulated signal:
5853 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
5857 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
5859 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
5866 Generate a FIR coefficients using frequency sampling method.
5868 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
5870 The filter accepts the following options:
5874 Set number of filter coefficents in output audio stream.
5875 Default value is 1025.
5878 Set frequency points from where magnitude and phase are set.
5879 This must be in non decreasing order, and first element must be 0, while last element
5880 must be 1. Elements are separated by white spaces.
5883 Set magnitude value for every frequency point set by @option{frequency}.
5884 Number of values must be same as number of frequency points.
5885 Values are separated by white spaces.
5888 Set phase value for every frequency point set by @option{frequency}.
5889 Number of values must be same as number of frequency points.
5890 Values are separated by white spaces.
5892 @item sample_rate, r
5893 Set sample rate, default is 44100.
5896 Set number of samples per each frame. Default is 1024.
5899 Set window function. Default is blackman.
5904 The null audio source, return unprocessed audio frames. It is mainly useful
5905 as a template and to be employed in analysis / debugging tools, or as
5906 the source for filters which ignore the input data (for example the sox
5909 This source accepts the following options:
5913 @item channel_layout, cl
5915 Specifies the channel layout, and can be either an integer or a string
5916 representing a channel layout. The default value of @var{channel_layout}
5919 Check the channel_layout_map definition in
5920 @file{libavutil/channel_layout.c} for the mapping between strings and
5921 channel layout values.
5923 @item sample_rate, r
5924 Specifies the sample rate, and defaults to 44100.
5927 Set the number of samples per requested frames.
5931 @subsection Examples
5935 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
5937 anullsrc=r=48000:cl=4
5941 Do the same operation with a more obvious syntax:
5943 anullsrc=r=48000:cl=mono
5947 All the parameters need to be explicitly defined.
5951 Synthesize a voice utterance using the libflite library.
5953 To enable compilation of this filter you need to configure FFmpeg with
5954 @code{--enable-libflite}.
5956 Note that versions of the flite library prior to 2.0 are not thread-safe.
5958 The filter accepts the following options:
5963 If set to 1, list the names of the available voices and exit
5964 immediately. Default value is 0.
5967 Set the maximum number of samples per frame. Default value is 512.
5970 Set the filename containing the text to speak.
5973 Set the text to speak.
5976 Set the voice to use for the speech synthesis. Default value is
5977 @code{kal}. See also the @var{list_voices} option.
5980 @subsection Examples
5984 Read from file @file{speech.txt}, and synthesize the text using the
5985 standard flite voice:
5987 flite=textfile=speech.txt
5991 Read the specified text selecting the @code{slt} voice:
5993 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
5997 Input text to ffmpeg:
5999 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6003 Make @file{ffplay} speak the specified text, using @code{flite} and
6004 the @code{lavfi} device:
6006 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6010 For more information about libflite, check:
6011 @url{http://www.festvox.org/flite/}
6015 Generate a noise audio signal.
6017 The filter accepts the following options:
6020 @item sample_rate, r
6021 Specify the sample rate. Default value is 48000 Hz.
6024 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6028 Specify the duration of the generated audio stream. Not specifying this option
6029 results in noise with an infinite length.
6031 @item color, colour, c
6032 Specify the color of noise. Available noise colors are white, pink, brown,
6033 blue, violet and velvet. Default color is white.
6036 Specify a value used to seed the PRNG.
6039 Set the number of samples per each output frame, default is 1024.
6042 @subsection Examples
6047 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6049 anoisesrc=d=60:c=pink:r=44100:a=0.5
6055 Generate odd-tap Hilbert transform FIR coefficients.
6057 The resulting stream can be used with @ref{afir} filter for phase-shifting
6058 the signal by 90 degrees.
6060 This is used in many matrix coding schemes and for analytic signal generation.
6061 The process is often written as a multiplication by i (or j), the imaginary unit.
6063 The filter accepts the following options:
6067 @item sample_rate, s
6068 Set sample rate, default is 44100.
6071 Set length of FIR filter, default is 22051.
6074 Set number of samples per each frame.
6077 Set window function to be used when generating FIR coefficients.
6082 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6084 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6086 The filter accepts the following options:
6089 @item sample_rate, r
6090 Set sample rate, default is 44100.
6093 Set number of samples per each frame. Default is 1024.
6096 Set high-pass frequency. Default is 0.
6099 Set low-pass frequency. Default is 0.
6100 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6101 is higher than 0 then filter will create band-pass filter coefficients,
6102 otherwise band-reject filter coefficients.
6105 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6108 Set Kaiser window beta.
6111 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6114 Enable rounding, by default is disabled.
6117 Set number of taps for high-pass filter.
6120 Set number of taps for low-pass filter.
6125 Generate an audio signal made of a sine wave with amplitude 1/8.
6127 The audio signal is bit-exact.
6129 The filter accepts the following options:
6134 Set the carrier frequency. Default is 440 Hz.
6136 @item beep_factor, b
6137 Enable a periodic beep every second with frequency @var{beep_factor} times
6138 the carrier frequency. Default is 0, meaning the beep is disabled.
6140 @item sample_rate, r
6141 Specify the sample rate, default is 44100.
6144 Specify the duration of the generated audio stream.
6146 @item samples_per_frame
6147 Set the number of samples per output frame.
6149 The expression can contain the following constants:
6153 The (sequential) number of the output audio frame, starting from 0.
6156 The PTS (Presentation TimeStamp) of the output audio frame,
6157 expressed in @var{TB} units.
6160 The PTS of the output audio frame, expressed in seconds.
6163 The timebase of the output audio frames.
6166 Default is @code{1024}.
6169 @subsection Examples
6174 Generate a simple 440 Hz sine wave:
6180 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6184 sine=frequency=220:beep_factor=4:duration=5
6188 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6191 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6195 @c man end AUDIO SOURCES
6197 @chapter Audio Sinks
6198 @c man begin AUDIO SINKS
6200 Below is a description of the currently available audio sinks.
6202 @section abuffersink
6204 Buffer audio frames, and make them available to the end of filter chain.
6206 This sink is mainly intended for programmatic use, in particular
6207 through the interface defined in @file{libavfilter/buffersink.h}
6208 or the options system.
6210 It accepts a pointer to an AVABufferSinkContext structure, which
6211 defines the incoming buffers' formats, to be passed as the opaque
6212 parameter to @code{avfilter_init_filter} for initialization.
6215 Null audio sink; do absolutely nothing with the input audio. It is
6216 mainly useful as a template and for use in analysis / debugging
6219 @c man end AUDIO SINKS
6221 @chapter Video Filters
6222 @c man begin VIDEO FILTERS
6224 When you configure your FFmpeg build, you can disable any of the
6225 existing filters using @code{--disable-filters}.
6226 The configure output will show the video filters included in your
6229 Below is a description of the currently available video filters.
6233 Mark a region of interest in a video frame.
6235 The frame data is passed through unchanged, but metadata is attached
6236 to the frame indicating regions of interest which can affect the
6237 behaviour of later encoding. Multiple regions can be marked by
6238 applying the filter multiple times.
6242 Region distance in pixels from the left edge of the frame.
6244 Region distance in pixels from the top edge of the frame.
6246 Region width in pixels.
6248 Region height in pixels.
6250 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6251 and may contain the following variables:
6254 Width of the input frame.
6256 Height of the input frame.
6260 Quantisation offset to apply within the region.
6262 This must be a real value in the range -1 to +1. A value of zero
6263 indicates no quality change. A negative value asks for better quality
6264 (less quantisation), while a positive value asks for worse quality
6265 (greater quantisation).
6267 The range is calibrated so that the extreme values indicate the
6268 largest possible offset - if the rest of the frame is encoded with the
6269 worst possible quality, an offset of -1 indicates that this region
6270 should be encoded with the best possible quality anyway. Intermediate
6271 values are then interpolated in some codec-dependent way.
6273 For example, in 10-bit H.264 the quantisation parameter varies between
6274 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6275 this region should be encoded with a QP around one-tenth of the full
6276 range better than the rest of the frame. So, if most of the frame
6277 were to be encoded with a QP of around 30, this region would get a QP
6278 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6279 An extreme value of -1 would indicate that this region should be
6280 encoded with the best possible quality regardless of the treatment of
6281 the rest of the frame - that is, should be encoded at a QP of -12.
6283 If set to true, remove any existing regions of interest marked on the
6284 frame before adding the new one.
6287 @subsection Examples
6291 Mark the centre quarter of the frame as interesting.
6293 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6296 Mark the 100-pixel-wide region on the left edge of the frame as very
6297 uninteresting (to be encoded at much lower quality than the rest of
6300 addroi=0:0:100:ih:+1/5
6304 @section alphaextract
6306 Extract the alpha component from the input as a grayscale video. This
6307 is especially useful with the @var{alphamerge} filter.
6311 Add or replace the alpha component of the primary input with the
6312 grayscale value of a second input. This is intended for use with
6313 @var{alphaextract} to allow the transmission or storage of frame
6314 sequences that have alpha in a format that doesn't support an alpha
6317 For example, to reconstruct full frames from a normal YUV-encoded video
6318 and a separate video created with @var{alphaextract}, you might use:
6320 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6323 Since this filter is designed for reconstruction, it operates on frame
6324 sequences without considering timestamps, and terminates when either
6325 input reaches end of stream. This will cause problems if your encoding
6326 pipeline drops frames. If you're trying to apply an image as an
6327 overlay to a video stream, consider the @var{overlay} filter instead.
6331 Amplify differences between current pixel and pixels of adjacent frames in
6332 same pixel location.
6334 This filter accepts the following options:
6338 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6339 For example radius of 3 will instruct filter to calculate average of 7 frames.
6342 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6345 Set threshold for difference amplification. Any difference greater or equal to
6346 this value will not alter source pixel. Default is 10.
6347 Allowed range is from 0 to 65535.
6350 Set tolerance for difference amplification. Any difference lower to
6351 this value will not alter source pixel. Default is 0.
6352 Allowed range is from 0 to 65535.
6355 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6356 This option controls maximum possible value that will decrease source pixel value.
6359 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6360 This option controls maximum possible value that will increase source pixel value.
6363 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6366 @subsection Commands
6368 This filter supports the following @ref{commands} that corresponds to option of same name:
6380 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6381 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6382 Substation Alpha) subtitles files.
6384 This filter accepts the following option in addition to the common options from
6385 the @ref{subtitles} filter:
6389 Set the shaping engine
6391 Available values are:
6394 The default libass shaping engine, which is the best available.
6396 Fast, font-agnostic shaper that can do only substitutions
6398 Slower shaper using OpenType for substitutions and positioning
6401 The default is @code{auto}.
6405 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6407 The filter accepts the following options:
6411 Set threshold A for 1st plane. Default is 0.02.
6412 Valid range is 0 to 0.3.
6415 Set threshold B for 1st plane. Default is 0.04.
6416 Valid range is 0 to 5.
6419 Set threshold A for 2nd plane. Default is 0.02.
6420 Valid range is 0 to 0.3.
6423 Set threshold B for 2nd plane. Default is 0.04.
6424 Valid range is 0 to 5.
6427 Set threshold A for 3rd plane. Default is 0.02.
6428 Valid range is 0 to 0.3.
6431 Set threshold B for 3rd plane. Default is 0.04.
6432 Valid range is 0 to 5.
6434 Threshold A is designed to react on abrupt changes in the input signal and
6435 threshold B is designed to react on continuous changes in the input signal.
6438 Set number of frames filter will use for averaging. Default is 9. Must be odd
6439 number in range [5, 129].
6442 Set what planes of frame filter will use for averaging. Default is all.
6445 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6446 Alternatively can be set to @code{s} serial.
6448 Parallel can be faster then serial, while other way around is never true.
6449 Parallel will abort early on first change being greater then thresholds, while serial
6450 will continue processing other side of frames if they are equal or bellow thresholds.
6453 @subsection Commands
6454 This filter supports same @ref{commands} as options except option @code{s}.
6455 The command accepts the same syntax of the corresponding option.
6459 Apply average blur filter.
6461 The filter accepts the following options:
6465 Set horizontal radius size.
6468 Set which planes to filter. By default all planes are filtered.
6471 Set vertical radius size, if zero it will be same as @code{sizeX}.
6472 Default is @code{0}.
6475 @subsection Commands
6476 This filter supports same commands as options.
6477 The command accepts the same syntax of the corresponding option.
6479 If the specified expression is not valid, it is kept at its current
6484 Compute the bounding box for the non-black pixels in the input frame
6487 This filter computes the bounding box containing all the pixels with a
6488 luminance value greater than the minimum allowed value.
6489 The parameters describing the bounding box are printed on the filter
6492 The filter accepts the following option:
6496 Set the minimal luminance value. Default is @code{16}.
6500 Apply bilateral filter, spatial smoothing while preserving edges.
6502 The filter accepts the following options:
6505 Set sigma of gaussian function to calculate spatial weight.
6506 Allowed range is 0 to 10. Default is 0.1.
6509 Set sigma of gaussian function to calculate range weight.
6510 Allowed range is 0 to 1. Default is 0.1.
6513 Set planes to filter. Default is first only.
6516 @section bitplanenoise
6518 Show and measure bit plane noise.
6520 The filter accepts the following options:
6524 Set which plane to analyze. Default is @code{1}.
6527 Filter out noisy pixels from @code{bitplane} set above.
6528 Default is disabled.
6531 @section blackdetect
6533 Detect video intervals that are (almost) completely black. Can be
6534 useful to detect chapter transitions, commercials, or invalid
6535 recordings. Output lines contains the time for the start, end and
6536 duration of the detected black interval expressed in seconds.
6538 In order to display the output lines, you need to set the loglevel at
6539 least to the AV_LOG_INFO value.
6541 The filter accepts the following options:
6544 @item black_min_duration, d
6545 Set the minimum detected black duration expressed in seconds. It must
6546 be a non-negative floating point number.
6548 Default value is 2.0.
6550 @item picture_black_ratio_th, pic_th
6551 Set the threshold for considering a picture "black".
6552 Express the minimum value for the ratio:
6554 @var{nb_black_pixels} / @var{nb_pixels}
6557 for which a picture is considered black.
6558 Default value is 0.98.
6560 @item pixel_black_th, pix_th
6561 Set the threshold for considering a pixel "black".
6563 The threshold expresses the maximum pixel luminance value for which a
6564 pixel is considered "black". The provided value is scaled according to
6565 the following equation:
6567 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6570 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6571 the input video format, the range is [0-255] for YUV full-range
6572 formats and [16-235] for YUV non full-range formats.
6574 Default value is 0.10.
6577 The following example sets the maximum pixel threshold to the minimum
6578 value, and detects only black intervals of 2 or more seconds:
6580 blackdetect=d=2:pix_th=0.00
6585 Detect frames that are (almost) completely black. Can be useful to
6586 detect chapter transitions or commercials. Output lines consist of
6587 the frame number of the detected frame, the percentage of blackness,
6588 the position in the file if known or -1 and the timestamp in seconds.
6590 In order to display the output lines, you need to set the loglevel at
6591 least to the AV_LOG_INFO value.
6593 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6594 The value represents the percentage of pixels in the picture that
6595 are below the threshold value.
6597 It accepts the following parameters:
6602 The percentage of the pixels that have to be below the threshold; it defaults to
6605 @item threshold, thresh
6606 The threshold below which a pixel value is considered black; it defaults to
6614 Blend two video frames into each other.
6616 The @code{blend} filter takes two input streams and outputs one
6617 stream, the first input is the "top" layer and second input is
6618 "bottom" layer. By default, the output terminates when the longest input terminates.
6620 The @code{tblend} (time blend) filter takes two consecutive frames
6621 from one single stream, and outputs the result obtained by blending
6622 the new frame on top of the old frame.
6624 A description of the accepted options follows.
6632 Set blend mode for specific pixel component or all pixel components in case
6633 of @var{all_mode}. Default value is @code{normal}.
6635 Available values for component modes are:
6677 Set blend opacity for specific pixel component or all pixel components in case
6678 of @var{all_opacity}. Only used in combination with pixel component blend modes.
6685 Set blend expression for specific pixel component or all pixel components in case
6686 of @var{all_expr}. Note that related mode options will be ignored if those are set.
6688 The expressions can use the following variables:
6692 The sequential number of the filtered frame, starting from @code{0}.
6696 the coordinates of the current sample
6700 the width and height of currently filtered plane
6704 Width and height scale for the plane being filtered. It is the
6705 ratio between the dimensions of the current plane to the luma plane,
6706 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
6707 the luma plane and @code{0.5,0.5} for the chroma planes.
6710 Time of the current frame, expressed in seconds.
6713 Value of pixel component at current location for first video frame (top layer).
6716 Value of pixel component at current location for second video frame (bottom layer).
6720 The @code{blend} filter also supports the @ref{framesync} options.
6722 @subsection Examples
6726 Apply transition from bottom layer to top layer in first 10 seconds:
6728 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
6732 Apply linear horizontal transition from top layer to bottom layer:
6734 blend=all_expr='A*(X/W)+B*(1-X/W)'
6738 Apply 1x1 checkerboard effect:
6740 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
6744 Apply uncover left effect:
6746 blend=all_expr='if(gte(N*SW+X,W),A,B)'
6750 Apply uncover down effect:
6752 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
6756 Apply uncover up-left effect:
6758 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
6762 Split diagonally video and shows top and bottom layer on each side:
6764 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
6768 Display differences between the current and the previous frame:
6770 tblend=all_mode=grainextract
6776 Denoise frames using Block-Matching 3D algorithm.
6778 The filter accepts the following options.
6782 Set denoising strength. Default value is 1.
6783 Allowed range is from 0 to 999.9.
6784 The denoising algorithm is very sensitive to sigma, so adjust it
6785 according to the source.
6788 Set local patch size. This sets dimensions in 2D.
6791 Set sliding step for processing blocks. Default value is 4.
6792 Allowed range is from 1 to 64.
6793 Smaller values allows processing more reference blocks and is slower.
6796 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
6797 When set to 1, no block matching is done. Larger values allows more blocks
6799 Allowed range is from 1 to 256.
6802 Set radius for search block matching. Default is 9.
6803 Allowed range is from 1 to INT32_MAX.
6806 Set step between two search locations for block matching. Default is 1.
6807 Allowed range is from 1 to 64. Smaller is slower.
6810 Set threshold of mean square error for block matching. Valid range is 0 to
6814 Set thresholding parameter for hard thresholding in 3D transformed domain.
6815 Larger values results in stronger hard-thresholding filtering in frequency
6819 Set filtering estimation mode. Can be @code{basic} or @code{final}.
6820 Default is @code{basic}.
6823 If enabled, filter will use 2nd stream for block matching.
6824 Default is disabled for @code{basic} value of @var{estim} option,
6825 and always enabled if value of @var{estim} is @code{final}.
6828 Set planes to filter. Default is all available except alpha.
6831 @subsection Examples
6835 Basic filtering with bm3d:
6837 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
6841 Same as above, but filtering only luma:
6843 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
6847 Same as above, but with both estimation modes:
6849 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
6853 Same as above, but prefilter with @ref{nlmeans} filter instead:
6855 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
6861 Apply a boxblur algorithm to the input video.
6863 It accepts the following parameters:
6867 @item luma_radius, lr
6868 @item luma_power, lp
6869 @item chroma_radius, cr
6870 @item chroma_power, cp
6871 @item alpha_radius, ar
6872 @item alpha_power, ap
6876 A description of the accepted options follows.
6879 @item luma_radius, lr
6880 @item chroma_radius, cr
6881 @item alpha_radius, ar
6882 Set an expression for the box radius in pixels used for blurring the
6883 corresponding input plane.
6885 The radius value must be a non-negative number, and must not be
6886 greater than the value of the expression @code{min(w,h)/2} for the
6887 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
6890 Default value for @option{luma_radius} is "2". If not specified,
6891 @option{chroma_radius} and @option{alpha_radius} default to the
6892 corresponding value set for @option{luma_radius}.
6894 The expressions can contain the following constants:
6898 The input width and height in pixels.
6902 The input chroma image width and height in pixels.
6906 The horizontal and vertical chroma subsample values. For example, for the
6907 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
6910 @item luma_power, lp
6911 @item chroma_power, cp
6912 @item alpha_power, ap
6913 Specify how many times the boxblur filter is applied to the
6914 corresponding plane.
6916 Default value for @option{luma_power} is 2. If not specified,
6917 @option{chroma_power} and @option{alpha_power} default to the
6918 corresponding value set for @option{luma_power}.
6920 A value of 0 will disable the effect.
6923 @subsection Examples
6927 Apply a boxblur filter with the luma, chroma, and alpha radii
6930 boxblur=luma_radius=2:luma_power=1
6935 Set the luma radius to 2, and alpha and chroma radius to 0:
6937 boxblur=2:1:cr=0:ar=0
6941 Set the luma and chroma radii to a fraction of the video dimension:
6943 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
6949 Deinterlace the input video ("bwdif" stands for "Bob Weaver
6950 Deinterlacing Filter").
6952 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
6953 interpolation algorithms.
6954 It accepts the following parameters:
6958 The interlacing mode to adopt. It accepts one of the following values:
6962 Output one frame for each frame.
6964 Output one frame for each field.
6967 The default value is @code{send_field}.
6970 The picture field parity assumed for the input interlaced video. It accepts one
6971 of the following values:
6975 Assume the top field is first.
6977 Assume the bottom field is first.
6979 Enable automatic detection of field parity.
6982 The default value is @code{auto}.
6983 If the interlacing is unknown or the decoder does not export this information,
6984 top field first will be assumed.
6987 Specify which frames to deinterlace. Accepts one of the following
6992 Deinterlace all frames.
6994 Only deinterlace frames marked as interlaced.
6997 The default value is @code{all}.
7002 Apply Contrast Adaptive Sharpen filter to video stream.
7004 The filter accepts the following options:
7008 Set the sharpening strength. Default value is 0.
7011 Set planes to filter. Default value is to filter all
7012 planes except alpha plane.
7016 Remove all color information for all colors except for certain one.
7018 The filter accepts the following options:
7022 The color which will not be replaced with neutral chroma.
7025 Similarity percentage with the above color.
7026 0.01 matches only the exact key color, while 1.0 matches everything.
7030 0.0 makes pixels either fully gray, or not gray at all.
7031 Higher values result in more preserved color.
7034 Signals that the color passed is already in YUV instead of RGB.
7036 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7037 This can be used to pass exact YUV values as hexadecimal numbers.
7040 @subsection Commands
7041 This filter supports same @ref{commands} as options.
7042 The command accepts the same syntax of the corresponding option.
7044 If the specified expression is not valid, it is kept at its current
7048 YUV colorspace color/chroma keying.
7050 The filter accepts the following options:
7054 The color which will be replaced with transparency.
7057 Similarity percentage with the key color.
7059 0.01 matches only the exact key color, while 1.0 matches everything.
7064 0.0 makes pixels either fully transparent, or not transparent at all.
7066 Higher values result in semi-transparent pixels, with a higher transparency
7067 the more similar the pixels color is to the key color.
7070 Signals that the color passed is already in YUV instead of RGB.
7072 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7073 This can be used to pass exact YUV values as hexadecimal numbers.
7076 @subsection Commands
7077 This filter supports same @ref{commands} as options.
7078 The command accepts the same syntax of the corresponding option.
7080 If the specified expression is not valid, it is kept at its current
7083 @subsection Examples
7087 Make every green pixel in the input image transparent:
7089 ffmpeg -i input.png -vf chromakey=green out.png
7093 Overlay a greenscreen-video on top of a static black background.
7095 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
7099 @section chromashift
7100 Shift chroma pixels horizontally and/or vertically.
7102 The filter accepts the following options:
7105 Set amount to shift chroma-blue horizontally.
7107 Set amount to shift chroma-blue vertically.
7109 Set amount to shift chroma-red horizontally.
7111 Set amount to shift chroma-red vertically.
7113 Set edge mode, can be @var{smear}, default, or @var{warp}.
7116 @subsection Commands
7118 This filter supports the all above options as @ref{commands}.
7122 Display CIE color diagram with pixels overlaid onto it.
7124 The filter accepts the following options:
7139 @item uhdtv, rec2020
7153 Set what gamuts to draw.
7155 See @code{system} option for available values.
7158 Set ciescope size, by default set to 512.
7161 Set intensity used to map input pixel values to CIE diagram.
7164 Set contrast used to draw tongue colors that are out of active color system gamut.
7167 Correct gamma displayed on scope, by default enabled.
7170 Show white point on CIE diagram, by default disabled.
7173 Set input gamma. Used only with XYZ input color space.
7178 Visualize information exported by some codecs.
7180 Some codecs can export information through frames using side-data or other
7181 means. For example, some MPEG based codecs export motion vectors through the
7182 @var{export_mvs} flag in the codec @option{flags2} option.
7184 The filter accepts the following option:
7188 Set motion vectors to visualize.
7190 Available flags for @var{mv} are:
7194 forward predicted MVs of P-frames
7196 forward predicted MVs of B-frames
7198 backward predicted MVs of B-frames
7202 Display quantization parameters using the chroma planes.
7205 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7207 Available flags for @var{mv_type} are:
7211 forward predicted MVs
7213 backward predicted MVs
7216 @item frame_type, ft
7217 Set frame type to visualize motion vectors of.
7219 Available flags for @var{frame_type} are:
7223 intra-coded frames (I-frames)
7225 predicted frames (P-frames)
7227 bi-directionally predicted frames (B-frames)
7231 @subsection Examples
7235 Visualize forward predicted MVs of all frames using @command{ffplay}:
7237 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7241 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7243 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7247 @section colorbalance
7248 Modify intensity of primary colors (red, green and blue) of input frames.
7250 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7251 regions for the red-cyan, green-magenta or blue-yellow balance.
7253 A positive adjustment value shifts the balance towards the primary color, a negative
7254 value towards the complementary color.
7256 The filter accepts the following options:
7262 Adjust red, green and blue shadows (darkest pixels).
7267 Adjust red, green and blue midtones (medium pixels).
7272 Adjust red, green and blue highlights (brightest pixels).
7274 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7277 Preserve lightness when changing color balance. Default is disabled.
7280 @subsection Examples
7284 Add red color cast to shadows:
7290 @subsection Commands
7292 This filter supports the all above options as @ref{commands}.
7294 @section colorchannelmixer
7296 Adjust video input frames by re-mixing color channels.
7298 This filter modifies a color channel by adding the values associated to
7299 the other channels of the same pixels. For example if the value to
7300 modify is red, the output value will be:
7302 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7305 The filter accepts the following options:
7312 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7313 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7319 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7320 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7326 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7327 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7333 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7334 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7336 Allowed ranges for options are @code{[-2.0, 2.0]}.
7339 @subsection Examples
7343 Convert source to grayscale:
7345 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7348 Simulate sepia tones:
7350 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7354 @subsection Commands
7356 This filter supports the all above options as @ref{commands}.
7359 RGB colorspace color keying.
7361 The filter accepts the following options:
7365 The color which will be replaced with transparency.
7368 Similarity percentage with the key color.
7370 0.01 matches only the exact key color, while 1.0 matches everything.
7375 0.0 makes pixels either fully transparent, or not transparent at all.
7377 Higher values result in semi-transparent pixels, with a higher transparency
7378 the more similar the pixels color is to the key color.
7381 @subsection Examples
7385 Make every green pixel in the input image transparent:
7387 ffmpeg -i input.png -vf colorkey=green out.png
7391 Overlay a greenscreen-video on top of a static background image.
7393 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
7397 @subsection Commands
7398 This filter supports same @ref{commands} as options.
7399 The command accepts the same syntax of the corresponding option.
7401 If the specified expression is not valid, it is kept at its current
7405 Remove all color information for all RGB colors except for certain one.
7407 The filter accepts the following options:
7411 The color which will not be replaced with neutral gray.
7414 Similarity percentage with the above color.
7415 0.01 matches only the exact key color, while 1.0 matches everything.
7418 Blend percentage. 0.0 makes pixels fully gray.
7419 Higher values result in more preserved color.
7422 @subsection Commands
7423 This filter supports same @ref{commands} as options.
7424 The command accepts the same syntax of the corresponding option.
7426 If the specified expression is not valid, it is kept at its current
7429 @section colorlevels
7431 Adjust video input frames using levels.
7433 The filter accepts the following options:
7440 Adjust red, green, blue and alpha input black point.
7441 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7447 Adjust red, green, blue and alpha input white point.
7448 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7450 Input levels are used to lighten highlights (bright tones), darken shadows
7451 (dark tones), change the balance of bright and dark tones.
7457 Adjust red, green, blue and alpha output black point.
7458 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7464 Adjust red, green, blue and alpha output white point.
7465 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7467 Output levels allows manual selection of a constrained output level range.
7470 @subsection Examples
7474 Make video output darker:
7476 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7482 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7486 Make video output lighter:
7488 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7492 Increase brightness:
7494 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7498 @subsection Commands
7500 This filter supports the all above options as @ref{commands}.
7502 @section colormatrix
7504 Convert color matrix.
7506 The filter accepts the following options:
7511 Specify the source and destination color matrix. Both values must be
7514 The accepted values are:
7542 For example to convert from BT.601 to SMPTE-240M, use the command:
7544 colormatrix=bt601:smpte240m
7549 Convert colorspace, transfer characteristics or color primaries.
7550 Input video needs to have an even size.
7552 The filter accepts the following options:
7557 Specify all color properties at once.
7559 The accepted values are:
7589 Specify output colorspace.
7591 The accepted values are:
7600 BT.470BG or BT.601-6 625
7603 SMPTE-170M or BT.601-6 525
7612 BT.2020 with non-constant luminance
7618 Specify output transfer characteristics.
7620 The accepted values are:
7632 Constant gamma of 2.2
7635 Constant gamma of 2.8
7638 SMPTE-170M, BT.601-6 625 or BT.601-6 525
7656 BT.2020 for 10-bits content
7659 BT.2020 for 12-bits content
7665 Specify output color primaries.
7667 The accepted values are:
7676 BT.470BG or BT.601-6 625
7679 SMPTE-170M or BT.601-6 525
7703 Specify output color range.
7705 The accepted values are:
7708 TV (restricted) range
7711 MPEG (restricted) range
7722 Specify output color format.
7724 The accepted values are:
7727 YUV 4:2:0 planar 8-bits
7730 YUV 4:2:0 planar 10-bits
7733 YUV 4:2:0 planar 12-bits
7736 YUV 4:2:2 planar 8-bits
7739 YUV 4:2:2 planar 10-bits
7742 YUV 4:2:2 planar 12-bits
7745 YUV 4:4:4 planar 8-bits
7748 YUV 4:4:4 planar 10-bits
7751 YUV 4:4:4 planar 12-bits
7756 Do a fast conversion, which skips gamma/primary correction. This will take
7757 significantly less CPU, but will be mathematically incorrect. To get output
7758 compatible with that produced by the colormatrix filter, use fast=1.
7761 Specify dithering mode.
7763 The accepted values are:
7769 Floyd-Steinberg dithering
7773 Whitepoint adaptation mode.
7775 The accepted values are:
7778 Bradford whitepoint adaptation
7781 von Kries whitepoint adaptation
7784 identity whitepoint adaptation (i.e. no whitepoint adaptation)
7788 Override all input properties at once. Same accepted values as @ref{all}.
7791 Override input colorspace. Same accepted values as @ref{space}.
7794 Override input color primaries. Same accepted values as @ref{primaries}.
7797 Override input transfer characteristics. Same accepted values as @ref{trc}.
7800 Override input color range. Same accepted values as @ref{range}.
7804 The filter converts the transfer characteristics, color space and color
7805 primaries to the specified user values. The output value, if not specified,
7806 is set to a default value based on the "all" property. If that property is
7807 also not specified, the filter will log an error. The output color range and
7808 format default to the same value as the input color range and format. The
7809 input transfer characteristics, color space, color primaries and color range
7810 should be set on the input data. If any of these are missing, the filter will
7811 log an error and no conversion will take place.
7813 For example to convert the input to SMPTE-240M, use the command:
7815 colorspace=smpte240m
7818 @section convolution
7820 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
7822 The filter accepts the following options:
7829 Set matrix for each plane.
7830 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
7831 and from 1 to 49 odd number of signed integers in @var{row} mode.
7837 Set multiplier for calculated value for each plane.
7838 If unset or 0, it will be sum of all matrix elements.
7844 Set bias for each plane. This value is added to the result of the multiplication.
7845 Useful for making the overall image brighter or darker. Default is 0.0.
7851 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
7852 Default is @var{square}.
7855 @subsection Examples
7861 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"
7867 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"
7873 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"
7879 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"
7883 Apply laplacian edge detector which includes diagonals:
7885 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"
7891 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"
7897 Apply 2D convolution of video stream in frequency domain using second stream
7900 The filter accepts the following options:
7904 Set which planes to process.
7907 Set which impulse video frames will be processed, can be @var{first}
7908 or @var{all}. Default is @var{all}.
7911 The @code{convolve} filter also supports the @ref{framesync} options.
7915 Copy the input video source unchanged to the output. This is mainly useful for
7920 Video filtering on GPU using Apple's CoreImage API on OSX.
7922 Hardware acceleration is based on an OpenGL context. Usually, this means it is
7923 processed by video hardware. However, software-based OpenGL implementations
7924 exist which means there is no guarantee for hardware processing. It depends on
7927 There are many filters and image generators provided by Apple that come with a
7928 large variety of options. The filter has to be referenced by its name along
7931 The coreimage filter accepts the following options:
7934 List all available filters and generators along with all their respective
7935 options as well as possible minimum and maximum values along with the default
7942 Specify all filters by their respective name and options.
7943 Use @var{list_filters} to determine all valid filter names and options.
7944 Numerical options are specified by a float value and are automatically clamped
7945 to their respective value range. Vector and color options have to be specified
7946 by a list of space separated float values. Character escaping has to be done.
7947 A special option name @code{default} is available to use default options for a
7950 It is required to specify either @code{default} or at least one of the filter options.
7951 All omitted options are used with their default values.
7952 The syntax of the filter string is as follows:
7954 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
7958 Specify a rectangle where the output of the filter chain is copied into the
7959 input image. It is given by a list of space separated float values:
7961 output_rect=x\ y\ width\ height
7963 If not given, the output rectangle equals the dimensions of the input image.
7964 The output rectangle is automatically cropped at the borders of the input
7965 image. Negative values are valid for each component.
7967 output_rect=25\ 25\ 100\ 100
7971 Several filters can be chained for successive processing without GPU-HOST
7972 transfers allowing for fast processing of complex filter chains.
7973 Currently, only filters with zero (generators) or exactly one (filters) input
7974 image and one output image are supported. Also, transition filters are not yet
7977 Some filters generate output images with additional padding depending on the
7978 respective filter kernel. The padding is automatically removed to ensure the
7979 filter output has the same size as the input image.
7981 For image generators, the size of the output image is determined by the
7982 previous output image of the filter chain or the input image of the whole
7983 filterchain, respectively. The generators do not use the pixel information of
7984 this image to generate their output. However, the generated output is
7985 blended onto this image, resulting in partial or complete coverage of the
7988 The @ref{coreimagesrc} video source can be used for generating input images
7989 which are directly fed into the filter chain. By using it, providing input
7990 images by another video source or an input video is not required.
7992 @subsection Examples
7997 List all filters available:
7999 coreimage=list_filters=true
8003 Use the CIBoxBlur filter with default options to blur an image:
8005 coreimage=filter=CIBoxBlur@@default
8009 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8010 its center at 100x100 and a radius of 50 pixels:
8012 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8016 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8017 given as complete and escaped command-line for Apple's standard bash shell:
8019 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8025 Cover a rectangular object
8027 It accepts the following options:
8031 Filepath of the optional cover image, needs to be in yuv420.
8036 It accepts the following values:
8039 cover it by the supplied image
8041 cover it by interpolating the surrounding pixels
8044 Default value is @var{blur}.
8047 @subsection Examples
8051 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8053 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8059 Crop the input video to given dimensions.
8061 It accepts the following parameters:
8065 The width of the output video. It defaults to @code{iw}.
8066 This expression is evaluated only once during the filter
8067 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8070 The height of the output video. It defaults to @code{ih}.
8071 This expression is evaluated only once during the filter
8072 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8075 The horizontal position, in the input video, of the left edge of the output
8076 video. It defaults to @code{(in_w-out_w)/2}.
8077 This expression is evaluated per-frame.
8080 The vertical position, in the input video, of the top edge of the output video.
8081 It defaults to @code{(in_h-out_h)/2}.
8082 This expression is evaluated per-frame.
8085 If set to 1 will force the output display aspect ratio
8086 to be the same of the input, by changing the output sample aspect
8087 ratio. It defaults to 0.
8090 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8091 width/height/x/y as specified and will not be rounded to nearest smaller value.
8095 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8096 expressions containing the following constants:
8101 The computed values for @var{x} and @var{y}. They are evaluated for
8106 The input width and height.
8110 These are the same as @var{in_w} and @var{in_h}.
8114 The output (cropped) width and height.
8118 These are the same as @var{out_w} and @var{out_h}.
8121 same as @var{iw} / @var{ih}
8124 input sample aspect ratio
8127 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8131 horizontal and vertical chroma subsample values. For example for the
8132 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8135 The number of the input frame, starting from 0.
8138 the position in the file of the input frame, NAN if unknown
8141 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8145 The expression for @var{out_w} may depend on the value of @var{out_h},
8146 and the expression for @var{out_h} may depend on @var{out_w}, but they
8147 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8148 evaluated after @var{out_w} and @var{out_h}.
8150 The @var{x} and @var{y} parameters specify the expressions for the
8151 position of the top-left corner of the output (non-cropped) area. They
8152 are evaluated for each frame. If the evaluated value is not valid, it
8153 is approximated to the nearest valid value.
8155 The expression for @var{x} may depend on @var{y}, and the expression
8156 for @var{y} may depend on @var{x}.
8158 @subsection Examples
8162 Crop area with size 100x100 at position (12,34).
8167 Using named options, the example above becomes:
8169 crop=w=100:h=100:x=12:y=34
8173 Crop the central input area with size 100x100:
8179 Crop the central input area with size 2/3 of the input video:
8181 crop=2/3*in_w:2/3*in_h
8185 Crop the input video central square:
8192 Delimit the rectangle with the top-left corner placed at position
8193 100:100 and the right-bottom corner corresponding to the right-bottom
8194 corner of the input image.
8196 crop=in_w-100:in_h-100:100:100
8200 Crop 10 pixels from the left and right borders, and 20 pixels from
8201 the top and bottom borders
8203 crop=in_w-2*10:in_h-2*20
8207 Keep only the bottom right quarter of the input image:
8209 crop=in_w/2:in_h/2:in_w/2:in_h/2
8213 Crop height for getting Greek harmony:
8215 crop=in_w:1/PHI*in_w
8219 Apply trembling effect:
8221 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)
8225 Apply erratic camera effect depending on timestamp:
8227 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)"
8231 Set x depending on the value of y:
8233 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8237 @subsection Commands
8239 This filter supports the following commands:
8245 Set width/height of the output video and the horizontal/vertical position
8247 The command accepts the same syntax of the corresponding option.
8249 If the specified expression is not valid, it is kept at its current
8255 Auto-detect the crop size.
8257 It calculates the necessary cropping parameters and prints the
8258 recommended parameters via the logging system. The detected dimensions
8259 correspond to the non-black area of the input video.
8261 It accepts the following parameters:
8266 Set higher black value threshold, which can be optionally specified
8267 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8268 value greater to the set value is considered non-black. It defaults to 24.
8269 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8270 on the bitdepth of the pixel format.
8273 The value which the width/height should be divisible by. It defaults to
8274 16. The offset is automatically adjusted to center the video. Use 2 to
8275 get only even dimensions (needed for 4:2:2 video). 16 is best when
8276 encoding to most video codecs.
8278 @item reset_count, reset
8279 Set the counter that determines after how many frames cropdetect will
8280 reset the previously detected largest video area and start over to
8281 detect the current optimal crop area. Default value is 0.
8283 This can be useful when channel logos distort the video area. 0
8284 indicates 'never reset', and returns the largest area encountered during
8291 Delay video filtering until a given wallclock timestamp. The filter first
8292 passes on @option{preroll} amount of frames, then it buffers at most
8293 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8294 it forwards the buffered frames and also any subsequent frames coming in its
8297 The filter can be used synchronize the output of multiple ffmpeg processes for
8298 realtime output devices like decklink. By putting the delay in the filtering
8299 chain and pre-buffering frames the process can pass on data to output almost
8300 immediately after the target wallclock timestamp is reached.
8302 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8308 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8311 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8314 The maximum duration of content to buffer before waiting for the cue expressed
8315 in seconds. Default is 0.
8322 Apply color adjustments using curves.
8324 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8325 component (red, green and blue) has its values defined by @var{N} key points
8326 tied from each other using a smooth curve. The x-axis represents the pixel
8327 values from the input frame, and the y-axis the new pixel values to be set for
8330 By default, a component curve is defined by the two points @var{(0;0)} and
8331 @var{(1;1)}. This creates a straight line where each original pixel value is
8332 "adjusted" to its own value, which means no change to the image.
8334 The filter allows you to redefine these two points and add some more. A new
8335 curve (using a natural cubic spline interpolation) will be define to pass
8336 smoothly through all these new coordinates. The new defined points needs to be
8337 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8338 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8339 the vector spaces, the values will be clipped accordingly.
8341 The filter accepts the following options:
8345 Select one of the available color presets. This option can be used in addition
8346 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8347 options takes priority on the preset values.
8348 Available presets are:
8351 @item color_negative
8354 @item increase_contrast
8356 @item linear_contrast
8357 @item medium_contrast
8359 @item strong_contrast
8362 Default is @code{none}.
8364 Set the master key points. These points will define a second pass mapping. It
8365 is sometimes called a "luminance" or "value" mapping. It can be used with
8366 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8367 post-processing LUT.
8369 Set the key points for the red component.
8371 Set the key points for the green component.
8373 Set the key points for the blue component.
8375 Set the key points for all components (not including master).
8376 Can be used in addition to the other key points component
8377 options. In this case, the unset component(s) will fallback on this
8378 @option{all} setting.
8380 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8382 Save Gnuplot script of the curves in specified file.
8385 To avoid some filtergraph syntax conflicts, each key points list need to be
8386 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8388 @subsection Examples
8392 Increase slightly the middle level of blue:
8394 curves=blue='0/0 0.5/0.58 1/1'
8400 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'
8402 Here we obtain the following coordinates for each components:
8405 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8407 @code{(0;0) (0.50;0.48) (1;1)}
8409 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8413 The previous example can also be achieved with the associated built-in preset:
8415 curves=preset=vintage
8425 Use a Photoshop preset and redefine the points of the green component:
8427 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8431 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8432 and @command{gnuplot}:
8434 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8435 gnuplot -p /tmp/curves.plt
8441 Video data analysis filter.
8443 This filter shows hexadecimal pixel values of part of video.
8445 The filter accepts the following options:
8449 Set output video size.
8452 Set x offset from where to pick pixels.
8455 Set y offset from where to pick pixels.
8458 Set scope mode, can be one of the following:
8461 Draw hexadecimal pixel values with white color on black background.
8464 Draw hexadecimal pixel values with input video pixel color on black
8468 Draw hexadecimal pixel values on color background picked from input video,
8469 the text color is picked in such way so its always visible.
8473 Draw rows and columns numbers on left and top of video.
8476 Set background opacity.
8479 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
8484 Denoise frames using 2D DCT (frequency domain filtering).
8486 This filter is not designed for real time.
8488 The filter accepts the following options:
8492 Set the noise sigma constant.
8494 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8495 coefficient (absolute value) below this threshold with be dropped.
8497 If you need a more advanced filtering, see @option{expr}.
8499 Default is @code{0}.
8502 Set number overlapping pixels for each block. Since the filter can be slow, you
8503 may want to reduce this value, at the cost of a less effective filter and the
8504 risk of various artefacts.
8506 If the overlapping value doesn't permit processing the whole input width or
8507 height, a warning will be displayed and according borders won't be denoised.
