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.
447 Simple audio dynamic range compression/expansion filter.
449 The filter accepts the following options:
453 Set contrast. Default is 33. Allowed range is between 0 and 100.
458 Copy the input audio source unchanged to the output. This is mainly useful for
463 Apply cross fade from one input audio stream to another input audio stream.
464 The cross fade is applied for specified duration near the end of first stream.
466 The filter accepts the following options:
470 Specify the number of samples for which the cross fade effect has to last.
471 At the end of the cross fade effect the first input audio will be completely
472 silent. Default is 44100.
475 Specify the duration of the cross fade effect. See
476 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
477 for the accepted syntax.
478 By default the duration is determined by @var{nb_samples}.
479 If set this option is used instead of @var{nb_samples}.
482 Should first stream end overlap with second stream start. Default is enabled.
485 Set curve for cross fade transition for first stream.
488 Set curve for cross fade transition for second stream.
490 For description of available curve types see @ref{afade} filter description.
497 Cross fade from one input to another:
499 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
503 Cross fade from one input to another but without overlapping:
505 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
510 Split audio stream into several bands.
512 This filter splits audio stream into two or more frequency ranges.
513 Summing all streams back will give flat output.
515 The filter accepts the following options:
519 Set split frequencies. Those must be positive and increasing.
522 Set filter order, can be @var{2nd}, @var{4th} or @var{8th}.
523 Default is @var{4th}.
528 Reduce audio bit resolution.
530 This filter is bit crusher with enhanced functionality. A bit crusher
531 is used to audibly reduce number of bits an audio signal is sampled
532 with. This doesn't change the bit depth at all, it just produces the
533 effect. Material reduced in bit depth sounds more harsh and "digital".
534 This filter is able to even round to continuous values instead of discrete
536 Additionally it has a D/C offset which results in different crushing of
537 the lower and the upper half of the signal.
538 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
540 Another feature of this filter is the logarithmic mode.
541 This setting switches from linear distances between bits to logarithmic ones.
542 The result is a much more "natural" sounding crusher which doesn't gate low
543 signals for example. The human ear has a logarithmic perception,
544 so this kind of crushing is much more pleasant.
545 Logarithmic crushing is also able to get anti-aliased.
547 The filter accepts the following options:
563 Can be linear: @code{lin} or logarithmic: @code{log}.
572 Set sample reduction.
575 Enable LFO. By default disabled.
586 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
590 Remove impulsive noise from input audio.
592 Samples detected as impulsive noise are replaced by interpolated samples using
593 autoregressive modelling.
597 Set window size, in milliseconds. Allowed range is from @code{10} to
598 @code{100}. Default value is @code{55} milliseconds.
599 This sets size of window which will be processed at once.
602 Set window overlap, in percentage of window size. Allowed range is from
603 @code{50} to @code{95}. Default value is @code{75} percent.
604 Setting this to a very high value increases impulsive noise removal but makes
605 whole process much slower.
608 Set autoregression order, in percentage of window size. Allowed range is from
609 @code{0} to @code{25}. Default value is @code{2} percent. This option also
610 controls quality of interpolated samples using neighbour good samples.
613 Set threshold value. Allowed range is from @code{1} to @code{100}.
614 Default value is @code{2}.
615 This controls the strength of impulsive noise which is going to be removed.
616 The lower value, the more samples will be detected as impulsive noise.
619 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
620 @code{10}. Default value is @code{2}.
621 If any two samples detected as noise are spaced less than this value then any
622 sample between those two samples will be also detected as noise.
627 It accepts the following values:
630 Select overlap-add method. Even not interpolated samples are slightly
631 changed with this method.
634 Select overlap-save method. Not interpolated samples remain unchanged.
637 Default value is @code{a}.
641 Remove clipped samples from input audio.
643 Samples detected as clipped are replaced by interpolated samples using
644 autoregressive modelling.
648 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
649 Default value is @code{55} milliseconds.
650 This sets size of window which will be processed at once.
653 Set window overlap, in percentage of window size. Allowed range is from @code{50}
654 to @code{95}. Default value is @code{75} percent.
657 Set autoregression order, in percentage of window size. Allowed range is from
658 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
659 quality of interpolated samples using neighbour good samples.
662 Set threshold value. Allowed range is from @code{1} to @code{100}.
663 Default value is @code{10}. Higher values make clip detection less aggressive.
666 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
667 Default value is @code{1000}. Higher values make clip detection less aggressive.
672 It accepts the following values:
675 Select overlap-add method. Even not interpolated samples are slightly changed
679 Select overlap-save method. Not interpolated samples remain unchanged.
682 Default value is @code{a}.
687 Delay one or more audio channels.
689 Samples in delayed channel are filled with silence.
691 The filter accepts the following option:
695 Set list of delays in milliseconds for each channel separated by '|'.
696 Unused delays will be silently ignored. If number of given delays is
697 smaller than number of channels all remaining channels will not be delayed.