8509 Default value is @var{blocksize}-1, which is the best possible setting.
8512 Set the coefficient factor expression.
8514 For each coefficient of a DCT block, this expression will be evaluated as a
8515 multiplier value for the coefficient.
8517 If this is option is set, the @option{sigma} option will be ignored.
8519 The absolute value of the coefficient can be accessed through the @var{c}
8523 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8524 @var{blocksize}, which is the width and height of the processed blocks.
8526 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8527 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8528 on the speed processing. Also, a larger block size does not necessarily means a
8532 @subsection Examples
8534 Apply a denoise with a @option{sigma} of @code{4.5}:
8539 The same operation can be achieved using the expression system:
8541 dctdnoiz=e='gte(c, 4.5*3)'
8544 Violent denoise using a block size of @code{16x16}:
8551 Remove banding artifacts from input video.
8552 It works by replacing banded pixels with average value of referenced pixels.
8554 The filter accepts the following options:
8561 Set banding detection threshold for each plane. Default is 0.02.
8562 Valid range is 0.00003 to 0.5.
8563 If difference between current pixel and reference pixel is less than threshold,
8564 it will be considered as banded.
8567 Banding detection range in pixels. Default is 16. If positive, random number
8568 in range 0 to set value will be used. If negative, exact absolute value
8570 The range defines square of four pixels around current pixel.
8573 Set direction in radians from which four pixel will be compared. If positive,
8574 random direction from 0 to set direction will be picked. If negative, exact of
8575 absolute value will be picked. For example direction 0, -PI or -2*PI radians
8576 will pick only pixels on same row and -PI/2 will pick only pixels on same
8580 If enabled, current pixel is compared with average value of all four
8581 surrounding pixels. The default is enabled. If disabled current pixel is
8582 compared with all four surrounding pixels. The pixel is considered banded
8583 if only all four differences with surrounding pixels are less than threshold.
8586 If enabled, current pixel is changed if and only if all pixel components are banded,
8587 e.g. banding detection threshold is triggered for all color components.
8588 The default is disabled.
8593 Remove blocking artifacts from input video.
8595 The filter accepts the following options:
8599 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
8600 This controls what kind of deblocking is applied.
8603 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
8609 Set blocking detection thresholds. Allowed range is 0 to 1.
8610 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
8611 Using higher threshold gives more deblocking strength.
8612 Setting @var{alpha} controls threshold detection at exact edge of block.
8613 Remaining options controls threshold detection near the edge. Each one for
8614 below/above or left/right. Setting any of those to @var{0} disables
8618 Set planes to filter. Default is to filter all available planes.
8621 @subsection Examples
8625 Deblock using weak filter and block size of 4 pixels.
8627 deblock=filter=weak:block=4
8631 Deblock using strong filter, block size of 4 pixels and custom thresholds for
8632 deblocking more edges.
8634 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
8638 Similar as above, but filter only first plane.
8640 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
8644 Similar as above, but filter only second and third plane.
8646 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
8653 Drop duplicated frames at regular intervals.
8655 The filter accepts the following options:
8659 Set the number of frames from which one will be dropped. Setting this to
8660 @var{N} means one frame in every batch of @var{N} frames will be dropped.
8661 Default is @code{5}.
8664 Set the threshold for duplicate detection. If the difference metric for a frame
8665 is less than or equal to this value, then it is declared as duplicate. Default
8669 Set scene change threshold. Default is @code{15}.
8673 Set the size of the x and y-axis blocks used during metric calculations.
8674 Larger blocks give better noise suppression, but also give worse detection of
8675 small movements. Must be a power of two. Default is @code{32}.
8678 Mark main input as a pre-processed input and activate clean source input
8679 stream. This allows the input to be pre-processed with various filters to help
8680 the metrics calculation while keeping the frame selection lossless. When set to
8681 @code{1}, the first stream is for the pre-processed input, and the second
8682 stream is the clean source from where the kept frames are chosen. Default is
8686 Set whether or not chroma is considered in the metric calculations. Default is
8692 Apply 2D deconvolution of video stream in frequency domain using second stream
8695 The filter accepts the following options:
8699 Set which planes to process.
8702 Set which impulse video frames will be processed, can be @var{first}
8703 or @var{all}. Default is @var{all}.
8706 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
8707 and height are not same and not power of 2 or if stream prior to convolving
8711 The @code{deconvolve} filter also supports the @ref{framesync} options.
8715 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
8717 It accepts the following options:
8721 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
8722 @var{rainbows} for cross-color reduction.
8725 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
8728 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
8731 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
8734 Set temporal chroma threshold. Lower values increases reduction of cross-color.
8739 Apply deflate effect to the video.
8741 This filter replaces the pixel by the local(3x3) average by taking into account
8742 only values lower than the pixel.
8744 It accepts the following options:
8751 Limit the maximum change for each plane, default is 65535.
8752 If 0, plane will remain unchanged.
8755 @subsection Commands
8757 This filter supports the all above options as @ref{commands}.
8761 Remove temporal frame luminance variations.
8763 It accepts the following options:
8767 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
8770 Set averaging mode to smooth temporal luminance variations.
8772 Available values are:
8797 Do not actually modify frame. Useful when one only wants metadata.
8802 Remove judder produced by partially interlaced telecined content.
8804 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
8805 source was partially telecined content then the output of @code{pullup,dejudder}
8806 will have a variable frame rate. May change the recorded frame rate of the
8807 container. Aside from that change, this filter will not affect constant frame
8810 The option available in this filter is:
8814 Specify the length of the window over which the judder repeats.
8816 Accepts any integer greater than 1. Useful values are:
8820 If the original was telecined from 24 to 30 fps (Film to NTSC).
8823 If the original was telecined from 25 to 30 fps (PAL to NTSC).
8826 If a mixture of the two.
8829 The default is @samp{4}.
8834 Suppress a TV station logo by a simple interpolation of the surrounding
8835 pixels. Just set a rectangle covering the logo and watch it disappear
8836 (and sometimes something even uglier appear - your mileage may vary).
8838 It accepts the following parameters:
8843 Specify the top left corner coordinates of the logo. They must be
8848 Specify the width and height of the logo to clear. They must be
8852 Specify the thickness of the fuzzy edge of the rectangle (added to
8853 @var{w} and @var{h}). The default value is 1. This option is
8854 deprecated, setting higher values should no longer be necessary and
8858 When set to 1, a green rectangle is drawn on the screen to simplify
8859 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
8860 The default value is 0.
8862 The rectangle is drawn on the outermost pixels which will be (partly)
8863 replaced with interpolated values. The values of the next pixels
8864 immediately outside this rectangle in each direction will be used to
8865 compute the interpolated pixel values inside the rectangle.
8869 @subsection Examples
8873 Set a rectangle covering the area with top left corner coordinates 0,0
8874 and size 100x77, and a band of size 10:
8876 delogo=x=0:y=0:w=100:h=77:band=10
8883 Remove the rain in the input image/video by applying the derain methods based on
8884 convolutional neural networks. Supported models:
8888 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
8889 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
8892 Training as well as model generation scripts are provided in
8893 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
8895 Native model files (.model) can be generated from TensorFlow model
8896 files (.pb) by using tools/python/convert.py
8898 The filter accepts the following options:
8902 Specify which filter to use. This option accepts the following values:
8906 Derain filter. To conduct derain filter, you need to use a derain model.
8909 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
8911 Default value is @samp{derain}.
8914 Specify which DNN backend to use for model loading and execution. This option accepts
8915 the following values:
8919 Native implementation of DNN loading and execution.
8922 TensorFlow backend. To enable this backend you
8923 need to install the TensorFlow for C library (see
8924 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
8925 @code{--enable-libtensorflow}
8927 Default value is @samp{native}.
8930 Set path to model file specifying network architecture and its parameters.
8931 Note that different backends use different file formats. TensorFlow and native
8932 backend can load files for only its format.
8937 Attempt to fix small changes in horizontal and/or vertical shift. This
8938 filter helps remove camera shake from hand-holding a camera, bumping a
8939 tripod, moving on a vehicle, etc.
8941 The filter accepts the following options:
8949 Specify a rectangular area where to limit the search for motion
8951 If desired the search for motion vectors can be limited to a
8952 rectangular area of the frame defined by its top left corner, width
8953 and height. These parameters have the same meaning as the drawbox
8954 filter which can be used to visualise the position of the bounding
8957 This is useful when simultaneous movement of subjects within the frame
8958 might be confused for camera motion by the motion vector search.
8960 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
8961 then the full frame is used. This allows later options to be set
8962 without specifying the bounding box for the motion vector search.
8964 Default - search the whole frame.
8968 Specify the maximum extent of movement in x and y directions in the
8969 range 0-64 pixels. Default 16.
8972 Specify how to generate pixels to fill blanks at the edge of the
8973 frame. Available values are:
8976 Fill zeroes at blank locations
8978 Original image at blank locations
8980 Extruded edge value at blank locations
8982 Mirrored edge at blank locations
8984 Default value is @samp{mirror}.
8987 Specify the blocksize to use for motion search. Range 4-128 pixels,
8991 Specify the contrast threshold for blocks. Only blocks with more than
8992 the specified contrast (difference between darkest and lightest
8993 pixels) will be considered. Range 1-255, default 125.
8996 Specify the search strategy. Available values are:
8999 Set exhaustive search
9001 Set less exhaustive search.
9003 Default value is @samp{exhaustive}.
9006 If set then a detailed log of the motion search is written to the
9013 Remove unwanted contamination of foreground colors, caused by reflected color of
9014 greenscreen or bluescreen.
9016 This filter accepts the following options:
9020 Set what type of despill to use.
9023 Set how spillmap will be generated.
9026 Set how much to get rid of still remaining spill.
9029 Controls amount of red in spill area.
9032 Controls amount of green in spill area.
9033 Should be -1 for greenscreen.
9036 Controls amount of blue in spill area.
9037 Should be -1 for bluescreen.
9040 Controls brightness of spill area, preserving colors.
9043 Modify alpha from generated spillmap.
9048 Apply an exact inverse of the telecine operation. It requires a predefined
9049 pattern specified using the pattern option which must be the same as that passed
9050 to the telecine filter.
9052 This filter accepts the following options:
9061 The default value is @code{top}.
9065 A string of numbers representing the pulldown pattern you wish to apply.
9066 The default value is @code{23}.
9069 A number representing position of the first frame with respect to the telecine
9070 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9075 Apply dilation effect to the video.
9077 This filter replaces the pixel by the local(3x3) maximum.
9079 It accepts the following options:
9086 Limit the maximum change for each plane, default is 65535.
9087 If 0, plane will remain unchanged.
9090 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9093 Flags to local 3x3 coordinates maps like this:
9100 @subsection Commands
9102 This filter supports the all above options as @ref{commands}.
9106 Displace pixels as indicated by second and third input stream.
9108 It takes three input streams and outputs one stream, the first input is the
9109 source, and second and third input are displacement maps.
9111 The second input specifies how much to displace pixels along the
9112 x-axis, while the third input specifies how much to displace pixels
9114 If one of displacement map streams terminates, last frame from that
9115 displacement map will be used.
9117 Note that once generated, displacements maps can be reused over and over again.
9119 A description of the accepted options follows.
9123 Set displace behavior for pixels that are out of range.
9125 Available values are:
9128 Missing pixels are replaced by black pixels.
9131 Adjacent pixels will spread out to replace missing pixels.
9134 Out of range pixels are wrapped so they point to pixels of other side.
9137 Out of range pixels will be replaced with mirrored pixels.
9139 Default is @samp{smear}.
9143 @subsection Examples
9147 Add ripple effect to rgb input of video size hd720:
9149 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
9153 Add wave effect to rgb input of video size hd720:
9155 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
9159 @anchor{dnn_processing}
9160 @section dnn_processing
9162 Do image processing with deep neural networks. It works together with another filter
9163 which converts the pixel format of the Frame to what the dnn network requires.
9165 The filter accepts the following options:
9169 Specify which DNN backend to use for model loading and execution. This option accepts
9170 the following values:
9174 Native implementation of DNN loading and execution.
9177 TensorFlow backend. To enable this backend you
9178 need to install the TensorFlow for C library (see
9179 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9180 @code{--enable-libtensorflow}
9183 Default value is @samp{native}.
9186 Set path to model file specifying network architecture and its parameters.
9187 Note that different backends use different file formats. TensorFlow and native
9188 backend can load files for only its format.
9190 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9193 Set the input name of the dnn network.
9196 Set the output name of the dnn network.
9200 @subsection Examples
9204 Halve the red channle of the frame with format rgb24:
9206 ffmpeg -i input.jpg -vf format=rgb24,dnn_processing=model=halve_first_channel.model:input=dnn_in:output=dnn_out:dnn_backend=native out.native.png
9210 Halve the pixel value of the frame with format gray32f:
9212 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
9216 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9218 ./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
9222 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9224 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9231 Draw a colored box on the input image.
9233 It accepts the following parameters:
9238 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9242 The expressions which specify the width and height of the box; if 0 they are interpreted as
9243 the input width and height. It defaults to 0.
9246 Specify the color of the box to write. For the general syntax of this option,
9247 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9248 value @code{invert} is used, the box edge color is the same as the
9249 video with inverted luma.
9252 The expression which sets the thickness of the box edge.
9253 A value of @code{fill} will create a filled box. Default value is @code{3}.
9255 See below for the list of accepted constants.
9258 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9259 will overwrite the video's color and alpha pixels.
9260 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9263 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9264 following constants:
9268 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9272 horizontal and vertical chroma subsample values. For example for the
9273 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9277 The input width and height.
9280 The input sample aspect ratio.
9284 The x and y offset coordinates where the box is drawn.
9288 The width and height of the drawn box.
9291 The thickness of the drawn box.
9293 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9294 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9298 @subsection Examples
9302 Draw a black box around the edge of the input image:
9308 Draw a box with color red and an opacity of 50%:
9310 drawbox=10:20:200:60:red@@0.5
9313 The previous example can be specified as:
9315 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9319 Fill the box with pink color:
9321 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9325 Draw a 2-pixel red 2.40:1 mask:
9327 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
9331 @subsection Commands
9332 This filter supports same commands as options.
9333 The command accepts the same syntax of the corresponding option.
9335 If the specified expression is not valid, it is kept at its current
9340 Draw a graph using input video metadata.
9342 It accepts the following parameters:
9346 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9349 Set 1st foreground color expression.
9352 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9355 Set 2nd foreground color expression.
9358 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9361 Set 3rd foreground color expression.
9364 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9367 Set 4th foreground color expression.
9370 Set minimal value of metadata value.
9373 Set maximal value of metadata value.
9376 Set graph background color. Default is white.
9381 Available values for mode is:
9388 Default is @code{line}.
9393 Available values for slide is:
9396 Draw new frame when right border is reached.
9399 Replace old columns with new ones.
9402 Scroll from right to left.
9405 Scroll from left to right.
9408 Draw single picture.
9411 Default is @code{frame}.
9414 Set size of graph video. For the syntax of this option, check the
9415 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9416 The default value is @code{900x256}.
9419 Set the output frame rate. Default value is @code{25}.
9421 The foreground color expressions can use the following variables:
9424 Minimal value of metadata value.
9427 Maximal value of metadata value.
9430 Current metadata key value.
9433 The color is defined as 0xAABBGGRR.
9436 Example using metadata from @ref{signalstats} filter:
9438 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9441 Example using metadata from @ref{ebur128} filter:
9443 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9448 Draw a grid on the input image.
9450 It accepts the following parameters:
9455 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9459 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9460 input width and height, respectively, minus @code{thickness}, so image gets
9461 framed. Default to 0.
9464 Specify the color of the grid. For the general syntax of this option,
9465 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9466 value @code{invert} is used, the grid color is the same as the
9467 video with inverted luma.
9470 The expression which sets the thickness of the grid line. Default value is @code{1}.
9472 See below for the list of accepted constants.
9475 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9476 will overwrite the video's color and alpha pixels.
9477 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9480 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9481 following constants:
9485 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9489 horizontal and vertical chroma subsample values. For example for the
9490 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9494 The input grid cell width and height.
9497 The input sample aspect ratio.
9501 The x and y coordinates of some point of grid intersection (meant to configure offset).
9505 The width and height of the drawn cell.
9508 The thickness of the drawn cell.
9510 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9511 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9515 @subsection Examples
9519 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9521 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9525 Draw a white 3x3 grid with an opacity of 50%:
9527 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
9531 @subsection Commands
9532 This filter supports same commands as options.
9533 The command accepts the same syntax of the corresponding option.
9535 If the specified expression is not valid, it is kept at its current
9541 Draw a text string or text from a specified file on top of a video, using the
9542 libfreetype library.
9544 To enable compilation of this filter, you need to configure FFmpeg with
9545 @code{--enable-libfreetype}.
9546 To enable default font fallback and the @var{font} option you need to
9547 configure FFmpeg with @code{--enable-libfontconfig}.
9548 To enable the @var{text_shaping} option, you need to configure FFmpeg with
9549 @code{--enable-libfribidi}.
9553 It accepts the following parameters:
9558 Used to draw a box around text using the background color.
9559 The value must be either 1 (enable) or 0 (disable).
9560 The default value of @var{box} is 0.
9563 Set the width of the border to be drawn around the box using @var{boxcolor}.
9564 The default value of @var{boxborderw} is 0.
9567 The color to be used for drawing box around text. For the syntax of this
9568 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9570 The default value of @var{boxcolor} is "white".
9573 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
9574 The default value of @var{line_spacing} is 0.
9577 Set the width of the border to be drawn around the text using @var{bordercolor}.
9578 The default value of @var{borderw} is 0.
9581 Set the color to be used for drawing border around text. For the syntax of this
9582 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9584 The default value of @var{bordercolor} is "black".
9587 Select how the @var{text} is expanded. Can be either @code{none},
9588 @code{strftime} (deprecated) or
9589 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
9593 Set a start time for the count. Value is in microseconds. Only applied
9594 in the deprecated strftime expansion mode. To emulate in normal expansion
9595 mode use the @code{pts} function, supplying the start time (in seconds)
9596 as the second argument.
9599 If true, check and fix text coords to avoid clipping.
9602 The color to be used for drawing fonts. For the syntax of this option, check
9603 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9605 The default value of @var{fontcolor} is "black".
9607 @item fontcolor_expr
9608 String which is expanded the same way as @var{text} to obtain dynamic
9609 @var{fontcolor} value. By default this option has empty value and is not
9610 processed. When this option is set, it overrides @var{fontcolor} option.
9613 The font family to be used for drawing text. By default Sans.
9616 The font file to be used for drawing text. The path must be included.
9617 This parameter is mandatory if the fontconfig support is disabled.
9620 Draw the text applying alpha blending. The value can
9621 be a number between 0.0 and 1.0.
9622 The expression accepts the same variables @var{x, y} as well.
9623 The default value is 1.
9624 Please see @var{fontcolor_expr}.
9627 The font size to be used for drawing text.
9628 The default value of @var{fontsize} is 16.
9631 If set to 1, attempt to shape the text (for example, reverse the order of
9632 right-to-left text and join Arabic characters) before drawing it.
9633 Otherwise, just draw the text exactly as given.
9634 By default 1 (if supported).
9637 The flags to be used for loading the fonts.
9639 The flags map the corresponding flags supported by libfreetype, and are
9640 a combination of the following values:
9647 @item vertical_layout
9648 @item force_autohint
9651 @item ignore_global_advance_width
9653 @item ignore_transform
9659 Default value is "default".
9661 For more information consult the documentation for the FT_LOAD_*
9665 The color to be used for drawing a shadow behind the drawn text. For the
9666 syntax of this option, check the @ref{color syntax,,"Color" section in the
9667 ffmpeg-utils manual,ffmpeg-utils}.
9669 The default value of @var{shadowcolor} is "black".
9673 The x and y offsets for the text shadow position with respect to the
9674 position of the text. They can be either positive or negative
9675 values. The default value for both is "0".
9678 The starting frame number for the n/frame_num variable. The default value
9682 The size in number of spaces to use for rendering the tab.
9686 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
9687 format. It can be used with or without text parameter. @var{timecode_rate}
9688 option must be specified.
9690 @item timecode_rate, rate, r
9691 Set the timecode frame rate (timecode only). Value will be rounded to nearest
9692 integer. Minimum value is "1".
9693 Drop-frame timecode is supported for frame rates 30 & 60.
9696 If set to 1, the output of the timecode option will wrap around at 24 hours.
9697 Default is 0 (disabled).
9700 The text string to be drawn. The text must be a sequence of UTF-8
9702 This parameter is mandatory if no file is specified with the parameter
9706 A text file containing text to be drawn. The text must be a sequence
9707 of UTF-8 encoded characters.
9709 This parameter is mandatory if no text string is specified with the
9710 parameter @var{text}.
9712 If both @var{text} and @var{textfile} are specified, an error is thrown.
9715 If set to 1, the @var{textfile} will be reloaded before each frame.
9716 Be sure to update it atomically, or it may be read partially, or even fail.
9720 The expressions which specify the offsets where text will be drawn
9721 within the video frame. They are relative to the top/left border of the
9724 The default value of @var{x} and @var{y} is "0".
9726 See below for the list of accepted constants and functions.
9729 The parameters for @var{x} and @var{y} are expressions containing the
9730 following constants and functions:
9734 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
9738 horizontal and vertical chroma subsample values. For example for the
9739 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9742 the height of each text line
9750 @item max_glyph_a, ascent
9751 the maximum distance from the baseline to the highest/upper grid
9752 coordinate used to place a glyph outline point, for all the rendered
9754 It is a positive value, due to the grid's orientation with the Y axis
9757 @item max_glyph_d, descent
9758 the maximum distance from the baseline to the lowest grid coordinate
9759 used to place a glyph outline point, for all the rendered glyphs.
9760 This is a negative value, due to the grid's orientation, with the Y axis
9764 maximum glyph height, that is the maximum height for all the glyphs
9765 contained in the rendered text, it is equivalent to @var{ascent} -
9769 maximum glyph width, that is the maximum width for all the glyphs
9770 contained in the rendered text
9773 the number of input frame, starting from 0
9775 @item rand(min, max)
9776 return a random number included between @var{min} and @var{max}
9779 The input sample aspect ratio.
9782 timestamp expressed in seconds, NAN if the input timestamp is unknown
9785 the height of the rendered text
9788 the width of the rendered text
9792 the x and y offset coordinates where the text is drawn.
9794 These parameters allow the @var{x} and @var{y} expressions to refer
9795 to each other, so you can for example specify @code{y=x/dar}.
9798 A one character description of the current frame's picture type.
9801 The current packet's position in the input file or stream
9802 (in bytes, from the start of the input). A value of -1 indicates
9803 this info is not available.
9806 The current packet's duration, in seconds.
9809 The current packet's size (in bytes).
9812 @anchor{drawtext_expansion}
9813 @subsection Text expansion
9815 If @option{expansion} is set to @code{strftime},
9816 the filter recognizes strftime() sequences in the provided text and
9817 expands them accordingly. Check the documentation of strftime(). This
9818 feature is deprecated.
9820 If @option{expansion} is set to @code{none}, the text is printed verbatim.
9822 If @option{expansion} is set to @code{normal} (which is the default),
9823 the following expansion mechanism is used.
9825 The backslash character @samp{\}, followed by any character, always expands to
9826 the second character.
9828 Sequences of the form @code{%@{...@}} are expanded. The text between the
9829 braces is a function name, possibly followed by arguments separated by ':'.
9830 If the arguments contain special characters or delimiters (':' or '@}'),
9831 they should be escaped.
9833 Note that they probably must also be escaped as the value for the
9834 @option{text} option in the filter argument string and as the filter
9835 argument in the filtergraph description, and possibly also for the shell,
9836 that makes up to four levels of escaping; using a text file avoids these
9839 The following functions are available:
9844 The expression evaluation result.
9846 It must take one argument specifying the expression to be evaluated,
9847 which accepts the same constants and functions as the @var{x} and
9848 @var{y} values. Note that not all constants should be used, for
9849 example the text size is not known when evaluating the expression, so
9850 the constants @var{text_w} and @var{text_h} will have an undefined
9853 @item expr_int_format, eif
9854 Evaluate the expression's value and output as formatted integer.
9856 The first argument is the expression to be evaluated, just as for the @var{expr} function.
9857 The second argument specifies the output format. Allowed values are @samp{x},
9858 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
9859 @code{printf} function.
9860 The third parameter is optional and sets the number of positions taken by the output.
9861 It can be used to add padding with zeros from the left.
9864 The time at which the filter is running, expressed in UTC.
9865 It can accept an argument: a strftime() format string.
9868 The time at which the filter is running, expressed in the local time zone.
9869 It can accept an argument: a strftime() format string.
9872 Frame metadata. Takes one or two arguments.
9874 The first argument is mandatory and specifies the metadata key.
9876 The second argument is optional and specifies a default value, used when the
9877 metadata key is not found or empty.
9879 Available metadata can be identified by inspecting entries
9880 starting with TAG included within each frame section
9881 printed by running @code{ffprobe -show_frames}.
9883 String metadata generated in filters leading to
9884 the drawtext filter are also available.
9887 The frame number, starting from 0.
9890 A one character description of the current picture type.
9893 The timestamp of the current frame.
9894 It can take up to three arguments.
9896 The first argument is the format of the timestamp; it defaults to @code{flt}
9897 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
9898 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
9899 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
9900 @code{localtime} stands for the timestamp of the frame formatted as
9901 local time zone time.
9903 The second argument is an offset added to the timestamp.
9905 If the format is set to @code{hms}, a third argument @code{24HH} may be
9906 supplied to present the hour part of the formatted timestamp in 24h format
9909 If the format is set to @code{localtime} or @code{gmtime},
9910 a third argument may be supplied: a strftime() format string.
9911 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
9914 @subsection Commands
9916 This filter supports altering parameters via commands:
9919 Alter existing filter parameters.
9921 Syntax for the argument is the same as for filter invocation, e.g.
9924 fontsize=56:fontcolor=green:text='Hello World'
9927 Full filter invocation with sendcmd would look like this:
9930 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
9934 If the entire argument can't be parsed or applied as valid values then the filter will
9935 continue with its existing parameters.
9937 @subsection Examples
9941 Draw "Test Text" with font FreeSerif, using the default values for the
9942 optional parameters.
9945 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
9949 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
9950 and y=50 (counting from the top-left corner of the screen), text is
9951 yellow with a red box around it. Both the text and the box have an
9955 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
9956 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
9959 Note that the double quotes are not necessary if spaces are not used
9960 within the parameter list.
9963 Show the text at the center of the video frame:
9965 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
9969 Show the text at a random position, switching to a new position every 30 seconds:
9971 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)"
9975 Show a text line sliding from right to left in the last row of the video
9976 frame. The file @file{LONG_LINE} is assumed to contain a single line
9979 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
9983 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
9985 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
9989 Draw a single green letter "g", at the center of the input video.
9990 The glyph baseline is placed at half screen height.
9992 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
9996 Show text for 1 second every 3 seconds:
9998 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10002 Use fontconfig to set the font. Note that the colons need to be escaped.
10004 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10008 Print the date of a real-time encoding (see strftime(3)):
10010 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10014 Show text fading in and out (appearing/disappearing):
10017 DS=1.0 # display start
10018 DE=10.0 # display end
10019 FID=1.5 # fade in duration
10020 FOD=5 # fade out duration
10021 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 @}"
10025 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10026 and the @option{fontsize} value are included in the @option{y} offset.
10028 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10029 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10033 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10034 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10035 must have option @option{-export_path_metadata 1} for the special metadata fields
10036 to be available for filters.
10038 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10043 For more information about libfreetype, check:
10044 @url{http://www.freetype.org/}.
10046 For more information about fontconfig, check:
10047 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10049 For more information about libfribidi, check:
10050 @url{http://fribidi.org/}.
10052 @section edgedetect
10054 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10056 The filter accepts the following options:
10061 Set low and high threshold values used by the Canny thresholding
10064 The high threshold selects the "strong" edge pixels, which are then
10065 connected through 8-connectivity with the "weak" edge pixels selected
10066 by the low threshold.
10068 @var{low} and @var{high} threshold values must be chosen in the range
10069 [0,1], and @var{low} should be lesser or equal to @var{high}.
10071 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10075 Define the drawing mode.
10079 Draw white/gray wires on black background.
10082 Mix the colors to create a paint/cartoon effect.
10085 Apply Canny edge detector on all selected planes.
10087 Default value is @var{wires}.
10090 Select planes for filtering. By default all available planes are filtered.
10093 @subsection Examples
10097 Standard edge detection with custom values for the hysteresis thresholding:
10099 edgedetect=low=0.1:high=0.4
10103 Painting effect without thresholding:
10105 edgedetect=mode=colormix:high=0
10111 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10113 For each input image, the filter will compute the optimal mapping from
10114 the input to the output given the codebook length, that is the number
10115 of distinct output colors.
10117 This filter accepts the following options.
10120 @item codebook_length, l
10121 Set codebook length. The value must be a positive integer, and
10122 represents the number of distinct output colors. Default value is 256.
10125 Set the maximum number of iterations to apply for computing the optimal
10126 mapping. The higher the value the better the result and the higher the
10127 computation time. Default value is 1.
10130 Set a random seed, must be an integer included between 0 and
10131 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10132 will try to use a good random seed on a best effort basis.
10135 Set pal8 output pixel format. This option does not work with codebook
10136 length greater than 256.
10141 Measure graylevel entropy in histogram of color channels of video frames.
10143 It accepts the following parameters:
10147 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10149 @var{diff} mode measures entropy of histogram delta values, absolute differences
10150 between neighbour histogram values.
10154 Set brightness, contrast, saturation and approximate gamma adjustment.
10156 The filter accepts the following options:
10160 Set the contrast expression. The value must be a float value in range
10161 @code{-1000.0} to @code{1000.0}. The default value is "1".
10164 Set the brightness expression. The value must be a float value in
10165 range @code{-1.0} to @code{1.0}. The default value is "0".
10168 Set the saturation expression. The value must be a float in
10169 range @code{0.0} to @code{3.0}. The default value is "1".
10172 Set the gamma expression. The value must be a float in range
10173 @code{0.1} to @code{10.0}. The default value is "1".
10176 Set the gamma expression for red. The value must be a float in
10177 range @code{0.1} to @code{10.0}. The default value is "1".
10180 Set the gamma expression for green. The value must be a float in range
10181 @code{0.1} to @code{10.0}. The default value is "1".
10184 Set the gamma expression for blue. The value must be a float in range
10185 @code{0.1} to @code{10.0}. The default value is "1".
10188 Set the gamma weight expression. It can be used to reduce the effect
10189 of a high gamma value on bright image areas, e.g. keep them from
10190 getting overamplified and just plain white. The value must be a float
10191 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10192 gamma correction all the way down while @code{1.0} leaves it at its
10193 full strength. Default is "1".
10196 Set when the expressions for brightness, contrast, saturation and
10197 gamma expressions are evaluated.
10199 It accepts the following values:
10202 only evaluate expressions once during the filter initialization or
10203 when a command is processed
10206 evaluate expressions for each incoming frame
10209 Default value is @samp{init}.
10212 The expressions accept the following parameters:
10215 frame count of the input frame starting from 0
10218 byte position of the corresponding packet in the input file, NAN if
10222 frame rate of the input video, NAN if the input frame rate is unknown
10225 timestamp expressed in seconds, NAN if the input timestamp is unknown
10228 @subsection Commands
10229 The filter supports the following commands:
10233 Set the contrast expression.
10236 Set the brightness expression.
10239 Set the saturation expression.
10242 Set the gamma expression.
10245 Set the gamma_r expression.
10248 Set gamma_g expression.
10251 Set gamma_b expression.
10254 Set gamma_weight expression.
10256 The command accepts the same syntax of the corresponding option.
10258 If the specified expression is not valid, it is kept at its current
10265 Apply erosion effect to the video.
10267 This filter replaces the pixel by the local(3x3) minimum.
10269 It accepts the following options:
10276 Limit the maximum change for each plane, default is 65535.
10277 If 0, plane will remain unchanged.
10280 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10283 Flags to local 3x3 coordinates maps like this:
10290 @subsection Commands
10292 This filter supports the all above options as @ref{commands}.
10294 @section extractplanes
10296 Extract color channel components from input video stream into
10297 separate grayscale video streams.
10299 The filter accepts the following option:
10303 Set plane(s) to extract.
10305 Available values for planes are:
10316 Choosing planes not available in the input will result in an error.
10317 That means you cannot select @code{r}, @code{g}, @code{b} planes
10318 with @code{y}, @code{u}, @code{v} planes at same time.
10321 @subsection Examples
10325 Extract luma, u and v color channel component from input video frame
10326 into 3 grayscale outputs:
10328 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
10334 Apply a fade-in/out effect to the input video.
10336 It accepts the following parameters:
10340 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10342 Default is @code{in}.
10344 @item start_frame, s
10345 Specify the number of the frame to start applying the fade
10346 effect at. Default is 0.
10349 The number of frames that the fade effect lasts. At the end of the
10350 fade-in effect, the output video will have the same intensity as the input video.
10351 At the end of the fade-out transition, the output video will be filled with the
10352 selected @option{color}.
10356 If set to 1, fade only alpha channel, if one exists on the input.
10357 Default value is 0.
10359 @item start_time, st
10360 Specify the timestamp (in seconds) of the frame to start to apply the fade
10361 effect. If both start_frame and start_time are specified, the fade will start at
10362 whichever comes last. Default is 0.
10365 The number of seconds for which the fade effect has to last. At the end of the
10366 fade-in effect the output video will have the same intensity as the input video,
10367 at the end of the fade-out transition the output video will be filled with the
10368 selected @option{color}.
10369 If both duration and nb_frames are specified, duration is used. Default is 0
10370 (nb_frames is used by default).
10373 Specify the color of the fade. Default is "black".
10376 @subsection Examples
10380 Fade in the first 30 frames of video:
10385 The command above is equivalent to:
10391 Fade out the last 45 frames of a 200-frame video:
10394 fade=type=out:start_frame=155:nb_frames=45
10398 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10400 fade=in:0:25, fade=out:975:25
10404 Make the first 5 frames yellow, then fade in from frame 5-24:
10406 fade=in:5:20:color=yellow
10410 Fade in alpha over first 25 frames of video:
10412 fade=in:0:25:alpha=1
10416 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10418 fade=t=in:st=5.5:d=0.5
10424 Denoise frames using 3D FFT (frequency domain filtering).
10426 The filter accepts the following options:
10430 Set the noise sigma constant. This sets denoising strength.
10431 Default value is 1. Allowed range is from 0 to 30.
10432 Using very high sigma with low overlap may give blocking artifacts.
10435 Set amount of denoising. By default all detected noise is reduced.
10436 Default value is 1. Allowed range is from 0 to 1.
10439 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10440 Actual size of block in pixels is 2 to power of @var{block}, so by default
10441 block size in pixels is 2^4 which is 16.
10444 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10447 Set number of previous frames to use for denoising. By default is set to 0.
10450 Set number of next frames to to use for denoising. By default is set to 0.
10453 Set planes which will be filtered, by default are all available filtered
10458 Apply arbitrary expressions to samples in frequency domain
10462 Adjust the dc value (gain) of the luma plane of the image. The filter
10463 accepts an integer value in range @code{0} to @code{1000}. The default
10464 value is set to @code{0}.
10467 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10468 filter accepts an integer value in range @code{0} to @code{1000}. The
10469 default value is set to @code{0}.
10472 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10473 filter accepts an integer value in range @code{0} to @code{1000}. The
10474 default value is set to @code{0}.
10477 Set the frequency domain weight expression for the luma plane.
10480 Set the frequency domain weight expression for the 1st chroma plane.
10483 Set the frequency domain weight expression for the 2nd chroma plane.
10486 Set when the expressions are evaluated.
10488 It accepts the following values:
10491 Only evaluate expressions once during the filter initialization.
10494 Evaluate expressions for each incoming frame.
10497 Default value is @samp{init}.
10499 The filter accepts the following variables:
10502 The coordinates of the current sample.
10506 The width and height of the image.
10509 The number of input frame, starting from 0.
10512 @subsection Examples
10518 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
10524 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
10530 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
10536 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
10543 Extract a single field from an interlaced image using stride
10544 arithmetic to avoid wasting CPU time. The output frames are marked as
10547 The filter accepts the following options:
10551 Specify whether to extract the top (if the value is @code{0} or
10552 @code{top}) or the bottom field (if the value is @code{1} or
10558 Create new frames by copying the top and bottom fields from surrounding frames
10559 supplied as numbers by the hint file.
10563 Set file containing hints: absolute/relative frame numbers.
10565 There must be one line for each frame in a clip. Each line must contain two
10566 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
10567 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
10568 is current frame number for @code{absolute} mode or out of [-1, 1] range
10569 for @code{relative} mode. First number tells from which frame to pick up top
10570 field and second number tells from which frame to pick up bottom field.
10572 If optionally followed by @code{+} output frame will be marked as interlaced,
10573 else if followed by @code{-} output frame will be marked as progressive, else
10574 it will be marked same as input frame.
10575 If optionally followed by @code{t} output frame will use only top field, or in
10576 case of @code{b} it will use only bottom field.
10577 If line starts with @code{#} or @code{;} that line is skipped.
10580 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
10583 Example of first several lines of @code{hint} file for @code{relative} mode:
10585 0,0 - # first frame
10586 1,0 - # second frame, use third's frame top field and second's frame bottom field
10587 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
10602 @section fieldmatch
10604 Field matching filter for inverse telecine. It is meant to reconstruct the
10605 progressive frames from a telecined stream. The filter does not drop duplicated
10606 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
10607 followed by a decimation filter such as @ref{decimate} in the filtergraph.
10609 The separation of the field matching and the decimation is notably motivated by
10610 the possibility of inserting a de-interlacing filter fallback between the two.
10611 If the source has mixed telecined and real interlaced content,
10612 @code{fieldmatch} will not be able to match fields for the interlaced parts.
10613 But these remaining combed frames will be marked as interlaced, and thus can be
10614 de-interlaced by a later filter such as @ref{yadif} before decimation.
10616 In addition to the various configuration options, @code{fieldmatch} can take an
10617 optional second stream, activated through the @option{ppsrc} option. If
10618 enabled, the frames reconstruction will be based on the fields and frames from
10619 this second stream. This allows the first input to be pre-processed in order to
10620 help the various algorithms of the filter, while keeping the output lossless
10621 (assuming the fields are matched properly). Typically, a field-aware denoiser,
10622 or brightness/contrast adjustments can help.
10624 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
10625 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
10626 which @code{fieldmatch} is based on. While the semantic and usage are very
10627 close, some behaviour and options names can differ.
10629 The @ref{decimate} filter currently only works for constant frame rate input.
10630 If your input has mixed telecined (30fps) and progressive content with a lower
10631 framerate like 24fps use the following filterchain to produce the necessary cfr
10632 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
10634 The filter accepts the following options:
10638 Specify the assumed field order of the input stream. Available values are:
10642 Auto detect parity (use FFmpeg's internal parity value).
10644 Assume bottom field first.
10646 Assume top field first.
10649 Note that it is sometimes recommended not to trust the parity announced by the
10652 Default value is @var{auto}.
10655 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
10656 sense that it won't risk creating jerkiness due to duplicate frames when
10657 possible, but if there are bad edits or blended fields it will end up
10658 outputting combed frames when a good match might actually exist. On the other
10659 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
10660 but will almost always find a good frame if there is one. The other values are
10661 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
10662 jerkiness and creating duplicate frames versus finding good matches in sections
10663 with bad edits, orphaned fields, blended fields, etc.
10665 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
10667 Available values are:
10671 2-way matching (p/c)
10673 2-way matching, and trying 3rd match if still combed (p/c + n)
10675 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
10677 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
10678 still combed (p/c + n + u/b)
10680 3-way matching (p/c/n)
10682 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
10683 detected as combed (p/c/n + u/b)
10686 The parenthesis at the end indicate the matches that would be used for that
10687 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
10690 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
10693 Default value is @var{pc_n}.
10696 Mark the main input stream as a pre-processed input, and enable the secondary
10697 input stream as the clean source to pick the fields from. See the filter
10698 introduction for more details. It is similar to the @option{clip2} feature from
10701 Default value is @code{0} (disabled).
10704 Set the field to match from. It is recommended to set this to the same value as
10705 @option{order} unless you experience matching failures with that setting. In
10706 certain circumstances changing the field that is used to match from can have a
10707 large impact on matching performance. Available values are:
10711 Automatic (same value as @option{order}).
10713 Match from the bottom field.
10715 Match from the top field.
10718 Default value is @var{auto}.
10721 Set whether or not chroma is included during the match comparisons. In most
10722 cases it is recommended to leave this enabled. You should set this to @code{0}
10723 only if your clip has bad chroma problems such as heavy rainbowing or other
10724 artifacts. Setting this to @code{0} could also be used to speed things up at
10725 the cost of some accuracy.
10727 Default value is @code{1}.
10731 These define an exclusion band which excludes the lines between @option{y0} and
10732 @option{y1} from being included in the field matching decision. An exclusion
10733 band can be used to ignore subtitles, a logo, or other things that may
10734 interfere with the matching. @option{y0} sets the starting scan line and
10735 @option{y1} sets the ending line; all lines in between @option{y0} and
10736 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
10737 @option{y0} and @option{y1} to the same value will disable the feature.
10738 @option{y0} and @option{y1} defaults to @code{0}.
10741 Set the scene change detection threshold as a percentage of maximum change on
10742 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
10743 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
10744 @option{scthresh} is @code{[0.0, 100.0]}.
10746 Default value is @code{12.0}.
10749 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
10750 account the combed scores of matches when deciding what match to use as the
10751 final match. Available values are:
10755 No final matching based on combed scores.
10757 Combed scores are only used when a scene change is detected.
10759 Use combed scores all the time.
10762 Default is @var{sc}.
10765 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
10766 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
10767 Available values are:
10771 No forced calculation.
10773 Force p/c/n calculations.
10775 Force p/c/n/u/b calculations.
10778 Default value is @var{none}.
10781 This is the area combing threshold used for combed frame detection. This
10782 essentially controls how "strong" or "visible" combing must be to be detected.
10783 Larger values mean combing must be more visible and smaller values mean combing
10784 can be less visible or strong and still be detected. Valid settings are from
10785 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
10786 be detected as combed). This is basically a pixel difference value. A good
10787 range is @code{[8, 12]}.
10789 Default value is @code{9}.
10792 Sets whether or not chroma is considered in the combed frame decision. Only
10793 disable this if your source has chroma problems (rainbowing, etc.) that are
10794 causing problems for the combed frame detection with chroma enabled. Actually,
10795 using @option{chroma}=@var{0} is usually more reliable, except for the case
10796 where there is chroma only combing in the source.
10798 Default value is @code{0}.
10802 Respectively set the x-axis and y-axis size of the window used during combed
10803 frame detection. This has to do with the size of the area in which
10804 @option{combpel} pixels are required to be detected as combed for a frame to be
10805 declared combed. See the @option{combpel} parameter description for more info.
10806 Possible values are any number that is a power of 2 starting at 4 and going up
10809 Default value is @code{16}.
10812 The number of combed pixels inside any of the @option{blocky} by
10813 @option{blockx} size blocks on the frame for the frame to be detected as
10814 combed. While @option{cthresh} controls how "visible" the combing must be, this
10815 setting controls "how much" combing there must be in any localized area (a
10816 window defined by the @option{blockx} and @option{blocky} settings) on the
10817 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
10818 which point no frames will ever be detected as combed). This setting is known
10819 as @option{MI} in TFM/VFM vocabulary.
10821 Default value is @code{80}.
10824 @anchor{p/c/n/u/b meaning}
10825 @subsection p/c/n/u/b meaning
10827 @subsubsection p/c/n
10829 We assume the following telecined stream:
10832 Top fields: 1 2 2 3 4
10833 Bottom fields: 1 2 3 4 4
10836 The numbers correspond to the progressive frame the fields relate to. Here, the
10837 first two frames are progressive, the 3rd and 4th are combed, and so on.
10839 When @code{fieldmatch} is configured to run a matching from bottom
10840 (@option{field}=@var{bottom}) this is how this input stream get transformed:
10845 B 1 2 3 4 4 <-- matching reference
10854 As a result of the field matching, we can see that some frames get duplicated.
10855 To perform a complete inverse telecine, you need to rely on a decimation filter
10856 after this operation. See for instance the @ref{decimate} filter.
10858 The same operation now matching from top fields (@option{field}=@var{top})
10863 T 1 2 2 3 4 <-- matching reference
10873 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
10874 basically, they refer to the frame and field of the opposite parity:
10877 @item @var{p} matches the field of the opposite parity in the previous frame
10878 @item @var{c} matches the field of the opposite parity in the current frame
10879 @item @var{n} matches the field of the opposite parity in the next frame
10884 The @var{u} and @var{b} matching are a bit special in the sense that they match
10885 from the opposite parity flag. In the following examples, we assume that we are
10886 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
10887 'x' is placed above and below each matched fields.
10889 With bottom matching (@option{field}=@var{bottom}):
10894 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10895 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10903 With top matching (@option{field}=@var{top}):
10908 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10909 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10917 @subsection Examples
10919 Simple IVTC of a top field first telecined stream:
10921 fieldmatch=order=tff:combmatch=none, decimate
10924 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
10926 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
10929 @section fieldorder
10931 Transform the field order of the input video.
10933 It accepts the following parameters:
10938 The output field order. Valid values are @var{tff} for top field first or @var{bff}
10939 for bottom field first.
10942 The default value is @samp{tff}.
10944 The transformation is done by shifting the picture content up or down
10945 by one line, and filling the remaining line with appropriate picture content.
10946 This method is consistent with most broadcast field order converters.
10948 If the input video is not flagged as being interlaced, or it is already
10949 flagged as being of the required output field order, then this filter does
10950 not alter the incoming video.
10952 It is very useful when converting to or from PAL DV material,
10953 which is bottom field first.
10957 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
10960 @section fifo, afifo
10962 Buffer input images and send them when they are requested.
10964 It is mainly useful when auto-inserted by the libavfilter
10967 It does not take parameters.
10969 @section fillborders
10971 Fill borders of the input video, without changing video stream dimensions.
10972 Sometimes video can have garbage at the four edges and you may not want to
10973 crop video input to keep size multiple of some number.
10975 This filter accepts the following options:
10979 Number of pixels to fill from left border.
10982 Number of pixels to fill from right border.
10985 Number of pixels to fill from top border.
10988 Number of pixels to fill from bottom border.
10993 It accepts the following values:
10996 fill pixels using outermost pixels
10999 fill pixels using mirroring
11002 fill pixels with constant value
11005 Default is @var{smear}.
11008 Set color for pixels in fixed mode. Default is @var{black}.
11011 @subsection Commands
11012 This filter supports same @ref{commands} as options.
11013 The command accepts the same syntax of the corresponding option.
11015 If the specified expression is not valid, it is kept at its current
11020 Find a rectangular object
11022 It accepts the following options:
11026 Filepath of the object image, needs to be in gray8.
11029 Detection threshold, default is 0.5.
11032 Number of mipmaps, default is 3.
11034 @item xmin, ymin, xmax, ymax
11035 Specifies the rectangle in which to search.
11038 @subsection Examples
11042 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11044 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11050 Flood area with values of same pixel components with another values.
11052 It accepts the following options:
11055 Set pixel x coordinate.
11058 Set pixel y coordinate.
11061 Set source #0 component value.
11064 Set source #1 component value.
11067 Set source #2 component value.
11070 Set source #3 component value.
11073 Set destination #0 component value.
11076 Set destination #1 component value.
11079 Set destination #2 component value.
11082 Set destination #3 component value.
11088 Convert the input video to one of the specified pixel formats.
11089 Libavfilter will try to pick one that is suitable as input to
11092 It accepts the following parameters:
11096 A '|'-separated list of pixel format names, such as
11097 "pix_fmts=yuv420p|monow|rgb24".
11101 @subsection Examples
11105 Convert the input video to the @var{yuv420p} format
11107 format=pix_fmts=yuv420p
11110 Convert the input video to any of the formats in the list
11112 format=pix_fmts=yuv420p|yuv444p|yuv410p
11119 Convert the video to specified constant frame rate by duplicating or dropping
11120 frames as necessary.
11122 It accepts the following parameters:
11126 The desired output frame rate. The default is @code{25}.
11129 Assume the first PTS should be the given value, in seconds. This allows for
11130 padding/trimming at the start of stream. By default, no assumption is made
11131 about the first frame's expected PTS, so no padding or trimming is done.
11132 For example, this could be set to 0 to pad the beginning with duplicates of
11133 the first frame if a video stream starts after the audio stream or to trim any
11134 frames with a negative PTS.
11137 Timestamp (PTS) rounding method.
11139 Possible values are:
11146 round towards -infinity
11148 round towards +infinity
11152 The default is @code{near}.
11155 Action performed when reading the last frame.
11157 Possible values are:
11160 Use same timestamp rounding method as used for other frames.
11162 Pass through last frame if input duration has not been reached yet.
11164 The default is @code{round}.
11168 Alternatively, the options can be specified as a flat string:
11169 @var{fps}[:@var{start_time}[:@var{round}]].
11171 See also the @ref{setpts} filter.
11173 @subsection Examples
11177 A typical usage in order to set the fps to 25:
11183 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11185 fps=fps=film:round=near
11191 Pack two different video streams into a stereoscopic video, setting proper
11192 metadata on supported codecs. The two views should have the same size and
11193 framerate and processing will stop when the shorter video ends. Please note
11194 that you may conveniently adjust view properties with the @ref{scale} and
11197 It accepts the following parameters:
11201 The desired packing format. Supported values are:
11206 The views are next to each other (default).
11209 The views are on top of each other.
11212 The views are packed by line.
11215 The views are packed by column.
11218 The views are temporally interleaved.
11227 # Convert left and right views into a frame-sequential video
11228 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11230 # Convert views into a side-by-side video with the same output resolution as the input
11231 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
11236 Change the frame rate by interpolating new video output frames from the source
11239 This filter is not designed to function correctly with interlaced media. If
11240 you wish to change the frame rate of interlaced media then you are required
11241 to deinterlace before this filter and re-interlace after this filter.
11243 A description of the accepted options follows.
11247 Specify the output frames per second. This option can also be specified
11248 as a value alone. The default is @code{50}.
11251 Specify the start of a range where the output frame will be created as a
11252 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11253 the default is @code{15}.
11256 Specify the end of a range where the output frame will be created as a
11257 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11258 the default is @code{240}.
11261 Specify the level at which a scene change is detected as a value between
11262 0 and 100 to indicate a new scene; a low value reflects a low
11263 probability for the current frame to introduce a new scene, while a higher
11264 value means the current frame is more likely to be one.
11265 The default is @code{8.2}.
11268 Specify flags influencing the filter process.
11270 Available value for @var{flags} is:
11273 @item scene_change_detect, scd
11274 Enable scene change detection using the value of the option @var{scene}.
11275 This flag is enabled by default.
11281 Select one frame every N-th frame.
11283 This filter accepts the following option:
11286 Select frame after every @code{step} frames.
11287 Allowed values are positive integers higher than 0. Default value is @code{1}.
11290 @section freezedetect
11292 Detect frozen video.
11294 This filter logs a message and sets frame metadata when it detects that the
11295 input video has no significant change in content during a specified duration.
11296 Video freeze detection calculates the mean average absolute difference of all
11297 the components of video frames and compares it to a noise floor.
11299 The printed times and duration are expressed in seconds. The
11300 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11301 whose timestamp equals or exceeds the detection duration and it contains the
11302 timestamp of the first frame of the freeze. The
11303 @code{lavfi.freezedetect.freeze_duration} and
11304 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11307 The filter accepts the following options:
11311 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11312 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11316 Set freeze duration until notification (default is 2 seconds).
11319 @section freezeframes
11321 Freeze video frames.
11323 This filter freezes video frames using frame from 2nd input.
11325 The filter accepts the following options:
11329 Set number of first frame from which to start freeze.
11332 Set number of last frame from which to end freeze.
11335 Set number of frame from 2nd input which will be used instead of replaced frames.
11341 Apply a frei0r effect to the input video.
11343 To enable the compilation of this filter, you need to install the frei0r
11344 header and configure FFmpeg with @code{--enable-frei0r}.
11346 It accepts the following parameters:
11351 The name of the frei0r effect to load. If the environment variable
11352 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11353 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11354 Otherwise, the standard frei0r paths are searched, in this order:
11355 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11356 @file{/usr/lib/frei0r-1/}.
11358 @item filter_params
11359 A '|'-separated list of parameters to pass to the frei0r effect.
11363 A frei0r effect parameter can be a boolean (its value is either
11364 "y" or "n"), a double, a color (specified as
11365 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11366 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11367 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11368 a position (specified as @var{X}/@var{Y}, where
11369 @var{X} and @var{Y} are floating point numbers) and/or a string.
11371 The number and types of parameters depend on the loaded effect. If an
11372 effect parameter is not specified, the default value is set.
11374 @subsection Examples
11378 Apply the distort0r effect, setting the first two double parameters:
11380 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11384 Apply the colordistance effect, taking a color as the first parameter:
11386 frei0r=colordistance:0.2/0.3/0.4
11387 frei0r=colordistance:violet
11388 frei0r=colordistance:0x112233
11392 Apply the perspective effect, specifying the top left and top right image
11395 frei0r=perspective:0.2/0.2|0.8/0.2
11399 For more information, see
11400 @url{http://frei0r.dyne.org}
11404 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11406 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11407 processing filter, one of them is performed once per block, not per pixel.
11408 This allows for much higher speed.
11410 The filter accepts the following options:
11414 Set quality. This option defines the number of levels for averaging. It accepts
11415 an integer in the range 4-5. Default value is @code{4}.
11418 Force a constant quantization parameter. It accepts an integer in range 0-63.
11419 If not set, the filter will use the QP from the video stream (if available).
11422 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11423 more details but also more artifacts, while higher values make the image smoother
11424 but also blurrier. Default value is @code{0} − PSNR optimal.
11426 @item use_bframe_qp
11427 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11428 option may cause flicker since the B-Frames have often larger QP. Default is
11429 @code{0} (not enabled).
11435 Apply Gaussian blur filter.
11437 The filter accepts the following options:
11441 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11444 Set number of steps for Gaussian approximation. Default is @code{1}.
11447 Set which planes to filter. By default all planes are filtered.
11450 Set vertical sigma, if negative it will be same as @code{sigma}.
11451 Default is @code{-1}.
11454 @subsection Commands
11455 This filter supports same commands as options.
11456 The command accepts the same syntax of the corresponding option.
11458 If the specified expression is not valid, it is kept at its current
11463 Apply generic equation to each pixel.
11465 The filter accepts the following options:
11468 @item lum_expr, lum
11469 Set the luminance expression.
11471 Set the chrominance blue expression.
11473 Set the chrominance red expression.
11474 @item alpha_expr, a
11475 Set the alpha expression.
11477 Set the red expression.
11478 @item green_expr, g
11479 Set the green expression.
11481 Set the blue expression.
11484 The colorspace is selected according to the specified options. If one
11485 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11486 options is specified, the filter will automatically select a YCbCr
11487 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11488 @option{blue_expr} options is specified, it will select an RGB
11491 If one of the chrominance expression is not defined, it falls back on the other
11492 one. If no alpha expression is specified it will evaluate to opaque value.
11493 If none of chrominance expressions are specified, they will evaluate
11494 to the luminance expression.
11496 The expressions can use the following variables and functions:
11500 The sequential number of the filtered frame, starting from @code{0}.
11504 The coordinates of the current sample.
11508 The width and height of the image.
11512 Width and height scale depending on the currently filtered plane. It is the
11513 ratio between the corresponding luma plane number of pixels and the current
11514 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11515 @code{0.5,0.5} for chroma planes.
11518 Time of the current frame, expressed in seconds.
11521 Return the value of the pixel at location (@var{x},@var{y}) of the current
11525 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
11529 Return the value of the pixel at location (@var{x},@var{y}) of the
11530 blue-difference chroma plane. Return 0 if there is no such plane.
11533 Return the value of the pixel at location (@var{x},@var{y}) of the
11534 red-difference chroma plane. Return 0 if there is no such plane.
11539 Return the value of the pixel at location (@var{x},@var{y}) of the
11540 red/green/blue component. Return 0 if there is no such component.
11543 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
11544 plane. Return 0 if there is no such plane.
11546 @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)
11547 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
11548 sums of samples within a rectangle. See the functions without the sum postfix.
11550 @item interpolation
11551 Set one of interpolation methods:
11556 Default is bilinear.
11559 For functions, if @var{x} and @var{y} are outside the area, the value will be
11560 automatically clipped to the closer edge.
11562 Please note that this filter can use multiple threads in which case each slice
11563 will have its own expression state. If you want to use only a single expression
11564 state because your expressions depend on previous state then you should limit
11565 the number of filter threads to 1.
11567 @subsection Examples
11571 Flip the image horizontally:
11577 Generate a bidimensional sine wave, with angle @code{PI/3} and a
11578 wavelength of 100 pixels:
11580 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
11584 Generate a fancy enigmatic moving light:
11586 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
11590 Generate a quick emboss effect:
11592 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
11596 Modify RGB components depending on pixel position:
11598 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
11602 Create a radial gradient that is the same size as the input (also see
11603 the @ref{vignette} filter):
11605 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
11611 Fix the banding artifacts that are sometimes introduced into nearly flat
11612 regions by truncation to 8-bit color depth.
11613 Interpolate the gradients that should go where the bands are, and
11616 It is designed for playback only. Do not use it prior to
11617 lossy compression, because compression tends to lose the dither and
11618 bring back the bands.
11620 It accepts the following parameters:
11625 The maximum amount by which the filter will change any one pixel. This is also
11626 the threshold for detecting nearly flat regions. Acceptable values range from
11627 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
11631 The neighborhood to fit the gradient to. A larger radius makes for smoother
11632 gradients, but also prevents the filter from modifying the pixels near detailed
11633 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
11634 values will be clipped to the valid range.
11638 Alternatively, the options can be specified as a flat string:
11639 @var{strength}[:@var{radius}]
11641 @subsection Examples
11645 Apply the filter with a @code{3.5} strength and radius of @code{8}:
11651 Specify radius, omitting the strength (which will fall-back to the default
11659 @anchor{graphmonitor}
11660 @section graphmonitor
11661 Show various filtergraph stats.
11663 With this filter one can debug complete filtergraph.
11664 Especially issues with links filling with queued frames.
11666 The filter accepts the following options:
11670 Set video output size. Default is @var{hd720}.
11673 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
11676 Set output mode, can be @var{fulll} or @var{compact}.
11677 In @var{compact} mode only filters with some queued frames have displayed stats.
11680 Set flags which enable which stats are shown in video.
11682 Available values for flags are:
11685 Display number of queued frames in each link.
11687 @item frame_count_in
11688 Display number of frames taken from filter.
11690 @item frame_count_out
11691 Display number of frames given out from filter.
11694 Display current filtered frame pts.
11697 Display current filtered frame time.
11700 Display time base for filter link.
11703 Display used format for filter link.
11706 Display video size or number of audio channels in case of audio used by filter link.
11709 Display video frame rate or sample rate in case of audio used by filter link.
11713 Set upper limit for video rate of output stream, Default value is @var{25}.
11714 This guarantee that output video frame rate will not be higher than this value.
11718 A color constancy variation filter which estimates scene illumination via grey edge algorithm
11719 and corrects the scene colors accordingly.
11721 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
11723 The filter accepts the following options:
11727 The order of differentiation to be applied on the scene. Must be chosen in the range
11728 [0,2] and default value is 1.
11731 The Minkowski parameter to be used for calculating the Minkowski distance. Must
11732 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
11733 max value instead of calculating Minkowski distance.
11736 The standard deviation of Gaussian blur to be applied on the scene. Must be
11737 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
11738 can't be equal to 0 if @var{difford} is greater than 0.
11741 @subsection Examples
11747 greyedge=difford=1:minknorm=5:sigma=2
11753 greyedge=difford=1:minknorm=0:sigma=2
11761 Apply a Hald CLUT to a video stream.
11763 First input is the video stream to process, and second one is the Hald CLUT.
11764 The Hald CLUT input can be a simple picture or a complete video stream.
11766 The filter accepts the following options:
11770 Force termination when the shortest input terminates. Default is @code{0}.
11772 Continue applying the last CLUT after the end of the stream. A value of
11773 @code{0} disable the filter after the last frame of the CLUT is reached.
11774 Default is @code{1}.
11777 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
11778 filters share the same internals).
11780 This filter also supports the @ref{framesync} options.
11782 More information about the Hald CLUT can be found on Eskil Steenberg's website
11783 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
11785 @subsection Workflow examples
11787 @subsubsection Hald CLUT video stream
11789 Generate an identity Hald CLUT stream altered with various effects:
11791 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
11794 Note: make sure you use a lossless codec.
11796 Then use it with @code{haldclut} to apply it on some random stream:
11798 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
11801 The Hald CLUT will be applied to the 10 first seconds (duration of
11802 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
11803 to the remaining frames of the @code{mandelbrot} stream.
11805 @subsubsection Hald CLUT with preview
11807 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
11808 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
11809 biggest possible square starting at the top left of the picture. The remaining
11810 padding pixels (bottom or right) will be ignored. This area can be used to add
11811 a preview of the Hald CLUT.
11813 Typically, the following generated Hald CLUT will be supported by the
11814 @code{haldclut} filter:
11817 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
11818 pad=iw+320 [padded_clut];
11819 smptebars=s=320x256, split [a][b];
11820 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
11821 [main][b] overlay=W-320" -frames:v 1 clut.png
11824 It contains the original and a preview of the effect of the CLUT: SMPTE color
11825 bars are displayed on the right-top, and below the same color bars processed by
11828 Then, the effect of this Hald CLUT can be visualized with:
11830 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
11835 Flip the input video horizontally.
11837 For example, to horizontally flip the input video with @command{ffmpeg}:
11839 ffmpeg -i in.avi -vf "hflip" out.avi
11843 This filter applies a global color histogram equalization on a
11846 It can be used to correct video that has a compressed range of pixel
11847 intensities. The filter redistributes the pixel intensities to
11848 equalize their distribution across the intensity range. It may be
11849 viewed as an "automatically adjusting contrast filter". This filter is
11850 useful only for correcting degraded or poorly captured source
11853 The filter accepts the following options:
11857 Determine the amount of equalization to be applied. As the strength
11858 is reduced, the distribution of pixel intensities more-and-more
11859 approaches that of the input frame. The value must be a float number
11860 in the range [0,1] and defaults to 0.200.
11863 Set the maximum intensity that can generated and scale the output
11864 values appropriately. The strength should be set as desired and then
11865 the intensity can be limited if needed to avoid washing-out. The value
11866 must be a float number in the range [0,1] and defaults to 0.210.
11869 Set the antibanding level. If enabled the filter will randomly vary
11870 the luminance of output pixels by a small amount to avoid banding of
11871 the histogram. Possible values are @code{none}, @code{weak} or
11872 @code{strong}. It defaults to @code{none}.
11878 Compute and draw a color distribution histogram for the input video.
11880 The computed histogram is a representation of the color component
11881 distribution in an image.
11883 Standard histogram displays the color components distribution in an image.
11884 Displays color graph for each color component. Shows distribution of
11885 the Y, U, V, A or R, G, B components, depending on input format, in the
11886 current frame. Below each graph a color component scale meter is shown.
11888 The filter accepts the following options:
11892 Set height of level. Default value is @code{200}.
11893 Allowed range is [50, 2048].
11896 Set height of color scale. Default value is @code{12}.
11897 Allowed range is [0, 40].
11901 It accepts the following values:
11904 Per color component graphs are placed below each other.
11907 Per color component graphs are placed side by side.
11910 Presents information identical to that in the @code{parade}, except
11911 that the graphs representing color components are superimposed directly
11914 Default is @code{stack}.
11917 Set mode. Can be either @code{linear}, or @code{logarithmic}.
11918 Default is @code{linear}.
11921 Set what color components to display.
11922 Default is @code{7}.
11925 Set foreground opacity. Default is @code{0.7}.
11928 Set background opacity. Default is @code{0.5}.
11931 @subsection Examples
11936 Calculate and draw histogram:
11938 ffplay -i input -vf histogram
11946 This is a high precision/quality 3d denoise filter. It aims to reduce
11947 image noise, producing smooth images and making still images really
11948 still. It should enhance compressibility.
11950 It accepts the following optional parameters:
11954 A non-negative floating point number which specifies spatial luma strength.
11955 It defaults to 4.0.
11957 @item chroma_spatial
11958 A non-negative floating point number which specifies spatial chroma strength.
11959 It defaults to 3.0*@var{luma_spatial}/4.0.
11962 A floating point number which specifies luma temporal strength. It defaults to
11963 6.0*@var{luma_spatial}/4.0.
11966 A floating point number which specifies chroma temporal strength. It defaults to
11967 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
11970 @subsection Commands
11971 This filter supports same @ref{commands} as options.
11972 The command accepts the same syntax of the corresponding option.
11974 If the specified expression is not valid, it is kept at its current
11977 @anchor{hwdownload}
11978 @section hwdownload
11980 Download hardware frames to system memory.
11982 The input must be in hardware frames, and the output a non-hardware format.
11983 Not all formats will be supported on the output - it may be necessary to insert
11984 an additional @option{format} filter immediately following in the graph to get
11985 the output in a supported format.
11989 Map hardware frames to system memory or to another device.
11991 This filter has several different modes of operation; which one is used depends
11992 on the input and output formats:
11995 Hardware frame input, normal frame output
11997 Map the input frames to system memory and pass them to the output. If the
11998 original hardware frame is later required (for example, after overlaying
11999 something else on part of it), the @option{hwmap} filter can be used again
12000 in the next mode to retrieve it.
12002 Normal frame input, hardware frame output
12004 If the input is actually a software-mapped hardware frame, then unmap it -
12005 that is, return the original hardware frame.
12007 Otherwise, a device must be provided. Create new hardware surfaces on that
12008 device for the output, then map them back to the software format at the input
12009 and give those frames to the preceding filter. This will then act like the
12010 @option{hwupload} filter, but may be able to avoid an additional copy when
12011 the input is already in a compatible format.
12013 Hardware frame input and output
12015 A device must be supplied for the output, either directly or with the
12016 @option{derive_device} option. The input and output devices must be of
12017 different types and compatible - the exact meaning of this is
12018 system-dependent, but typically it means that they must refer to the same
12019 underlying hardware context (for example, refer to the same graphics card).
12021 If the input frames were originally created on the output device, then unmap
12022 to retrieve the original frames.
12024 Otherwise, map the frames to the output device - create new hardware frames
12025 on the output corresponding to the frames on the input.
12028 The following additional parameters are accepted:
12032 Set the frame mapping mode. Some combination of:
12035 The mapped frame should be readable.
12037 The mapped frame should be writeable.
12039 The mapping will always overwrite the entire frame.
12041 This may improve performance in some cases, as the original contents of the
12042 frame need not be loaded.
12044 The mapping must not involve any copying.
12046 Indirect mappings to copies of frames are created in some cases where either
12047 direct mapping is not possible or it would have unexpected properties.
12048 Setting this flag ensures that the mapping is direct and will fail if that is
12051 Defaults to @var{read+write} if not specified.
12053 @item derive_device @var{type}
12054 Rather than using the device supplied at initialisation, instead derive a new
12055 device of type @var{type} from the device the input frames exist on.
12058 In a hardware to hardware mapping, map in reverse - create frames in the sink
12059 and map them back to the source. This may be necessary in some cases where
12060 a mapping in one direction is required but only the opposite direction is
12061 supported by the devices being used.
12063 This option is dangerous - it may break the preceding filter in undefined
12064 ways if there are any additional constraints on that filter's output.
12065 Do not use it without fully understanding the implications of its use.
12071 Upload system memory frames to hardware surfaces.
12073 The device to upload to must be supplied when the filter is initialised. If
12074 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12075 option or with the @option{derive_device} option. The input and output devices
12076 must be of different types and compatible - the exact meaning of this is
12077 system-dependent, but typically it means that they must refer to the same
12078 underlying hardware context (for example, refer to the same graphics card).
12080 The following additional parameters are accepted:
12083 @item derive_device @var{type}
12084 Rather than using the device supplied at initialisation, instead derive a new
12085 device of type @var{type} from the device the input frames exist on.
12088 @anchor{hwupload_cuda}
12089 @section hwupload_cuda
12091 Upload system memory frames to a CUDA device.
12093 It accepts the following optional parameters:
12097 The number of the CUDA device to use
12102 Apply a high-quality magnification filter designed for pixel art. This filter
12103 was originally created by Maxim Stepin.
12105 It accepts the following option:
12109 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12110 @code{hq3x} and @code{4} for @code{hq4x}.
12111 Default is @code{3}.
12115 Stack input videos horizontally.
12117 All streams must be of same pixel format and of same height.
12119 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12120 to create same output.
12122 The filter accepts the following option:
12126 Set number of input streams. Default is 2.
12129 If set to 1, force the output to terminate when the shortest input
12130 terminates. Default value is 0.
12135 Modify the hue and/or the saturation of the input.
12137 It accepts the following parameters:
12141 Specify the hue angle as a number of degrees. It accepts an expression,
12142 and defaults to "0".
12145 Specify the saturation in the [-10,10] range. It accepts an expression and
12149 Specify the hue angle as a number of radians. It accepts an
12150 expression, and defaults to "0".
12153 Specify the brightness in the [-10,10] range. It accepts an expression and
12157 @option{h} and @option{H} are mutually exclusive, and can't be
12158 specified at the same time.
12160 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12161 expressions containing the following constants:
12165 frame count of the input frame starting from 0
12168 presentation timestamp of the input frame expressed in time base units
12171 frame rate of the input video, NAN if the input frame rate is unknown
12174 timestamp expressed in seconds, NAN if the input timestamp is unknown
12177 time base of the input video
12180 @subsection Examples
12184 Set the hue to 90 degrees and the saturation to 1.0:
12190 Same command but expressing the hue in radians:
12196 Rotate hue and make the saturation swing between 0
12197 and 2 over a period of 1 second:
12199 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12203 Apply a 3 seconds saturation fade-in effect starting at 0:
12205 hue="s=min(t/3\,1)"
12208 The general fade-in expression can be written as:
12210 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12214 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12216 hue="s=max(0\, min(1\, (8-t)/3))"
12219 The general fade-out expression can be written as:
12221 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12226 @subsection Commands
12228 This filter supports the following commands:
12234 Modify the hue and/or the saturation and/or brightness of the input video.
12235 The command accepts the same syntax of the corresponding option.
12237 If the specified expression is not valid, it is kept at its current
12241 @section hysteresis
12243 Grow first stream into second stream by connecting components.
12244 This makes it possible to build more robust edge masks.
12246 This filter accepts the following options:
12250 Set which planes will be processed as bitmap, unprocessed planes will be
12251 copied from first stream.
12252 By default value 0xf, all planes will be processed.
12255 Set threshold which is used in filtering. If pixel component value is higher than
12256 this value filter algorithm for connecting components is activated.
12257 By default value is 0.
12260 The @code{hysteresis} filter also supports the @ref{framesync} options.
12264 Detect video interlacing type.
12266 This filter tries to detect if the input frames are interlaced, progressive,
12267 top or bottom field first. It will also try to detect fields that are
12268 repeated between adjacent frames (a sign of telecine).
12270 Single frame detection considers only immediately adjacent frames when classifying each frame.
12271 Multiple frame detection incorporates the classification history of previous frames.
12273 The filter will log these metadata values:
12276 @item single.current_frame
12277 Detected type of current frame using single-frame detection. One of:
12278 ``tff'' (top field first), ``bff'' (bottom field first),
12279 ``progressive'', or ``undetermined''
12282 Cumulative number of frames detected as top field first using single-frame detection.
12285 Cumulative number of frames detected as top field first using multiple-frame detection.
12288 Cumulative number of frames detected as bottom field first using single-frame detection.
12290 @item multiple.current_frame
12291 Detected type of current frame using multiple-frame detection. One of:
12292 ``tff'' (top field first), ``bff'' (bottom field first),
12293 ``progressive'', or ``undetermined''
12296 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12298 @item single.progressive
12299 Cumulative number of frames detected as progressive using single-frame detection.
12301 @item multiple.progressive
12302 Cumulative number of frames detected as progressive using multiple-frame detection.
12304 @item single.undetermined
12305 Cumulative number of frames that could not be classified using single-frame detection.
12307 @item multiple.undetermined
12308 Cumulative number of frames that could not be classified using multiple-frame detection.
12310 @item repeated.current_frame
12311 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12313 @item repeated.neither
12314 Cumulative number of frames with no repeated field.
12317 Cumulative number of frames with the top field repeated from the previous frame's top field.
12319 @item repeated.bottom
12320 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12323 The filter accepts the following options:
12327 Set interlacing threshold.
12329 Set progressive threshold.
12331 Threshold for repeated field detection.
12333 Number of frames after which a given frame's contribution to the
12334 statistics is halved (i.e., it contributes only 0.5 to its
12335 classification). The default of 0 means that all frames seen are given
12336 full weight of 1.0 forever.
12337 @item analyze_interlaced_flag
12338 When this is not 0 then idet will use the specified number of frames to determine
12339 if the interlaced flag is accurate, it will not count undetermined frames.
12340 If the flag is found to be accurate it will be used without any further
12341 computations, if it is found to be inaccurate it will be cleared without any
12342 further computations. This allows inserting the idet filter as a low computational
12343 method to clean up the interlaced flag
12348 Deinterleave or interleave fields.
12350 This filter allows one to process interlaced images fields without
12351 deinterlacing them. Deinterleaving splits the input frame into 2
12352 fields (so called half pictures). Odd lines are moved to the top
12353 half of the output image, even lines to the bottom half.
12354 You can process (filter) them independently and then re-interleave them.
12356 The filter accepts the following options:
12360 @item chroma_mode, c
12361 @item alpha_mode, a
12362 Available values for @var{luma_mode}, @var{chroma_mode} and
12363 @var{alpha_mode} are:
12369 @item deinterleave, d
12370 Deinterleave fields, placing one above the other.
12372 @item interleave, i
12373 Interleave fields. Reverse the effect of deinterleaving.
12375 Default value is @code{none}.
12377 @item luma_swap, ls
12378 @item chroma_swap, cs
12379 @item alpha_swap, as
12380 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12383 @subsection Commands
12385 This filter supports the all above options as @ref{commands}.
12389 Apply inflate effect to the video.
12391 This filter replaces the pixel by the local(3x3) average by taking into account
12392 only values higher than the pixel.
12394 It accepts the following options:
12401 Limit the maximum change for each plane, default is 65535.
12402 If 0, plane will remain unchanged.
12405 @subsection Commands
12407 This filter supports the all above options as @ref{commands}.
12411 Simple interlacing filter from progressive contents. This interleaves upper (or
12412 lower) lines from odd frames with lower (or upper) lines from even frames,
12413 halving the frame rate and preserving image height.
12416 Original Original New Frame
12417 Frame 'j' Frame 'j+1' (tff)
12418 ========== =========== ==================
12419 Line 0 --------------------> Frame 'j' Line 0
12420 Line 1 Line 1 ----> Frame 'j+1' Line 1
12421 Line 2 ---------------------> Frame 'j' Line 2
12422 Line 3 Line 3 ----> Frame 'j+1' Line 3
12424 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12427 It accepts the following optional parameters:
12431 This determines whether the interlaced frame is taken from the even
12432 (tff - default) or odd (bff) lines of the progressive frame.
12435 Vertical lowpass filter to avoid twitter interlacing and
12436 reduce moire patterns.
12440 Disable vertical lowpass filter
12443 Enable linear filter (default)
12446 Enable complex filter. This will slightly less reduce twitter and moire
12447 but better retain detail and subjective sharpness impression.
12454 Deinterlace input video by applying Donald Graft's adaptive kernel
12455 deinterling. Work on interlaced parts of a video to produce
12456 progressive frames.
12458 The description of the accepted parameters follows.
12462 Set the threshold which affects the filter's tolerance when
12463 determining if a pixel line must be processed. It must be an integer
12464 in the range [0,255] and defaults to 10. A value of 0 will result in
12465 applying the process on every pixels.
12468 Paint pixels exceeding the threshold value to white if set to 1.
12472 Set the fields order. Swap fields if set to 1, leave fields alone if
12476 Enable additional sharpening if set to 1. Default is 0.
12479 Enable twoway sharpening if set to 1. Default is 0.
12482 @subsection Examples
12486 Apply default values:
12488 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12492 Enable additional sharpening:
12498 Paint processed pixels in white:
12506 Slowly update darker pixels.
12508 This filter makes short flashes of light appear longer.
12509 This filter accepts the following options:
12513 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
12516 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
12519 @section lenscorrection
12521 Correct radial lens distortion
12523 This filter can be used to correct for radial distortion as can result from the use
12524 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
12525 one can use tools available for example as part of opencv or simply trial-and-error.
12526 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
12527 and extract the k1 and k2 coefficients from the resulting matrix.
12529 Note that effectively the same filter is available in the open-source tools Krita and
12530 Digikam from the KDE project.
12532 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
12533 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
12534 brightness distribution, so you may want to use both filters together in certain
12535 cases, though you will have to take care of ordering, i.e. whether vignetting should
12536 be applied before or after lens correction.
12538 @subsection Options
12540 The filter accepts the following options:
12544 Relative x-coordinate of the focal point of the image, and thereby the center of the
12545 distortion. This value has a range [0,1] and is expressed as fractions of the image
12546 width. Default is 0.5.
12548 Relative y-coordinate of the focal point of the image, and thereby the center of the
12549 distortion. This value has a range [0,1] and is expressed as fractions of the image
12550 height. Default is 0.5.
12552 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
12553 no correction. Default is 0.
12555 Coefficient of the double quadratic correction term. This value has a range [-1,1].
12556 0 means no correction. Default is 0.
12559 The formula that generates the correction is:
12561 @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)
12563 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
12564 distances from the focal point in the source and target images, respectively.
12568 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
12570 The @code{lensfun} filter requires the camera make, camera model, and lens model
12571 to apply the lens correction. The filter will load the lensfun database and
12572 query it to find the corresponding camera and lens entries in the database. As
12573 long as these entries can be found with the given options, the filter can
12574 perform corrections on frames. Note that incomplete strings will result in the
12575 filter choosing the best match with the given options, and the filter will
12576 output the chosen camera and lens models (logged with level "info"). You must
12577 provide the make, camera model, and lens model as they are required.
12579 The filter accepts the following options:
12583 The make of the camera (for example, "Canon"). This option is required.
12586 The model of the camera (for example, "Canon EOS 100D"). This option is
12590 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
12591 option is required.
12594 The type of correction to apply. The following values are valid options:
12598 Enables fixing lens vignetting.
12601 Enables fixing lens geometry. This is the default.
12604 Enables fixing chromatic aberrations.
12607 Enables fixing lens vignetting and lens geometry.
12610 Enables fixing lens vignetting and chromatic aberrations.
12613 Enables fixing both lens geometry and chromatic aberrations.
12616 Enables all possible corrections.
12620 The focal length of the image/video (zoom; expected constant for video). For
12621 example, a 18--55mm lens has focal length range of [18--55], so a value in that
12622 range should be chosen when using that lens. Default 18.
12625 The aperture of the image/video (expected constant for video). Note that
12626 aperture is only used for vignetting correction. Default 3.5.
12628 @item focus_distance
12629 The focus distance of the image/video (expected constant for video). Note that
12630 focus distance is only used for vignetting and only slightly affects the
12631 vignetting correction process. If unknown, leave it at the default value (which
12635 The scale factor which is applied after transformation. After correction the
12636 video is no longer necessarily rectangular. This parameter controls how much of
12637 the resulting image is visible. The value 0 means that a value will be chosen
12638 automatically such that there is little or no unmapped area in the output
12639 image. 1.0 means that no additional scaling is done. Lower values may result
12640 in more of the corrected image being visible, while higher values may avoid
12641 unmapped areas in the output.
12643 @item target_geometry
12644 The target geometry of the output image/video. The following values are valid
12648 @item rectilinear (default)
12651 @item equirectangular
12652 @item fisheye_orthographic
12653 @item fisheye_stereographic
12654 @item fisheye_equisolid
12655 @item fisheye_thoby
12658 Apply the reverse of image correction (instead of correcting distortion, apply
12661 @item interpolation
12662 The type of interpolation used when correcting distortion. The following values
12667 @item linear (default)
12672 @subsection Examples
12676 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
12677 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
12681 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
12685 Apply the same as before, but only for the first 5 seconds of video.
12688 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
12695 Obtain the VMAF (Video Multi-Method Assessment Fusion)
12696 score between two input videos.
12698 The obtained VMAF score is printed through the logging system.
12700 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
12701 After installing the library it can be enabled using:
12702 @code{./configure --enable-libvmaf --enable-version3}.
12703 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
12705 The filter has following options:
12709 Set the model path which is to be used for SVM.
12710 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
12713 Set the file path to be used to store logs.
12716 Set the format of the log file (xml or json).
12718 @item enable_transform
12719 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
12720 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
12721 Default value: @code{false}
12724 Invokes the phone model which will generate VMAF scores higher than in the
12725 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
12726 Default value: @code{false}
12729 Enables computing psnr along with vmaf.
12730 Default value: @code{false}
12733 Enables computing ssim along with vmaf.
12734 Default value: @code{false}
12737 Enables computing ms_ssim along with vmaf.
12738 Default value: @code{false}
12741 Set the pool method to be used for computing vmaf.
12742 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
12745 Set number of threads to be used when computing vmaf.
12746 Default value: @code{0}, which makes use of all available logical processors.
12749 Set interval for frame subsampling used when computing vmaf.
12750 Default value: @code{1}
12752 @item enable_conf_interval
12753 Enables confidence interval.
12754 Default value: @code{false}
12757 This filter also supports the @ref{framesync} options.
12759 @subsection Examples
12762 On the below examples the input file @file{main.mpg} being processed is
12763 compared with the reference file @file{ref.mpg}.
12766 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
12770 Example with options:
12772 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
12776 Example with options and different containers:
12778 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 -
12784 Limits the pixel components values to the specified range [min, max].
12786 The filter accepts the following options:
12790 Lower bound. Defaults to the lowest allowed value for the input.
12793 Upper bound. Defaults to the highest allowed value for the input.
12796 Specify which planes will be processed. Defaults to all available.
12803 The filter accepts the following options:
12807 Set the number of loops. Setting this value to -1 will result in infinite loops.
12811 Set maximal size in number of frames. Default is 0.
12814 Set first frame of loop. Default is 0.
12817 @subsection Examples
12821 Loop single first frame infinitely:
12823 loop=loop=-1:size=1:start=0
12827 Loop single first frame 10 times:
12829 loop=loop=10:size=1:start=0
12833 Loop 10 first frames 5 times:
12835 loop=loop=5:size=10:start=0
12841 Apply a 1D LUT to an input video.
12843 The filter accepts the following options:
12847 Set the 1D LUT file name.
12849 Currently supported formats:
12858 Select interpolation mode.
12860 Available values are:
12864 Use values from the nearest defined point.
12866 Interpolate values using the linear interpolation.
12868 Interpolate values using the cosine interpolation.
12870 Interpolate values using the cubic interpolation.
12872 Interpolate values using the spline interpolation.
12879 Apply a 3D LUT to an input video.
12881 The filter accepts the following options:
12885 Set the 3D LUT file name.
12887 Currently supported formats:
12901 Select interpolation mode.
12903 Available values are:
12907 Use values from the nearest defined point.
12909 Interpolate values using the 8 points defining a cube.
12911 Interpolate values using a tetrahedron.
12917 Turn certain luma values into transparency.
12919 The filter accepts the following options:
12923 Set the luma which will be used as base for transparency.
12924 Default value is @code{0}.
12927 Set the range of luma values to be keyed out.
12928 Default value is @code{0.01}.
12931 Set the range of softness. Default value is @code{0}.
12932 Use this to control gradual transition from zero to full transparency.
12935 @subsection Commands
12936 This filter supports same @ref{commands} as options.
12937 The command accepts the same syntax of the corresponding option.
12939 If the specified expression is not valid, it is kept at its current
12942 @section lut, lutrgb, lutyuv
12944 Compute a look-up table for binding each pixel component input value
12945 to an output value, and apply it to the input video.
12947 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
12948 to an RGB input video.
12950 These filters accept the following parameters:
12953 set first pixel component expression
12955 set second pixel component expression
12957 set third pixel component expression
12959 set fourth pixel component expression, corresponds to the alpha component
12962 set red component expression
12964 set green component expression
12966 set blue component expression
12968 alpha component expression
12971 set Y/luminance component expression
12973 set U/Cb component expression
12975 set V/Cr component expression
12978 Each of them specifies the expression to use for computing the lookup table for
12979 the corresponding pixel component values.
12981 The exact component associated to each of the @var{c*} options depends on the
12984 The @var{lut} filter requires either YUV or RGB pixel formats in input,
12985 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
12987 The expressions can contain the following constants and functions:
12992 The input width and height.
12995 The input value for the pixel component.
12998 The input value, clipped to the @var{minval}-@var{maxval} range.
13001 The maximum value for the pixel component.
13004 The minimum value for the pixel component.
13007 The negated value for the pixel component value, clipped to the
13008 @var{minval}-@var{maxval} range; it corresponds to the expression
13009 "maxval-clipval+minval".
13012 The computed value in @var{val}, clipped to the
13013 @var{minval}-@var{maxval} range.
13015 @item gammaval(gamma)
13016 The computed gamma correction value of the pixel component value,
13017 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13019 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13023 All expressions default to "val".
13025 @subsection Examples
13029 Negate input video:
13031 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13032 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13035 The above is the same as:
13037 lutrgb="r=negval:g=negval:b=negval"
13038 lutyuv="y=negval:u=negval:v=negval"
13048 Remove chroma components, turning the video into a graytone image:
13050 lutyuv="u=128:v=128"
13054 Apply a luma burning effect:
13060 Remove green and blue components:
13066 Set a constant alpha channel value on input:
13068 format=rgba,lutrgb=a="maxval-minval/2"
13072 Correct luminance gamma by a factor of 0.5:
13074 lutyuv=y=gammaval(0.5)
13078 Discard least significant bits of luma:
13080 lutyuv=y='bitand(val, 128+64+32)'
13084 Technicolor like effect:
13086 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13090 @section lut2, tlut2
13092 The @code{lut2} filter takes two input streams and outputs one
13095 The @code{tlut2} (time lut2) filter takes two consecutive frames
13096 from one single stream.
13098 This filter accepts the following parameters:
13101 set first pixel component expression
13103 set second pixel component expression
13105 set third pixel component expression
13107 set fourth pixel component expression, corresponds to the alpha component
13110 set output bit depth, only available for @code{lut2} filter. By default is 0,
13111 which means bit depth is automatically picked from first input format.
13114 The @code{lut2} filter also supports the @ref{framesync} options.
13116 Each of them specifies the expression to use for computing the lookup table for
13117 the corresponding pixel component values.
13119 The exact component associated to each of the @var{c*} options depends on the
13122 The expressions can contain the following constants:
13127 The input width and height.
13130 The first input value for the pixel component.
13133 The second input value for the pixel component.
13136 The first input video bit depth.
13139 The second input video bit depth.
13142 All expressions default to "x".
13144 @subsection Examples
13148 Highlight differences between two RGB video streams:
13150 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)'
13154 Highlight differences between two YUV video streams:
13156 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)'
13160 Show max difference between two video streams:
13162 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)))'
13166 @section maskedclamp
13168 Clamp the first input stream with the second input and third input stream.
13170 Returns the value of first stream to be between second input
13171 stream - @code{undershoot} and third input stream + @code{overshoot}.
13173 This filter accepts the following options:
13176 Default value is @code{0}.
13179 Default value is @code{0}.
13182 Set which planes will be processed as bitmap, unprocessed planes will be
13183 copied from first stream.