698 If you want to delay exact number of samples, append 'S' to number.
699 If you want instead to delay in seconds, append 's' to number.
702 Use last set delay for all remaining channels. By default is disabled.
703 This option if enabled changes how option @code{delays} is interpreted.
710 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
711 the second channel (and any other channels that may be present) unchanged.
717 Delay second channel by 500 samples, the third channel by 700 samples and leave
718 the first channel (and any other channels that may be present) unchanged.
724 Delay all channels by same number of samples:
726 adelay=delays=64S:all=1
730 @section aderivative, aintegral
732 Compute derivative/integral of audio stream.
734 Applying both filters one after another produces original audio.
738 Apply echoing to the input audio.
740 Echoes are reflected sound and can occur naturally amongst mountains
741 (and sometimes large buildings) when talking or shouting; digital echo
742 effects emulate this behaviour and are often used to help fill out the
743 sound of a single instrument or vocal. The time difference between the
744 original signal and the reflection is the @code{delay}, and the
745 loudness of the reflected signal is the @code{decay}.
746 Multiple echoes can have different delays and decays.
748 A description of the accepted parameters follows.
752 Set input gain of reflected signal. Default is @code{0.6}.
755 Set output gain of reflected signal. Default is @code{0.3}.
758 Set list of time intervals in milliseconds between original signal and reflections
759 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
760 Default is @code{1000}.
763 Set list of loudness of reflected signals separated by '|'.
764 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
765 Default is @code{0.5}.
772 Make it sound as if there are twice as many instruments as are actually playing:
774 aecho=0.8:0.88:60:0.4
778 If delay is very short, then it sounds like a (metallic) robot playing music:
784 A longer delay will sound like an open air concert in the mountains:
786 aecho=0.8:0.9:1000:0.3
790 Same as above but with one more mountain:
792 aecho=0.8:0.9:1000|1800:0.3|0.25
797 Audio emphasis filter creates or restores material directly taken from LPs or
798 emphased CDs with different filter curves. E.g. to store music on vinyl the
799 signal has to be altered by a filter first to even out the disadvantages of
800 this recording medium.
801 Once the material is played back the inverse filter has to be applied to
802 restore the distortion of the frequency response.
804 The filter accepts the following options:
814 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
815 use @code{production} mode. Default is @code{reproduction} mode.
818 Set filter type. Selects medium. Can be one of the following:
830 select Compact Disc (CD).
836 select 50µs (FM-KF).
838 select 75µs (FM-KF).
844 Modify an audio signal according to the specified expressions.
846 This filter accepts one or more expressions (one for each channel),
847 which are evaluated and used to modify a corresponding audio signal.
849 It accepts the following parameters:
853 Set the '|'-separated expressions list for each separate channel. If
854 the number of input channels is greater than the number of
855 expressions, the last specified expression is used for the remaining
858 @item channel_layout, c
859 Set output channel layout. If not specified, the channel layout is
860 specified by the number of expressions. If set to @samp{same}, it will
861 use by default the same input channel layout.
864 Each expression in @var{exprs} can contain the following constants and functions:
868 channel number of the current expression
871 number of the evaluated sample, starting from 0
877 time of the evaluated sample expressed in seconds
880 @item nb_out_channels
881 input and output number of channels
884 the value of input channel with number @var{CH}
887 Note: this filter is slow. For faster processing you should use a
896 aeval=val(ch)/2:c=same
900 Invert phase of the second channel:
909 Apply fade-in/out effect to input audio.
911 A description of the accepted parameters follows.
915 Specify the effect type, can be either @code{in} for fade-in, or
916 @code{out} for a fade-out effect. Default is @code{in}.
918 @item start_sample, ss
919 Specify the number of the start sample for starting to apply the fade
920 effect. Default is 0.
923 Specify the number of samples for which the fade effect has to last. At
924 the end of the fade-in effect the output audio will have the same
925 volume as the input audio, at the end of the fade-out transition
926 the output audio will be silence. Default is 44100.
929 Specify the start time of the fade effect. Default is 0.
930 The value must be specified as a time duration; see
931 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
932 for the accepted syntax.
933 If set this option is used instead of @var{start_sample}.
936 Specify the duration of the fade effect. See
937 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
938 for the accepted syntax.