13184 By default value 0xf, all planes will be processed.
13189 Merge the second and third input stream into output stream using absolute differences
13190 between second input stream and first input stream and absolute difference between
13191 third input stream and first input stream. The picked value will be from second input
13192 stream if second absolute difference is greater than first one or from third input stream
13195 This filter accepts the following options:
13198 Set which planes will be processed as bitmap, unprocessed planes will be
13199 copied from first stream.
13200 By default value 0xf, all planes will be processed.
13203 @section maskedmerge
13205 Merge the first input stream with the second input stream using per pixel
13206 weights in the third input stream.
13208 A value of 0 in the third stream pixel component means that pixel component
13209 from first stream is returned unchanged, while maximum value (eg. 255 for
13210 8-bit videos) means that pixel component from second stream is returned
13211 unchanged. Intermediate values define the amount of merging between both
13212 input stream's pixel components.
13214 This filter accepts the following options:
13217 Set which planes will be processed as bitmap, unprocessed planes will be
13218 copied from first stream.
13219 By default value 0xf, all planes will be processed.
13224 Merge the second and third input stream into output stream using absolute differences
13225 between second input stream and first input stream and absolute difference between
13226 third input stream and first input stream. The picked value will be from second input
13227 stream if second absolute difference is less than first one or from third input stream
13230 This filter accepts the following options:
13233 Set which planes will be processed as bitmap, unprocessed planes will be
13234 copied from first stream.
13235 By default value 0xf, all planes will be processed.
13239 Create mask from input video.
13241 For example it is useful to create motion masks after @code{tblend} filter.
13243 This filter accepts the following options:
13247 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13250 Set high threshold. Any pixel component higher than this value will be set to max value
13251 allowed for current pixel format.
13254 Set planes to filter, by default all available planes are filtered.
13257 Fill all frame pixels with this value.
13260 Set max average pixel value for frame. If sum of all pixel components is higher that this
13261 average, output frame will be completely filled with value set by @var{fill} option.
13262 Typically useful for scene changes when used in combination with @code{tblend} filter.
13267 Apply motion-compensation deinterlacing.
13269 It needs one field per frame as input and must thus be used together
13270 with yadif=1/3 or equivalent.
13272 This filter accepts the following options:
13275 Set the deinterlacing mode.
13277 It accepts one of the following values:
13282 use iterative motion estimation
13284 like @samp{slow}, but use multiple reference frames.
13286 Default value is @samp{fast}.
13289 Set the picture field parity assumed for the input video. It must be
13290 one of the following values:
13294 assume top field first
13296 assume bottom field first
13299 Default value is @samp{bff}.
13302 Set per-block quantization parameter (QP) used by the internal
13305 Higher values should result in a smoother motion vector field but less
13306 optimal individual vectors. Default value is 1.
13311 Pick median pixel from certain rectangle defined by radius.
13313 This filter accepts the following options:
13317 Set horizontal radius size. Default value is @code{1}.
13318 Allowed range is integer from 1 to 127.
13321 Set which planes to process. Default is @code{15}, which is all available planes.
13324 Set vertical radius size. Default value is @code{0}.
13325 Allowed range is integer from 0 to 127.
13326 If it is 0, value will be picked from horizontal @code{radius} option.
13329 Set median percentile. Default value is @code{0.5}.
13330 Default value of @code{0.5} will pick always median values, while @code{0} will pick
13331 minimum values, and @code{1} maximum values.
13334 @subsection Commands
13335 This filter supports same @ref{commands} as options.
13336 The command accepts the same syntax of the corresponding option.
13338 If the specified expression is not valid, it is kept at its current
13341 @section mergeplanes
13343 Merge color channel components from several video streams.
13345 The filter accepts up to 4 input streams, and merge selected input
13346 planes to the output video.
13348 This filter accepts the following options:
13351 Set input to output plane mapping. Default is @code{0}.
13353 The mappings is specified as a bitmap. It should be specified as a
13354 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13355 mapping for the first plane of the output stream. 'A' sets the number of
13356 the input stream to use (from 0 to 3), and 'a' the plane number of the
13357 corresponding input to use (from 0 to 3). The rest of the mappings is
13358 similar, 'Bb' describes the mapping for the output stream second
13359 plane, 'Cc' describes the mapping for the output stream third plane and
13360 'Dd' describes the mapping for the output stream fourth plane.
13363 Set output pixel format. Default is @code{yuva444p}.
13366 @subsection Examples
13370 Merge three gray video streams of same width and height into single video stream:
13372 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13376 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13378 [a0][a1]mergeplanes=0x00010210:yuva444p
13382 Swap Y and A plane in yuva444p stream:
13384 format=yuva444p,mergeplanes=0x03010200:yuva444p
13388 Swap U and V plane in yuv420p stream:
13390 format=yuv420p,mergeplanes=0x000201:yuv420p
13394 Cast a rgb24 clip to yuv444p:
13396 format=rgb24,mergeplanes=0x000102:yuv444p
13402 Estimate and export motion vectors using block matching algorithms.
13403 Motion vectors are stored in frame side data to be used by other filters.
13405 This filter accepts the following options:
13408 Specify the motion estimation method. Accepts one of the following values:
13412 Exhaustive search algorithm.
13414 Three step search algorithm.
13416 Two dimensional logarithmic search algorithm.
13418 New three step search algorithm.
13420 Four step search algorithm.
13422 Diamond search algorithm.
13424 Hexagon-based search algorithm.
13426 Enhanced predictive zonal search algorithm.
13428 Uneven multi-hexagon search algorithm.
13430 Default value is @samp{esa}.
13433 Macroblock size. Default @code{16}.
13436 Search parameter. Default @code{7}.
13439 @section midequalizer
13441 Apply Midway Image Equalization effect using two video streams.
13443 Midway Image Equalization adjusts a pair of images to have the same
13444 histogram, while maintaining their dynamics as much as possible. It's
13445 useful for e.g. matching exposures from a pair of stereo cameras.
13447 This filter has two inputs and one output, which must be of same pixel format, but
13448 may be of different sizes. The output of filter is first input adjusted with
13449 midway histogram of both inputs.
13451 This filter accepts the following option:
13455 Set which planes to process. Default is @code{15}, which is all available planes.
13458 @section minterpolate
13460 Convert the video to specified frame rate using motion interpolation.
13462 This filter accepts the following options:
13465 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}.
13468 Motion interpolation mode. Following values are accepted:
13471 Duplicate previous or next frame for interpolating new ones.
13473 Blend source frames. Interpolated frame is mean of previous and next frames.
13475 Motion compensated interpolation. Following options are effective when this mode is selected:
13479 Motion compensation mode. Following values are accepted:
13482 Overlapped block motion compensation.
13484 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13486 Default mode is @samp{obmc}.
13489 Motion estimation mode. Following values are accepted:
13492 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
13494 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
13496 Default mode is @samp{bilat}.
13499 The algorithm to be used for motion estimation. Following values are accepted:
13502 Exhaustive search algorithm.
13504 Three step search algorithm.
13506 Two dimensional logarithmic search algorithm.
13508 New three step search algorithm.
13510 Four step search algorithm.
13512 Diamond search algorithm.
13514 Hexagon-based search algorithm.
13516 Enhanced predictive zonal search algorithm.
13518 Uneven multi-hexagon search algorithm.
13520 Default algorithm is @samp{epzs}.
13523 Macroblock size. Default @code{16}.
13526 Motion estimation search parameter. Default @code{32}.
13529 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).
13534 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:
13537 Disable scene change detection.
13539 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
13541 Default method is @samp{fdiff}.
13543 @item scd_threshold
13544 Scene change detection threshold. Default is @code{5.0}.
13549 Mix several video input streams into one video stream.
13551 A description of the accepted options follows.
13555 The number of inputs. If unspecified, it defaults to 2.
13558 Specify weight of each input video stream as sequence.
13559 Each weight is separated by space. If number of weights
13560 is smaller than number of @var{frames} last specified
13561 weight will be used for all remaining unset weights.
13564 Specify scale, if it is set it will be multiplied with sum
13565 of each weight multiplied with pixel values to give final destination
13566 pixel value. By default @var{scale} is auto scaled to sum of weights.
13569 Specify how end of stream is determined.
13572 The duration of the longest input. (default)
13575 The duration of the shortest input.
13578 The duration of the first input.
13582 @section mpdecimate
13584 Drop frames that do not differ greatly from the previous frame in
13585 order to reduce frame rate.
13587 The main use of this filter is for very-low-bitrate encoding
13588 (e.g. streaming over dialup modem), but it could in theory be used for
13589 fixing movies that were inverse-telecined incorrectly.
13591 A description of the accepted options follows.
13595 Set the maximum number of consecutive frames which can be dropped (if
13596 positive), or the minimum interval between dropped frames (if
13597 negative). If the value is 0, the frame is dropped disregarding the
13598 number of previous sequentially dropped frames.
13600 Default value is 0.
13605 Set the dropping threshold values.
13607 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
13608 represent actual pixel value differences, so a threshold of 64
13609 corresponds to 1 unit of difference for each pixel, or the same spread
13610 out differently over the block.
13612 A frame is a candidate for dropping if no 8x8 blocks differ by more
13613 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
13614 meaning the whole image) differ by more than a threshold of @option{lo}.
13616 Default value for @option{hi} is 64*12, default value for @option{lo} is
13617 64*5, and default value for @option{frac} is 0.33.
13623 Negate (invert) the input video.
13625 It accepts the following option:
13630 With value 1, it negates the alpha component, if present. Default value is 0.
13636 Denoise frames using Non-Local Means algorithm.
13638 Each pixel is adjusted by looking for other pixels with similar contexts. This
13639 context similarity is defined by comparing their surrounding patches of size
13640 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
13643 Note that the research area defines centers for patches, which means some
13644 patches will be made of pixels outside that research area.
13646 The filter accepts the following options.
13650 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
13653 Set patch size. Default is 7. Must be odd number in range [0, 99].
13656 Same as @option{p} but for chroma planes.
13658 The default value is @var{0} and means automatic.
13661 Set research size. Default is 15. Must be odd number in range [0, 99].
13664 Same as @option{r} but for chroma planes.
13666 The default value is @var{0} and means automatic.
13671 Deinterlace video using neural network edge directed interpolation.
13673 This filter accepts the following options:
13677 Mandatory option, without binary file filter can not work.
13678 Currently file can be found here:
13679 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
13682 Set which frames to deinterlace, by default it is @code{all}.
13683 Can be @code{all} or @code{interlaced}.
13686 Set mode of operation.
13688 Can be one of the following:
13692 Use frame flags, both fields.
13694 Use frame flags, single field.
13696 Use top field only.
13698 Use bottom field only.
13700 Use both fields, top first.
13702 Use both fields, bottom first.
13706 Set which planes to process, by default filter process all frames.
13709 Set size of local neighborhood around each pixel, used by the predictor neural
13712 Can be one of the following:
13725 Set the number of neurons in predictor neural network.
13726 Can be one of the following:
13737 Controls the number of different neural network predictions that are blended
13738 together to compute the final output value. Can be @code{fast}, default or
13742 Set which set of weights to use in the predictor.
13743 Can be one of the following:
13747 weights trained to minimize absolute error
13749 weights trained to minimize squared error
13753 Controls whether or not the prescreener neural network is used to decide
13754 which pixels should be processed by the predictor neural network and which
13755 can be handled by simple cubic interpolation.
13756 The prescreener is trained to know whether cubic interpolation will be
13757 sufficient for a pixel or whether it should be predicted by the predictor nn.
13758 The computational complexity of the prescreener nn is much less than that of
13759 the predictor nn. Since most pixels can be handled by cubic interpolation,
13760 using the prescreener generally results in much faster processing.
13761 The prescreener is pretty accurate, so the difference between using it and not
13762 using it is almost always unnoticeable.
13764 Can be one of the following:
13772 Default is @code{new}.
13775 Set various debugging flags.
13780 Force libavfilter not to use any of the specified pixel formats for the
13781 input to the next filter.
13783 It accepts the following parameters:
13787 A '|'-separated list of pixel format names, such as
13788 pix_fmts=yuv420p|monow|rgb24".
13792 @subsection Examples
13796 Force libavfilter to use a format different from @var{yuv420p} for the
13797 input to the vflip filter:
13799 noformat=pix_fmts=yuv420p,vflip
13803 Convert the input video to any of the formats not contained in the list:
13805 noformat=yuv420p|yuv444p|yuv410p
13811 Add noise on video input frame.
13813 The filter accepts the following options:
13821 Set noise seed for specific pixel component or all pixel components in case
13822 of @var{all_seed}. Default value is @code{123457}.
13824 @item all_strength, alls
13825 @item c0_strength, c0s
13826 @item c1_strength, c1s
13827 @item c2_strength, c2s
13828 @item c3_strength, c3s
13829 Set noise strength for specific pixel component or all pixel components in case
13830 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
13832 @item all_flags, allf
13833 @item c0_flags, c0f
13834 @item c1_flags, c1f
13835 @item c2_flags, c2f
13836 @item c3_flags, c3f
13837 Set pixel component flags or set flags for all components if @var{all_flags}.
13838 Available values for component flags are:
13841 averaged temporal noise (smoother)
13843 mix random noise with a (semi)regular pattern
13845 temporal noise (noise pattern changes between frames)
13847 uniform noise (gaussian otherwise)
13851 @subsection Examples
13853 Add temporal and uniform noise to input video:
13855 noise=alls=20:allf=t+u
13860 Normalize RGB video (aka histogram stretching, contrast stretching).
13861 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
13863 For each channel of each frame, the filter computes the input range and maps
13864 it linearly to the user-specified output range. The output range defaults
13865 to the full dynamic range from pure black to pure white.
13867 Temporal smoothing can be used on the input range to reduce flickering (rapid
13868 changes in brightness) caused when small dark or bright objects enter or leave
13869 the scene. This is similar to the auto-exposure (automatic gain control) on a
13870 video camera, and, like a video camera, it may cause a period of over- or
13871 under-exposure of the video.
13873 The R,G,B channels can be normalized independently, which may cause some
13874 color shifting, or linked together as a single channel, which prevents
13875 color shifting. Linked normalization preserves hue. Independent normalization
13876 does not, so it can be used to remove some color casts. Independent and linked
13877 normalization can be combined in any ratio.
13879 The normalize filter accepts the following options:
13884 Colors which define the output range. The minimum input value is mapped to
13885 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
13886 The defaults are black and white respectively. Specifying white for
13887 @var{blackpt} and black for @var{whitept} will give color-inverted,
13888 normalized video. Shades of grey can be used to reduce the dynamic range
13889 (contrast). Specifying saturated colors here can create some interesting
13893 The number of previous frames to use for temporal smoothing. The input range
13894 of each channel is smoothed using a rolling average over the current frame
13895 and the @var{smoothing} previous frames. The default is 0 (no temporal
13899 Controls the ratio of independent (color shifting) channel normalization to
13900 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
13901 independent. Defaults to 1.0 (fully independent).
13904 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
13905 expensive no-op. Defaults to 1.0 (full strength).
13909 @subsection Commands
13910 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
13911 The command accepts the same syntax of the corresponding option.
13913 If the specified expression is not valid, it is kept at its current
13916 @subsection Examples
13918 Stretch video contrast to use the full dynamic range, with no temporal
13919 smoothing; may flicker depending on the source content:
13921 normalize=blackpt=black:whitept=white:smoothing=0
13924 As above, but with 50 frames of temporal smoothing; flicker should be
13925 reduced, depending on the source content:
13927 normalize=blackpt=black:whitept=white:smoothing=50
13930 As above, but with hue-preserving linked channel normalization:
13932 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
13935 As above, but with half strength:
13937 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
13940 Map the darkest input color to red, the brightest input color to cyan:
13942 normalize=blackpt=red:whitept=cyan
13947 Pass the video source unchanged to the output.
13950 Optical Character Recognition
13952 This filter uses Tesseract for optical character recognition. To enable
13953 compilation of this filter, you need to configure FFmpeg with
13954 @code{--enable-libtesseract}.
13956 It accepts the following options:
13960 Set datapath to tesseract data. Default is to use whatever was
13961 set at installation.
13964 Set language, default is "eng".
13967 Set character whitelist.
13970 Set character blacklist.
13973 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
13974 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
13978 Apply a video transform using libopencv.
13980 To enable this filter, install the libopencv library and headers and
13981 configure FFmpeg with @code{--enable-libopencv}.
13983 It accepts the following parameters:
13988 The name of the libopencv filter to apply.
13990 @item filter_params
13991 The parameters to pass to the libopencv filter. If not specified, the default
13992 values are assumed.
13996 Refer to the official libopencv documentation for more precise
13998 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14000 Several libopencv filters are supported; see the following subsections.
14005 Dilate an image by using a specific structuring element.
14006 It corresponds to the libopencv function @code{cvDilate}.
14008 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14010 @var{struct_el} represents a structuring element, and has the syntax:
14011 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14013 @var{cols} and @var{rows} represent the number of columns and rows of
14014 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14015 point, and @var{shape} the shape for the structuring element. @var{shape}
14016 must be "rect", "cross", "ellipse", or "custom".
14018 If the value for @var{shape} is "custom", it must be followed by a
14019 string of the form "=@var{filename}". The file with name
14020 @var{filename} is assumed to represent a binary image, with each
14021 printable character corresponding to a bright pixel. When a custom
14022 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14023 or columns and rows of the read file are assumed instead.
14025 The default value for @var{struct_el} is "3x3+0x0/rect".
14027 @var{nb_iterations} specifies the number of times the transform is
14028 applied to the image, and defaults to 1.
14032 # Use the default values
14035 # Dilate using a structuring element with a 5x5 cross, iterating two times
14036 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14038 # Read the shape from the file diamond.shape, iterating two times.
14039 # The file diamond.shape may contain a pattern of characters like this
14045 # The specified columns and rows are ignored
14046 # but the anchor point coordinates are not
14047 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14052 Erode an image by using a specific structuring element.
14053 It corresponds to the libopencv function @code{cvErode}.
14055 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14056 with the same syntax and semantics as the @ref{dilate} filter.
14060 Smooth the input video.
14062 The filter takes the following parameters:
14063 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14065 @var{type} is the type of smooth filter to apply, and must be one of
14066 the following values: "blur", "blur_no_scale", "median", "gaussian",
14067 or "bilateral". The default value is "gaussian".
14069 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14070 depends on the smooth type. @var{param1} and
14071 @var{param2} accept integer positive values or 0. @var{param3} and
14072 @var{param4} accept floating point values.
14074 The default value for @var{param1} is 3. The default value for the
14075 other parameters is 0.
14077 These parameters correspond to the parameters assigned to the
14078 libopencv function @code{cvSmooth}.
14080 @section oscilloscope
14082 2D Video Oscilloscope.
14084 Useful to measure spatial impulse, step responses, chroma delays, etc.
14086 It accepts the following parameters:
14090 Set scope center x position.
14093 Set scope center y position.
14096 Set scope size, relative to frame diagonal.
14099 Set scope tilt/rotation.
14105 Set trace center x position.
14108 Set trace center y position.
14111 Set trace width, relative to width of frame.
14114 Set trace height, relative to height of frame.
14117 Set which components to trace. By default it traces first three components.
14120 Draw trace grid. By default is enabled.
14123 Draw some statistics. By default is enabled.
14126 Draw scope. By default is enabled.
14129 @subsection Commands
14130 This filter supports same @ref{commands} as options.
14131 The command accepts the same syntax of the corresponding option.
14133 If the specified expression is not valid, it is kept at its current
14136 @subsection Examples
14140 Inspect full first row of video frame.
14142 oscilloscope=x=0.5:y=0:s=1
14146 Inspect full last row of video frame.
14148 oscilloscope=x=0.5:y=1:s=1
14152 Inspect full 5th line of video frame of height 1080.
14154 oscilloscope=x=0.5:y=5/1080:s=1
14158 Inspect full last column of video frame.
14160 oscilloscope=x=1:y=0.5:s=1:t=1
14168 Overlay one video on top of another.
14170 It takes two inputs and has one output. The first input is the "main"
14171 video on which the second input is overlaid.
14173 It accepts the following parameters:
14175 A description of the accepted options follows.
14180 Set the expression for the x and y coordinates of the overlaid video
14181 on the main video. Default value is "0" for both expressions. In case
14182 the expression is invalid, it is set to a huge value (meaning that the
14183 overlay will not be displayed within the output visible area).
14186 See @ref{framesync}.
14189 Set when the expressions for @option{x}, and @option{y} are evaluated.
14191 It accepts the following values:
14194 only evaluate expressions once during the filter initialization or
14195 when a command is processed
14198 evaluate expressions for each incoming frame
14201 Default value is @samp{frame}.
14204 See @ref{framesync}.
14207 Set the format for the output video.
14209 It accepts the following values:
14212 force YUV420 output
14215 force YUV422 output
14218 force YUV444 output
14221 force packed RGB output
14224 force planar RGB output
14227 automatically pick format
14230 Default value is @samp{yuv420}.
14233 See @ref{framesync}.
14236 Set format of alpha of the overlaid video, it can be @var{straight} or
14237 @var{premultiplied}. Default is @var{straight}.
14240 The @option{x}, and @option{y} expressions can contain the following
14246 The main input width and height.
14250 The overlay input width and height.
14254 The computed values for @var{x} and @var{y}. They are evaluated for
14259 horizontal and vertical chroma subsample values of the output
14260 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14264 the number of input frame, starting from 0
14267 the position in the file of the input frame, NAN if unknown
14270 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14274 This filter also supports the @ref{framesync} options.
14276 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14277 when evaluation is done @emph{per frame}, and will evaluate to NAN
14278 when @option{eval} is set to @samp{init}.
14280 Be aware that frames are taken from each input video in timestamp
14281 order, hence, if their initial timestamps differ, it is a good idea
14282 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14283 have them begin in the same zero timestamp, as the example for
14284 the @var{movie} filter does.
14286 You can chain together more overlays but you should test the
14287 efficiency of such approach.
14289 @subsection Commands
14291 This filter supports the following commands:
14295 Modify the x and y of the overlay input.
14296 The command accepts the same syntax of the corresponding option.
14298 If the specified expression is not valid, it is kept at its current
14302 @subsection Examples
14306 Draw the overlay at 10 pixels from the bottom right corner of the main
14309 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14312 Using named options the example above becomes:
14314 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14318 Insert a transparent PNG logo in the bottom left corner of the input,
14319 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14321 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14325 Insert 2 different transparent PNG logos (second logo on bottom
14326 right corner) using the @command{ffmpeg} tool:
14328 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
14332 Add a transparent color layer on top of the main video; @code{WxH}
14333 must specify the size of the main input to the overlay filter:
14335 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14339 Play an original video and a filtered version (here with the deshake
14340 filter) side by side using the @command{ffplay} tool:
14342 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14345 The above command is the same as:
14347 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14351 Make a sliding overlay appearing from the left to the right top part of the
14352 screen starting since time 2:
14354 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14358 Compose output by putting two input videos side to side:
14360 ffmpeg -i left.avi -i right.avi -filter_complex "
14361 nullsrc=size=200x100 [background];
14362 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14363 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14364 [background][left] overlay=shortest=1 [background+left];
14365 [background+left][right] overlay=shortest=1:x=100 [left+right]
14370 Mask 10-20 seconds of a video by applying the delogo filter to a section
14372 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14373 -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]'
14378 Chain several overlays in cascade:
14380 nullsrc=s=200x200 [bg];
14381 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14382 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14383 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14384 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14385 [in3] null, [mid2] overlay=100:100 [out0]
14392 Apply Overcomplete Wavelet denoiser.
14394 The filter accepts the following options:
14400 Larger depth values will denoise lower frequency components more, but
14401 slow down filtering.
14403 Must be an int in the range 8-16, default is @code{8}.
14405 @item luma_strength, ls
14408 Must be a double value in the range 0-1000, default is @code{1.0}.
14410 @item chroma_strength, cs
14411 Set chroma strength.
14413 Must be a double value in the range 0-1000, default is @code{1.0}.
14419 Add paddings to the input image, and place the original input at the
14420 provided @var{x}, @var{y} coordinates.
14422 It accepts the following parameters:
14427 Specify an expression for the size of the output image with the
14428 paddings added. If the value for @var{width} or @var{height} is 0, the
14429 corresponding input size is used for the output.
14431 The @var{width} expression can reference the value set by the
14432 @var{height} expression, and vice versa.
14434 The default value of @var{width} and @var{height} is 0.
14438 Specify the offsets to place the input image at within the padded area,
14439 with respect to the top/left border of the output image.
14441 The @var{x} expression can reference the value set by the @var{y}
14442 expression, and vice versa.
14444 The default value of @var{x} and @var{y} is 0.
14446 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14447 so the input image is centered on the padded area.
14450 Specify the color of the padded area. For the syntax of this option,
14451 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14452 manual,ffmpeg-utils}.
14454 The default value of @var{color} is "black".
14457 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
14459 It accepts the following values:
14463 Only evaluate expressions once during the filter initialization or when
14464 a command is processed.
14467 Evaluate expressions for each incoming frame.
14471 Default value is @samp{init}.
14474 Pad to aspect instead to a resolution.
14478 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
14479 options are expressions containing the following constants:
14484 The input video width and height.
14488 These are the same as @var{in_w} and @var{in_h}.
14492 The output width and height (the size of the padded area), as
14493 specified by the @var{width} and @var{height} expressions.
14497 These are the same as @var{out_w} and @var{out_h}.
14501 The x and y offsets as specified by the @var{x} and @var{y}
14502 expressions, or NAN if not yet specified.
14505 same as @var{iw} / @var{ih}
14508 input sample aspect ratio
14511 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
14515 The horizontal and vertical chroma subsample values. For example for the
14516 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
14519 @subsection Examples
14523 Add paddings with the color "violet" to the input video. The output video
14524 size is 640x480, and the top-left corner of the input video is placed at
14527 pad=640:480:0:40:violet
14530 The example above is equivalent to the following command:
14532 pad=width=640:height=480:x=0:y=40:color=violet
14536 Pad the input to get an output with dimensions increased by 3/2,
14537 and put the input video at the center of the padded area:
14539 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
14543 Pad the input to get a squared output with size equal to the maximum
14544 value between the input width and height, and put the input video at
14545 the center of the padded area:
14547 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
14551 Pad the input to get a final w/h ratio of 16:9:
14553 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
14557 In case of anamorphic video, in order to set the output display aspect
14558 correctly, it is necessary to use @var{sar} in the expression,
14559 according to the relation:
14561 (ih * X / ih) * sar = output_dar
14562 X = output_dar / sar
14565 Thus the previous example needs to be modified to:
14567 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
14571 Double the output size and put the input video in the bottom-right
14572 corner of the output padded area:
14574 pad="2*iw:2*ih:ow-iw:oh-ih"
14578 @anchor{palettegen}
14579 @section palettegen
14581 Generate one palette for a whole video stream.
14583 It accepts the following options:
14587 Set the maximum number of colors to quantize in the palette.
14588 Note: the palette will still contain 256 colors; the unused palette entries
14591 @item reserve_transparent
14592 Create a palette of 255 colors maximum and reserve the last one for
14593 transparency. Reserving the transparency color is useful for GIF optimization.
14594 If not set, the maximum of colors in the palette will be 256. You probably want
14595 to disable this option for a standalone image.
14598 @item transparency_color
14599 Set the color that will be used as background for transparency.
14602 Set statistics mode.
14604 It accepts the following values:
14607 Compute full frame histograms.
14609 Compute histograms only for the part that differs from previous frame. This
14610 might be relevant to give more importance to the moving part of your input if
14611 the background is static.
14613 Compute new histogram for each frame.
14616 Default value is @var{full}.
14619 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
14620 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
14621 color quantization of the palette. This information is also visible at
14622 @var{info} logging level.
14624 @subsection Examples
14628 Generate a representative palette of a given video using @command{ffmpeg}:
14630 ffmpeg -i input.mkv -vf palettegen palette.png
14634 @section paletteuse
14636 Use a palette to downsample an input video stream.
14638 The filter takes two inputs: one video stream and a palette. The palette must
14639 be a 256 pixels image.
14641 It accepts the following options:
14645 Select dithering mode. Available algorithms are:
14648 Ordered 8x8 bayer dithering (deterministic)
14650 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
14651 Note: this dithering is sometimes considered "wrong" and is included as a
14653 @item floyd_steinberg
14654 Floyd and Steingberg dithering (error diffusion)
14656 Frankie Sierra dithering v2 (error diffusion)
14658 Frankie Sierra dithering v2 "Lite" (error diffusion)
14661 Default is @var{sierra2_4a}.
14664 When @var{bayer} dithering is selected, this option defines the scale of the
14665 pattern (how much the crosshatch pattern is visible). A low value means more
14666 visible pattern for less banding, and higher value means less visible pattern
14667 at the cost of more banding.
14669 The option must be an integer value in the range [0,5]. Default is @var{2}.
14672 If set, define the zone to process
14676 Only the changing rectangle will be reprocessed. This is similar to GIF
14677 cropping/offsetting compression mechanism. This option can be useful for speed
14678 if only a part of the image is changing, and has use cases such as limiting the
14679 scope of the error diffusal @option{dither} to the rectangle that bounds the
14680 moving scene (it leads to more deterministic output if the scene doesn't change
14681 much, and as a result less moving noise and better GIF compression).
14684 Default is @var{none}.
14687 Take new palette for each output frame.
14689 @item alpha_threshold
14690 Sets the alpha threshold for transparency. Alpha values above this threshold
14691 will be treated as completely opaque, and values below this threshold will be
14692 treated as completely transparent.
14694 The option must be an integer value in the range [0,255]. Default is @var{128}.
14697 @subsection Examples
14701 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
14702 using @command{ffmpeg}:
14704 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
14708 @section perspective
14710 Correct perspective of video not recorded perpendicular to the screen.
14712 A description of the accepted parameters follows.
14723 Set coordinates expression for top left, top right, bottom left and bottom right corners.
14724 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
14725 If the @code{sense} option is set to @code{source}, then the specified points will be sent
14726 to the corners of the destination. If the @code{sense} option is set to @code{destination},
14727 then the corners of the source will be sent to the specified coordinates.
14729 The expressions can use the following variables:
14734 the width and height of video frame.
14738 Output frame count.
14741 @item interpolation
14742 Set interpolation for perspective correction.
14744 It accepts the following values:
14750 Default value is @samp{linear}.
14753 Set interpretation of coordinate options.
14755 It accepts the following values:
14759 Send point in the source specified by the given coordinates to
14760 the corners of the destination.
14762 @item 1, destination
14764 Send the corners of the source to the point in the destination specified
14765 by the given coordinates.
14767 Default value is @samp{source}.
14771 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
14773 It accepts the following values:
14776 only evaluate expressions once during the filter initialization or
14777 when a command is processed
14780 evaluate expressions for each incoming frame
14783 Default value is @samp{init}.
14788 Delay interlaced video by one field time so that the field order changes.
14790 The intended use is to fix PAL movies that have been captured with the
14791 opposite field order to the film-to-video transfer.
14793 A description of the accepted parameters follows.
14799 It accepts the following values:
14802 Capture field order top-first, transfer bottom-first.
14803 Filter will delay the bottom field.
14806 Capture field order bottom-first, transfer top-first.
14807 Filter will delay the top field.
14810 Capture and transfer with the same field order. This mode only exists
14811 for the documentation of the other options to refer to, but if you
14812 actually select it, the filter will faithfully do nothing.
14815 Capture field order determined automatically by field flags, transfer
14817 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
14818 basis using field flags. If no field information is available,
14819 then this works just like @samp{u}.
14822 Capture unknown or varying, transfer opposite.
14823 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
14824 analyzing the images and selecting the alternative that produces best
14825 match between the fields.
14828 Capture top-first, transfer unknown or varying.
14829 Filter selects among @samp{t} and @samp{p} using image analysis.
14832 Capture bottom-first, transfer unknown or varying.
14833 Filter selects among @samp{b} and @samp{p} using image analysis.
14836 Capture determined by field flags, transfer unknown or varying.
14837 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
14838 image analysis. If no field information is available, then this works just
14839 like @samp{U}. This is the default mode.
14842 Both capture and transfer unknown or varying.
14843 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
14847 @section photosensitivity
14848 Reduce various flashes in video, so to help users with epilepsy.
14850 It accepts the following options:
14853 Set how many frames to use when filtering. Default is 30.
14856 Set detection threshold factor. Default is 1.
14860 Set how many pixels to skip when sampling frames. Default is 1.
14861 Allowed range is from 1 to 1024.
14864 Leave frames unchanged. Default is disabled.
14867 @section pixdesctest
14869 Pixel format descriptor test filter, mainly useful for internal
14870 testing. The output video should be equal to the input video.
14874 format=monow, pixdesctest
14877 can be used to test the monowhite pixel format descriptor definition.
14881 Display sample values of color channels. Mainly useful for checking color
14882 and levels. Minimum supported resolution is 640x480.
14884 The filters accept the following options:
14888 Set scope X position, relative offset on X axis.
14891 Set scope Y position, relative offset on Y axis.
14900 Set window opacity. This window also holds statistics about pixel area.
14903 Set window X position, relative offset on X axis.
14906 Set window Y position, relative offset on Y axis.
14911 Enable the specified chain of postprocessing subfilters using libpostproc. This
14912 library should be automatically selected with a GPL build (@code{--enable-gpl}).
14913 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
14914 Each subfilter and some options have a short and a long name that can be used
14915 interchangeably, i.e. dr/dering are the same.
14917 The filters accept the following options:
14921 Set postprocessing subfilters string.
14924 All subfilters share common options to determine their scope:
14928 Honor the quality commands for this subfilter.
14931 Do chrominance filtering, too (default).
14934 Do luminance filtering only (no chrominance).
14937 Do chrominance filtering only (no luminance).
14940 These options can be appended after the subfilter name, separated by a '|'.
14942 Available subfilters are:
14945 @item hb/hdeblock[|difference[|flatness]]
14946 Horizontal deblocking filter
14949 Difference factor where higher values mean more deblocking (default: @code{32}).
14951 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14954 @item vb/vdeblock[|difference[|flatness]]
14955 Vertical deblocking filter
14958 Difference factor where higher values mean more deblocking (default: @code{32}).
14960 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14963 @item ha/hadeblock[|difference[|flatness]]
14964 Accurate horizontal deblocking filter
14967 Difference factor where higher values mean more deblocking (default: @code{32}).
14969 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14972 @item va/vadeblock[|difference[|flatness]]
14973 Accurate vertical deblocking filter
14976 Difference factor where higher values mean more deblocking (default: @code{32}).
14978 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14982 The horizontal and vertical deblocking filters share the difference and
14983 flatness values so you cannot set different horizontal and vertical
14987 @item h1/x1hdeblock
14988 Experimental horizontal deblocking filter
14990 @item v1/x1vdeblock
14991 Experimental vertical deblocking filter
14996 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
14999 larger -> stronger filtering
15001 larger -> stronger filtering
15003 larger -> stronger filtering
15006 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15009 Stretch luminance to @code{0-255}.
15012 @item lb/linblenddeint
15013 Linear blend deinterlacing filter that deinterlaces the given block by
15014 filtering all lines with a @code{(1 2 1)} filter.
15016 @item li/linipoldeint
15017 Linear interpolating deinterlacing filter that deinterlaces the given block by
15018 linearly interpolating every second line.
15020 @item ci/cubicipoldeint
15021 Cubic interpolating deinterlacing filter deinterlaces the given block by
15022 cubically interpolating every second line.
15024 @item md/mediandeint
15025 Median deinterlacing filter that deinterlaces the given block by applying a
15026 median filter to every second line.
15028 @item fd/ffmpegdeint
15029 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15030 second line with a @code{(-1 4 2 4 -1)} filter.
15033 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15034 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15036 @item fq/forceQuant[|quantizer]
15037 Overrides the quantizer table from the input with the constant quantizer you
15045 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15048 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15051 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15054 @subsection Examples
15058 Apply horizontal and vertical deblocking, deringing and automatic
15059 brightness/contrast:
15065 Apply default filters without brightness/contrast correction:
15071 Apply default filters and temporal denoiser:
15073 pp=default/tmpnoise|1|2|3
15077 Apply deblocking on luminance only, and switch vertical deblocking on or off
15078 automatically depending on available CPU time:
15085 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15086 similar to spp = 6 with 7 point DCT, where only the center sample is
15089 The filter accepts the following options:
15093 Force a constant quantization parameter. It accepts an integer in range
15094 0 to 63. If not set, the filter will use the QP from the video stream
15098 Set thresholding mode. Available modes are:
15102 Set hard thresholding.
15104 Set soft thresholding (better de-ringing effect, but likely blurrier).
15106 Set medium thresholding (good results, default).
15110 @section premultiply
15111 Apply alpha premultiply effect to input video stream using first plane
15112 of second stream as alpha.
15114 Both streams must have same dimensions and same pixel format.
15116 The filter accepts the following option:
15120 Set which planes will be processed, unprocessed planes will be copied.
15121 By default value 0xf, all planes will be processed.
15124 Do not require 2nd input for processing, instead use alpha plane from input stream.
15128 Apply prewitt operator to input video stream.
15130 The filter accepts the following option:
15134 Set which planes will be processed, unprocessed planes will be copied.
15135 By default value 0xf, all planes will be processed.
15138 Set value which will be multiplied with filtered result.
15141 Set value which will be added to filtered result.
15144 @section pseudocolor
15146 Alter frame colors in video with pseudocolors.
15148 This filter accepts the following options:
15152 set pixel first component expression
15155 set pixel second component expression
15158 set pixel third component expression
15161 set pixel fourth component expression, corresponds to the alpha component
15164 set component to use as base for altering colors
15167 Each of them specifies the expression to use for computing the lookup table for
15168 the corresponding pixel component values.
15170 The expressions can contain the following constants and functions:
15175 The input width and height.
15178 The input value for the pixel component.
15180 @item ymin, umin, vmin, amin
15181 The minimum allowed component value.
15183 @item ymax, umax, vmax, amax
15184 The maximum allowed component value.
15187 All expressions default to "val".
15189 @subsection Examples
15193 Change too high luma values to gradient:
15195 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'"
15201 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15202 Ratio) between two input videos.
15204 This filter takes in input two input videos, the first input is
15205 considered the "main" source and is passed unchanged to the
15206 output. The second input is used as a "reference" video for computing
15209 Both video inputs must have the same resolution and pixel format for
15210 this filter to work correctly. Also it assumes that both inputs
15211 have the same number of frames, which are compared one by one.
15213 The obtained average PSNR is printed through the logging system.
15215 The filter stores the accumulated MSE (mean squared error) of each
15216 frame, and at the end of the processing it is averaged across all frames
15217 equally, and the following formula is applied to obtain the PSNR:
15220 PSNR = 10*log10(MAX^2/MSE)
15223 Where MAX is the average of the maximum values of each component of the
15226 The description of the accepted parameters follows.
15229 @item stats_file, f
15230 If specified the filter will use the named file to save the PSNR of
15231 each individual frame. When filename equals "-" the data is sent to
15234 @item stats_version
15235 Specifies which version of the stats file format to use. Details of
15236 each format are written below.
15237 Default value is 1.
15239 @item stats_add_max
15240 Determines whether the max value is output to the stats log.
15241 Default value is 0.
15242 Requires stats_version >= 2. If this is set and stats_version < 2,
15243 the filter will return an error.
15246 This filter also supports the @ref{framesync} options.
15248 The file printed if @var{stats_file} is selected, contains a sequence of
15249 key/value pairs of the form @var{key}:@var{value} for each compared
15252 If a @var{stats_version} greater than 1 is specified, a header line precedes
15253 the list of per-frame-pair stats, with key value pairs following the frame
15254 format with the following parameters:
15257 @item psnr_log_version
15258 The version of the log file format. Will match @var{stats_version}.
15261 A comma separated list of the per-frame-pair parameters included in
15265 A description of each shown per-frame-pair parameter follows:
15269 sequential number of the input frame, starting from 1
15272 Mean Square Error pixel-by-pixel average difference of the compared
15273 frames, averaged over all the image components.
15275 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15276 Mean Square Error pixel-by-pixel average difference of the compared
15277 frames for the component specified by the suffix.
15279 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15280 Peak Signal to Noise ratio of the compared frames for the component
15281 specified by the suffix.
15283 @item max_avg, max_y, max_u, max_v
15284 Maximum allowed value for each channel, and average over all
15288 @subsection Examples
15293 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15294 [main][ref] psnr="stats_file=stats.log" [out]
15297 On this example the input file being processed is compared with the
15298 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15299 is stored in @file{stats.log}.
15302 Another example with different containers:
15304 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 -
15311 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15312 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15315 The pullup filter is designed to take advantage of future context in making
15316 its decisions. This filter is stateless in the sense that it does not lock
15317 onto a pattern to follow, but it instead looks forward to the following
15318 fields in order to identify matches and rebuild progressive frames.
15320 To produce content with an even framerate, insert the fps filter after
15321 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15322 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15324 The filter accepts the following options:
15331 These options set the amount of "junk" to ignore at the left, right, top, and
15332 bottom of the image, respectively. Left and right are in units of 8 pixels,
15333 while top and bottom are in units of 2 lines.
15334 The default is 8 pixels on each side.
15337 Set the strict breaks. Setting this option to 1 will reduce the chances of
15338 filter generating an occasional mismatched frame, but it may also cause an
15339 excessive number of frames to be dropped during high motion sequences.
15340 Conversely, setting it to -1 will make filter match fields more easily.
15341 This may help processing of video where there is slight blurring between
15342 the fields, but may also cause there to be interlaced frames in the output.
15343 Default value is @code{0}.
15346 Set the metric plane to use. It accepts the following values:
15352 Use chroma blue plane.
15355 Use chroma red plane.
15358 This option may be set to use chroma plane instead of the default luma plane
15359 for doing filter's computations. This may improve accuracy on very clean
15360 source material, but more likely will decrease accuracy, especially if there
15361 is chroma noise (rainbow effect) or any grayscale video.
15362 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15363 load and make pullup usable in realtime on slow machines.
15366 For best results (without duplicated frames in the output file) it is
15367 necessary to change the output frame rate. For example, to inverse
15368 telecine NTSC input:
15370 ffmpeg -i input -vf pullup -r 24000/1001 ...
15375 Change video quantization parameters (QP).
15377 The filter accepts the following option:
15381 Set expression for quantization parameter.
15384 The expression is evaluated through the eval API and can contain, among others,
15385 the following constants:
15389 1 if index is not 129, 0 otherwise.
15392 Sequential index starting from -129 to 128.
15395 @subsection Examples
15399 Some equation like:
15407 Flush video frames from internal cache of frames into a random order.
15408 No frame is discarded.
15409 Inspired by @ref{frei0r} nervous filter.
15413 Set size in number of frames of internal cache, in range from @code{2} to
15414 @code{512}. Default is @code{30}.
15417 Set seed for random number generator, must be an integer included between
15418 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15419 less than @code{0}, the filter will try to use a good random seed on a
15423 @section readeia608
15425 Read closed captioning (EIA-608) information from the top lines of a video frame.
15427 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15428 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15429 with EIA-608 data (starting from 0). A description of each metadata value follows:
15432 @item lavfi.readeia608.X.cc
15433 The two bytes stored as EIA-608 data (printed in hexadecimal).
15435 @item lavfi.readeia608.X.line
15436 The number of the line on which the EIA-608 data was identified and read.
15439 This filter accepts the following options:
15443 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15446 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15449 Set the ratio of width reserved for sync code detection.
15450 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
15453 Enable checking the parity bit. In the event of a parity error, the filter will output
15454 @code{0x00} for that character. Default is false.
15457 Lowpass lines prior to further processing. Default is enabled.
15460 @subsection Examples
15464 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
15466 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
15472 Read vertical interval timecode (VITC) information from the top lines of a
15475 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
15476 timecode value, if a valid timecode has been detected. Further metadata key
15477 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
15478 timecode data has been found or not.
15480 This filter accepts the following options:
15484 Set the maximum number of lines to scan for VITC data. If the value is set to
15485 @code{-1} the full video frame is scanned. Default is @code{45}.
15488 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
15489 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
15492 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
15493 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
15496 @subsection Examples
15500 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
15501 draw @code{--:--:--:--} as a placeholder:
15503 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
15509 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
15511 Destination pixel at position (X, Y) will be picked from source (x, y) position
15512 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
15513 value for pixel will be used for destination pixel.
15515 Xmap and Ymap input video streams must be of same dimensions. Output video stream
15516 will have Xmap/Ymap video stream dimensions.
15517 Xmap and Ymap input video streams are 16bit depth, single channel.
15521 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
15522 Default is @code{color}.
15525 Specify the color of the unmapped pixels. For the syntax of this option,
15526 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15527 manual,ffmpeg-utils}. Default color is @code{black}.
15530 @section removegrain
15532 The removegrain filter is a spatial denoiser for progressive video.
15536 Set mode for the first plane.
15539 Set mode for the second plane.
15542 Set mode for the third plane.
15545 Set mode for the fourth plane.
15548 Range of mode is from 0 to 24. Description of each mode follows:
15552 Leave input plane unchanged. Default.
15555 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
15558 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
15561 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
15564 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
15565 This is equivalent to a median filter.
15568 Line-sensitive clipping giving the minimal change.
15571 Line-sensitive clipping, intermediate.
15574 Line-sensitive clipping, intermediate.
15577 Line-sensitive clipping, intermediate.
15580 Line-sensitive clipping on a line where the neighbours pixels are the closest.
15583 Replaces the target pixel with the closest neighbour.
15586 [1 2 1] horizontal and vertical kernel blur.
15592 Bob mode, interpolates top field from the line where the neighbours
15593 pixels are the closest.
15596 Bob mode, interpolates bottom field from the line where the neighbours
15597 pixels are the closest.
15600 Bob mode, interpolates top field. Same as 13 but with a more complicated
15601 interpolation formula.
15604 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
15605 interpolation formula.
15608 Clips the pixel with the minimum and maximum of respectively the maximum and
15609 minimum of each pair of opposite neighbour pixels.
15612 Line-sensitive clipping using opposite neighbours whose greatest distance from
15613 the current pixel is minimal.
15616 Replaces the pixel with the average of its 8 neighbours.
15619 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
15622 Clips pixels using the averages of opposite neighbour.
15625 Same as mode 21 but simpler and faster.
15628 Small edge and halo removal, but reputed useless.
15634 @section removelogo
15636 Suppress a TV station logo, using an image file to determine which
15637 pixels comprise the logo. It works by filling in the pixels that
15638 comprise the logo with neighboring pixels.
15640 The filter accepts the following options:
15644 Set the filter bitmap file, which can be any image format supported by
15645 libavformat. The width and height of the image file must match those of the
15646 video stream being processed.
15649 Pixels in the provided bitmap image with a value of zero are not
15650 considered part of the logo, non-zero pixels are considered part of
15651 the logo. If you use white (255) for the logo and black (0) for the
15652 rest, you will be safe. For making the filter bitmap, it is
15653 recommended to take a screen capture of a black frame with the logo
15654 visible, and then using a threshold filter followed by the erode
15655 filter once or twice.
15657 If needed, little splotches can be fixed manually. Remember that if
15658 logo pixels are not covered, the filter quality will be much
15659 reduced. Marking too many pixels as part of the logo does not hurt as
15660 much, but it will increase the amount of blurring needed to cover over
15661 the image and will destroy more information than necessary, and extra
15662 pixels will slow things down on a large logo.
15664 @section repeatfields
15666 This filter uses the repeat_field flag from the Video ES headers and hard repeats
15667 fields based on its value.
15671 Reverse a video clip.
15673 Warning: This filter requires memory to buffer the entire clip, so trimming
15676 @subsection Examples
15680 Take the first 5 seconds of a clip, and reverse it.
15687 Shift R/G/B/A pixels horizontally and/or vertically.
15689 The filter accepts the following options:
15692 Set amount to shift red horizontally.
15694 Set amount to shift red vertically.
15696 Set amount to shift green horizontally.
15698 Set amount to shift green vertically.
15700 Set amount to shift blue horizontally.
15702 Set amount to shift blue vertically.
15704 Set amount to shift alpha horizontally.
15706 Set amount to shift alpha vertically.
15708 Set edge mode, can be @var{smear}, default, or @var{warp}.
15711 @subsection Commands
15713 This filter supports the all above options as @ref{commands}.
15716 Apply roberts cross operator to input video stream.
15718 The filter accepts the following option:
15722 Set which planes will be processed, unprocessed planes will be copied.
15723 By default value 0xf, all planes will be processed.
15726 Set value which will be multiplied with filtered result.
15729 Set value which will be added to filtered result.
15734 Rotate video by an arbitrary angle expressed in radians.
15736 The filter accepts the following options:
15738 A description of the optional parameters follows.
15741 Set an expression for the angle by which to rotate the input video
15742 clockwise, expressed as a number of radians. A negative value will
15743 result in a counter-clockwise rotation. By default it is set to "0".
15745 This expression is evaluated for each frame.
15748 Set the output width expression, default value is "iw".
15749 This expression is evaluated just once during configuration.
15752 Set the output height expression, default value is "ih".
15753 This expression is evaluated just once during configuration.
15756 Enable bilinear interpolation if set to 1, a value of 0 disables
15757 it. Default value is 1.
15760 Set the color used to fill the output area not covered by the rotated
15761 image. For the general syntax of this option, check the
15762 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
15763 If the special value "none" is selected then no
15764 background is printed (useful for example if the background is never shown).
15766 Default value is "black".
15769 The expressions for the angle and the output size can contain the
15770 following constants and functions:
15774 sequential number of the input frame, starting from 0. It is always NAN
15775 before the first frame is filtered.
15778 time in seconds of the input frame, it is set to 0 when the filter is
15779 configured. It is always NAN before the first frame is filtered.
15783 horizontal and vertical chroma subsample values. For example for the
15784 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15788 the input video width and height
15792 the output width and height, that is the size of the padded area as
15793 specified by the @var{width} and @var{height} expressions
15797 the minimal width/height required for completely containing the input
15798 video rotated by @var{a} radians.
15800 These are only available when computing the @option{out_w} and
15801 @option{out_h} expressions.
15804 @subsection Examples
15808 Rotate the input by PI/6 radians clockwise:
15814 Rotate the input by PI/6 radians counter-clockwise:
15820 Rotate the input by 45 degrees clockwise:
15826 Apply a constant rotation with period T, starting from an angle of PI/3:
15828 rotate=PI/3+2*PI*t/T
15832 Make the input video rotation oscillating with a period of T
15833 seconds and an amplitude of A radians:
15835 rotate=A*sin(2*PI/T*t)
15839 Rotate the video, output size is chosen so that the whole rotating
15840 input video is always completely contained in the output:
15842 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
15846 Rotate the video, reduce the output size so that no background is ever
15849 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
15853 @subsection Commands
15855 The filter supports the following commands:
15859 Set the angle expression.
15860 The command accepts the same syntax of the corresponding option.
15862 If the specified expression is not valid, it is kept at its current
15868 Apply Shape Adaptive Blur.
15870 The filter accepts the following options:
15873 @item luma_radius, lr
15874 Set luma blur filter strength, must be a value in range 0.1-4.0, default
15875 value is 1.0. A greater value will result in a more blurred image, and
15876 in slower processing.
15878 @item luma_pre_filter_radius, lpfr
15879 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
15882 @item luma_strength, ls
15883 Set luma maximum difference between pixels to still be considered, must
15884 be a value in the 0.1-100.0 range, default value is 1.0.
15886 @item chroma_radius, cr
15887 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
15888 greater value will result in a more blurred image, and in slower
15891 @item chroma_pre_filter_radius, cpfr
15892 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
15894 @item chroma_strength, cs
15895 Set chroma maximum difference between pixels to still be considered,
15896 must be a value in the -0.9-100.0 range.
15899 Each chroma option value, if not explicitly specified, is set to the
15900 corresponding luma option value.
15905 Scale (resize) the input video, using the libswscale library.
15907 The scale filter forces the output display aspect ratio to be the same
15908 of the input, by changing the output sample aspect ratio.
15910 If the input image format is different from the format requested by
15911 the next filter, the scale filter will convert the input to the
15914 @subsection Options
15915 The filter accepts the following options, or any of the options
15916 supported by the libswscale scaler.
15918 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
15919 the complete list of scaler options.
15924 Set the output video dimension expression. Default value is the input
15927 If the @var{width} or @var{w} value is 0, the input width is used for
15928 the output. If the @var{height} or @var{h} value is 0, the input height
15929 is used for the output.
15931 If one and only one of the values is -n with n >= 1, the scale filter
15932 will use a value that maintains the aspect ratio of the input image,
15933 calculated from the other specified dimension. After that it will,
15934 however, make sure that the calculated dimension is divisible by n and
15935 adjust the value if necessary.
15937 If both values are -n with n >= 1, the behavior will be identical to
15938 both values being set to 0 as previously detailed.
15940 See below for the list of accepted constants for use in the dimension
15944 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
15948 Only evaluate expressions once during the filter initialization or when a command is processed.
15951 Evaluate expressions for each incoming frame.
15955 Default value is @samp{init}.
15959 Set the interlacing mode. It accepts the following values:
15963 Force interlaced aware scaling.
15966 Do not apply interlaced scaling.
15969 Select interlaced aware scaling depending on whether the source frames
15970 are flagged as interlaced or not.
15973 Default value is @samp{0}.
15976 Set libswscale scaling flags. See
15977 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15978 complete list of values. If not explicitly specified the filter applies
15982 @item param0, param1
15983 Set libswscale input parameters for scaling algorithms that need them. See
15984 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15985 complete documentation. If not explicitly specified the filter applies
15991 Set the video size. For the syntax of this option, check the
15992 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
15994 @item in_color_matrix
15995 @item out_color_matrix
15996 Set in/output YCbCr color space type.
15998 This allows the autodetected value to be overridden as well as allows forcing
15999 a specific value used for the output and encoder.
16001 If not specified, the color space type depends on the pixel format.
16007 Choose automatically.
16010 Format conforming to International Telecommunication Union (ITU)
16011 Recommendation BT.709.
16014 Set color space conforming to the United States Federal Communications
16015 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16020 Set color space conforming to:
16024 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16027 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16030 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16035 Set color space conforming to SMPTE ST 240:1999.
16038 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16043 Set in/output YCbCr sample range.
16045 This allows the autodetected value to be overridden as well as allows forcing
16046 a specific value used for the output and encoder. If not specified, the
16047 range depends on the pixel format. Possible values:
16051 Choose automatically.
16054 Set full range (0-255 in case of 8-bit luma).
16056 @item mpeg/limited/tv
16057 Set "MPEG" range (16-235 in case of 8-bit luma).
16060 @item force_original_aspect_ratio
16061 Enable decreasing or increasing output video width or height if necessary to
16062 keep the original aspect ratio. Possible values:
16066 Scale the video as specified and disable this feature.
16069 The output video dimensions will automatically be decreased if needed.
16072 The output video dimensions will automatically be increased if needed.
16076 One useful instance of this option is that when you know a specific device's
16077 maximum allowed resolution, you can use this to limit the output video to
16078 that, while retaining the aspect ratio. For example, device A allows
16079 1280x720 playback, and your video is 1920x800. Using this option (set it to
16080 decrease) and specifying 1280x720 to the command line makes the output
16083 Please note that this is a different thing than specifying -1 for @option{w}
16084 or @option{h}, you still need to specify the output resolution for this option
16087 @item force_divisible_by
16088 Ensures that both the output dimensions, width and height, are divisible by the
16089 given integer when used together with @option{force_original_aspect_ratio}. This
16090 works similar to using @code{-n} in the @option{w} and @option{h} options.
16092 This option respects the value set for @option{force_original_aspect_ratio},
16093 increasing or decreasing the resolution accordingly. The video's aspect ratio
16094 may be slightly modified.
16096 This option can be handy if you need to have a video fit within or exceed
16097 a defined resolution using @option{force_original_aspect_ratio} but also have
16098 encoder restrictions on width or height divisibility.
16102 The values of the @option{w} and @option{h} options are expressions
16103 containing the following constants:
16108 The input width and height
16112 These are the same as @var{in_w} and @var{in_h}.
16116 The output (scaled) width and height
16120 These are the same as @var{out_w} and @var{out_h}
16123 The same as @var{iw} / @var{ih}
16126 input sample aspect ratio
16129 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16133 horizontal and vertical input chroma subsample values. For example for the
16134 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16138 horizontal and vertical output chroma subsample values. For example for the
16139 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16142 The (sequential) number of the input frame, starting from 0.
16143 Only available with @code{eval=frame}.
16146 The presentation timestamp of the input frame, expressed as a number of
16147 seconds. Only available with @code{eval=frame}.
16150 The position (byte offset) of the frame in the input stream, or NaN if
16151 this information is unavailable and/or meaningless (for example in case of synthetic video).
16152 Only available with @code{eval=frame}.
16155 @subsection Examples
16159 Scale the input video to a size of 200x100
16164 This is equivalent to:
16175 Specify a size abbreviation for the output size:
16180 which can also be written as:
16186 Scale the input to 2x:
16188 scale=w=2*iw:h=2*ih
16192 The above is the same as:
16194 scale=2*in_w:2*in_h
16198 Scale the input to 2x with forced interlaced scaling:
16200 scale=2*iw:2*ih:interl=1
16204 Scale the input to half size:
16206 scale=w=iw/2:h=ih/2
16210 Increase the width, and set the height to the same size:
16216 Seek Greek harmony:
16223 Increase the height, and set the width to 3/2 of the height:
16225 scale=w=3/2*oh:h=3/5*ih
16229 Increase the size, making the size a multiple of the chroma
16232 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16236 Increase the width to a maximum of 500 pixels,
16237 keeping the same aspect ratio as the input:
16239 scale=w='min(500\, iw*3/2):h=-1'
16243 Make pixels square by combining scale and setsar:
16245 scale='trunc(ih*dar):ih',setsar=1/1
16249 Make pixels square by combining scale and setsar,
16250 making sure the resulting resolution is even (required by some codecs):
16252 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16256 @subsection Commands
16258 This filter supports the following commands:
16262 Set the output video dimension expression.
16263 The command accepts the same syntax of the corresponding option.
16265 If the specified expression is not valid, it is kept at its current
16271 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16272 format conversion on CUDA video frames. Setting the output width and height
16273 works in the same way as for the @var{scale} filter.
16275 The following additional options are accepted:
16278 The pixel format of the output CUDA frames. If set to the string "same" (the
16279 default), the input format will be kept. Note that automatic format negotiation
16280 and conversion is not yet supported for hardware frames
16283 The interpolation algorithm used for resizing. One of the following:
16290 @item cubic2p_bspline
16291 2-parameter cubic (B=1, C=0)
16293 @item cubic2p_catmullrom
16294 2-parameter cubic (B=0, C=1/2)
16296 @item cubic2p_b05c03
16297 2-parameter cubic (B=1/2, C=3/10)
16305 @item force_original_aspect_ratio
16306 Enable decreasing or increasing output video width or height if necessary to
16307 keep the original aspect ratio. Possible values:
16311 Scale the video as specified and disable this feature.
16314 The output video dimensions will automatically be decreased if needed.
16317 The output video dimensions will automatically be increased if needed.
16321 One useful instance of this option is that when you know a specific device's
16322 maximum allowed resolution, you can use this to limit the output video to
16323 that, while retaining the aspect ratio. For example, device A allows
16324 1280x720 playback, and your video is 1920x800. Using this option (set it to
16325 decrease) and specifying 1280x720 to the command line makes the output
16328 Please note that this is a different thing than specifying -1 for @option{w}
16329 or @option{h}, you still need to specify the output resolution for this option
16332 @item force_divisible_by
16333 Ensures that both the output dimensions, width and height, are divisible by the
16334 given integer when used together with @option{force_original_aspect_ratio}. This
16335 works similar to using @code{-n} in the @option{w} and @option{h} options.
16337 This option respects the value set for @option{force_original_aspect_ratio},
16338 increasing or decreasing the resolution accordingly. The video's aspect ratio
16339 may be slightly modified.
16341 This option can be handy if you need to have a video fit within or exceed
16342 a defined resolution using @option{force_original_aspect_ratio} but also have
16343 encoder restrictions on width or height divisibility.
16349 Scale (resize) the input video, based on a reference video.
16351 See the scale filter for available options, scale2ref supports the same but
16352 uses the reference video instead of the main input as basis. scale2ref also
16353 supports the following additional constants for the @option{w} and
16354 @option{h} options:
16359 The main input video's width and height
16362 The same as @var{main_w} / @var{main_h}
16365 The main input video's sample aspect ratio
16367 @item main_dar, mdar
16368 The main input video's display aspect ratio. Calculated from
16369 @code{(main_w / main_h) * main_sar}.
16373 The main input video's horizontal and vertical chroma subsample values.
16374 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16378 The (sequential) number of the main input frame, starting from 0.
16379 Only available with @code{eval=frame}.
16382 The presentation timestamp of the main input frame, expressed as a number of
16383 seconds. Only available with @code{eval=frame}.
16386 The position (byte offset) of the frame in the main input stream, or NaN if
16387 this information is unavailable and/or meaningless (for example in case of synthetic video).
16388 Only available with @code{eval=frame}.
16391 @subsection Examples
16395 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16397 'scale2ref[b][a];[a][b]overlay'
16401 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16403 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16407 @subsection Commands
16409 This filter supports the following commands:
16413 Set the output video dimension expression.
16414 The command accepts the same syntax of the corresponding option.
16416 If the specified expression is not valid, it is kept at its current
16421 Scroll input video horizontally and/or vertically by constant speed.
16423 The filter accepts the following options:
16425 @item horizontal, h
16426 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16427 Negative values changes scrolling direction.
16430 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16431 Negative values changes scrolling direction.
16434 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16437 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16440 @subsection Commands
16442 This filter supports the following @ref{commands}:
16444 @item horizontal, h
16445 Set the horizontal scrolling speed.
16447 Set the vertical scrolling speed.
16450 @anchor{selectivecolor}
16451 @section selectivecolor
16453 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
16454 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
16455 by the "purity" of the color (that is, how saturated it already is).
16457 This filter is similar to the Adobe Photoshop Selective Color tool.
16459 The filter accepts the following options:
16462 @item correction_method
16463 Select color correction method.
16465 Available values are:
16468 Specified adjustments are applied "as-is" (added/subtracted to original pixel
16471 Specified adjustments are relative to the original component value.
16473 Default is @code{absolute}.
16475 Adjustments for red pixels (pixels where the red component is the maximum)
16477 Adjustments for yellow pixels (pixels where the blue component is the minimum)
16479 Adjustments for green pixels (pixels where the green component is the maximum)
16481 Adjustments for cyan pixels (pixels where the red component is the minimum)
16483 Adjustments for blue pixels (pixels where the blue component is the maximum)
16485 Adjustments for magenta pixels (pixels where the green component is the minimum)
16487 Adjustments for white pixels (pixels where all components are greater than 128)
16489 Adjustments for all pixels except pure black and pure white
16491 Adjustments for black pixels (pixels where all components are lesser than 128)
16493 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
16496 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
16497 4 space separated floating point adjustment values in the [-1,1] range,
16498 respectively to adjust the amount of cyan, magenta, yellow and black for the
16499 pixels of its range.
16501 @subsection Examples
16505 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
16506 increase magenta by 27% in blue areas:
16508 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
16512 Use a Photoshop selective color preset:
16514 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
16518 @anchor{separatefields}
16519 @section separatefields
16521 The @code{separatefields} takes a frame-based video input and splits
16522 each frame into its components fields, producing a new half height clip
16523 with twice the frame rate and twice the frame count.
16525 This filter use field-dominance information in frame to decide which
16526 of each pair of fields to place first in the output.
16527 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
16529 @section setdar, setsar
16531 The @code{setdar} filter sets the Display Aspect Ratio for the filter
16534 This is done by changing the specified Sample (aka Pixel) Aspect
16535 Ratio, according to the following equation:
16537 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
16540 Keep in mind that the @code{setdar} filter does not modify the pixel
16541 dimensions of the video frame. Also, the display aspect ratio set by
16542 this filter may be changed by later filters in the filterchain,
16543 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
16546 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
16547 the filter output video.
16549 Note that as a consequence of the application of this filter, the
16550 output display aspect ratio will change according to the equation
16553 Keep in mind that the sample aspect ratio set by the @code{setsar}
16554 filter may be changed by later filters in the filterchain, e.g. if
16555 another "setsar" or a "setdar" filter is applied.
16557 It accepts the following parameters:
16560 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
16561 Set the aspect ratio used by the filter.
16563 The parameter can be a floating point number string, an expression, or
16564 a string of the form @var{num}:@var{den}, where @var{num} and
16565 @var{den} are the numerator and denominator of the aspect ratio. If
16566 the parameter is not specified, it is assumed the value "0".
16567 In case the form "@var{num}:@var{den}" is used, the @code{:} character
16571 Set the maximum integer value to use for expressing numerator and
16572 denominator when reducing the expressed aspect ratio to a rational.
16573 Default value is @code{100}.
16577 The parameter @var{sar} is an expression containing
16578 the following constants:
16582 These are approximated values for the mathematical constants e
16583 (Euler's number), pi (Greek pi), and phi (the golden ratio).
16586 The input width and height.
16589 These are the same as @var{w} / @var{h}.
16592 The input sample aspect ratio.
16595 The input display aspect ratio. It is the same as
16596 (@var{w} / @var{h}) * @var{sar}.
16599 Horizontal and vertical chroma subsample values. For example, for the
16600 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16603 @subsection Examples
16608 To change the display aspect ratio to 16:9, specify one of the following:
16615 To change the sample aspect ratio to 10:11, specify:
16621 To set a display aspect ratio of 16:9, and specify a maximum integer value of
16622 1000 in the aspect ratio reduction, use the command:
16624 setdar=ratio=16/9:max=1000
16632 Force field for the output video frame.
16634 The @code{setfield} filter marks the interlace type field for the
16635 output frames. It does not change the input frame, but only sets the
16636 corresponding property, which affects how the frame is treated by
16637 following filters (e.g. @code{fieldorder} or @code{yadif}).
16639 The filter accepts the following options:
16644 Available values are:
16648 Keep the same field property.
16651 Mark the frame as bottom-field-first.
16654 Mark the frame as top-field-first.
16657 Mark the frame as progressive.
16664 Force frame parameter for the output video frame.
16666 The @code{setparams} filter marks interlace and color range for the
16667 output frames. It does not change the input frame, but only sets the
16668 corresponding property, which affects how the frame is treated by
16673 Available values are:
16677 Keep the same field property (default).
16680 Mark the frame as bottom-field-first.
16683 Mark the frame as top-field-first.
16686 Mark the frame as progressive.
16690 Available values are:
16694 Keep the same color range property (default).
16696 @item unspecified, unknown
16697 Mark the frame as unspecified color range.
16699 @item limited, tv, mpeg
16700 Mark the frame as limited range.
16702 @item full, pc, jpeg
16703 Mark the frame as full range.
16706 @item color_primaries
16707 Set the color primaries.
16708 Available values are:
16712 Keep the same color primaries property (default).
16729 Set the color transfer.
16730 Available values are:
16734 Keep the same color trc property (default).
16756 Set the colorspace.
16757 Available values are:
16761 Keep the same colorspace property (default).
16774 @item chroma-derived-nc
16775 @item chroma-derived-c
16782 Show a line containing various information for each input video frame.
16783 The input video is not modified.
16785 This filter supports the following options:
16789 Calculate checksums of each plane. By default enabled.
16792 The shown line contains a sequence of key/value pairs of the form
16793 @var{key}:@var{value}.
16795 The following values are shown in the output:
16799 The (sequential) number of the input frame, starting from 0.
16802 The Presentation TimeStamp of the input frame, expressed as a number of
16803 time base units. The time base unit depends on the filter input pad.
16806 The Presentation TimeStamp of the input frame, expressed as a number of
16810 The position of the frame in the input stream, or -1 if this information is
16811 unavailable and/or meaningless (for example in case of synthetic video).
16814 The pixel format name.
16817 The sample aspect ratio of the input frame, expressed in the form
16818 @var{num}/@var{den}.
16821 The size of the input frame. For the syntax of this option, check the
16822 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16825 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
16826 for bottom field first).
16829 This is 1 if the frame is a key frame, 0 otherwise.
16832 The picture type of the input frame ("I" for an I-frame, "P" for a
16833 P-frame, "B" for a B-frame, or "?" for an unknown type).
16834 Also refer to the documentation of the @code{AVPictureType} enum and of
16835 the @code{av_get_picture_type_char} function defined in
16836 @file{libavutil/avutil.h}.
16839 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
16841 @item plane_checksum
16842 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
16843 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
16846 The mean value of pixels in each plane of the input frame, expressed in the form
16847 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
16850 The standard deviation of pixel values in each plane of the input frame, expressed
16851 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
16855 @section showpalette
16857 Displays the 256 colors palette of each frame. This filter is only relevant for
16858 @var{pal8} pixel format frames.
16860 It accepts the following option:
16864 Set the size of the box used to represent one palette color entry. Default is
16865 @code{30} (for a @code{30x30} pixel box).
16868 @section shuffleframes
16870 Reorder and/or duplicate and/or drop video frames.
16872 It accepts the following parameters:
16876 Set the destination indexes of input frames.
16877 This is space or '|' separated list of indexes that maps input frames to output
16878 frames. Number of indexes also sets maximal value that each index may have.
16879 '-1' index have special meaning and that is to drop frame.
16882 The first frame has the index 0. The default is to keep the input unchanged.
16884 @subsection Examples
16888 Swap second and third frame of every three frames of the input:
16890 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
16894 Swap 10th and 1st frame of every ten frames of the input:
16896 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
16900 @section shuffleplanes
16902 Reorder and/or duplicate video planes.
16904 It accepts the following parameters:
16909 The index of the input plane to be used as the first output plane.
16912 The index of the input plane to be used as the second output plane.
16915 The index of the input plane to be used as the third output plane.
16918 The index of the input plane to be used as the fourth output plane.
16922 The first plane has the index 0. The default is to keep the input unchanged.
16924 @subsection Examples
16928 Swap the second and third planes of the input:
16930 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
16934 @anchor{signalstats}
16935 @section signalstats
16936 Evaluate various visual metrics that assist in determining issues associated
16937 with the digitization of analog video media.
16939 By default the filter will log these metadata values:
16943 Display the minimal Y value contained within the input frame. Expressed in
16947 Display the Y value at the 10% percentile within the input frame. Expressed in
16951 Display the average Y value within the input frame. Expressed in range of
16955 Display the Y value at the 90% percentile within the input frame. Expressed in
16959 Display the maximum Y value contained within the input frame. Expressed in
16963 Display the minimal U value contained within the input frame. Expressed in
16967 Display the U value at the 10% percentile within the input frame. Expressed in
16971 Display the average U value within the input frame. Expressed in range of
16975 Display the U value at the 90% percentile within the input frame. Expressed in
16979 Display the maximum U value contained within the input frame. Expressed in
16983 Display the minimal V value contained within the input frame. Expressed in
16987 Display the V value at the 10% percentile within the input frame. Expressed in
16991 Display the average V value within the input frame. Expressed in range of
16995 Display the V value at the 90% percentile within the input frame. Expressed in
16999 Display the maximum V value contained within the input frame. Expressed in
17003 Display the minimal saturation value contained within the input frame.
17004 Expressed in range of [0-~181.02].
17007 Display the saturation value at the 10% percentile within the input frame.
17008 Expressed in range of [0-~181.02].
17011 Display the average saturation value within the input frame. Expressed in range
17015 Display the saturation value at the 90% percentile within the input frame.
17016 Expressed in range of [0-~181.02].
17019 Display the maximum saturation value contained within the input frame.
17020 Expressed in range of [0-~181.02].
17023 Display the median value for hue within the input frame. Expressed in range of
17027 Display the average value for hue within the input frame. Expressed in range of
17031 Display the average of sample value difference between all values of the Y
17032 plane in the current frame and corresponding values of the previous input frame.
17033 Expressed in range of [0-255].
17036 Display the average of sample value difference between all values of the U
17037 plane in the current frame and corresponding values of the previous input frame.
17038 Expressed in range of [0-255].
17041 Display the average of sample value difference between all values of the V
17042 plane in the current frame and corresponding values of the previous input frame.
17043 Expressed in range of [0-255].
17046 Display bit depth of Y plane in current frame.
17047 Expressed in range of [0-16].
17050 Display bit depth of U plane in current frame.
17051 Expressed in range of [0-16].
17054 Display bit depth of V plane in current frame.
17055 Expressed in range of [0-16].
17058 The filter accepts the following options:
17064 @option{stat} specify an additional form of image analysis.
17065 @option{out} output video with the specified type of pixel highlighted.
17067 Both options accept the following values:
17071 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17072 unlike the neighboring pixels of the same field. Examples of temporal outliers
17073 include the results of video dropouts, head clogs, or tape tracking issues.
17076 Identify @var{vertical line repetition}. Vertical line repetition includes
17077 similar rows of pixels within a frame. In born-digital video vertical line
17078 repetition is common, but this pattern is uncommon in video digitized from an
17079 analog source. When it occurs in video that results from the digitization of an
17080 analog source it can indicate concealment from a dropout compensator.
17083 Identify pixels that fall outside of legal broadcast range.
17087 Set the highlight color for the @option{out} option. The default color is
17091 @subsection Examples
17095 Output data of various video metrics:
17097 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17101 Output specific data about the minimum and maximum values of the Y plane per frame:
17103 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
17107 Playback video while highlighting pixels that are outside of broadcast range in red.
17109 ffplay example.mov -vf signalstats="out=brng:color=red"
17113 Playback video with signalstats metadata drawn over the frame.
17115 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
17118 The contents of signalstat_drawtext.txt used in the command are:
17121 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
17122 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
17123 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
17124 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
17132 Calculates the MPEG-7 Video Signature. The filter can handle more than one
17133 input. In this case the matching between the inputs can be calculated additionally.
17134 The filter always passes through the first input. The signature of each stream can
17135 be written into a file.
17137 It accepts the following options:
17141 Enable or disable the matching process.
17143 Available values are:
17147 Disable the calculation of a matching (default).
17149 Calculate the matching for the whole video and output whether the whole video
17150 matches or only parts.
17152 Calculate only until a matching is found or the video ends. Should be faster in
17157 Set the number of inputs. The option value must be a non negative integer.
17158 Default value is 1.
17161 Set the path to which the output is written. If there is more than one input,
17162 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
17163 integer), that will be replaced with the input number. If no filename is
17164 specified, no output will be written. This is the default.
17167 Choose the output format.
17169 Available values are:
17173 Use the specified binary representation (default).
17175 Use the specified xml representation.
17179 Set threshold to detect one word as similar. The option value must be an integer
17180 greater than zero. The default value is 9000.
17183 Set threshold to detect all words as similar. The option value must be an integer
17184 greater than zero. The default value is 60000.
17187 Set threshold to detect frames as similar. The option value must be an integer
17188 greater than zero. The default value is 116.
17191 Set the minimum length of a sequence in frames to recognize it as matching
17192 sequence. The option value must be a non negative integer value.
17193 The default value is 0.
17196 Set the minimum relation, that matching frames to all frames must have.
17197 The option value must be a double value between 0 and 1. The default value is 0.5.
17200 @subsection Examples
17204 To calculate the signature of an input video and store it in signature.bin:
17206 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17210 To detect whether two videos match and store the signatures in XML format in
17211 signature0.xml and signature1.xml:
17213 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 -
17221 Blur the input video without impacting the outlines.
17223 It accepts the following options:
17226 @item luma_radius, lr
17227 Set the luma radius. The option value must be a float number in
17228 the range [0.1,5.0] that specifies the variance of the gaussian filter
17229 used to blur the image (slower if larger). Default value is 1.0.
17231 @item luma_strength, ls
17232 Set the luma strength. The option value must be a float number
17233 in the range [-1.0,1.0] that configures the blurring. A value included
17234 in [0.0,1.0] will blur the image whereas a value included in
17235 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17237 @item luma_threshold, lt
17238 Set the luma threshold used as a coefficient to determine
17239 whether a pixel should be blurred or not. The option value must be an
17240 integer in the range [-30,30]. A value of 0 will filter all the image,
17241 a value included in [0,30] will filter flat areas and a value included
17242 in [-30,0] will filter edges. Default value is 0.
17244 @item chroma_radius, cr
17245 Set the chroma radius. The option value must be a float number in
17246 the range [0.1,5.0] that specifies the variance of the gaussian filter
17247 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17249 @item chroma_strength, cs
17250 Set the chroma strength. The option value must be a float number
17251 in the range [-1.0,1.0] that configures the blurring. A value included
17252 in [0.0,1.0] will blur the image whereas a value included in
17253 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17255 @item chroma_threshold, ct
17256 Set the chroma threshold used as a coefficient to determine
17257 whether a pixel should be blurred or not. The option value must be an
17258 integer in the range [-30,30]. A value of 0 will filter all the image,
17259 a value included in [0,30] will filter flat areas and a value included
17260 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17263 If a chroma option is not explicitly set, the corresponding luma value
17267 Apply sobel operator to input video stream.
17269 The filter accepts the following option:
17273 Set which planes will be processed, unprocessed planes will be copied.
17274 By default value 0xf, all planes will be processed.
17277 Set value which will be multiplied with filtered result.
17280 Set value which will be added to filtered result.
17286 Apply a simple postprocessing filter that compresses and decompresses the image
17287 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17288 and average the results.
17290 The filter accepts the following options:
17294 Set quality. This option defines the number of levels for averaging. It accepts
17295 an integer in the range 0-6. If set to @code{0}, the filter will have no
17296 effect. A value of @code{6} means the higher quality. For each increment of
17297 that value the speed drops by a factor of approximately 2. Default value is
17301 Force a constant quantization parameter. If not set, the filter will use the QP
17302 from the video stream (if available).
17305 Set thresholding mode. Available modes are:
17309 Set hard thresholding (default).
17311 Set soft thresholding (better de-ringing effect, but likely blurrier).
17314 @item use_bframe_qp
17315 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17316 option may cause flicker since the B-Frames have often larger QP. Default is
17317 @code{0} (not enabled).
17320 @subsection Commands
17322 This filter supports the following commands:
17324 @item quality, level
17325 Set quality level. The value @code{max} can be used to set the maximum level,
17326 currently @code{6}.
17332 Scale the input by applying one of the super-resolution methods based on
17333 convolutional neural networks. Supported models:
17337 Super-Resolution Convolutional Neural Network model (SRCNN).
17338 See @url{https://arxiv.org/abs/1501.00092}.
17341 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17342 See @url{https://arxiv.org/abs/1609.05158}.
17345 Training scripts as well as scripts for model file (.pb) saving can be found at
17346 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17347 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17349 Native model files (.model) can be generated from TensorFlow model
17350 files (.pb) by using tools/python/convert.py
17352 The filter accepts the following options:
17356 Specify which DNN backend to use for model loading and execution. This option accepts
17357 the following values:
17361 Native implementation of DNN loading and execution.
17364 TensorFlow backend. To enable this backend you
17365 need to install the TensorFlow for C library (see
17366 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17367 @code{--enable-libtensorflow}
17370 Default value is @samp{native}.
17373 Set path to model file specifying network architecture and its parameters.
17374 Note that different backends use different file formats. TensorFlow backend
17375 can load files for both formats, while native backend can load files for only
17379 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17380 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17381 input upscaled using bicubic upscaling with proper scale factor.
17384 This feature can also be finished with @ref{dnn_processing} filter.
17388 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17390 This filter takes in input two input videos, the first input is
17391 considered the "main" source and is passed unchanged to the
17392 output. The second input is used as a "reference" video for computing
17395 Both video inputs must have the same resolution and pixel format for
17396 this filter to work correctly. Also it assumes that both inputs
17397 have the same number of frames, which are compared one by one.
17399 The filter stores the calculated SSIM of each frame.
17401 The description of the accepted parameters follows.
17404 @item stats_file, f
17405 If specified the filter will use the named file to save the SSIM of
17406 each individual frame. When filename equals "-" the data is sent to
17410 The file printed if @var{stats_file} is selected, contains a sequence of
17411 key/value pairs of the form @var{key}:@var{value} for each compared
17414 A description of each shown parameter follows:
17418 sequential number of the input frame, starting from 1
17420 @item Y, U, V, R, G, B
17421 SSIM of the compared frames for the component specified by the suffix.
17424 SSIM of the compared frames for the whole frame.
17427 Same as above but in dB representation.
17430 This filter also supports the @ref{framesync} options.
17432 @subsection Examples
17437 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
17438 [main][ref] ssim="stats_file=stats.log" [out]
17441 On this example the input file being processed is compared with the
17442 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
17443 is stored in @file{stats.log}.
17446 Another example with both psnr and ssim at same time:
17448 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
17452 Another example with different containers:
17454 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 -
17460 Convert between different stereoscopic image formats.
17462 The filters accept the following options:
17466 Set stereoscopic image format of input.
17468 Available values for input image formats are:
17471 side by side parallel (left eye left, right eye right)
17474 side by side crosseye (right eye left, left eye right)
17477 side by side parallel with half width resolution
17478 (left eye left, right eye right)
17481 side by side crosseye with half width resolution
17482 (right eye left, left eye right)
17486 above-below (left eye above, right eye below)
17490 above-below (right eye above, left eye below)
17494 above-below with half height resolution
17495 (left eye above, right eye below)
17499 above-below with half height resolution
17500 (right eye above, left eye below)
17503 alternating frames (left eye first, right eye second)
17506 alternating frames (right eye first, left eye second)
17509 interleaved rows (left eye has top row, right eye starts on next row)
17512 interleaved rows (right eye has top row, left eye starts on next row)
17515 interleaved columns, left eye first
17518 interleaved columns, right eye first
17520 Default value is @samp{sbsl}.
17524 Set stereoscopic image format of output.
17528 side by side parallel (left eye left, right eye right)
17531 side by side crosseye (right eye left, left eye right)
17534 side by side parallel with half width resolution
17535 (left eye left, right eye right)
17538 side by side crosseye with half width resolution
17539 (right eye left, left eye right)
17543 above-below (left eye above, right eye below)
17547 above-below (right eye above, left eye below)
17551 above-below with half height resolution
17552 (left eye above, right eye below)
17556 above-below with half height resolution
17557 (right eye above, left eye below)
17560 alternating frames (left eye first, right eye second)
17563 alternating frames (right eye first, left eye second)
17566 interleaved rows (left eye has top row, right eye starts on next row)
17569 interleaved rows (right eye has top row, left eye starts on next row)
17572 anaglyph red/blue gray
17573 (red filter on left eye, blue filter on right eye)
17576 anaglyph red/green gray
17577 (red filter on left eye, green filter on right eye)
17580 anaglyph red/cyan gray
17581 (red filter on left eye, cyan filter on right eye)
17584 anaglyph red/cyan half colored
17585 (red filter on left eye, cyan filter on right eye)
17588 anaglyph red/cyan color
17589 (red filter on left eye, cyan filter on right eye)
17592 anaglyph red/cyan color optimized with the least squares projection of dubois
17593 (red filter on left eye, cyan filter on right eye)
17596 anaglyph green/magenta gray
17597 (green filter on left eye, magenta filter on right eye)
17600 anaglyph green/magenta half colored
17601 (green filter on left eye, magenta filter on right eye)
17604 anaglyph green/magenta colored
17605 (green filter on left eye, magenta filter on right eye)
17608 anaglyph green/magenta color optimized with the least squares projection of dubois
17609 (green filter on left eye, magenta filter on right eye)
17612 anaglyph yellow/blue gray
17613 (yellow filter on left eye, blue filter on right eye)
17616 anaglyph yellow/blue half colored
17617 (yellow filter on left eye, blue filter on right eye)
17620 anaglyph yellow/blue colored
17621 (yellow filter on left eye, blue filter on right eye)
17624 anaglyph yellow/blue color optimized with the least squares projection of dubois
17625 (yellow filter on left eye, blue filter on right eye)
17628 mono output (left eye only)
17631 mono output (right eye only)
17634 checkerboard, left eye first
17637 checkerboard, right eye first
17640 interleaved columns, left eye first
17643 interleaved columns, right eye first
17649 Default value is @samp{arcd}.
17652 @subsection Examples
17656 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
17662 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
17668 @section streamselect, astreamselect
17669 Select video or audio streams.
17671 The filter accepts the following options:
17675 Set number of inputs. Default is 2.
17678 Set input indexes to remap to outputs.
17681 @subsection Commands
17683 The @code{streamselect} and @code{astreamselect} filter supports the following
17688 Set input indexes to remap to outputs.
17691 @subsection Examples
17695 Select first 5 seconds 1st stream and rest of time 2nd stream:
17697 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
17701 Same as above, but for audio:
17703 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
17710 Draw subtitles on top of input video using the libass library.
17712 To enable compilation of this filter you need to configure FFmpeg with
17713 @code{--enable-libass}. This filter also requires a build with libavcodec and
17714 libavformat to convert the passed subtitles file to ASS (Advanced Substation
17715 Alpha) subtitles format.
17717 The filter accepts the following options:
17721 Set the filename of the subtitle file to read. It must be specified.
17723 @item original_size
17724 Specify the size of the original video, the video for which the ASS file
17725 was composed. For the syntax of this option, check the
17726 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17727 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
17728 correctly scale the fonts if the aspect ratio has been changed.
17731 Set a directory path containing fonts that can be used by the filter.
17732 These fonts will be used in addition to whatever the font provider uses.
17735 Process alpha channel, by default alpha channel is untouched.
17738 Set subtitles input character encoding. @code{subtitles} filter only. Only
17739 useful if not UTF-8.
17741 @item stream_index, si
17742 Set subtitles stream index. @code{subtitles} filter only.
17745 Override default style or script info parameters of the subtitles. It accepts a
17746 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
17749 If the first key is not specified, it is assumed that the first value
17750 specifies the @option{filename}.
17752 For example, to render the file @file{sub.srt} on top of the input
17753 video, use the command:
17758 which is equivalent to:
17760 subtitles=filename=sub.srt
17763 To render the default subtitles stream from file @file{video.mkv}, use:
17765 subtitles=video.mkv
17768 To render the second subtitles stream from that file, use:
17770 subtitles=video.mkv:si=1
17773 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
17774 @code{DejaVu Serif}, use:
17776 subtitles=sub.srt:force_style='FontName=DejaVu Serif,PrimaryColour=&HCCFF0000'
17779 @section super2xsai
17781 Scale the input by 2x and smooth using the Super2xSaI (Scale and
17782 Interpolate) pixel art scaling algorithm.
17784 Useful for enlarging pixel art images without reducing sharpness.
17788 Swap two rectangular objects in video.
17790 This filter accepts the following options:
17800 Set 1st rect x coordinate.
17803 Set 1st rect y coordinate.
17806 Set 2nd rect x coordinate.
17809 Set 2nd rect y coordinate.
17811 All expressions are evaluated once for each frame.
17814 The all options are expressions containing the following constants:
17819 The input width and height.
17822 same as @var{w} / @var{h}
17825 input sample aspect ratio
17828 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
17831 The number of the input frame, starting from 0.
17834 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
17837 the position in the file of the input frame, NAN if unknown
17844 Blend successive video frames.
17850 Apply telecine process to the video.
17852 This filter accepts the following options:
17861 The default value is @code{top}.
17865 A string of numbers representing the pulldown pattern you wish to apply.
17866 The default value is @code{23}.
17870 Some typical patterns:
17875 24p: 2332 (preferred)
17882 24p: 222222222223 ("Euro pulldown")
17887 @section thistogram
17889 Compute and draw a color distribution histogram for the input video across time.
17891 Unlike @ref{histogram} video filter which only shows histogram of single input frame
17892 at certain time, this filter shows also past histograms of number of frames defined
17893 by @code{width} option.
17895 The computed histogram is a representation of the color component
17896 distribution in an image.