939 At the end of the fade-in effect the output audio will have the same
940 volume as the input audio, at the end of the fade-out transition
941 the output audio will be silence.
942 By default the duration is determined by @var{nb_samples}.
943 If set this option is used instead of @var{nb_samples}.
946 Set curve for fade transition.
948 It accepts the following values:
951 select triangular, linear slope (default)
953 select quarter of sine wave
955 select half of sine wave
957 select exponential sine wave
961 select inverted parabola
975 select inverted quarter of sine wave
977 select inverted half of sine wave
979 select double-exponential seat
981 select double-exponential sigmoid
983 select logistic sigmoid
993 Fade in first 15 seconds of audio:
999 Fade out last 25 seconds of a 900 seconds audio:
1001 afade=t=out:st=875:d=25
1006 Denoise audio samples with FFT.
1008 A description of the accepted parameters follows.
1012 Set the noise reduction in dB, allowed range is 0.01 to 97.
1013 Default value is 12 dB.
1016 Set the noise floor in dB, allowed range is -80 to -20.
1017 Default value is -50 dB.
1022 It accepts the following values:
1031 Select shellac noise.
1034 Select custom noise, defined in @code{bn} option.
1036 Default value is white noise.
1040 Set custom band noise for every one of 15 bands.
1041 Bands are separated by ' ' or '|'.
1044 Set the residual floor in dB, allowed range is -80 to -20.
1045 Default value is -38 dB.
1048 Enable noise tracking. By default is disabled.
1049 With this enabled, noise floor is automatically adjusted.
1052 Enable residual tracking. By default is disabled.
1055 Set the output mode.
1057 It accepts the following values:
1060 Pass input unchanged.
1063 Pass noise filtered out.
1068 Default value is @var{o}.
1072 @subsection Commands
1074 This filter supports the following commands:
1076 @item sample_noise, sn
1077 Start or stop measuring noise profile.
1078 Syntax for the command is : "start" or "stop" string.
1079 After measuring noise profile is stopped it will be
1080 automatically applied in filtering.
1082 @item noise_reduction, nr
1083 Change noise reduction. Argument is single float number.
1084 Syntax for the command is : "@var{noise_reduction}"
1086 @item noise_floor, nf
1087 Change noise floor. Argument is single float number.
1088 Syntax for the command is : "@var{noise_floor}"
1090 @item output_mode, om
1091 Change output mode operation.
1092 Syntax for the command is : "i", "o" or "n" string.
1096 Apply arbitrary expressions to samples in frequency domain.
1100 Set frequency domain real expression for each separate channel separated
1101 by '|'. Default is "re".
1102 If the number of input channels is greater than the number of
1103 expressions, the last specified expression is used for the remaining
1107 Set frequency domain imaginary expression for each separate channel
1108 separated by '|'. Default is "im".
1110 Each expression in @var{real} and @var{imag} can contain the following
1111 constants and functions:
1118 current frequency bin number
1121 number of available bins
1124 channel number of the current expression
1133 current real part of frequency bin of current channel
1136 current imaginary part of frequency bin of current channel
1139 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1142 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1146 Set window size. Allowed range is from 16 to 131072.
1147 Default is @code{4096}
1150 Set window function. Default is @code{hann}.
1153 Set window overlap. If set to 1, the recommended overlap for selected
1154 window function will be picked. Default is @code{0.75}.
1157 @subsection Examples
1161 Leave almost only low frequencies in audio:
1163 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1167 Apply robotize effect:
1169 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1173 Apply whisper effect:
1175 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"
1182 Apply an arbitrary Frequency Impulse Response filter.
1184 This filter is designed for applying long FIR filters,
1185 up to 60 seconds long.
1187 It can be used as component for digital crossover filters,
1188 room equalization, cross talk cancellation, wavefield synthesis,
1189 auralization, ambiophonics, ambisonics and spatialization.
1191 This filter uses the second stream as FIR coefficients.
1192 If the second stream holds a single channel, it will be used
1193 for all input channels in the first stream, otherwise
1194 the number of channels in the second stream must be same as
1195 the number of channels in the first stream.
1197 It accepts the following parameters:
1201 Set dry gain. This sets input gain.
1204 Set wet gain. This sets final output gain.
1207 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1210 Enable applying gain measured from power of IR.
1212 Set which approach to use for auto gain measurement.
1216 Do not apply any gain.
1219 select peak gain, very conservative approach. This is default value.
1222 select DC gain, limited application.
1225 select gain to noise approach, this is most popular one.
1229 Set gain to be applied to IR coefficients before filtering.
1230 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1233 Set format of IR stream. Can be @code{mono} or @code{input}.