17898 The filter accepts the following options:
17902 Set width of single color component output. Default value is @code{0}.
17903 Value of @code{0} means width will be picked from input video.
17904 This also set number of passed histograms to keep.
17905 Allowed range is [0, 8192].
17907 @item display_mode, d
17909 It accepts the following values:
17912 Per color component graphs are placed below each other.
17915 Per color component graphs are placed side by side.
17918 Presents information identical to that in the @code{parade}, except
17919 that the graphs representing color components are superimposed directly
17922 Default is @code{stack}.
17924 @item levels_mode, m
17925 Set mode. Can be either @code{linear}, or @code{logarithmic}.
17926 Default is @code{linear}.
17928 @item components, c
17929 Set what color components to display.
17930 Default is @code{7}.
17933 Set background opacity. Default is @code{0.9}.
17936 Show envelope. Default is disabled.
17939 Set envelope color. Default is @code{gold}.
17944 Apply threshold effect to video stream.
17946 This filter needs four video streams to perform thresholding.
17947 First stream is stream we are filtering.
17948 Second stream is holding threshold values, third stream is holding min values,
17949 and last, fourth stream is holding max values.
17951 The filter accepts the following option:
17955 Set which planes will be processed, unprocessed planes will be copied.
17956 By default value 0xf, all planes will be processed.
17959 For example if first stream pixel's component value is less then threshold value
17960 of pixel component from 2nd threshold stream, third stream value will picked,
17961 otherwise fourth stream pixel component value will be picked.
17963 Using color source filter one can perform various types of thresholding:
17965 @subsection Examples
17969 Binary threshold, using gray color as threshold:
17971 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
17975 Inverted binary threshold, using gray color as threshold:
17977 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
17981 Truncate binary threshold, using gray color as threshold:
17983 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
17987 Threshold to zero, using gray color as threshold:
17989 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
17993 Inverted threshold to zero, using gray color as threshold:
17995 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18000 Select the most representative frame in a given sequence of consecutive frames.
18002 The filter accepts the following options:
18006 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18007 will pick one of them, and then handle the next batch of @var{n} frames until
18008 the end. Default is @code{100}.
18011 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18012 value will result in a higher memory usage, so a high value is not recommended.
18014 @subsection Examples
18018 Extract one picture each 50 frames:
18024 Complete example of a thumbnail creation with @command{ffmpeg}:
18026 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18032 Tile several successive frames together.
18034 The filter accepts the following options:
18039 Set the grid size (i.e. the number of lines and columns). For the syntax of
18040 this option, check the
18041 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18044 Set the maximum number of frames to render in the given area. It must be less
18045 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18046 the area will be used.
18049 Set the outer border margin in pixels.
18052 Set the inner border thickness (i.e. the number of pixels between frames). For
18053 more advanced padding options (such as having different values for the edges),
18054 refer to the pad video filter.
18057 Specify the color of the unused area. For the syntax of this option, check the
18058 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18059 The default value of @var{color} is "black".
18062 Set the number of frames to overlap when tiling several successive frames together.
18063 The value must be between @code{0} and @var{nb_frames - 1}.
18066 Set the number of frames to initially be empty before displaying first output frame.
18067 This controls how soon will one get first output frame.
18068 The value must be between @code{0} and @var{nb_frames - 1}.
18071 @subsection Examples
18075 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
18077 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
18079 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
18080 duplicating each output frame to accommodate the originally detected frame
18084 Display @code{5} pictures in an area of @code{3x2} frames,
18085 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
18086 mixed flat and named options:
18088 tile=3x2:nb_frames=5:padding=7:margin=2
18092 @section tinterlace
18094 Perform various types of temporal field interlacing.
18096 Frames are counted starting from 1, so the first input frame is
18099 The filter accepts the following options:
18104 Specify the mode of the interlacing. This option can also be specified
18105 as a value alone. See below for a list of values for this option.
18107 Available values are:
18111 Move odd frames into the upper field, even into the lower field,
18112 generating a double height frame at half frame rate.
18116 Frame 1 Frame 2 Frame 3 Frame 4
18118 11111 22222 33333 44444
18119 11111 22222 33333 44444
18120 11111 22222 33333 44444
18121 11111 22222 33333 44444
18135 Only output odd frames, even frames are dropped, generating a frame with
18136 unchanged height at half frame rate.
18141 Frame 1 Frame 2 Frame 3 Frame 4
18143 11111 22222 33333 44444
18144 11111 22222 33333 44444
18145 11111 22222 33333 44444
18146 11111 22222 33333 44444
18156 Only output even frames, odd frames are dropped, generating a frame with
18157 unchanged height at half frame rate.
18162 Frame 1 Frame 2 Frame 3 Frame 4
18164 11111 22222 33333 44444
18165 11111 22222 33333 44444
18166 11111 22222 33333 44444
18167 11111 22222 33333 44444
18177 Expand each frame to full height, but pad alternate lines with black,
18178 generating a frame with double height at the same input frame rate.
18183 Frame 1 Frame 2 Frame 3 Frame 4
18185 11111 22222 33333 44444
18186 11111 22222 33333 44444
18187 11111 22222 33333 44444
18188 11111 22222 33333 44444
18191 11111 ..... 33333 .....
18192 ..... 22222 ..... 44444
18193 11111 ..... 33333 .....
18194 ..... 22222 ..... 44444
18195 11111 ..... 33333 .....
18196 ..... 22222 ..... 44444
18197 11111 ..... 33333 .....
18198 ..... 22222 ..... 44444
18202 @item interleave_top, 4
18203 Interleave the upper field from odd frames with the lower field from
18204 even frames, generating a frame with unchanged height at half frame rate.
18209 Frame 1 Frame 2 Frame 3 Frame 4
18211 11111<- 22222 33333<- 44444
18212 11111 22222<- 33333 44444<-
18213 11111<- 22222 33333<- 44444
18214 11111 22222<- 33333 44444<-
18224 @item interleave_bottom, 5
18225 Interleave the lower field from odd frames with the upper field from
18226 even frames, generating a frame with unchanged height at half frame rate.
18231 Frame 1 Frame 2 Frame 3 Frame 4
18233 11111 22222<- 33333 44444<-
18234 11111<- 22222 33333<- 44444
18235 11111 22222<- 33333 44444<-
18236 11111<- 22222 33333<- 44444
18246 @item interlacex2, 6
18247 Double frame rate with unchanged height. Frames are inserted each
18248 containing the second temporal field from the previous input frame and
18249 the first temporal field from the next input frame. This mode relies on
18250 the top_field_first flag. Useful for interlaced video displays with no
18251 field synchronisation.
18256 Frame 1 Frame 2 Frame 3 Frame 4
18258 11111 22222 33333 44444
18259 11111 22222 33333 44444
18260 11111 22222 33333 44444
18261 11111 22222 33333 44444
18264 11111 22222 22222 33333 33333 44444 44444
18265 11111 11111 22222 22222 33333 33333 44444
18266 11111 22222 22222 33333 33333 44444 44444
18267 11111 11111 22222 22222 33333 33333 44444
18272 Move odd frames into the upper field, even into the lower field,
18273 generating a double height frame at same frame rate.
18278 Frame 1 Frame 2 Frame 3 Frame 4
18280 11111 22222 33333 44444
18281 11111 22222 33333 44444
18282 11111 22222 33333 44444
18283 11111 22222 33333 44444
18286 11111 33333 33333 55555
18287 22222 22222 44444 44444
18288 11111 33333 33333 55555
18289 22222 22222 44444 44444
18290 11111 33333 33333 55555
18291 22222 22222 44444 44444
18292 11111 33333 33333 55555
18293 22222 22222 44444 44444
18298 Numeric values are deprecated but are accepted for backward
18299 compatibility reasons.
18301 Default mode is @code{merge}.
18304 Specify flags influencing the filter process.
18306 Available value for @var{flags} is:
18309 @item low_pass_filter, vlpf
18310 Enable linear vertical low-pass filtering in the filter.
18311 Vertical low-pass filtering is required when creating an interlaced
18312 destination from a progressive source which contains high-frequency
18313 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18316 @item complex_filter, cvlpf
18317 Enable complex vertical low-pass filtering.
18318 This will slightly less reduce interlace 'twitter' and Moire
18319 patterning but better retain detail and subjective sharpness impression.
18322 Bypass already interlaced frames, only adjust the frame rate.
18325 Vertical low-pass filtering and bypassing already interlaced frames can only be
18326 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
18332 Mix successive video frames.
18334 A description of the accepted options follows.
18338 The number of successive frames to mix. If unspecified, it defaults to 3.
18341 Specify weight of each input video frame.
18342 Each weight is separated by space. If number of weights is smaller than
18343 number of @var{frames} last specified weight will be used for all remaining
18347 Specify scale, if it is set it will be multiplied with sum
18348 of each weight multiplied with pixel values to give final destination
18349 pixel value. By default @var{scale} is auto scaled to sum of weights.
18352 @subsection Examples
18356 Average 7 successive frames:
18358 tmix=frames=7:weights="1 1 1 1 1 1 1"
18362 Apply simple temporal convolution:
18364 tmix=frames=3:weights="-1 3 -1"
18368 Similar as above but only showing temporal differences:
18370 tmix=frames=3:weights="-1 2 -1":scale=1
18376 Tone map colors from different dynamic ranges.
18378 This filter expects data in single precision floating point, as it needs to
18379 operate on (and can output) out-of-range values. Another filter, such as
18380 @ref{zscale}, is needed to convert the resulting frame to a usable format.
18382 The tonemapping algorithms implemented only work on linear light, so input
18383 data should be linearized beforehand (and possibly correctly tagged).
18386 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
18389 @subsection Options
18390 The filter accepts the following options.
18394 Set the tone map algorithm to use.
18396 Possible values are:
18399 Do not apply any tone map, only desaturate overbright pixels.
18402 Hard-clip any out-of-range values. Use it for perfect color accuracy for
18403 in-range values, while distorting out-of-range values.
18406 Stretch the entire reference gamut to a linear multiple of the display.
18409 Fit a logarithmic transfer between the tone curves.
18412 Preserve overall image brightness with a simple curve, using nonlinear
18413 contrast, which results in flattening details and degrading color accuracy.
18416 Preserve both dark and bright details better than @var{reinhard}, at the cost
18417 of slightly darkening everything. Use it when detail preservation is more
18418 important than color and brightness accuracy.
18421 Smoothly map out-of-range values, while retaining contrast and colors for
18422 in-range material as much as possible. Use it when color accuracy is more
18423 important than detail preservation.
18429 Tune the tone mapping algorithm.
18431 This affects the following algorithms:
18437 Specifies the scale factor to use while stretching.
18441 Specifies the exponent of the function.
18445 Specify an extra linear coefficient to multiply into the signal before clipping.
18449 Specify the local contrast coefficient at the display peak.
18450 Default to 0.5, which means that in-gamut values will be about half as bright
18457 Specify the transition point from linear to mobius transform. Every value
18458 below this point is guaranteed to be mapped 1:1. The higher the value, the
18459 more accurate the result will be, at the cost of losing bright details.
18460 Default to 0.3, which due to the steep initial slope still preserves in-range
18461 colors fairly accurately.
18465 Apply desaturation for highlights that exceed this level of brightness. The
18466 higher the parameter, the more color information will be preserved. This
18467 setting helps prevent unnaturally blown-out colors for super-highlights, by
18468 (smoothly) turning into white instead. This makes images feel more natural,
18469 at the cost of reducing information about out-of-range colors.
18471 The default of 2.0 is somewhat conservative and will mostly just apply to
18472 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
18474 This option works only if the input frame has a supported color tag.
18477 Override signal/nominal/reference peak with this value. Useful when the
18478 embedded peak information in display metadata is not reliable or when tone
18479 mapping from a lower range to a higher range.
18484 Temporarily pad video frames.
18486 The filter accepts the following options:
18490 Specify number of delay frames before input video stream.
18493 Specify number of padding frames after input video stream.
18494 Set to -1 to pad indefinitely.
18497 Set kind of frames added to beginning of stream.
18498 Can be either @var{add} or @var{clone}.
18499 With @var{add} frames of solid-color are added.
18500 With @var{clone} frames are clones of first frame.
18503 Set kind of frames added to end of stream.
18504 Can be either @var{add} or @var{clone}.
18505 With @var{add} frames of solid-color are added.
18506 With @var{clone} frames are clones of last frame.
18508 @item start_duration, stop_duration
18509 Specify the duration of the start/stop delay. See
18510 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18511 for the accepted syntax.
18512 These options override @var{start} and @var{stop}.
18515 Specify the color of the padded area. For the syntax of this option,
18516 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
18517 manual,ffmpeg-utils}.
18519 The default value of @var{color} is "black".
18525 Transpose rows with columns in the input video and optionally flip it.
18527 It accepts the following parameters:
18532 Specify the transposition direction.
18534 Can assume the following values:
18536 @item 0, 4, cclock_flip
18537 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
18545 Rotate by 90 degrees clockwise, that is:
18553 Rotate by 90 degrees counterclockwise, that is:
18560 @item 3, 7, clock_flip
18561 Rotate by 90 degrees clockwise and vertically flip, that is:
18569 For values between 4-7, the transposition is only done if the input
18570 video geometry is portrait and not landscape. These values are
18571 deprecated, the @code{passthrough} option should be used instead.
18573 Numerical values are deprecated, and should be dropped in favor of
18574 symbolic constants.
18577 Do not apply the transposition if the input geometry matches the one
18578 specified by the specified value. It accepts the following values:
18581 Always apply transposition.
18583 Preserve portrait geometry (when @var{height} >= @var{width}).
18585 Preserve landscape geometry (when @var{width} >= @var{height}).
18588 Default value is @code{none}.
18591 For example to rotate by 90 degrees clockwise and preserve portrait
18594 transpose=dir=1:passthrough=portrait
18597 The command above can also be specified as:
18599 transpose=1:portrait
18602 @section transpose_npp
18604 Transpose rows with columns in the input video and optionally flip it.
18605 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
18607 It accepts the following parameters:
18612 Specify the transposition direction.
18614 Can assume the following values:
18617 Rotate by 90 degrees counterclockwise and vertically flip. (default)
18620 Rotate by 90 degrees clockwise.
18623 Rotate by 90 degrees counterclockwise.
18626 Rotate by 90 degrees clockwise and vertically flip.
18630 Do not apply the transposition if the input geometry matches the one
18631 specified by the specified value. It accepts the following values:
18634 Always apply transposition. (default)
18636 Preserve portrait geometry (when @var{height} >= @var{width}).
18638 Preserve landscape geometry (when @var{width} >= @var{height}).
18644 Trim the input so that the output contains one continuous subpart of the input.
18646 It accepts the following parameters:
18649 Specify the time of the start of the kept section, i.e. the frame with the
18650 timestamp @var{start} will be the first frame in the output.
18653 Specify the time of the first frame that will be dropped, i.e. the frame
18654 immediately preceding the one with the timestamp @var{end} will be the last
18655 frame in the output.
18658 This is the same as @var{start}, except this option sets the start timestamp
18659 in timebase units instead of seconds.
18662 This is the same as @var{end}, except this option sets the end timestamp
18663 in timebase units instead of seconds.
18666 The maximum duration of the output in seconds.
18669 The number of the first frame that should be passed to the output.
18672 The number of the first frame that should be dropped.
18675 @option{start}, @option{end}, and @option{duration} are expressed as time
18676 duration specifications; see
18677 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18678 for the accepted syntax.
18680 Note that the first two sets of the start/end options and the @option{duration}
18681 option look at the frame timestamp, while the _frame variants simply count the
18682 frames that pass through the filter. Also note that this filter does not modify
18683 the timestamps. If you wish for the output timestamps to start at zero, insert a
18684 setpts filter after the trim filter.
18686 If multiple start or end options are set, this filter tries to be greedy and
18687 keep all the frames that match at least one of the specified constraints. To keep
18688 only the part that matches all the constraints at once, chain multiple trim
18691 The defaults are such that all the input is kept. So it is possible to set e.g.
18692 just the end values to keep everything before the specified time.
18697 Drop everything except the second minute of input:
18699 ffmpeg -i INPUT -vf trim=60:120
18703 Keep only the first second:
18705 ffmpeg -i INPUT -vf trim=duration=1
18710 @section unpremultiply
18711 Apply alpha unpremultiply effect to input video stream using first plane
18712 of second stream as alpha.
18714 Both streams must have same dimensions and same pixel format.
18716 The filter accepts the following option:
18720 Set which planes will be processed, unprocessed planes will be copied.
18721 By default value 0xf, all planes will be processed.
18723 If the format has 1 or 2 components, then luma is bit 0.
18724 If the format has 3 or 4 components:
18725 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
18726 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
18727 If present, the alpha channel is always the last bit.
18730 Do not require 2nd input for processing, instead use alpha plane from input stream.
18736 Sharpen or blur the input video.
18738 It accepts the following parameters:
18741 @item luma_msize_x, lx
18742 Set the luma matrix horizontal size. It must be an odd integer between
18743 3 and 23. The default value is 5.
18745 @item luma_msize_y, ly
18746 Set the luma matrix vertical size. It must be an odd integer between 3
18747 and 23. The default value is 5.
18749 @item luma_amount, la
18750 Set the luma effect strength. It must be a floating point number, reasonable
18751 values lay between -1.5 and 1.5.
18753 Negative values will blur the input video, while positive values will
18754 sharpen it, a value of zero will disable the effect.
18756 Default value is 1.0.
18758 @item chroma_msize_x, cx
18759 Set the chroma matrix horizontal size. It must be an odd integer
18760 between 3 and 23. The default value is 5.
18762 @item chroma_msize_y, cy
18763 Set the chroma matrix vertical size. It must be an odd integer
18764 between 3 and 23. The default value is 5.
18766 @item chroma_amount, ca
18767 Set the chroma effect strength. It must be a floating point number, reasonable
18768 values lay between -1.5 and 1.5.
18770 Negative values will blur the input video, while positive values will
18771 sharpen it, a value of zero will disable the effect.
18773 Default value is 0.0.
18777 All parameters are optional and default to the equivalent of the
18778 string '5:5:1.0:5:5:0.0'.
18780 @subsection Examples
18784 Apply strong luma sharpen effect:
18786 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
18790 Apply a strong blur of both luma and chroma parameters:
18792 unsharp=7:7:-2:7:7:-2
18798 Apply ultra slow/simple postprocessing filter that compresses and decompresses
18799 the image at several (or - in the case of @option{quality} level @code{8} - all)
18800 shifts and average the results.
18802 The way this differs from the behavior of spp is that uspp actually encodes &
18803 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
18804 DCT similar to MJPEG.
18806 The filter accepts the following options:
18810 Set quality. This option defines the number of levels for averaging. It accepts
18811 an integer in the range 0-8. If set to @code{0}, the filter will have no
18812 effect. A value of @code{8} means the higher quality. For each increment of
18813 that value the speed drops by a factor of approximately 2. Default value is
18817 Force a constant quantization parameter. If not set, the filter will use the QP
18818 from the video stream (if available).
18823 Convert 360 videos between various formats.
18825 The filter accepts the following options:
18831 Set format of the input/output video.
18839 Equirectangular projection.
18844 Cubemap with 3x2/6x1/1x6 layout.
18846 Format specific options:
18851 Set padding proportion for the input/output cubemap. Values in decimals.
18858 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)
18861 Default value is @b{@samp{0}}.
18865 Set fixed padding for the input/output cubemap. Values in pixels.
18867 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
18871 Set order of faces for the input/output cubemap. Choose one direction for each position.
18873 Designation of directions:
18889 Default value is @b{@samp{rludfb}}.
18893 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
18895 Designation of angles:
18898 0 degrees clockwise
18900 90 degrees clockwise
18902 180 degrees clockwise
18904 270 degrees clockwise
18907 Default value is @b{@samp{000000}}.
18911 Equi-Angular Cubemap.
18918 Format specific options:
18923 Set output horizontal/vertical/diagonal field of view. Values in degrees.
18925 If diagonal field of view is set it overrides horizontal and vertical field of view.
18930 Set input horizontal/vertical/diagonal field of view. Values in degrees.
18932 If diagonal field of view is set it overrides horizontal and vertical field of view.
18938 Format specific options:
18942 Set padding proportion. Values in decimals.
18952 Default value is @b{@samp{0}}.
18958 Facebook's 360 formats.
18961 Stereographic format.
18963 Format specific options:
18968 Set output horizontal/vertical/diagonal field of view. Values in degrees.
18970 If diagonal field of view is set it overrides horizontal and vertical field of view.
18975 Set input horizontal/vertical/diagonal field of view. Values in degrees.
18977 If diagonal field of view is set it overrides horizontal and vertical field of view.
18984 Ball format, gives significant distortion toward the back.
18987 Hammer-Aitoff map projection format.
18990 Sinusoidal map projection format.
18993 Fisheye projection.
18995 Format specific options:
19000 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19002 If diagonal field of view is set it overrides horizontal and vertical field of view.
19007 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19009 If diagonal field of view is set it overrides horizontal and vertical field of view.
19013 Pannini projection. @i{(output only)}
19015 Format specific options:
19018 Set pannini parameter.
19022 Cylindrical projection.
19024 Format specific options:
19029 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19031 If diagonal field of view is set it overrides horizontal and vertical field of view.
19036 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19038 If diagonal field of view is set it overrides horizontal and vertical field of view.
19042 Perspective projection. @i{(output only)}
19044 Format specific options:
19047 Set perspective parameter.
19051 Tetrahedron projection.
19054 Truncated square pyramid projection.
19058 Half equirectangular projection.
19062 Set interpolation method.@*
19063 @i{Note: more complex interpolation methods require much more memory to run.}
19073 Bilinear interpolation.
19076 Bicubic interpolation.
19079 Lanczos interpolation.
19082 Spline16 interpolation.
19085 Gaussian interpolation.
19088 Default value is @b{@samp{line}}.
19092 Set the output video resolution.
19094 Default resolution depends on formats.
19098 Set the input/output stereo format.
19109 Default value is @b{@samp{2d}} for input and output format.
19114 Set rotation for the output video. Values in degrees.
19117 Set rotation order for the output video. Choose one item for each position.
19128 Default value is @b{@samp{ypr}}.
19133 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
19137 Set if input video is flipped horizontally/vertically. Boolean values.
19140 Set if input video is transposed. Boolean value, by default disabled.
19143 Set if output video needs to be transposed. Boolean value, by default disabled.
19146 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
19149 @subsection Examples
19153 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
19155 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
19158 Extract back view of Equi-Angular Cubemap:
19160 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
19163 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
19165 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
19169 @subsection Commands
19171 This filter supports subset of above options as @ref{commands}.
19173 @section vaguedenoiser
19175 Apply a wavelet based denoiser.
19177 It transforms each frame from the video input into the wavelet domain,
19178 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
19179 the obtained coefficients. It does an inverse wavelet transform after.
19180 Due to wavelet properties, it should give a nice smoothed result, and
19181 reduced noise, without blurring picture features.
19183 This filter accepts the following options:
19187 The filtering strength. The higher, the more filtered the video will be.
19188 Hard thresholding can use a higher threshold than soft thresholding
19189 before the video looks overfiltered. Default value is 2.
19192 The filtering method the filter will use.
19194 It accepts the following values:
19197 All values under the threshold will be zeroed.
19200 All values under the threshold will be zeroed. All values above will be
19201 reduced by the threshold.
19204 Scales or nullifies coefficients - intermediary between (more) soft and
19205 (less) hard thresholding.
19208 Default is garrote.
19211 Number of times, the wavelet will decompose the picture. Picture can't
19212 be decomposed beyond a particular point (typically, 8 for a 640x480
19213 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
19216 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
19219 A list of the planes to process. By default all planes are processed.
19222 @section vectorscope
19224 Display 2 color component values in the two dimensional graph (which is called
19227 This filter accepts the following options:
19231 Set vectorscope mode.
19233 It accepts the following values:
19237 Gray values are displayed on graph, higher brightness means more pixels have
19238 same component color value on location in graph. This is the default mode.
19241 Gray values are displayed on graph. Surrounding pixels values which are not
19242 present in video frame are drawn in gradient of 2 color components which are
19243 set by option @code{x} and @code{y}. The 3rd color component is static.
19246 Actual color components values present in video frame are displayed on graph.
19249 Similar as color2 but higher frequency of same values @code{x} and @code{y}
19250 on graph increases value of another color component, which is luminance by
19251 default values of @code{x} and @code{y}.
19254 Actual colors present in video frame are displayed on graph. If two different
19255 colors map to same position on graph then color with higher value of component
19256 not present in graph is picked.
19259 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
19260 component picked from radial gradient.
19264 Set which color component will be represented on X-axis. Default is @code{1}.
19267 Set which color component will be represented on Y-axis. Default is @code{2}.
19270 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
19271 of color component which represents frequency of (X, Y) location in graph.
19276 No envelope, this is default.
19279 Instant envelope, even darkest single pixel will be clearly highlighted.
19282 Hold maximum and minimum values presented in graph over time. This way you
19283 can still spot out of range values without constantly looking at vectorscope.
19286 Peak and instant envelope combined together.
19290 Set what kind of graticule to draw.
19299 Set graticule opacity.
19302 Set graticule flags.
19306 Draw graticule for white point.
19309 Draw graticule for black point.
19312 Draw color points short names.
19316 Set background opacity.
19318 @item lthreshold, l
19319 Set low threshold for color component not represented on X or Y axis.
19320 Values lower than this value will be ignored. Default is 0.
19321 Note this value is multiplied with actual max possible value one pixel component
19322 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
19325 @item hthreshold, h
19326 Set high threshold for color component not represented on X or Y axis.
19327 Values higher than this value will be ignored. Default is 1.
19328 Note this value is multiplied with actual max possible value one pixel component
19329 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
19330 is 0.9 * 255 = 230.
19332 @item colorspace, c
19333 Set what kind of colorspace to use when drawing graticule.
19343 Set color tint for gray/tint vectorscope mode. By default both options are zero.
19344 This means no tint, and output will remain gray.
19347 @anchor{vidstabdetect}
19348 @section vidstabdetect
19350 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
19351 @ref{vidstabtransform} for pass 2.
19353 This filter generates a file with relative translation and rotation
19354 transform information about subsequent frames, which is then used by
19355 the @ref{vidstabtransform} filter.
19357 To enable compilation of this filter you need to configure FFmpeg with
19358 @code{--enable-libvidstab}.
19360 This filter accepts the following options:
19364 Set the path to the file used to write the transforms information.
19365 Default value is @file{transforms.trf}.
19368 Set how shaky the video is and how quick the camera is. It accepts an
19369 integer in the range 1-10, a value of 1 means little shakiness, a
19370 value of 10 means strong shakiness. Default value is 5.
19373 Set the accuracy of the detection process. It must be a value in the
19374 range 1-15. A value of 1 means low accuracy, a value of 15 means high
19375 accuracy. Default value is 15.
19378 Set stepsize of the search process. The region around minimum is
19379 scanned with 1 pixel resolution. Default value is 6.
19382 Set minimum contrast. Below this value a local measurement field is
19383 discarded. Must be a floating point value in the range 0-1. Default
19387 Set reference frame number for tripod mode.
19389 If enabled, the motion of the frames is compared to a reference frame
19390 in the filtered stream, identified by the specified number. The idea
19391 is to compensate all movements in a more-or-less static scene and keep
19392 the camera view absolutely still.
19394 If set to 0, it is disabled. The frames are counted starting from 1.
19397 Show fields and transforms in the resulting frames. It accepts an
19398 integer in the range 0-2. Default value is 0, which disables any
19402 @subsection Examples
19406 Use default values:
19412 Analyze strongly shaky movie and put the results in file
19413 @file{mytransforms.trf}:
19415 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
19419 Visualize the result of internal transformations in the resulting
19422 vidstabdetect=show=1
19426 Analyze a video with medium shakiness using @command{ffmpeg}:
19428 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
19432 @anchor{vidstabtransform}
19433 @section vidstabtransform
19435 Video stabilization/deshaking: pass 2 of 2,
19436 see @ref{vidstabdetect} for pass 1.
19438 Read a file with transform information for each frame and
19439 apply/compensate them. Together with the @ref{vidstabdetect}
19440 filter this can be used to deshake videos. See also
19441 @url{http://public.hronopik.de/vid.stab}. It is important to also use
19442 the @ref{unsharp} filter, see below.
19444 To enable compilation of this filter you need to configure FFmpeg with
19445 @code{--enable-libvidstab}.
19447 @subsection Options
19451 Set path to the file used to read the transforms. Default value is
19452 @file{transforms.trf}.
19455 Set the number of frames (value*2 + 1) used for lowpass filtering the
19456 camera movements. Default value is 10.
19458 For example a number of 10 means that 21 frames are used (10 in the
19459 past and 10 in the future) to smoothen the motion in the video. A
19460 larger value leads to a smoother video, but limits the acceleration of
19461 the camera (pan/tilt movements). 0 is a special case where a static
19462 camera is simulated.
19465 Set the camera path optimization algorithm.
19467 Accepted values are:
19470 gaussian kernel low-pass filter on camera motion (default)
19472 averaging on transformations
19476 Set maximal number of pixels to translate frames. Default value is -1,
19480 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
19481 value is -1, meaning no limit.
19484 Specify how to deal with borders that may be visible due to movement
19487 Available values are:
19490 keep image information from previous frame (default)
19492 fill the border black
19496 Invert transforms if set to 1. Default value is 0.
19499 Consider transforms as relative to previous frame if set to 1,
19500 absolute if set to 0. Default value is 0.
19503 Set percentage to zoom. A positive value will result in a zoom-in
19504 effect, a negative value in a zoom-out effect. Default value is 0 (no
19508 Set optimal zooming to avoid borders.
19510 Accepted values are:
19515 optimal static zoom value is determined (only very strong movements
19516 will lead to visible borders) (default)
19518 optimal adaptive zoom value is determined (no borders will be
19519 visible), see @option{zoomspeed}
19522 Note that the value given at zoom is added to the one calculated here.
19525 Set percent to zoom maximally each frame (enabled when
19526 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
19530 Specify type of interpolation.
19532 Available values are:
19537 linear only horizontal
19539 linear in both directions (default)
19541 cubic in both directions (slow)
19545 Enable virtual tripod mode if set to 1, which is equivalent to
19546 @code{relative=0:smoothing=0}. Default value is 0.
19548 Use also @code{tripod} option of @ref{vidstabdetect}.
19551 Increase log verbosity if set to 1. Also the detected global motions
19552 are written to the temporary file @file{global_motions.trf}. Default
19556 @subsection Examples
19560 Use @command{ffmpeg} for a typical stabilization with default values:
19562 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
19565 Note the use of the @ref{unsharp} filter which is always recommended.
19568 Zoom in a bit more and load transform data from a given file:
19570 vidstabtransform=zoom=5:input="mytransforms.trf"
19574 Smoothen the video even more:
19576 vidstabtransform=smoothing=30
19582 Flip the input video vertically.
19584 For example, to vertically flip a video with @command{ffmpeg}:
19586 ffmpeg -i in.avi -vf "vflip" out.avi
19591 Detect variable frame rate video.
19593 This filter tries to detect if the input is variable or constant frame rate.
19595 At end it will output number of frames detected as having variable delta pts,
19596 and ones with constant delta pts.
19597 If there was frames with variable delta, than it will also show min, max and
19598 average delta encountered.
19602 Boost or alter saturation.
19604 The filter accepts the following options:
19607 Set strength of boost if positive value or strength of alter if negative value.
19608 Default is 0. Allowed range is from -2 to 2.
19611 Set the red balance. Default is 1. Allowed range is from -10 to 10.
19614 Set the green balance. Default is 1. Allowed range is from -10 to 10.
19617 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
19620 Set the red luma coefficient.
19623 Set the green luma coefficient.
19626 Set the blue luma coefficient.
19629 If @code{intensity} is negative and this is set to 1, colors will change,
19630 otherwise colors will be less saturated, more towards gray.
19633 @subsection Commands
19635 This filter supports the all above options as @ref{commands}.
19640 Make or reverse a natural vignetting effect.
19642 The filter accepts the following options:
19646 Set lens angle expression as a number of radians.
19648 The value is clipped in the @code{[0,PI/2]} range.
19650 Default value: @code{"PI/5"}
19654 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
19658 Set forward/backward mode.
19660 Available modes are:
19663 The larger the distance from the central point, the darker the image becomes.
19666 The larger the distance from the central point, the brighter the image becomes.
19667 This can be used to reverse a vignette effect, though there is no automatic
19668 detection to extract the lens @option{angle} and other settings (yet). It can
19669 also be used to create a burning effect.
19672 Default value is @samp{forward}.
19675 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
19677 It accepts the following values:
19680 Evaluate expressions only once during the filter initialization.
19683 Evaluate expressions for each incoming frame. This is way slower than the
19684 @samp{init} mode since it requires all the scalers to be re-computed, but it
19685 allows advanced dynamic expressions.
19688 Default value is @samp{init}.
19691 Set dithering to reduce the circular banding effects. Default is @code{1}
19695 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
19696 Setting this value to the SAR of the input will make a rectangular vignetting
19697 following the dimensions of the video.
19699 Default is @code{1/1}.
19702 @subsection Expressions
19704 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
19705 following parameters.
19710 input width and height
19713 the number of input frame, starting from 0
19716 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
19717 @var{TB} units, NAN if undefined
19720 frame rate of the input video, NAN if the input frame rate is unknown
19723 the PTS (Presentation TimeStamp) of the filtered video frame,
19724 expressed in seconds, NAN if undefined
19727 time base of the input video
19731 @subsection Examples
19735 Apply simple strong vignetting effect:
19741 Make a flickering vignetting:
19743 vignette='PI/4+random(1)*PI/50':eval=frame
19748 @section vmafmotion
19750 Obtain the average VMAF motion score of a video.
19751 It is one of the component metrics of VMAF.
19753 The obtained average motion score is printed through the logging system.
19755 The filter accepts the following options:
19759 If specified, the filter will use the named file to save the motion score of
19760 each frame with respect to the previous frame.
19761 When filename equals "-" the data is sent to standard output.
19766 ffmpeg -i ref.mpg -vf vmafmotion -f null -
19770 Stack input videos vertically.
19772 All streams must be of same pixel format and of same width.
19774 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
19775 to create same output.
19777 The filter accepts the following options:
19781 Set number of input streams. Default is 2.
19784 If set to 1, force the output to terminate when the shortest input
19785 terminates. Default value is 0.
19790 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
19791 Deinterlacing Filter").
19793 Based on the process described by Martin Weston for BBC R&D, and
19794 implemented based on the de-interlace algorithm written by Jim
19795 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
19796 uses filter coefficients calculated by BBC R&D.
19798 This filter uses field-dominance information in frame to decide which
19799 of each pair of fields to place first in the output.
19800 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
19802 There are two sets of filter coefficients, so called "simple"
19803 and "complex". Which set of filter coefficients is used can
19804 be set by passing an optional parameter:
19808 Set the interlacing filter coefficients. Accepts one of the following values:
19812 Simple filter coefficient set.
19814 More-complex filter coefficient set.
19816 Default value is @samp{complex}.
19819 Specify which frames to deinterlace. Accepts one of the following values:
19823 Deinterlace all frames,
19825 Only deinterlace frames marked as interlaced.
19828 Default value is @samp{all}.
19832 Video waveform monitor.
19834 The waveform monitor plots color component intensity. By default luminance
19835 only. Each column of the waveform corresponds to a column of pixels in the
19838 It accepts the following options:
19842 Can be either @code{row}, or @code{column}. Default is @code{column}.
19843 In row mode, the graph on the left side represents color component value 0 and
19844 the right side represents value = 255. In column mode, the top side represents
19845 color component value = 0 and bottom side represents value = 255.
19848 Set intensity. Smaller values are useful to find out how many values of the same
19849 luminance are distributed across input rows/columns.
19850 Default value is @code{0.04}. Allowed range is [0, 1].
19853 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
19854 In mirrored mode, higher values will be represented on the left
19855 side for @code{row} mode and at the top for @code{column} mode. Default is
19856 @code{1} (mirrored).
19860 It accepts the following values:
19863 Presents information identical to that in the @code{parade}, except
19864 that the graphs representing color components are superimposed directly
19867 This display mode makes it easier to spot relative differences or similarities
19868 in overlapping areas of the color components that are supposed to be identical,
19869 such as neutral whites, grays, or blacks.
19872 Display separate graph for the color components side by side in
19873 @code{row} mode or one below the other in @code{column} mode.
19876 Display separate graph for the color components side by side in
19877 @code{column} mode or one below the other in @code{row} mode.
19879 Using this display mode makes it easy to spot color casts in the highlights
19880 and shadows of an image, by comparing the contours of the top and the bottom
19881 graphs of each waveform. Since whites, grays, and blacks are characterized
19882 by exactly equal amounts of red, green, and blue, neutral areas of the picture
19883 should display three waveforms of roughly equal width/height. If not, the
19884 correction is easy to perform by making level adjustments the three waveforms.
19886 Default is @code{stack}.
19888 @item components, c
19889 Set which color components to display. Default is 1, which means only luminance
19890 or red color component if input is in RGB colorspace. If is set for example to
19891 7 it will display all 3 (if) available color components.
19896 No envelope, this is default.
19899 Instant envelope, minimum and maximum values presented in graph will be easily
19900 visible even with small @code{step} value.
19903 Hold minimum and maximum values presented in graph across time. This way you
19904 can still spot out of range values without constantly looking at waveforms.
19907 Peak and instant envelope combined together.
19913 No filtering, this is default.
19916 Luma and chroma combined together.
19919 Similar as above, but shows difference between blue and red chroma.
19922 Similar as above, but use different colors.
19925 Similar as above, but again with different colors.
19928 Displays only chroma.
19931 Displays actual color value on waveform.
19934 Similar as above, but with luma showing frequency of chroma values.
19938 Set which graticule to display.
19942 Do not display graticule.
19945 Display green graticule showing legal broadcast ranges.
19948 Display orange graticule showing legal broadcast ranges.
19951 Display invert graticule showing legal broadcast ranges.
19955 Set graticule opacity.
19958 Set graticule flags.
19962 Draw numbers above lines. By default enabled.
19965 Draw dots instead of lines.
19969 Set scale used for displaying graticule.
19976 Default is digital.
19979 Set background opacity.
19983 Set tint for output.
19984 Only used with lowpass filter and when display is not overlay and input
19985 pixel formats are not RGB.
19988 @section weave, doubleweave
19990 The @code{weave} takes a field-based video input and join
19991 each two sequential fields into single frame, producing a new double
19992 height clip with half the frame rate and half the frame count.
19994 The @code{doubleweave} works same as @code{weave} but without
19995 halving frame rate and frame count.
19997 It accepts the following option:
20001 Set first field. Available values are:
20005 Set the frame as top-field-first.
20008 Set the frame as bottom-field-first.
20012 @subsection Examples
20016 Interlace video using @ref{select} and @ref{separatefields} filter:
20018 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
20023 Apply the xBR high-quality magnification filter which is designed for pixel
20024 art. It follows a set of edge-detection rules, see
20025 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
20027 It accepts the following option:
20031 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
20032 @code{3xBR} and @code{4} for @code{4xBR}.
20033 Default is @code{3}.
20038 Apply cross fade from one input video stream to another input video stream.
20039 The cross fade is applied for specified duration.
20041 The filter accepts the following options:
20045 Set one of available transition effects:
20081 Default transition effect is fade.
20084 Set cross fade duration in seconds.
20085 Default duration is 1 second.
20088 Set cross fade start relative to first input stream in seconds.
20089 Default offset is 0.
20092 Set expression for custom transition effect.
20094 The expressions can use the following variables and functions:
20099 The coordinates of the current sample.
20103 The width and height of the image.
20106 Progress of transition effect.
20109 Currently processed plane.
20112 Return value of first input at current location and plane.
20115 Return value of second input at current location and plane.
20121 Return the value of the pixel at location (@var{x},@var{y}) of the
20122 first/second/third/fourth component of first input.
20128 Return the value of the pixel at location (@var{x},@var{y}) of the
20129 first/second/third/fourth component of second input.
20133 @subsection Examples
20137 Cross fade from one input video to another input video, with fade transition and duration of transition
20138 of 2 seconds starting at offset of 5 seconds:
20140 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
20145 Pick median pixels from several input videos.
20147 The filter accepts the following options:
20151 Set number of inputs.
20152 Default is 3. Allowed range is from 3 to 255.
20153 If number of inputs is even number, than result will be mean value between two median values.
20156 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
20159 Set median percentile. Default value is @code{0.5}.
20160 Default value of @code{0.5} will pick always median values, while @code{0} will pick
20161 minimum values, and @code{1} maximum values.
20165 Stack video inputs into custom layout.
20167 All streams must be of same pixel format.
20169 The filter accepts the following options:
20173 Set number of input streams. Default is 2.
20176 Specify layout of inputs.
20177 This option requires the desired layout configuration to be explicitly set by the user.
20178 This sets position of each video input in output. Each input
20179 is separated by '|'.
20180 The first number represents the column, and the second number represents the row.
20181 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
20182 where X is video input from which to take width or height.
20183 Multiple values can be used when separated by '+'. In such
20184 case values are summed together.
20186 Note that if inputs are of different sizes gaps may appear, as not all of
20187 the output video frame will be filled. Similarly, videos can overlap each
20188 other if their position doesn't leave enough space for the full frame of
20191 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
20192 a layout must be set by the user.
20195 If set to 1, force the output to terminate when the shortest input
20196 terminates. Default value is 0.
20199 If set to valid color, all unused pixels will be filled with that color.
20200 By default fill is set to none, so it is disabled.
20203 @subsection Examples
20207 Display 4 inputs into 2x2 grid.
20211 input1(0, 0) | input3(w0, 0)
20212 input2(0, h0) | input4(w0, h0)
20216 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
20219 Note that if inputs are of different sizes, gaps or overlaps may occur.
20222 Display 4 inputs into 1x4 grid.
20229 input4(0, h0+h1+h2)
20233 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
20236 Note that if inputs are of different widths, unused space will appear.
20239 Display 9 inputs into 3x3 grid.
20243 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
20244 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
20245 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
20249 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
20252 Note that if inputs are of different sizes, gaps or overlaps may occur.
20255 Display 16 inputs into 4x4 grid.
20259 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
20260 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
20261 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
20262 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
20266 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|
20267 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
20270 Note that if inputs are of different sizes, gaps or overlaps may occur.
20277 Deinterlace the input video ("yadif" means "yet another deinterlacing
20280 It accepts the following parameters:
20286 The interlacing mode to adopt. It accepts one of the following values:
20289 @item 0, send_frame
20290 Output one frame for each frame.
20291 @item 1, send_field
20292 Output one frame for each field.
20293 @item 2, send_frame_nospatial
20294 Like @code{send_frame}, but it skips the spatial interlacing check.
20295 @item 3, send_field_nospatial
20296 Like @code{send_field}, but it skips the spatial interlacing check.
20299 The default value is @code{send_frame}.
20302 The picture field parity assumed for the input interlaced video. It accepts one
20303 of the following values:
20307 Assume the top field is first.
20309 Assume the bottom field is first.
20311 Enable automatic detection of field parity.
20314 The default value is @code{auto}.
20315 If the interlacing is unknown or the decoder does not export this information,
20316 top field first will be assumed.
20319 Specify which frames to deinterlace. Accepts one of the following
20324 Deinterlace all frames.
20325 @item 1, interlaced
20326 Only deinterlace frames marked as interlaced.
20329 The default value is @code{all}.
20332 @section yadif_cuda
20334 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
20335 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
20338 It accepts the following parameters:
20344 The interlacing mode to adopt. It accepts one of the following values:
20347 @item 0, send_frame
20348 Output one frame for each frame.
20349 @item 1, send_field
20350 Output one frame for each field.
20351 @item 2, send_frame_nospatial
20352 Like @code{send_frame}, but it skips the spatial interlacing check.
20353 @item 3, send_field_nospatial
20354 Like @code{send_field}, but it skips the spatial interlacing check.
20357 The default value is @code{send_frame}.
20360 The picture field parity assumed for the input interlaced video. It accepts one
20361 of the following values:
20365 Assume the top field is first.
20367 Assume the bottom field is first.
20369 Enable automatic detection of field parity.
20372 The default value is @code{auto}.
20373 If the interlacing is unknown or the decoder does not export this information,
20374 top field first will be assumed.