1234 Default is @code{input}.
1237 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1238 Allowed range is 0.1 to 60 seconds.
1241 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1242 By default it is disabled.
1245 Set for which IR channel to display frequency response. By default is first channel
1246 displayed. This option is used only when @var{response} is enabled.
1249 Set video stream size. This option is used only when @var{response} is enabled.
1252 Set video stream frame rate. This option is used only when @var{response} is enabled.
1255 Set minimal partition size used for convolution. Default is @var{8192}.
1256 Allowed range is from @var{8} to @var{32768}.
1257 Lower values decreases latency at cost of higher CPU usage.
1260 Set maximal partition size used for convolution. Default is @var{8192}.
1261 Allowed range is from @var{8} to @var{32768}.
1262 Lower values may increase CPU usage.
1265 @subsection Examples
1269 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1271 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1278 Set output format constraints for the input audio. The framework will
1279 negotiate the most appropriate format to minimize conversions.
1281 It accepts the following parameters:
1285 A '|'-separated list of requested sample formats.
1288 A '|'-separated list of requested sample rates.
1290 @item channel_layouts
1291 A '|'-separated list of requested channel layouts.
1293 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1294 for the required syntax.
1297 If a parameter is omitted, all values are allowed.
1299 Force the output to either unsigned 8-bit or signed 16-bit stereo
1301 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1306 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1307 processing reduces disturbing noise between useful signals.
1309 Gating is done by detecting the volume below a chosen level @var{threshold}
1310 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1311 floor is set via @var{range}. Because an exact manipulation of the signal
1312 would cause distortion of the waveform the reduction can be levelled over
1313 time. This is done by setting @var{attack} and @var{release}.
1315 @var{attack} determines how long the signal has to fall below the threshold
1316 before any reduction will occur and @var{release} sets the time the signal
1317 has to rise above the threshold to reduce the reduction again.
1318 Shorter signals than the chosen attack time will be left untouched.
1322 Set input level before filtering.
1323 Default is 1. Allowed range is from 0.015625 to 64.
1326 Set the mode of operation. Can be @code{upward} or @code{downward}.
1327 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1328 will be amplified, expanding dynamic range in upward direction.
1329 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1332 Set the level of gain reduction when the signal is below the threshold.
1333 Default is 0.06125. Allowed range is from 0 to 1.
1334 Setting this to 0 disables reduction and then filter behaves like expander.
1337 If a signal rises above this level the gain reduction is released.
1338 Default is 0.125. Allowed range is from 0 to 1.
1341 Set a ratio by which the signal is reduced.
1342 Default is 2. Allowed range is from 1 to 9000.
1345 Amount of milliseconds the signal has to rise above the threshold before gain
1347 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1350 Amount of milliseconds the signal has to fall below the threshold before the
1351 reduction is increased again. Default is 250 milliseconds.
1352 Allowed range is from 0.01 to 9000.
1355 Set amount of amplification of signal after processing.
1356 Default is 1. Allowed range is from 1 to 64.
1359 Curve the sharp knee around the threshold to enter gain reduction more softly.
1360 Default is 2.828427125. Allowed range is from 1 to 8.
1363 Choose if exact signal should be taken for detection or an RMS like one.
1364 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1367 Choose if the average level between all channels or the louder channel affects
1369 Default is @code{average}. Can be @code{average} or @code{maximum}.
1374 Apply an arbitrary Infinite Impulse Response filter.
1376 It accepts the following parameters:
1380 Set numerator/zeros coefficients.
1383 Set denominator/poles coefficients.
1395 Set coefficients format.
1401 Z-plane zeros/poles, cartesian (default)
1403 Z-plane zeros/poles, polar radians
1405 Z-plane zeros/poles, polar degrees
1409 Set kind of processing.
1410 Can be @code{d} - direct or @code{s} - serial cascading. Default is @code{s}.
1413 Set filtering precision.
1417 double-precision floating-point (default)
1419 single-precision floating-point
1427 How much to use filtered signal in output. Default is 1.
1428 Range is between 0 and 1.
1431 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1432 By default it is disabled.
1435 Set for which IR channel to display frequency response. By default is first channel
1436 displayed. This option is used only when @var{response} is enabled.
1439 Set video stream size. This option is used only when @var{response} is enabled.
1442 Coefficients in @code{tf} format are separated by spaces and are in ascending
1445 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1446 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1449 Different coefficients and gains can be provided for every channel, in such case
1450 use '|' to separate coefficients or gains. Last provided coefficients will be
1451 used for all remaining channels.