20377 Specify which frames to deinterlace. Accepts one of the following
20382 Deinterlace all frames.
20383 @item 1, interlaced
20384 Only deinterlace frames marked as interlaced.
20387 The default value is @code{all}.
20392 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
20393 The algorithm is described in
20394 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
20396 It accepts the following parameters:
20400 Set the window radius. Default value is 3.
20403 Set which planes to filter. Default is only the first plane.
20406 Set blur strength. Default value is 128.
20409 @subsection Commands
20410 This filter supports same @ref{commands} as options.
20414 Apply Zoom & Pan effect.
20416 This filter accepts the following options:
20420 Set the zoom expression. Range is 1-10. Default is 1.
20424 Set the x and y expression. Default is 0.
20427 Set the duration expression in number of frames.
20428 This sets for how many number of frames effect will last for
20429 single input image.
20432 Set the output image size, default is 'hd720'.
20435 Set the output frame rate, default is '25'.
20438 Each expression can contain the following constants:
20457 Output frame count.
20461 Last calculated 'x' and 'y' position from 'x' and 'y' expression
20462 for current input frame.
20466 'x' and 'y' of last output frame of previous input frame or 0 when there was
20467 not yet such frame (first input frame).
20470 Last calculated zoom from 'z' expression for current input frame.
20473 Last calculated zoom of last output frame of previous input frame.
20476 Number of output frames for current input frame. Calculated from 'd' expression
20477 for each input frame.
20480 number of output frames created for previous input frame
20483 Rational number: input width / input height
20486 sample aspect ratio
20489 display aspect ratio
20493 @subsection Examples
20497 Zoom-in up to 1.5 and pan at same time to some spot near center of picture:
20499 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
20503 Zoom-in up to 1.5 and pan always at center of picture:
20505 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20509 Same as above but without pausing:
20511 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20517 Scale (resize) the input video, using the z.lib library:
20518 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
20519 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
20521 The zscale filter forces the output display aspect ratio to be the same
20522 as the input, by changing the output sample aspect ratio.
20524 If the input image format is different from the format requested by
20525 the next filter, the zscale filter will convert the input to the
20528 @subsection Options
20529 The filter accepts the following options.
20534 Set the output video dimension expression. Default value is the input
20537 If the @var{width} or @var{w} value is 0, the input width is used for
20538 the output. If the @var{height} or @var{h} value is 0, the input height
20539 is used for the output.
20541 If one and only one of the values is -n with n >= 1, the zscale filter
20542 will use a value that maintains the aspect ratio of the input image,
20543 calculated from the other specified dimension. After that it will,
20544 however, make sure that the calculated dimension is divisible by n and
20545 adjust the value if necessary.
20547 If both values are -n with n >= 1, the behavior will be identical to
20548 both values being set to 0 as previously detailed.
20550 See below for the list of accepted constants for use in the dimension
20554 Set the video size. For the syntax of this option, check the
20555 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
20558 Set the dither type.
20560 Possible values are:
20565 @item error_diffusion
20571 Set the resize filter type.
20573 Possible values are:
20583 Default is bilinear.
20586 Set the color range.
20588 Possible values are:
20595 Default is same as input.
20598 Set the color primaries.
20600 Possible values are:
20610 Default is same as input.
20613 Set the transfer characteristics.
20615 Possible values are:
20629 Default is same as input.
20632 Set the colorspace matrix.
20634 Possible value are:
20645 Default is same as input.
20648 Set the input color range.
20650 Possible values are:
20657 Default is same as input.
20659 @item primariesin, pin
20660 Set the input color primaries.
20662 Possible values are:
20672 Default is same as input.
20674 @item transferin, tin
20675 Set the input transfer characteristics.
20677 Possible values are:
20688 Default is same as input.
20690 @item matrixin, min
20691 Set the input colorspace matrix.
20693 Possible value are:
20705 Set the output chroma location.
20707 Possible values are:
20718 @item chromalin, cin
20719 Set the input chroma location.
20721 Possible values are:
20733 Set the nominal peak luminance.
20736 The values of the @option{w} and @option{h} options are expressions
20737 containing the following constants:
20742 The input width and height
20746 These are the same as @var{in_w} and @var{in_h}.
20750 The output (scaled) width and height
20754 These are the same as @var{out_w} and @var{out_h}
20757 The same as @var{iw} / @var{ih}
20760 input sample aspect ratio
20763 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
20767 horizontal and vertical input chroma subsample values. For example for the
20768 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20772 horizontal and vertical output chroma subsample values. For example for the
20773 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20776 @subsection Commands
20778 This filter supports the following commands:
20782 Set the output video dimension expression.
20783 The command accepts the same syntax of the corresponding option.
20785 If the specified expression is not valid, it is kept at its current
20789 @c man end VIDEO FILTERS
20791 @chapter OpenCL Video Filters
20792 @c man begin OPENCL VIDEO FILTERS
20794 Below is a description of the currently available OpenCL video filters.
20796 To enable compilation of these filters you need to configure FFmpeg with
20797 @code{--enable-opencl}.
20799 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
20802 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
20803 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
20804 given device parameters.
20806 @item -filter_hw_device @var{name}
20807 Pass the hardware device called @var{name} to all filters in any filter graph.
20811 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
20815 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
20817 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
20821 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.
20823 @section avgblur_opencl
20825 Apply average blur filter.
20827 The filter accepts the following options:
20831 Set horizontal radius size.
20832 Range is @code{[1, 1024]} and default value is @code{1}.
20835 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20838 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
20841 @subsection Example
20845 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.
20847 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
20851 @section boxblur_opencl
20853 Apply a boxblur algorithm to the input video.
20855 It accepts the following parameters:
20859 @item luma_radius, lr
20860 @item luma_power, lp
20861 @item chroma_radius, cr
20862 @item chroma_power, cp
20863 @item alpha_radius, ar
20864 @item alpha_power, ap
20868 A description of the accepted options follows.
20871 @item luma_radius, lr
20872 @item chroma_radius, cr
20873 @item alpha_radius, ar
20874 Set an expression for the box radius in pixels used for blurring the
20875 corresponding input plane.
20877 The radius value must be a non-negative number, and must not be
20878 greater than the value of the expression @code{min(w,h)/2} for the
20879 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
20882 Default value for @option{luma_radius} is "2". If not specified,
20883 @option{chroma_radius} and @option{alpha_radius} default to the
20884 corresponding value set for @option{luma_radius}.
20886 The expressions can contain the following constants:
20890 The input width and height in pixels.
20894 The input chroma image width and height in pixels.
20898 The horizontal and vertical chroma subsample values. For example, for the
20899 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
20902 @item luma_power, lp
20903 @item chroma_power, cp
20904 @item alpha_power, ap
20905 Specify how many times the boxblur filter is applied to the
20906 corresponding plane.
20908 Default value for @option{luma_power} is 2. If not specified,
20909 @option{chroma_power} and @option{alpha_power} default to the
20910 corresponding value set for @option{luma_power}.
20912 A value of 0 will disable the effect.
20915 @subsection Examples
20917 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.
20921 Apply a boxblur filter with the luma, chroma, and alpha radius
20922 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.
20924 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
20925 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
20929 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.
20931 For the luma plane, a 2x2 box radius will be run once.
20933 For the chroma plane, a 4x4 box radius will be run 5 times.
20935 For the alpha plane, a 3x3 box radius will be run 7 times.
20937 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
20941 @section colorkey_opencl
20942 RGB colorspace color keying.
20944 The filter accepts the following options:
20948 The color which will be replaced with transparency.
20951 Similarity percentage with the key color.
20953 0.01 matches only the exact key color, while 1.0 matches everything.
20958 0.0 makes pixels either fully transparent, or not transparent at all.
20960 Higher values result in semi-transparent pixels, with a higher transparency
20961 the more similar the pixels color is to the key color.
20964 @subsection Examples
20968 Make every semi-green pixel in the input transparent with some slight blending:
20970 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
20974 @section convolution_opencl
20976 Apply convolution of 3x3, 5x5, 7x7 matrix.
20978 The filter accepts the following options:
20985 Set matrix for each plane.
20986 Matrix is sequence of 9, 25 or 49 signed numbers.
20987 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
20993 Set multiplier for calculated value for each plane.
20994 If unset or 0, it will be sum of all matrix elements.
20995 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
21001 Set bias for each plane. This value is added to the result of the multiplication.
21002 Useful for making the overall image brighter or darker.
21003 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
21007 @subsection Examples
21013 -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
21019 -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
21023 Apply edge enhance:
21025 -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
21031 -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
21035 Apply laplacian edge detector which includes diagonals:
21037 -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
21043 -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
21047 @section erosion_opencl
21049 Apply erosion effect to the video.
21051 This filter replaces the pixel by the local(3x3) minimum.
21053 It accepts the following options:
21060 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21061 If @code{0}, plane will remain unchanged.
21064 Flag which specifies the pixel to refer to.
21065 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21067 Flags to local 3x3 coordinates region centered on @code{x}:
21076 @subsection Example
21080 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.
21082 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21086 @section deshake_opencl
21087 Feature-point based video stabilization filter.
21089 The filter accepts the following options:
21093 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
21096 Whether or not additional debug info should be displayed, both in the processed output and in the console.
21098 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
21100 Viewing point matches in the output video is only supported for RGB input.
21102 Defaults to @code{0}.
21104 @item adaptive_crop
21105 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
21107 Defaults to @code{1}.
21109 @item refine_features
21110 Whether or not feature points should be refined at a sub-pixel level.
21112 This can be turned off for a slight performance gain at the cost of precision.
21114 Defaults to @code{1}.
21116 @item smooth_strength
21117 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
21119 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
21121 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
21123 Defaults to @code{0.0}.
21125 @item smooth_window_multiplier
21126 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
21128 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
21130 Acceptable values range from @code{0.1} to @code{10.0}.
21132 Larger values increase the amount of motion data available for determining how to smooth the camera path,
21133 potentially improving smoothness, but also increase latency and memory usage.
21135 Defaults to @code{2.0}.
21139 @subsection Examples
21143 Stabilize a video with a fixed, medium smoothing strength:
21145 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
21149 Stabilize a video with debugging (both in console and in rendered video):
21151 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
21155 @section dilation_opencl
21157 Apply dilation effect to the video.
21159 This filter replaces the pixel by the local(3x3) maximum.
21161 It accepts the following options:
21168 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
21169 If @code{0}, plane will remain unchanged.
21172 Flag which specifies the pixel to refer to.
21173 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
21175 Flags to local 3x3 coordinates region centered on @code{x}:
21184 @subsection Example
21188 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.
21190 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
21194 @section nlmeans_opencl
21196 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
21198 @section overlay_opencl
21200 Overlay one video on top of another.
21202 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
21203 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
21205 The filter accepts the following options:
21210 Set the x coordinate of the overlaid video on the main video.
21211 Default value is @code{0}.
21214 Set the y coordinate of the overlaid video on the main video.
21215 Default value is @code{0}.
21219 @subsection Examples
21223 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
21225 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21228 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
21230 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
21235 @section pad_opencl
21237 Add paddings to the input image, and place the original input at the
21238 provided @var{x}, @var{y} coordinates.
21240 It accepts the following options:
21245 Specify an expression for the size of the output image with the
21246 paddings added. If the value for @var{width} or @var{height} is 0, the
21247 corresponding input size is used for the output.
21249 The @var{width} expression can reference the value set by the
21250 @var{height} expression, and vice versa.
21252 The default value of @var{width} and @var{height} is 0.
21256 Specify the offsets to place the input image at within the padded area,
21257 with respect to the top/left border of the output image.
21259 The @var{x} expression can reference the value set by the @var{y}
21260 expression, and vice versa.
21262 The default value of @var{x} and @var{y} is 0.
21264 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
21265 so the input image is centered on the padded area.
21268 Specify the color of the padded area. For the syntax of this option,
21269 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
21270 manual,ffmpeg-utils}.
21273 Pad to an aspect instead to a resolution.
21276 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
21277 options are expressions containing the following constants:
21282 The input video width and height.
21286 These are the same as @var{in_w} and @var{in_h}.
21290 The output width and height (the size of the padded area), as
21291 specified by the @var{width} and @var{height} expressions.
21295 These are the same as @var{out_w} and @var{out_h}.
21299 The x and y offsets as specified by the @var{x} and @var{y}
21300 expressions, or NAN if not yet specified.
21303 same as @var{iw} / @var{ih}
21306 input sample aspect ratio
21309 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
21312 @section prewitt_opencl
21314 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
21316 The filter accepts the following option:
21320 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21323 Set value which will be multiplied with filtered result.
21324 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
21327 Set value which will be added to filtered result.
21328 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
21331 @subsection Example
21335 Apply the Prewitt operator with scale set to 2 and delta set to 10.
21337 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
21341 @anchor{program_opencl}
21342 @section program_opencl
21344 Filter video using an OpenCL program.
21349 OpenCL program source file.
21352 Kernel name in program.
21355 Number of inputs to the filter. Defaults to 1.
21358 Size of output frames. Defaults to the same as the first input.
21362 The @code{program_opencl} filter also supports the @ref{framesync} options.
21364 The program source file must contain a kernel function with the given name,
21365 which will be run once for each plane of the output. Each run on a plane
21366 gets enqueued as a separate 2D global NDRange with one work-item for each
21367 pixel to be generated. The global ID offset for each work-item is therefore
21368 the coordinates of a pixel in the destination image.
21370 The kernel function needs to take the following arguments:
21373 Destination image, @var{__write_only image2d_t}.
21375 This image will become the output; the kernel should write all of it.
21377 Frame index, @var{unsigned int}.
21379 This is a counter starting from zero and increasing by one for each frame.
21381 Source images, @var{__read_only image2d_t}.
21383 These are the most recent images on each input. The kernel may read from
21384 them to generate the output, but they can't be written to.
21391 Copy the input to the output (output must be the same size as the input).
21393 __kernel void copy(__write_only image2d_t destination,
21394 unsigned int index,
21395 __read_only image2d_t source)
21397 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
21399 int2 location = (int2)(get_global_id(0), get_global_id(1));
21401 float4 value = read_imagef(source, sampler, location);
21403 write_imagef(destination, location, value);
21408 Apply a simple transformation, rotating the input by an amount increasing
21409 with the index counter. Pixel values are linearly interpolated by the
21410 sampler, and the output need not have the same dimensions as the input.
21412 __kernel void rotate_image(__write_only image2d_t dst,
21413 unsigned int index,
21414 __read_only image2d_t src)
21416 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21417 CLK_FILTER_LINEAR);
21419 float angle = (float)index / 100.0f;
21421 float2 dst_dim = convert_float2(get_image_dim(dst));
21422 float2 src_dim = convert_float2(get_image_dim(src));
21424 float2 dst_cen = dst_dim / 2.0f;
21425 float2 src_cen = src_dim / 2.0f;
21427 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
21429 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
21431 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
21432 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
21434 src_pos = src_pos * src_dim / dst_dim;
21436 float2 src_loc = src_pos + src_cen;
21438 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
21439 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
21440 write_imagef(dst, dst_loc, 0.5f);
21442 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
21447 Blend two inputs together, with the amount of each input used varying
21448 with the index counter.
21450 __kernel void blend_images(__write_only image2d_t dst,
21451 unsigned int index,
21452 __read_only image2d_t src1,
21453 __read_only image2d_t src2)
21455 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21456 CLK_FILTER_LINEAR);
21458 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
21460 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
21461 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
21462 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
21464 float4 val1 = read_imagef(src1, sampler, src1_loc);
21465 float4 val2 = read_imagef(src2, sampler, src2_loc);
21467 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
21473 @section roberts_opencl
21474 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
21476 The filter accepts the following option:
21480 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21483 Set value which will be multiplied with filtered result.
21484 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
21487 Set value which will be added to filtered result.
21488 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
21491 @subsection Example
21495 Apply the Roberts cross operator with scale set to 2 and delta set to 10
21497 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
21501 @section sobel_opencl
21503 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
21505 The filter accepts the following option:
21509 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21512 Set value which will be multiplied with filtered result.
21513 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
21516 Set value which will be added to filtered result.
21517 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
21520 @subsection Example
21524 Apply sobel operator with scale set to 2 and delta set to 10
21526 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
21530 @section tonemap_opencl
21532 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
21534 It accepts the following parameters:
21538 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
21541 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
21544 Apply desaturation for highlights that exceed this level of brightness. The
21545 higher the parameter, the more color information will be preserved. This
21546 setting helps prevent unnaturally blown-out colors for super-highlights, by
21547 (smoothly) turning into white instead. This makes images feel more natural,
21548 at the cost of reducing information about out-of-range colors.
21550 The default value is 0.5, and the algorithm here is a little different from
21551 the cpu version tonemap currently. A setting of 0.0 disables this option.
21554 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
21555 is used to detect whether the scene has changed or not. If the distance between
21556 the current frame average brightness and the current running average exceeds
21557 a threshold value, we would re-calculate scene average and peak brightness.
21558 The default value is 0.2.
21561 Specify the output pixel format.
21563 Currently supported formats are:
21570 Set the output color range.
21572 Possible values are:
21578 Default is same as input.
21581 Set the output color primaries.
21583 Possible values are:
21589 Default is same as input.
21592 Set the output transfer characteristics.
21594 Possible values are:
21603 Set the output colorspace matrix.
21605 Possible value are:
21611 Default is same as input.
21615 @subsection Example
21619 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
21621 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
21625 @section unsharp_opencl
21627 Sharpen or blur the input video.
21629 It accepts the following parameters:
21632 @item luma_msize_x, lx
21633 Set the luma matrix horizontal size.
21634 Range is @code{[1, 23]} and default value is @code{5}.
21636 @item luma_msize_y, ly
21637 Set the luma matrix vertical size.
21638 Range is @code{[1, 23]} and default value is @code{5}.
21640 @item luma_amount, la
21641 Set the luma effect strength.
21642 Range is @code{[-10, 10]} and default value is @code{1.0}.
21644 Negative values will blur the input video, while positive values will
21645 sharpen it, a value of zero will disable the effect.
21647 @item chroma_msize_x, cx
21648 Set the chroma matrix horizontal size.
21649 Range is @code{[1, 23]} and default value is @code{5}.
21651 @item chroma_msize_y, cy
21652 Set the chroma matrix vertical size.
21653 Range is @code{[1, 23]} and default value is @code{5}.
21655 @item chroma_amount, ca
21656 Set the chroma effect strength.
21657 Range is @code{[-10, 10]} and default value is @code{0.0}.
21659 Negative values will blur the input video, while positive values will
21660 sharpen it, a value of zero will disable the effect.
21664 All parameters are optional and default to the equivalent of the
21665 string '5:5:1.0:5:5:0.0'.
21667 @subsection Examples
21671 Apply strong luma sharpen effect:
21673 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
21677 Apply a strong blur of both luma and chroma parameters:
21679 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
21683 @section xfade_opencl
21685 Cross fade two videos with custom transition effect by using OpenCL.
21687 It accepts the following options:
21691 Set one of possible transition effects.
21695 Select custom transition effect, the actual transition description
21696 will be picked from source and kernel options.
21708 Default transition is fade.
21712 OpenCL program source file for custom transition.
21715 Set name of kernel to use for custom transition from program source file.
21718 Set duration of video transition.
21721 Set time of start of transition relative to first video.
21724 The program source file must contain a kernel function with the given name,
21725 which will be run once for each plane of the output. Each run on a plane
21726 gets enqueued as a separate 2D global NDRange with one work-item for each
21727 pixel to be generated. The global ID offset for each work-item is therefore
21728 the coordinates of a pixel in the destination image.
21730 The kernel function needs to take the following arguments:
21733 Destination image, @var{__write_only image2d_t}.
21735 This image will become the output; the kernel should write all of it.
21738 First Source image, @var{__read_only image2d_t}.
21739 Second Source image, @var{__read_only image2d_t}.
21741 These are the most recent images on each input. The kernel may read from
21742 them to generate the output, but they can't be written to.
21745 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
21752 Apply dots curtain transition effect:
21754 __kernel void blend_images(__write_only image2d_t dst,
21755 __read_only image2d_t src1,
21756 __read_only image2d_t src2,
21759 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
21760 CLK_FILTER_LINEAR);
21761 int2 p = (int2)(get_global_id(0), get_global_id(1));
21762 float2 rp = (float2)(get_global_id(0), get_global_id(1));
21763 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
21766 float2 dots = (float2)(20.0, 20.0);
21767 float2 center = (float2)(0,0);
21770 float4 val1 = read_imagef(src1, sampler, p);
21771 float4 val2 = read_imagef(src2, sampler, p);
21772 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
21774 write_imagef(dst, p, next ? val1 : val2);
21780 @c man end OPENCL VIDEO FILTERS
21782 @chapter VAAPI Video Filters
21783 @c man begin VAAPI VIDEO FILTERS
21785 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
21787 To enable compilation of these filters you need to configure FFmpeg with
21788 @code{--enable-vaapi}.
21790 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}
21792 @section tonemap_vaapi
21794 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
21795 It maps the dynamic range of HDR10 content to the SDR content.
21796 It currently only accepts HDR10 as input.
21798 It accepts the following parameters:
21802 Specify the output pixel format.
21804 Currently supported formats are:
21813 Set the output color primaries.
21815 Default is same as input.
21818 Set the output transfer characteristics.
21823 Set the output colorspace matrix.
21825 Default is same as input.
21829 @subsection Example
21833 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
21835 tonemap_vaapi=format=p010:t=bt2020-10
21839 @c man end VAAPI VIDEO FILTERS
21841 @chapter Video Sources
21842 @c man begin VIDEO SOURCES
21844 Below is a description of the currently available video sources.
21848 Buffer video frames, and make them available to the filter chain.
21850 This source is mainly intended for a programmatic use, in particular
21851 through the interface defined in @file{libavfilter/vsrc_buffer.h}.
21853 It accepts the following parameters:
21858 Specify the size (width and height) of the buffered video frames. For the
21859 syntax of this option, check the
21860 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21863 The input video width.
21866 The input video height.
21869 A string representing the pixel format of the buffered video frames.
21870 It may be a number corresponding to a pixel format, or a pixel format
21874 Specify the timebase assumed by the timestamps of the buffered frames.
21877 Specify the frame rate expected for the video stream.
21879 @item pixel_aspect, sar
21880 The sample (pixel) aspect ratio of the input video.
21883 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
21884 to the filtergraph description to specify swscale flags for automatically
21885 inserted scalers. See @ref{Filtergraph syntax}.
21887 @item hw_frames_ctx
21888 When using a hardware pixel format, this should be a reference to an
21889 AVHWFramesContext describing input frames.
21894 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
21897 will instruct the source to accept video frames with size 320x240 and
21898 with format "yuv410p", assuming 1/24 as the timestamps timebase and
21899 square pixels (1:1 sample aspect ratio).
21900 Since the pixel format with name "yuv410p" corresponds to the number 6
21901 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
21902 this example corresponds to:
21904 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
21907 Alternatively, the options can be specified as a flat string, but this
21908 syntax is deprecated:
21910 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
21914 Create a pattern generated by an elementary cellular automaton.
21916 The initial state of the cellular automaton can be defined through the
21917 @option{filename} and @option{pattern} options. If such options are
21918 not specified an initial state is created randomly.
21920 At each new frame a new row in the video is filled with the result of
21921 the cellular automaton next generation. The behavior when the whole
21922 frame is filled is defined by the @option{scroll} option.
21924 This source accepts the following options:
21928 Read the initial cellular automaton state, i.e. the starting row, from
21929 the specified file.
21930 In the file, each non-whitespace character is considered an alive
21931 cell, a newline will terminate the row, and further characters in the
21932 file will be ignored.
21935 Read the initial cellular automaton state, i.e. the starting row, from
21936 the specified string.
21938 Each non-whitespace character in the string is considered an alive
21939 cell, a newline will terminate the row, and further characters in the
21940 string will be ignored.
21943 Set the video rate, that is the number of frames generated per second.
21946 @item random_fill_ratio, ratio
21947 Set the random fill ratio for the initial cellular automaton row. It
21948 is a floating point number value ranging from 0 to 1, defaults to
21951 This option is ignored when a file or a pattern is specified.
21953 @item random_seed, seed
21954 Set the seed for filling randomly the initial row, must be an integer
21955 included between 0 and UINT32_MAX. If not specified, or if explicitly
21956 set to -1, the filter will try to use a good random seed on a best
21960 Set the cellular automaton rule, it is a number ranging from 0 to 255.
21961 Default value is 110.
21964 Set the size of the output video. For the syntax of this option, check the
21965 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21967 If @option{filename} or @option{pattern} is specified, the size is set
21968 by default to the width of the specified initial state row, and the
21969 height is set to @var{width} * PHI.
21971 If @option{size} is set, it must contain the width of the specified
21972 pattern string, and the specified pattern will be centered in the
21975 If a filename or a pattern string is not specified, the size value
21976 defaults to "320x518" (used for a randomly generated initial state).
21979 If set to 1, scroll the output upward when all the rows in the output
21980 have been already filled. If set to 0, the new generated row will be
21981 written over the top row just after the bottom row is filled.
21984 @item start_full, full
21985 If set to 1, completely fill the output with generated rows before
21986 outputting the first frame.
21987 This is the default behavior, for disabling set the value to 0.
21990 If set to 1, stitch the left and right row edges together.
21991 This is the default behavior, for disabling set the value to 0.
21994 @subsection Examples
21998 Read the initial state from @file{pattern}, and specify an output of
22001 cellauto=f=pattern:s=200x400
22005 Generate a random initial row with a width of 200 cells, with a fill
22008 cellauto=ratio=2/3:s=200x200
22012 Create a pattern generated by rule 18 starting by a single alive cell
22013 centered on an initial row with width 100:
22015 cellauto=p=@@:s=100x400:full=0:rule=18
22019 Specify a more elaborated initial pattern:
22021 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
22026 @anchor{coreimagesrc}
22027 @section coreimagesrc
22028 Video source generated on GPU using Apple's CoreImage API on OSX.
22030 This video source is a specialized version of the @ref{coreimage} video filter.
22031 Use a core image generator at the beginning of the applied filterchain to
22032 generate the content.
22034 The coreimagesrc video source accepts the following options:
22036 @item list_generators
22037 List all available generators along with all their respective options as well as
22038 possible minimum and maximum values along with the default values.
22040 list_generators=true
22044 Specify the size of the sourced video. For the syntax of this option, check the
22045 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22046 The default value is @code{320x240}.
22049 Specify the frame rate of the sourced video, as the number of frames
22050 generated per second. It has to be a string in the format
22051 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22052 number or a valid video frame rate abbreviation. The default value is
22056 Set the sample aspect ratio of the sourced video.
22059 Set the duration of the sourced video. See
22060 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22061 for the accepted syntax.
22063 If not specified, or the expressed duration is negative, the video is
22064 supposed to be generated forever.
22067 Additionally, all options of the @ref{coreimage} video filter are accepted.
22068 A complete filterchain can be used for further processing of the
22069 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
22070 and examples for details.
22072 @subsection Examples
22077 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
22078 given as complete and escaped command-line for Apple's standard bash shell:
22080 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
22082 This example is equivalent to the QRCode example of @ref{coreimage} without the
22083 need for a nullsrc video source.
22087 @section mandelbrot
22089 Generate a Mandelbrot set fractal, and progressively zoom towards the
22090 point specified with @var{start_x} and @var{start_y}.
22092 This source accepts the following options:
22097 Set the terminal pts value. Default value is 400.
22100 Set the terminal scale value.
22101 Must be a floating point value. Default value is 0.3.
22104 Set the inner coloring mode, that is the algorithm used to draw the
22105 Mandelbrot fractal internal region.
22107 It shall assume one of the following values:
22112 Show time until convergence.
22114 Set color based on point closest to the origin of the iterations.
22119 Default value is @var{mincol}.
22122 Set the bailout value. Default value is 10.0.
22125 Set the maximum of iterations performed by the rendering
22126 algorithm. Default value is 7189.
22129 Set outer coloring mode.
22130 It shall assume one of following values:
22132 @item iteration_count
22133 Set iteration count mode.
22134 @item normalized_iteration_count
22135 set normalized iteration count mode.
22137 Default value is @var{normalized_iteration_count}.
22140 Set frame rate, expressed as number of frames per second. Default
22144 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22145 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22148 Set the initial scale value. Default value is 3.0.
22151 Set the initial x position. Must be a floating point value between
22152 -100 and 100. Default value is -0.743643887037158704752191506114774.
22155 Set the initial y position. Must be a floating point value between
22156 -100 and 100. Default value is -0.131825904205311970493132056385139.
22161 Generate various test patterns, as generated by the MPlayer test filter.
22163 The size of the generated video is fixed, and is 256x256.
22164 This source is useful in particular for testing encoding features.
22166 This source accepts the following options:
22171 Specify the frame rate of the sourced video, as the number of frames
22172 generated per second. It has to be a string in the format
22173 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22174 number or a valid video frame rate abbreviation. The default value is
22178 Set the duration of the sourced video. See
22179 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22180 for the accepted syntax.
22182 If not specified, or the expressed duration is negative, the video is
22183 supposed to be generated forever.
22187 Set the number or the name of the test to perform. Supported tests are:
22201 @item max_frames, m
22202 Set the maximum number of frames generated for each test, default value is 30.
22206 Default value is "all", which will cycle through the list of all tests.
22211 mptestsrc=t=dc_luma
22214 will generate a "dc_luma" test pattern.
22216 @section frei0r_src
22218 Provide a frei0r source.
22220 To enable compilation of this filter you need to install the frei0r
22221 header and configure FFmpeg with @code{--enable-frei0r}.
22223 This source accepts the following parameters:
22228 The size of the video to generate. For the syntax of this option, check the
22229 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22232 The framerate of the generated video. It may be a string of the form
22233 @var{num}/@var{den} or a frame rate abbreviation.
22236 The name to the frei0r source to load. For more information regarding frei0r and
22237 how to set the parameters, read the @ref{frei0r} section in the video filters
22240 @item filter_params
22241 A '|'-separated list of parameters to pass to the frei0r source.
22245 For example, to generate a frei0r partik0l source with size 200x200
22246 and frame rate 10 which is overlaid on the overlay filter main input:
22248 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
22253 Generate a life pattern.
22255 This source is based on a generalization of John Conway's life game.
22257 The sourced input represents a life grid, each pixel represents a cell
22258 which can be in one of two possible states, alive or dead. Every cell
22259 interacts with its eight neighbours, which are the cells that are
22260 horizontally, vertically, or diagonally adjacent.
22262 At each interaction the grid evolves according to the adopted rule,
22263 which specifies the number of neighbor alive cells which will make a
22264 cell stay alive or born. The @option{rule} option allows one to specify
22267 This source accepts the following options:
22271 Set the file from which to read the initial grid state. In the file,
22272 each non-whitespace character is considered an alive cell, and newline
22273 is used to delimit the end of each row.
22275 If this option is not specified, the initial grid is generated
22279 Set the video rate, that is the number of frames generated per second.
22282 @item random_fill_ratio, ratio
22283 Set the random fill ratio for the initial random grid. It is a
22284 floating point number value ranging from 0 to 1, defaults to 1/PHI.
22285 It is ignored when a file is specified.
22287 @item random_seed, seed
22288 Set the seed for filling the initial random grid, must be an integer
22289 included between 0 and UINT32_MAX. If not specified, or if explicitly
22290 set to -1, the filter will try to use a good random seed on a best
22296 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
22297 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
22298 @var{NS} specifies the number of alive neighbor cells which make a
22299 live cell stay alive, and @var{NB} the number of alive neighbor cells
22300 which make a dead cell to become alive (i.e. to "born").
22301 "s" and "b" can be used in place of "S" and "B", respectively.
22303 Alternatively a rule can be specified by an 18-bits integer. The 9
22304 high order bits are used to encode the next cell state if it is alive
22305 for each number of neighbor alive cells, the low order bits specify
22306 the rule for "borning" new cells. Higher order bits encode for an
22307 higher number of neighbor cells.
22308 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
22309 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
22311 Default value is "S23/B3", which is the original Conway's game of life
22312 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
22313 cells, and will born a new cell if there are three alive cells around
22317 Set the size of the output video. For the syntax of this option, check the
22318 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22320 If @option{filename} is specified, the size is set by default to the
22321 same size of the input file. If @option{size} is set, it must contain
22322 the size specified in the input file, and the initial grid defined in
22323 that file is centered in the larger resulting area.
22325 If a filename is not specified, the size value defaults to "320x240"
22326 (used for a randomly generated initial grid).
22329 If set to 1, stitch the left and right grid edges together, and the
22330 top and bottom edges also. Defaults to 1.
22333 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
22334 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
22335 value from 0 to 255.
22338 Set the color of living (or new born) cells.
22341 Set the color of dead cells. If @option{mold} is set, this is the first color
22342 used to represent a dead cell.
22345 Set mold color, for definitely dead and moldy cells.
22347 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
22348 ffmpeg-utils manual,ffmpeg-utils}.
22351 @subsection Examples
22355 Read a grid from @file{pattern}, and center it on a grid of size
22358 life=f=pattern:s=300x300
22362 Generate a random grid of size 200x200, with a fill ratio of 2/3:
22364 life=ratio=2/3:s=200x200
22368 Specify a custom rule for evolving a randomly generated grid:
22374 Full example with slow death effect (mold) using @command{ffplay}:
22376 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
22383 @anchor{haldclutsrc}
22386 @anchor{pal100bars}
22387 @anchor{rgbtestsrc}
22389 @anchor{smptehdbars}
22392 @anchor{yuvtestsrc}
22393 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
22395 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
22397 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
22399 The @code{color} source provides an uniformly colored input.
22401 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
22402 @ref{haldclut} filter.
22404 The @code{nullsrc} source returns unprocessed video frames. It is
22405 mainly useful to be employed in analysis / debugging tools, or as the
22406 source for filters which ignore the input data.
22408 The @code{pal75bars} source generates a color bars pattern, based on
22409 EBU PAL recommendations with 75% color levels.
22411 The @code{pal100bars} source generates a color bars pattern, based on
22412 EBU PAL recommendations with 100% color levels.
22414 The @code{rgbtestsrc} source generates an RGB test pattern useful for
22415 detecting RGB vs BGR issues. You should see a red, green and blue
22416 stripe from top to bottom.
22418 The @code{smptebars} source generates a color bars pattern, based on
22419 the SMPTE Engineering Guideline EG 1-1990.
22421 The @code{smptehdbars} source generates a color bars pattern, based on
22422 the SMPTE RP 219-2002.
22424 The @code{testsrc} source generates a test video pattern, showing a
22425 color pattern, a scrolling gradient and a timestamp. This is mainly
22426 intended for testing purposes.
22428 The @code{testsrc2} source is similar to testsrc, but supports more
22429 pixel formats instead of just @code{rgb24}. This allows using it as an
22430 input for other tests without requiring a format conversion.
22432 The @code{yuvtestsrc} source generates an YUV test pattern. You should
22433 see a y, cb and cr stripe from top to bottom.
22435 The sources accept the following parameters:
22440 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
22441 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
22442 pixels to be used as identity matrix for 3D lookup tables. Each component is
22443 coded on a @code{1/(N*N)} scale.
22446 Specify the color of the source, only available in the @code{color}
22447 source. For the syntax of this option, check the
22448 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
22451 Specify the size of the sourced video. For the syntax of this option, check the
22452 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22453 The default value is @code{320x240}.
22455 This option is not available with the @code{allrgb}, @code{allyuv}, and
22456 @code{haldclutsrc} filters.
22459 Specify the frame rate of the sourced video, as the number of frames
22460 generated per second. It has to be a string in the format
22461 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
22462 number or a valid video frame rate abbreviation. The default value is
22466 Set the duration of the sourced video. See
22467 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
22468 for the accepted syntax.
22470 If not specified, or the expressed duration is negative, the video is
22471 supposed to be generated forever.
22474 Set the sample aspect ratio of the sourced video.
22477 Specify the alpha (opacity) of the background, only available in the
22478 @code{testsrc2} source. The value must be between 0 (fully transparent) and
22479 255 (fully opaque, the default).
22482 Set the number of decimals to show in the timestamp, only available in the
22483 @code{testsrc} source.
22485 The displayed timestamp value will correspond to the original
22486 timestamp value multiplied by the power of 10 of the specified
22487 value. Default value is 0.
22490 @subsection Examples
22494 Generate a video with a duration of 5.3 seconds, with size
22495 176x144 and a frame rate of 10 frames per second:
22497 testsrc=duration=5.3:size=qcif:rate=10
22501 The following graph description will generate a red source
22502 with an opacity of 0.2, with size "qcif" and a frame rate of 10
22505 color=c=red@@0.2:s=qcif:r=10
22509 If the input content is to be ignored, @code{nullsrc} can be used. The
22510 following command generates noise in the luminance plane by employing
22511 the @code{geq} filter:
22513 nullsrc=s=256x256, geq=random(1)*255:128:128
22517 @subsection Commands
22519 The @code{color} source supports the following commands:
22523 Set the color of the created image. Accepts the same syntax of the
22524 corresponding @option{color} option.
22529 Generate video using an OpenCL program.
22534 OpenCL program source file.
22537 Kernel name in program.
22540 Size of frames to generate. This must be set.
22543 Pixel format to use for the generated frames. This must be set.
22546 Number of frames generated every second. Default value is '25'.
22550 For details of how the program loading works, see the @ref{program_opencl}
22557 Generate a colour ramp by setting pixel values from the position of the pixel
22558 in the output image. (Note that this will work with all pixel formats, but
22559 the generated output will not be the same.)
22561 __kernel void ramp(__write_only image2d_t dst,
22562 unsigned int index)
22564 int2 loc = (int2)(get_global_id(0), get_global_id(1));
22567 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
22569 write_imagef(dst, loc, val);
22574 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
22576 __kernel void sierpinski_carpet(__write_only image2d_t dst,
22577 unsigned int index)
22579 int2 loc = (int2)(get_global_id(0), get_global_id(1));
22581 float4 value = 0.0f;
22582 int x = loc.x + index;
22583 int y = loc.y + index;
22584 while (x > 0 || y > 0) {
22585 if (x % 3 == 1 && y % 3 == 1) {
22593 write_imagef(dst, loc, value);
22599 @section sierpinski
22601 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
22603 This source accepts the following options:
22607 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
22608 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
22611 Set frame rate, expressed as number of frames per second. Default
22615 Set seed which is used for random panning.
22618 Set max jump for single pan destination. Allowed range is from 1 to 10000.
22621 Set fractal type, can be default @code{carpet} or @code{triangle}.
22624 @c man end VIDEO SOURCES
22626 @chapter Video Sinks
22627 @c man begin VIDEO SINKS
22629 Below is a description of the currently available video sinks.
22631 @section buffersink
22633 Buffer video frames, and make them available to the end of the filter
22636 This sink is mainly intended for programmatic use, in particular
22637 through the interface defined in @file{libavfilter/buffersink.h}
22638 or the options system.
22640 It accepts a pointer to an AVBufferSinkContext structure, which
22641 defines the incoming buffers' formats, to be passed as the opaque
22642 parameter to @code{avfilter_init_filter} for initialization.
22646 Null video sink: do absolutely nothing with the input video. It is
22647 mainly useful as a template and for use in analysis / debugging
22650 @c man end VIDEO SINKS
22652 @chapter Multimedia Filters
22653 @c man begin MULTIMEDIA FILTERS
22655 Below is a description of the currently available multimedia filters.
22659 Convert input audio to a video output, displaying the audio bit scope.
22661 The filter accepts the following options:
22665 Set frame rate, expressed as number of frames per second. Default
22669 Specify the video size for the output. For the syntax of this option, check the
22670 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22671 Default value is @code{1024x256}.
22674 Specify list of colors separated by space or by '|' which will be used to
22675 draw channels. Unrecognized or missing colors will be replaced
22679 @section adrawgraph
22680 Draw a graph using input audio metadata.
22682 See @ref{drawgraph}
22684 @section agraphmonitor
22686 See @ref{graphmonitor}.
22688 @section ahistogram
22690 Convert input audio to a video output, displaying the volume histogram.
22692 The filter accepts the following options:
22696 Specify how histogram is calculated.
22698 It accepts the following values:
22701 Use single histogram for all channels.
22703 Use separate histogram for each channel.
22705 Default is @code{single}.
22708 Set frame rate, expressed as number of frames per second. Default
22712 Specify the video size for the output. For the syntax of this option, check the
22713 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22714 Default value is @code{hd720}.
22719 It accepts the following values:
22730 reverse logarithmic
22732 Default is @code{log}.
22735 Set amplitude scale.
22737 It accepts the following values:
22744 Default is @code{log}.
22747 Set how much frames to accumulate in histogram.
22748 Default is 1. Setting this to -1 accumulates all frames.
22751 Set histogram ratio of window height.
22754 Set sonogram sliding.
22756 It accepts the following values:
22759 replace old rows with new ones.
22761 scroll from top to bottom.
22763 Default is @code{replace}.
22766 @section aphasemeter
22768 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
22769 representing mean phase of current audio frame. A video output can also be produced and is
22770 enabled by default. The audio is passed through as first output.
22772 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
22773 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
22774 and @code{1} means channels are in phase.
22776 The filter accepts the following options, all related to its video output:
22780 Set the output frame rate. Default value is @code{25}.
22783 Set the video size for the output. For the syntax of this option, check the
22784 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22785 Default value is @code{800x400}.
22790 Specify the red, green, blue contrast. Default values are @code{2},
22791 @code{7} and @code{1}.
22792 Allowed range is @code{[0, 255]}.
22795 Set color which will be used for drawing median phase. If color is
22796 @code{none} which is default, no median phase value will be drawn.
22799 Enable video output. Default is enabled.
22802 @section avectorscope
22804 Convert input audio to a video output, representing the audio vector
22807 The filter is used to measure the difference between channels of stereo
22808 audio stream. A monaural signal, consisting of identical left and right
22809 signal, results in straight vertical line. Any stereo separation is visible
22810 as a deviation from this line, creating a Lissajous figure.
22811 If the straight (or deviation from it) but horizontal line appears this
22812 indicates that the left and right channels are out of phase.
22814 The filter accepts the following options:
22818 Set the vectorscope mode.
22820 Available values are:
22823 Lissajous rotated by 45 degrees.
22826 Same as above but not rotated.
22829 Shape resembling half of circle.
22832 Default value is @samp{lissajous}.
22835 Set the video size for the output. For the syntax of this option, check the
22836 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22837 Default value is @code{400x400}.
22840 Set the output frame rate. Default value is @code{25}.
22846 Specify the red, green, blue and alpha contrast. Default values are @code{40},
22847 @code{160}, @code{80} and @code{255}.
22848 Allowed range is @code{[0, 255]}.
22854 Specify the red, green, blue and alpha fade. Default values are @code{15},
22855 @code{10}, @code{5} and @code{5}.
22856 Allowed range is @code{[0, 255]}.
22859 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
22860 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
22863 Set the vectorscope drawing mode.
22865 Available values are:
22868 Draw dot for each sample.
22871 Draw line between previous and current sample.
22874 Default value is @samp{dot}.
22877 Specify amplitude scale of audio samples.
22879 Available values are:
22895 Swap left channel axis with right channel axis.
22905 Mirror only x axis.
22908 Mirror only y axis.
22916 @subsection Examples
22920 Complete example using @command{ffplay}:
22922 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
22923 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
22927 @section bench, abench
22929 Benchmark part of a filtergraph.
22931 The filter accepts the following options:
22935 Start or stop a timer.
22937 Available values are:
22940 Get the current time, set it as frame metadata (using the key
22941 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
22944 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
22945 the input frame metadata to get the time difference. Time difference, average,
22946 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
22947 @code{min}) are then printed. The timestamps are expressed in seconds.
22951 @subsection Examples
22955 Benchmark @ref{selectivecolor} filter:
22957 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
22963 Concatenate audio and video streams, joining them together one after the
22966 The filter works on segments of synchronized video and audio streams. All
22967 segments must have the same number of streams of each type, and that will
22968 also be the number of streams at output.
22970 The filter accepts the following options:
22975 Set the number of segments. Default is 2.
22978 Set the number of output video streams, that is also the number of video
22979 streams in each segment. Default is 1.
22982 Set the number of output audio streams, that is also the number of audio
22983 streams in each segment. Default is 0.
22986 Activate unsafe mode: do not fail if segments have a different format.
22990 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
22991 @var{a} audio outputs.
22993 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
22994 segment, in the same order as the outputs, then the inputs for the second
22997 Related streams do not always have exactly the same duration, for various
22998 reasons including codec frame size or sloppy authoring. For that reason,
22999 related synchronized streams (e.g. a video and its audio track) should be
23000 concatenated at once. The concat filter will use the duration of the longest
23001 stream in each segment (except the last one), and if necessary pad shorter
23002 audio streams with silence.
23004 For this filter to work correctly, all segments must start at timestamp 0.
23006 All corresponding streams must have the same parameters in all segments; the
23007 filtering system will automatically select a common pixel format for video
23008 streams, and a common sample format, sample rate and channel layout for
23009 audio streams, but other settings, such as resolution, must be converted
23010 explicitly by the user.
23012 Different frame rates are acceptable but will result in variable frame rate
23013 at output; be sure to configure the output file to handle it.
23015 @subsection Examples
23019 Concatenate an opening, an episode and an ending, all in bilingual version
23020 (video in stream 0, audio in streams 1 and 2):
23022 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
23023 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
23024 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
23025 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
23029 Concatenate two parts, handling audio and video separately, using the
23030 (a)movie sources, and adjusting the resolution:
23032 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
23033 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
23034 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
23036 Note that a desync will happen at the stitch if the audio and video streams
23037 do not have exactly the same duration in the first file.
23041 @subsection Commands
23043 This filter supports the following commands:
23046 Close the current segment and step to the next one
23052 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
23053 level. By default, it logs a message at a frequency of 10Hz with the
23054 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
23055 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
23057 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
23058 sample format is double-precision floating point. The input stream will be converted to
23059 this specification, if needed. Users may need to insert aformat and/or aresample filters
23060 after this filter to obtain the original parameters.
23062 The filter also has a video output (see the @var{video} option) with a real
23063 time graph to observe the loudness evolution. The graphic contains the logged
23064 message mentioned above, so it is not printed anymore when this option is set,
23065 unless the verbose logging is set. The main graphing area contains the
23066 short-term loudness (3 seconds of analysis), and the gauge on the right is for
23067 the momentary loudness (400 milliseconds), but can optionally be configured
23068 to instead display short-term loudness (see @var{gauge}).
23070 The green area marks a +/- 1LU target range around the target loudness
23071 (-23LUFS by default, unless modified through @var{target}).
23073 More information about the Loudness Recommendation EBU R128 on
23074 @url{http://tech.ebu.ch/loudness}.
23076 The filter accepts the following options:
23081 Activate the video output. The audio stream is passed unchanged whether this
23082 option is set or no. The video stream will be the first output stream if
23083 activated. Default is @code{0}.
23086 Set the video size. This option is for video only. For the syntax of this
23088 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23089 Default and minimum resolution is @code{640x480}.
23092 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
23093 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
23094 other integer value between this range is allowed.
23097 Set metadata injection. If set to @code{1}, the audio input will be segmented
23098 into 100ms output frames, each of them containing various loudness information
23099 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
23101 Default is @code{0}.
23104 Force the frame logging level.
23106 Available values are:
23109 information logging level
23111 verbose logging level
23114 By default, the logging level is set to @var{info}. If the @option{video} or
23115 the @option{metadata} options are set, it switches to @var{verbose}.
23120 Available modes can be cumulated (the option is a @code{flag} type). Possible
23124 Disable any peak mode (default).
23126 Enable sample-peak mode.
23128 Simple peak mode looking for the higher sample value. It logs a message
23129 for sample-peak (identified by @code{SPK}).
23131 Enable true-peak mode.
23133 If enabled, the peak lookup is done on an over-sampled version of the input
23134 stream for better peak accuracy. It logs a message for true-peak.
23135 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
23136 This mode requires a build with @code{libswresample}.
23140 Treat mono input files as "dual mono". If a mono file is intended for playback
23141 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
23142 If set to @code{true}, this option will compensate for this effect.
23143 Multi-channel input files are not affected by this option.
23146 Set a specific pan law to be used for the measurement of dual mono files.
23147 This parameter is optional, and has a default value of -3.01dB.
23150 Set a specific target level (in LUFS) used as relative zero in the visualization.
23151 This parameter is optional and has a default value of -23LUFS as specified
23152 by EBU R128. However, material published online may prefer a level of -16LUFS
23153 (e.g. for use with podcasts or video platforms).
23156 Set the value displayed by the gauge. Valid values are @code{momentary} and s
23157 @code{shortterm}. By default the momentary value will be used, but in certain
23158 scenarios it may be more useful to observe the short term value instead (e.g.
23162 Sets the display scale for the loudness. Valid parameters are @code{absolute}
23163 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
23164 video output, not the summary or continuous log output.
23167 @subsection Examples
23171 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
23173 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
23177 Run an analysis with @command{ffmpeg}:
23179 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
23183 @section interleave, ainterleave
23185 Temporally interleave frames from several inputs.
23187 @code{interleave} works with video inputs, @code{ainterleave} with audio.
23189 These filters read frames from several inputs and send the oldest
23190 queued frame to the output.
23192 Input streams must have well defined, monotonically increasing frame
23195 In order to submit one frame to output, these filters need to enqueue
23196 at least one frame for each input, so they cannot work in case one
23197 input is not yet terminated and will not receive incoming frames.
23199 For example consider the case when one input is a @code{select} filter
23200 which always drops input frames. The @code{interleave} filter will keep
23201 reading from that input, but it will never be able to send new frames
23202 to output until the input sends an end-of-stream signal.
23204 Also, depending on inputs synchronization, the filters will drop
23205 frames in case one input receives more frames than the other ones, and
23206 the queue is already filled.
23208 These filters accept the following options:
23212 Set the number of different inputs, it is 2 by default.
23215 @subsection Examples
23219 Interleave frames belonging to different streams using @command{ffmpeg}:
23221 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
23225 Add flickering blur effect:
23227 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
23231 @section metadata, ametadata
23233 Manipulate frame metadata.
23235 This filter accepts the following options:
23239 Set mode of operation of the filter.
23241 Can be one of the following:
23245 If both @code{value} and @code{key} is set, select frames
23246 which have such metadata. If only @code{key} is set, select
23247 every frame that has such key in metadata.
23250 Add new metadata @code{key} and @code{value}. If key is already available
23254 Modify value of already present key.
23257 If @code{value} is set, delete only keys that have such value.
23258 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
23262 Print key and its value if metadata was found. If @code{key} is not set print all
23263 metadata values available in frame.
23267 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
23270 Set metadata value which will be used. This option is mandatory for
23271 @code{modify} and @code{add} mode.
23274 Which function to use when comparing metadata value and @code{value}.
23276 Can be one of following:
23280 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
23283 Values are interpreted as strings, returns true if metadata value starts with
23284 the @code{value} option string.
23287 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
23290 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
23293 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
23296 Values are interpreted as floats, returns true if expression from option @code{expr}
23300 Values are interpreted as strings, returns true if metadata value ends with
23301 the @code{value} option string.
23305 Set expression which is used when @code{function} is set to @code{expr}.
23306 The expression is evaluated through the eval API and can contain the following
23311 Float representation of @code{value} from metadata key.
23314 Float representation of @code{value} as supplied by user in @code{value} option.
23318 If specified in @code{print} mode, output is written to the named file. Instead of
23319 plain filename any writable url can be specified. Filename ``-'' is a shorthand
23320 for standard output. If @code{file} option is not set, output is written to the log
23321 with AV_LOG_INFO loglevel.
23324 Reduces buffering in print mode when output is written to a URL set using @var{file}.
23328 @subsection Examples
23332 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
23335 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
23338 Print silencedetect output to file @file{metadata.txt}.
23340 silencedetect,ametadata=mode=print:file=metadata.txt
23343 Direct all metadata to a pipe with file descriptor 4.
23345 metadata=mode=print:file='pipe\:4'
23349 @section perms, aperms
23351 Set read/write permissions for the output frames.
23353 These filters are mainly aimed at developers to test direct path in the
23354 following filter in the filtergraph.
23356 The filters accept the following options:
23360 Select the permissions mode.
23362 It accepts the following values:
23365 Do nothing. This is the default.
23367 Set all the output frames read-only.
23369 Set all the output frames directly writable.
23371 Make the frame read-only if writable, and writable if read-only.
23373 Set each output frame read-only or writable randomly.
23377 Set the seed for the @var{random} mode, must be an integer included between
23378 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
23379 @code{-1}, the filter will try to use a good random seed on a best effort
23383 Note: in case of auto-inserted filter between the permission filter and the
23384 following one, the permission might not be received as expected in that
23385 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
23386 perms/aperms filter can avoid this problem.
23388 @section realtime, arealtime
23390 Slow down filtering to match real time approximately.
23392 These filters will pause the filtering for a variable amount of time to
23393 match the output rate with the input timestamps.
23394 They are similar to the @option{re} option to @code{ffmpeg}.
23396 They accept the following options:
23400 Time limit for the pauses. Any pause longer than that will be considered
23401 a timestamp discontinuity and reset the timer. Default is 2 seconds.
23403 Speed factor for processing. The value must be a float larger than zero.
23404 Values larger than 1.0 will result in faster than realtime processing,
23405 smaller will slow processing down. The @var{limit} is automatically adapted
23406 accordingly. Default is 1.0.
23408 A processing speed faster than what is possible without these filters cannot
23413 @section select, aselect
23415 Select frames to pass in output.
23417 This filter accepts the following options:
23422 Set expression, which is evaluated for each input frame.
23424 If the expression is evaluated to zero, the frame is discarded.
23426 If the evaluation result is negative or NaN, the frame is sent to the
23427 first output; otherwise it is sent to the output with index
23428 @code{ceil(val)-1}, assuming that the input index starts from 0.
23430 For example a value of @code{1.2} corresponds to the output with index
23431 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
23434 Set the number of outputs. The output to which to send the selected
23435 frame is based on the result of the evaluation. Default value is 1.
23438 The expression can contain the following constants:
23442 The (sequential) number of the filtered frame, starting from 0.
23445 The (sequential) number of the selected frame, starting from 0.
23447 @item prev_selected_n
23448 The sequential number of the last selected frame. It's NAN if undefined.
23451 The timebase of the input timestamps.
23454 The PTS (Presentation TimeStamp) of the filtered video frame,
23455 expressed in @var{TB} units. It's NAN if undefined.
23458 The PTS of the filtered video frame,
23459 expressed in seconds. It's NAN if undefined.
23462 The PTS of the previously filtered video frame. It's NAN if undefined.
23464 @item prev_selected_pts
23465 The PTS of the last previously filtered video frame. It's NAN if undefined.
23467 @item prev_selected_t
23468 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
23471 The PTS of the first video frame in the video. It's NAN if undefined.
23474 The time of the first video frame in the video. It's NAN if undefined.
23476 @item pict_type @emph{(video only)}
23477 The type of the filtered frame. It can assume one of the following
23489 @item interlace_type @emph{(video only)}
23490 The frame interlace type. It can assume one of the following values:
23493 The frame is progressive (not interlaced).
23495 The frame is top-field-first.
23497 The frame is bottom-field-first.
23500 @item consumed_sample_n @emph{(audio only)}
23501 the number of selected samples before the current frame
23503 @item samples_n @emph{(audio only)}
23504 the number of samples in the current frame
23506 @item sample_rate @emph{(audio only)}
23507 the input sample rate
23510 This is 1 if the filtered frame is a key-frame, 0 otherwise.
23513 the position in the file of the filtered frame, -1 if the information
23514 is not available (e.g. for synthetic video)
23516 @item scene @emph{(video only)}
23517 value between 0 and 1 to indicate a new scene; a low value reflects a low
23518 probability for the current frame to introduce a new scene, while a higher
23519 value means the current frame is more likely to be one (see the example below)
23521 @item concatdec_select
23522 The concat demuxer can select only part of a concat input file by setting an
23523 inpoint and an outpoint, but the output packets may not be entirely contained
23524 in the selected interval. By using this variable, it is possible to skip frames
23525 generated by the concat demuxer which are not exactly contained in the selected
23528 This works by comparing the frame pts against the @var{lavf.concat.start_time}
23529 and the @var{lavf.concat.duration} packet metadata values which are also
23530 present in the decoded frames.
23532 The @var{concatdec_select} variable is -1 if the frame pts is at least
23533 start_time and either the duration metadata is missing or the frame pts is less
23534 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
23537 That basically means that an input frame is selected if its pts is within the
23538 interval set by the concat demuxer.
23542 The default value of the select expression is "1".
23544 @subsection Examples
23548 Select all frames in input:
23553 The example above is the same as:
23565 Select only I-frames:
23567 select='eq(pict_type\,I)'
23571 Select one frame every 100:
23573 select='not(mod(n\,100))'
23577 Select only frames contained in the 10-20 time interval:
23579 select=between(t\,10\,20)
23583 Select only I-frames contained in the 10-20 time interval:
23585 select=between(t\,10\,20)*eq(pict_type\,I)
23589 Select frames with a minimum distance of 10 seconds:
23591 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
23595 Use aselect to select only audio frames with samples number > 100:
23597 aselect='gt(samples_n\,100)'
23601 Create a mosaic of the first scenes:
23603 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
23606 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
23610 Send even and odd frames to separate outputs, and compose them:
23612 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
23616 Select useful frames from an ffconcat file which is using inpoints and
23617 outpoints but where the source files are not intra frame only.
23619 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
23623 @section sendcmd, asendcmd
23625 Send commands to filters in the filtergraph.
23627 These filters read commands to be sent to other filters in the
23630 @code{sendcmd} must be inserted between two video filters,
23631 @code{asendcmd} must be inserted between two audio filters, but apart
23632 from that they act the same way.
23634 The specification of commands can be provided in the filter arguments
23635 with the @var{commands} option, or in a file specified by the
23636 @var{filename} option.
23638 These filters accept the following options:
23641 Set the commands to be read and sent to the other filters.
23643 Set the filename of the commands to be read and sent to the other
23647 @subsection Commands syntax
23649 A commands description consists of a sequence of interval
23650 specifications, comprising a list of commands to be executed when a
23651 particular event related to that interval occurs. The occurring event
23652 is typically the current frame time entering or leaving a given time
23655 An interval is specified by the following syntax:
23657 @var{START}[-@var{END}] @var{COMMANDS};
23660 The time interval is specified by the @var{START} and @var{END} times.
23661 @var{END} is optional and defaults to the maximum time.
23663 The current frame time is considered within the specified interval if
23664 it is included in the interval [@var{START}, @var{END}), that is when
23665 the time is greater or equal to @var{START} and is lesser than
23668 @var{COMMANDS} consists of a sequence of one or more command
23669 specifications, separated by ",", relating to that interval. The
23670 syntax of a command specification is given by:
23672 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
23675 @var{FLAGS} is optional and specifies the type of events relating to
23676 the time interval which enable sending the specified command, and must
23677 be a non-null sequence of identifier flags separated by "+" or "|" and
23678 enclosed between "[" and "]".
23680 The following flags are recognized:
23683 The command is sent when the current frame timestamp enters the
23684 specified interval. In other words, the command is sent when the
23685 previous frame timestamp was not in the given interval, and the
23689 The command is sent when the current frame timestamp leaves the
23690 specified interval. In other words, the command is sent when the
23691 previous frame timestamp was in the given interval, and the
23695 The command @var{ARG} is interpreted as expression and result of
23696 expression is passed as @var{ARG}.
23698 The expression is evaluated through the eval API and can contain the following
23703 Original position in the file of the frame, or undefined if undefined
23704 for the current frame.
23707 The presentation timestamp in input.
23710 The count of the input frame for video or audio, starting from 0.
23713 The time in seconds of the current frame.
23718 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
23721 @var{TARGET} specifies the target of the command, usually the name of
23722 the filter class or a specific filter instance name.
23724 @var{COMMAND} specifies the name of the command for the target filter.
23726 @var{ARG} is optional and specifies the optional list of argument for
23727 the given @var{COMMAND}.
23729 Between one interval specification and another, whitespaces, or
23730 sequences of characters starting with @code{#} until the end of line,
23731 are ignored and can be used to annotate comments.
23733 A simplified BNF description of the commands specification syntax
23736 @var{COMMAND_FLAG} ::= "enter" | "leave"
23737 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
23738 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
23739 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
23740 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
23741 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
23744 @subsection Examples
23748 Specify audio tempo change at second 4:
23750 asendcmd=c='4.0 atempo tempo 1.5',atempo
23754 Target a specific filter instance:
23756 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
23760 Specify a list of drawtext and hue commands in a file.
23762 # show text in the interval 5-10
23763 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
23764 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
23766 # desaturate the image in the interval 15-20
23767 15.0-20.0 [enter] hue s 0,
23768 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
23770 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
23772 # apply an exponential saturation fade-out effect, starting from time 25
23773 25 [enter] hue s exp(25-t)
23776 A filtergraph allowing to read and process the above command list
23777 stored in a file @file{test.cmd}, can be specified with:
23779 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
23784 @section setpts, asetpts
23786 Change the PTS (presentation timestamp) of the input frames.
23788 @code{setpts} works on video frames, @code{asetpts} on audio frames.
23790 This filter accepts the following options:
23795 The expression which is evaluated for each frame to construct its timestamp.
23799 The expression is evaluated through the eval API and can contain the following
23803 @item FRAME_RATE, FR
23804 frame rate, only defined for constant frame-rate video
23807 The presentation timestamp in input
23810 The count of the input frame for video or the number of consumed samples,
23811 not including the current frame for audio, starting from 0.
23813 @item NB_CONSUMED_SAMPLES
23814 The number of consumed samples, not including the current frame (only
23817 @item NB_SAMPLES, S
23818 The number of samples in the current frame (only audio)
23820 @item SAMPLE_RATE, SR
23821 The audio sample rate.
23824 The PTS of the first frame.
23827 the time in seconds of the first frame
23830 State whether the current frame is interlaced.
23833 the time in seconds of the current frame
23836 original position in the file of the frame, or undefined if undefined
23837 for the current frame
23840 The previous input PTS.
23843 previous input time in seconds
23846 The previous output PTS.
23849 previous output time in seconds
23852 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
23856 The wallclock (RTC) time at the start of the movie in microseconds.
23859 The timebase of the input timestamps.
23863 @subsection Examples
23867 Start counting PTS from zero
23869 setpts=PTS-STARTPTS
23873 Apply fast motion effect:
23879 Apply slow motion effect:
23885 Set fixed rate of 25 frames per second:
23891 Set fixed rate 25 fps with some jitter:
23893 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
23897 Apply an offset of 10 seconds to the input PTS:
23903 Generate timestamps from a "live source" and rebase onto the current timebase:
23905 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
23909 Generate timestamps by counting samples:
23918 Force color range for the output video frame.
23920 The @code{setrange} filter marks the color range property for the
23921 output frames. It does not change the input frame, but only sets the
23922 corresponding property, which affects how the frame is treated by
23925 The filter accepts the following options:
23930 Available values are:
23934 Keep the same color range property.
23936 @item unspecified, unknown
23937 Set the color range as unspecified.
23939 @item limited, tv, mpeg
23940 Set the color range as limited.
23942 @item full, pc, jpeg
23943 Set the color range as full.
23947 @section settb, asettb
23949 Set the timebase to use for the output frames timestamps.
23950 It is mainly useful for testing timebase configuration.
23952 It accepts the following parameters:
23957 The expression which is evaluated into the output timebase.
23961 The value for @option{tb} is an arithmetic expression representing a
23962 rational. The expression can contain the constants "AVTB" (the default
23963 timebase), "intb" (the input timebase) and "sr" (the sample rate,
23964 audio only). Default value is "intb".
23966 @subsection Examples
23970 Set the timebase to 1/25:
23976 Set the timebase to 1/10:
23982 Set the timebase to 1001/1000:
23988 Set the timebase to 2*intb:
23994 Set the default timebase value:
24001 Convert input audio to a video output representing frequency spectrum
24002 logarithmically using Brown-Puckette constant Q transform algorithm with
24003 direct frequency domain coefficient calculation (but the transform itself
24004 is not really constant Q, instead the Q factor is actually variable/clamped),
24005 with musical tone scale, from E0 to D#10.
24007 The filter accepts the following options:
24011 Specify the video size for the output. It must be even. For the syntax of this option,
24012 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24013 Default value is @code{1920x1080}.
24016 Set the output frame rate. Default value is @code{25}.
24019 Set the bargraph height. It must be even. Default value is @code{-1} which
24020 computes the bargraph height automatically.
24023 Set the axis height. It must be even. Default value is @code{-1} which computes
24024 the axis height automatically.
24027 Set the sonogram height. It must be even. Default value is @code{-1} which
24028 computes the sonogram height automatically.
24031 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
24032 instead. Default value is @code{1}.
24034 @item sono_v, volume
24035 Specify the sonogram volume expression. It can contain variables:
24038 the @var{bar_v} evaluated expression
24039 @item frequency, freq, f
24040 the frequency where it is evaluated
24041 @item timeclamp, tc
24042 the value of @var{timeclamp} option
24046 @item a_weighting(f)
24047 A-weighting of equal loudness
24048 @item b_weighting(f)
24049 B-weighting of equal loudness
24050 @item c_weighting(f)
24051 C-weighting of equal loudness.
24053 Default value is @code{16}.
24055 @item bar_v, volume2
24056 Specify the bargraph volume expression. It can contain variables:
24059 the @var{sono_v} evaluated expression
24060 @item frequency, freq, f
24061 the frequency where it is evaluated
24062 @item timeclamp, tc
24063 the value of @var{timeclamp} option
24067 @item a_weighting(f)
24068 A-weighting of equal loudness
24069 @item b_weighting(f)
24070 B-weighting of equal loudness
24071 @item c_weighting(f)
24072 C-weighting of equal loudness.
24074 Default value is @code{sono_v}.
24076 @item sono_g, gamma
24077 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
24078 higher gamma makes the spectrum having more range. Default value is @code{3}.
24079 Acceptable range is @code{[1, 7]}.
24081 @item bar_g, gamma2
24082 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
24086 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
24087 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
24089 @item timeclamp, tc
24090 Specify the transform timeclamp. At low frequency, there is trade-off between
24091 accuracy in time domain and frequency domain. If timeclamp is lower,
24092 event in time domain is represented more accurately (such as fast bass drum),
24093 otherwise event in frequency domain is represented more accurately
24094 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
24097 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
24098 limits future samples by applying asymmetric windowing in time domain, useful
24099 when low latency is required. Accepted range is @code{[0, 1]}.
24102 Specify the transform base frequency. Default value is @code{20.01523126408007475},
24103 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
24106 Specify the transform end frequency. Default value is @code{20495.59681441799654},
24107 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
24110 This option is deprecated and ignored.
24113 Specify the transform length in time domain. Use this option to control accuracy
24114 trade-off between time domain and frequency domain at every frequency sample.
24115 It can contain variables:
24117 @item frequency, freq, f
24118 the frequency where it is evaluated
24119 @item timeclamp, tc
24120 the value of @var{timeclamp} option.
24122 Default value is @code{384*tc/(384+tc*f)}.
24125 Specify the transform count for every video frame. Default value is @code{6}.
24126 Acceptable range is @code{[1, 30]}.
24129 Specify the transform count for every single pixel. Default value is @code{0},
24130 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
24133 Specify font file for use with freetype to draw the axis. If not specified,
24134 use embedded font. Note that drawing with font file or embedded font is not
24135 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
24139 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
24140 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
24144 Specify font color expression. This is arithmetic expression that should return
24145 integer value 0xRRGGBB. It can contain variables:
24147 @item frequency, freq, f
24148 the frequency where it is evaluated
24149 @item timeclamp, tc
24150 the value of @var{timeclamp} option
24155 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
24156 @item r(x), g(x), b(x)
24157 red, green, and blue value of intensity x.
24159 Default value is @code{st(0, (midi(f)-59.5)/12);
24160 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
24161 r(1-ld(1)) + b(ld(1))}.
24164 Specify image file to draw the axis. This option override @var{fontfile} and
24165 @var{fontcolor} option.
24168 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
24169 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
24170 Default value is @code{1}.
24173 Set colorspace. The accepted values are:
24176 Unspecified (default)
24185 BT.470BG or BT.601-6 625
24188 SMPTE-170M or BT.601-6 525
24194 BT.2020 with non-constant luminance
24199 Set spectrogram color scheme. This is list of floating point values with format
24200 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
24201 The default is @code{1|0.5|0|0|0.5|1}.
24205 @subsection Examples
24209 Playing audio while showing the spectrum:
24211 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
24215 Same as above, but with frame rate 30 fps:
24217 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
24221 Playing at 1280x720:
24223 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
24227 Disable sonogram display:
24233 A1 and its harmonics: A1, A2, (near)E3, A3:
24235 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),
24236 asplit[a][out1]; [a] showcqt [out0]'
24240 Same as above, but with more accuracy in frequency domain:
24242 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),
24243 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
24249 bar_v=10:sono_v=bar_v*a_weighting(f)
24253 Custom gamma, now spectrum is linear to the amplitude.
24259 Custom tlength equation:
24261 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)))'
24265 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
24267 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
24271 Custom font using fontconfig:
24273 font='Courier New,Monospace,mono|bold'
24277 Custom frequency range with custom axis using image file:
24279 axisfile=myaxis.png:basefreq=40:endfreq=10000
24285 Convert input audio to video output representing the audio power spectrum.
24286 Audio amplitude is on Y-axis while frequency is on X-axis.
24288 The filter accepts the following options:
24292 Specify size of video. For the syntax of this option, check the
24293 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24294 Default is @code{1024x512}.
24298 This set how each frequency bin will be represented.
24300 It accepts the following values:
24306 Default is @code{bar}.
24309 Set amplitude scale.
24311 It accepts the following values:
24325 Default is @code{log}.
24328 Set frequency scale.
24330 It accepts the following values:
24339 Reverse logarithmic scale.
24341 Default is @code{lin}.
24344 Set window size. Allowed range is from 16 to 65536.
24346 Default is @code{2048}
24349 Set windowing function.
24351 It accepts the following values:
24374 Default is @code{hanning}.
24377 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
24378 which means optimal overlap for selected window function will be picked.
24381 Set time averaging. Setting this to 0 will display current maximal peaks.
24382 Default is @code{1}, which means time averaging is disabled.
24385 Specify list of colors separated by space or by '|' which will be used to
24386 draw channel frequencies. Unrecognized or missing colors will be replaced
24390 Set channel display mode.
24392 It accepts the following values:
24397 Default is @code{combined}.
24400 Set minimum amplitude used in @code{log} amplitude scaler.
24404 @section showspatial
24406 Convert stereo input audio to a video output, representing the spatial relationship
24407 between two channels.
24409 The filter accepts the following options:
24413 Specify the video size for the output. For the syntax of this option, check the
24414 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24415 Default value is @code{512x512}.
24418 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
24421 Set window function.
24423 It accepts the following values:
24448 Default value is @code{hann}.
24451 Set ratio of overlap window. Default value is @code{0.5}.
24452 When value is @code{1} overlap is set to recommended size for specific
24453 window function currently used.
24456 @anchor{showspectrum}
24457 @section showspectrum
24459 Convert input audio to a video output, representing the audio frequency
24462 The filter accepts the following options:
24466 Specify the video size for the output. For the syntax of this option, check the
24467 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24468 Default value is @code{640x512}.
24471 Specify how the spectrum should slide along the window.
24473 It accepts the following values:
24476 the samples start again on the left when they reach the right
24478 the samples scroll from right to left
24480 frames are only produced when the samples reach the right
24482 the samples scroll from left to right
24485 Default value is @code{replace}.
24488 Specify display mode.
24490 It accepts the following values:
24493 all channels are displayed in the same row
24495 all channels are displayed in separate rows
24498 Default value is @samp{combined}.
24501 Specify display color mode.
24503 It accepts the following values:
24506 each channel is displayed in a separate color
24508 each channel is displayed using the same color scheme
24510 each channel is displayed using the rainbow color scheme
24512 each channel is displayed using the moreland color scheme
24514 each channel is displayed using the nebulae color scheme
24516 each channel is displayed using the fire color scheme
24518 each channel is displayed using the fiery color scheme
24520 each channel is displayed using the fruit color scheme
24522 each channel is displayed using the cool color scheme
24524 each channel is displayed using the magma color scheme
24526 each channel is displayed using the green color scheme
24528 each channel is displayed using the viridis color scheme
24530 each channel is displayed using the plasma color scheme
24532 each channel is displayed using the cividis color scheme
24534 each channel is displayed using the terrain color scheme
24537 Default value is @samp{channel}.
24540 Specify scale used for calculating intensity color values.
24542 It accepts the following values:
24547 square root, default
24558 Default value is @samp{sqrt}.
24561 Specify frequency scale.
24563 It accepts the following values:
24571 Default value is @samp{lin}.
24574 Set saturation modifier for displayed colors. Negative values provide
24575 alternative color scheme. @code{0} is no saturation at all.
24576 Saturation must be in [-10.0, 10.0] range.
24577 Default value is @code{1}.
24580 Set window function.
24582 It accepts the following values:
24607 Default value is @code{hann}.
24610 Set orientation of time vs frequency axis. Can be @code{vertical} or
24611 @code{horizontal}. Default is @code{vertical}.
24614 Set ratio of overlap window. Default value is @code{0}.
24615 When value is @code{1} overlap is set to recommended size for specific
24616 window function currently used.
24619 Set scale gain for calculating intensity color values.
24620 Default value is @code{1}.
24623 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
24626 Set color rotation, must be in [-1.0, 1.0] range.
24627 Default value is @code{0}.
24630 Set start frequency from which to display spectrogram. Default is @code{0}.
24633 Set stop frequency to which to display spectrogram. Default is @code{0}.
24636 Set upper frame rate limit. Default is @code{auto}, unlimited.
24639 Draw time and frequency axes and legends. Default is disabled.
24642 The usage is very similar to the showwaves filter; see the examples in that
24645 @subsection Examples
24649 Large window with logarithmic color scaling:
24651 showspectrum=s=1280x480:scale=log
24655 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
24657 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24658 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
24662 @section showspectrumpic
24664 Convert input audio to a single video frame, representing the audio frequency
24667 The filter accepts the following options:
24671 Specify the video size for the output. For the syntax of this option, check the
24672 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24673 Default value is @code{4096x2048}.
24676 Specify display mode.
24678 It accepts the following values:
24681 all channels are displayed in the same row
24683 all channels are displayed in separate rows
24685 Default value is @samp{combined}.
24688 Specify display color mode.
24690 It accepts the following values:
24693 each channel is displayed in a separate color
24695 each channel is displayed using the same color scheme
24697 each channel is displayed using the rainbow color scheme
24699 each channel is displayed using the moreland color scheme
24701 each channel is displayed using the nebulae color scheme
24703 each channel is displayed using the fire color scheme
24705 each channel is displayed using the fiery color scheme
24707 each channel is displayed using the fruit color scheme
24709 each channel is displayed using the cool color scheme
24711 each channel is displayed using the magma color scheme
24713 each channel is displayed using the green color scheme
24715 each channel is displayed using the viridis color scheme
24717 each channel is displayed using the plasma color scheme
24719 each channel is displayed using the cividis color scheme
24721 each channel is displayed using the terrain color scheme
24723 Default value is @samp{intensity}.
24726 Specify scale used for calculating intensity color values.
24728 It accepts the following values:
24733 square root, default
24743 Default value is @samp{log}.
24746 Specify frequency scale.
24748 It accepts the following values:
24756 Default value is @samp{lin}.
24759 Set saturation modifier for displayed colors. Negative values provide
24760 alternative color scheme. @code{0} is no saturation at all.
24761 Saturation must be in [-10.0, 10.0] range.
24762 Default value is @code{1}.
24765 Set window function.
24767 It accepts the following values:
24791 Default value is @code{hann}.
24794 Set orientation of time vs frequency axis. Can be @code{vertical} or
24795 @code{horizontal}. Default is @code{vertical}.
24798 Set scale gain for calculating intensity color values.
24799 Default value is @code{1}.
24802 Draw time and frequency axes and legends. Default is enabled.
24805 Set color rotation, must be in [-1.0, 1.0] range.
24806 Default value is @code{0}.
24809 Set start frequency from which to display spectrogram. Default is @code{0}.
24812 Set stop frequency to which to display spectrogram. Default is @code{0}.
24815 @subsection Examples
24819 Extract an audio spectrogram of a whole audio track
24820 in a 1024x1024 picture using @command{ffmpeg}:
24822 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
24826 @section showvolume
24828 Convert input audio volume to a video output.
24830 The filter accepts the following options:
24837 Set border width, allowed range is [0, 5]. Default is 1.
24840 Set channel width, allowed range is [80, 8192]. Default is 400.
24843 Set channel height, allowed range is [1, 900]. Default is 20.
24846 Set fade, allowed range is [0, 1]. Default is 0.95.
24849 Set volume color expression.
24851 The expression can use the following variables:
24855 Current max volume of channel in dB.
24861 Current channel number, starting from 0.
24865 If set, displays channel names. Default is enabled.
24868 If set, displays volume values. Default is enabled.
24871 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
24872 default is @code{h}.
24875 Set step size, allowed range is [0, 5]. Default is 0, which means
24879 Set background opacity, allowed range is [0, 1]. Default is 0.
24882 Set metering mode, can be peak: @code{p} or rms: @code{r},
24883 default is @code{p}.
24886 Set display scale, can be linear: @code{lin} or log: @code{log},
24887 default is @code{lin}.
24891 If set to > 0., display a line for the max level
24892 in the previous seconds.
24893 default is disabled: @code{0.}
24896 The color of the max line. Use when @code{dm} option is set to > 0.
24897 default is: @code{orange}
24902 Convert input audio to a video output, representing the samples waves.
24904 The filter accepts the following options:
24908 Specify the video size for the output. For the syntax of this option, check the
24909 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24910 Default value is @code{600x240}.
24915 Available values are:
24918 Draw a point for each sample.
24921 Draw a vertical line for each sample.
24924 Draw a point for each sample and a line between them.
24927 Draw a centered vertical line for each sample.
24930 Default value is @code{point}.
24933 Set the number of samples which are printed on the same column. A
24934 larger value will decrease the frame rate. Must be a positive
24935 integer. This option can be set only if the value for @var{rate}
24936 is not explicitly specified.
24939 Set the (approximate) output frame rate. This is done by setting the
24940 option @var{n}. Default value is "25".
24942 @item split_channels
24943 Set if channels should be drawn separately or overlap. Default value is 0.
24946 Set colors separated by '|' which are going to be used for drawing of each channel.
24949 Set amplitude scale.
24951 Available values are:
24969 Set the draw mode. This is mostly useful to set for high @var{n}.
24971 Available values are:
24974 Scale pixel values for each drawn sample.
24977 Draw every sample directly.
24980 Default value is @code{scale}.
24983 @subsection Examples
24987 Output the input file audio and the corresponding video representation
24990 amovie=a.mp3,asplit[out0],showwaves[out1]
24994 Create a synthetic signal and show it with showwaves, forcing a
24995 frame rate of 30 frames per second:
24997 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
25001 @section showwavespic
25003 Convert input audio to a single video frame, representing the samples waves.
25005 The filter accepts the following options:
25009 Specify the video size for the output. For the syntax of this option, check the
25010 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25011 Default value is @code{600x240}.
25013 @item split_channels
25014 Set if channels should be drawn separately or overlap. Default value is 0.
25017 Set colors separated by '|' which are going to be used for drawing of each channel.
25020 Set amplitude scale.
25022 Available values are:
25042 Available values are:
25045 Scale pixel values for each drawn sample.
25048 Draw every sample directly.
25051 Default value is @code{scale}.
25054 @subsection Examples
25058 Extract a channel split representation of the wave form of a whole audio track
25059 in a 1024x800 picture using @command{ffmpeg}:
25061 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
25065 @section sidedata, asidedata
25067 Delete frame side data, or select frames based on it.
25069 This filter accepts the following options:
25073 Set mode of operation of the filter.
25075 Can be one of the following:
25079 Select every frame with side data of @code{type}.
25082 Delete side data of @code{type}. If @code{type} is not set, delete all side
25088 Set side data type used with all modes. Must be set for @code{select} mode. For
25089 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
25090 in @file{libavutil/frame.h}. For example, to choose
25091 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
25095 @section spectrumsynth
25097 Synthesize audio from 2 input video spectrums, first input stream represents
25098 magnitude across time and second represents phase across time.
25099 The filter will transform from frequency domain as displayed in videos back
25100 to time domain as presented in audio output.
25102 This filter is primarily created for reversing processed @ref{showspectrum}
25103 filter outputs, but can synthesize sound from other spectrograms too.
25104 But in such case results are going to be poor if the phase data is not
25105 available, because in such cases phase data need to be recreated, usually
25106 it's just recreated from random noise.
25107 For best results use gray only output (@code{channel} color mode in
25108 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
25109 @code{lin} scale for phase video. To produce phase, for 2nd video, use
25110 @code{data} option. Inputs videos should generally use @code{fullframe}
25111 slide mode as that saves resources needed for decoding video.
25113 The filter accepts the following options:
25117 Specify sample rate of output audio, the sample rate of audio from which
25118 spectrum was generated may differ.
25121 Set number of channels represented in input video spectrums.
25124 Set scale which was used when generating magnitude input spectrum.
25125 Can be @code{lin} or @code{log}. Default is @code{log}.
25128 Set slide which was used when generating inputs spectrums.
25129 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
25130 Default is @code{fullframe}.
25133 Set window function used for resynthesis.
25136 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25137 which means optimal overlap for selected window function will be picked.
25140 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
25141 Default is @code{vertical}.
25144 @subsection Examples
25148 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
25149 then resynthesize videos back to audio with spectrumsynth:
25151 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
25152 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
25153 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
25157 @section split, asplit
25159 Split input into several identical outputs.
25161 @code{asplit} works with audio input, @code{split} with video.
25163 The filter accepts a single parameter which specifies the number of outputs. If
25164 unspecified, it defaults to 2.
25166 @subsection Examples
25170 Create two separate outputs from the same input:
25172 [in] split [out0][out1]
25176 To create 3 or more outputs, you need to specify the number of
25179 [in] asplit=3 [out0][out1][out2]
25183 Create two separate outputs from the same input, one cropped and
25186 [in] split [splitout1][splitout2];
25187 [splitout1] crop=100:100:0:0 [cropout];
25188 [splitout2] pad=200:200:100:100 [padout];
25192 Create 5 copies of the input audio with @command{ffmpeg}:
25194 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
25200 Receive commands sent through a libzmq client, and forward them to
25201 filters in the filtergraph.
25203 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
25204 must be inserted between two video filters, @code{azmq} between two
25205 audio filters. Both are capable to send messages to any filter type.
25207 To enable these filters you need to install the libzmq library and
25208 headers and configure FFmpeg with @code{--enable-libzmq}.
25210 For more information about libzmq see:
25211 @url{http://www.zeromq.org/}
25213 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
25214 receives messages sent through a network interface defined by the
25215 @option{bind_address} (or the abbreviation "@option{b}") option.
25216 Default value of this option is @file{tcp://localhost:5555}. You may
25217 want to alter this value to your needs, but do not forget to escape any
25218 ':' signs (see @ref{filtergraph escaping}).
25220 The received message must be in the form:
25222 @var{TARGET} @var{COMMAND} [@var{ARG}]
25225 @var{TARGET} specifies the target of the command, usually the name of
25226 the filter class or a specific filter instance name. The default
25227 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
25228 but you can override this by using the @samp{filter_name@@id} syntax
25229 (see @ref{Filtergraph syntax}).
25231 @var{COMMAND} specifies the name of the command for the target filter.
25233 @var{ARG} is optional and specifies the optional argument list for the
25234 given @var{COMMAND}.
25236 Upon reception, the message is processed and the corresponding command
25237 is injected into the filtergraph. Depending on the result, the filter
25238 will send a reply to the client, adopting the format:
25240 @var{ERROR_CODE} @var{ERROR_REASON}
25244 @var{MESSAGE} is optional.
25246 @subsection Examples
25248 Look at @file{tools/zmqsend} for an example of a zmq client which can
25249 be used to send commands processed by these filters.
25251 Consider the following filtergraph generated by @command{ffplay}.
25252 In this example the last overlay filter has an instance name. All other
25253 filters will have default instance names.
25256 ffplay -dumpgraph 1 -f lavfi "
25257 color=s=100x100:c=red [l];
25258 color=s=100x100:c=blue [r];
25259 nullsrc=s=200x100, zmq [bg];
25260 [bg][l] overlay [bg+l];
25261 [bg+l][r] overlay@@my=x=100 "
25264 To change the color of the left side of the video, the following
25265 command can be used:
25267 echo Parsed_color_0 c yellow | tools/zmqsend
25270 To change the right side:
25272 echo Parsed_color_1 c pink | tools/zmqsend
25275 To change the position of the right side:
25277 echo overlay@@my x 150 | tools/zmqsend
25281 @c man end MULTIMEDIA FILTERS
25283 @chapter Multimedia Sources
25284 @c man begin MULTIMEDIA SOURCES
25286 Below is a description of the currently available multimedia sources.
25290 This is the same as @ref{movie} source, except it selects an audio
25296 Read audio and/or video stream(s) from a movie container.
25298 It accepts the following parameters:
25302 The name of the resource to read (not necessarily a file; it can also be a
25303 device or a stream accessed through some protocol).
25305 @item format_name, f
25306 Specifies the format assumed for the movie to read, and can be either
25307 the name of a container or an input device. If not specified, the
25308 format is guessed from @var{movie_name} or by probing.
25310 @item seek_point, sp
25311 Specifies the seek point in seconds. The frames will be output
25312 starting from this seek point. The parameter is evaluated with
25313 @code{av_strtod}, so the numerical value may be suffixed by an IS
25314 postfix. The default value is "0".
25317 Specifies the streams to read. Several streams can be specified,
25318 separated by "+". The source will then have as many outputs, in the
25319 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
25320 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
25321 respectively the default (best suited) video and audio stream. Default
25322 is "dv", or "da" if the filter is called as "amovie".
25324 @item stream_index, si
25325 Specifies the index of the video stream to read. If the value is -1,
25326 the most suitable video stream will be automatically selected. The default
25327 value is "-1". Deprecated. If the filter is called "amovie", it will select
25328 audio instead of video.
25331 Specifies how many times to read the stream in sequence.
25332 If the value is 0, the stream will be looped infinitely.
25333 Default value is "1".
25335 Note that when the movie is looped the source timestamps are not
25336 changed, so it will generate non monotonically increasing timestamps.
25338 @item discontinuity
25339 Specifies the time difference between frames above which the point is
25340 considered a timestamp discontinuity which is removed by adjusting the later
25344 It allows overlaying a second video on top of the main input of
25345 a filtergraph, as shown in this graph:
25347 input -----------> deltapts0 --> overlay --> output
25350 movie --> scale--> deltapts1 -------+
25352 @subsection Examples
25356 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
25357 on top of the input labelled "in":
25359 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
25360 [in] setpts=PTS-STARTPTS [main];
25361 [main][over] overlay=16:16 [out]
25365 Read from a video4linux2 device, and overlay it on top of the input
25368 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
25369 [in] setpts=PTS-STARTPTS [main];
25370 [main][over] overlay=16:16 [out]
25374 Read the first video stream and the audio stream with id 0x81 from
25375 dvd.vob; the video is connected to the pad named "video" and the audio is
25376 connected to the pad named "audio":
25378 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
25382 @subsection Commands
25384 Both movie and amovie support the following commands:
25387 Perform seek using "av_seek_frame".
25388 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
25391 @var{stream_index}: If stream_index is -1, a default
25392 stream is selected, and @var{timestamp} is automatically converted
25393 from AV_TIME_BASE units to the stream specific time_base.
25395 @var{timestamp}: Timestamp in AVStream.time_base units
25396 or, if no stream is specified, in AV_TIME_BASE units.
25398 @var{flags}: Flags which select direction and seeking mode.
25402 Get movie duration in AV_TIME_BASE units.
25406 @c man end MULTIMEDIA SOURCES