1453 @subsection Examples
1457 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1459 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
1463 Same as above but in @code{zp} format:
1465 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
1471 The limiter prevents an input signal from rising over a desired threshold.
1472 This limiter uses lookahead technology to prevent your signal from distorting.
1473 It means that there is a small delay after the signal is processed. Keep in mind
1474 that the delay it produces is the attack time you set.
1476 The filter accepts the following options:
1480 Set input gain. Default is 1.
1483 Set output gain. Default is 1.
1486 Don't let signals above this level pass the limiter. Default is 1.
1489 The limiter will reach its attenuation level in this amount of time in
1490 milliseconds. Default is 5 milliseconds.
1493 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1494 Default is 50 milliseconds.
1497 When gain reduction is always needed ASC takes care of releasing to an
1498 average reduction level rather than reaching a reduction of 0 in the release
1502 Select how much the release time is affected by ASC, 0 means nearly no changes
1503 in release time while 1 produces higher release times.
1506 Auto level output signal. Default is enabled.
1507 This normalizes audio back to 0dB if enabled.
1510 Depending on picked setting it is recommended to upsample input 2x or 4x times
1511 with @ref{aresample} before applying this filter.
1515 Apply a two-pole all-pass filter with central frequency (in Hz)
1516 @var{frequency}, and filter-width @var{width}.
1517 An all-pass filter changes the audio's frequency to phase relationship
1518 without changing its frequency to amplitude relationship.
1520 The filter accepts the following options:
1524 Set frequency in Hz.
1527 Set method to specify band-width of filter.
1542 Specify the band-width of a filter in width_type units.
1545 How much to use filtered signal in output. Default is 1.
1546 Range is between 0 and 1.
1549 Specify which channels to filter, by default all available are filtered.
1552 @subsection Commands
1554 This filter supports the following commands:
1557 Change allpass frequency.
1558 Syntax for the command is : "@var{frequency}"
1561 Change allpass width_type.
1562 Syntax for the command is : "@var{width_type}"
1565 Change allpass width.
1566 Syntax for the command is : "@var{width}"
1570 Syntax for the command is : "@var{mix}"
1577 The filter accepts the following options:
1581 Set the number of loops. Setting this value to -1 will result in infinite loops.
1585 Set maximal number of samples. Default is 0.
1588 Set first sample of loop. Default is 0.
1594 Merge two or more audio streams into a single multi-channel stream.
1596 The filter accepts the following options:
1601 Set the number of inputs. Default is 2.
1605 If the channel layouts of the inputs are disjoint, and therefore compatible,
1606 the channel layout of the output will be set accordingly and the channels
1607 will be reordered as necessary. If the channel layouts of the inputs are not
1608 disjoint, the output will have all the channels of the first input then all
1609 the channels of the second input, in that order, and the channel layout of
1610 the output will be the default value corresponding to the total number of
1613 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1614 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1615 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1616 first input, b1 is the first channel of the second input).
1618 On the other hand, if both input are in stereo, the output channels will be
1619 in the default order: a1, a2, b1, b2, and the channel layout will be
1620 arbitrarily set to 4.0, which may or may not be the expected value.
1622 All inputs must have the same sample rate, and format.
1624 If inputs do not have the same duration, the output will stop with the
1627 @subsection Examples
1631 Merge two mono files into a stereo stream:
1633 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1637 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1639 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
1645 Mixes multiple audio inputs into a single output.
1647 Note that this filter only supports float samples (the @var{amerge}
1648 and @var{pan} audio filters support many formats). If the @var{amix}
1649 input has integer samples then @ref{aresample} will be automatically
1650 inserted to perform the conversion to float samples.
1654 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1656 will mix 3 input audio streams to a single output with the same duration as the
1657 first input and a dropout transition time of 3 seconds.
1659 It accepts the following parameters:
1663 The number of inputs. If unspecified, it defaults to 2.
1666 How to determine the end-of-stream.
1670 The duration of the longest input. (default)
1673 The duration of the shortest input.
1676 The duration of the first input.
1680 @item dropout_transition
1681 The transition time, in seconds, for volume renormalization when an input
1682 stream ends. The default value is 2 seconds.
1685 Specify weight of each input audio stream as sequence.
1686 Each weight is separated by space. By default all inputs have same weight.
1691 Multiply first audio stream with second audio stream and store result
1692 in output audio stream. Multiplication is done by multiplying each
1693 sample from first stream with sample at same position from second stream.
1695 With this element-wise multiplication one can create amplitude fades and
1696 amplitude modulations.
1698 @section anequalizer
1700 High-order parametric multiband equalizer for each channel.
1702 It accepts the following parameters:
1706 This option string is in format:
1707 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1708 Each equalizer band is separated by '|'.
1712 Set channel number to which equalization will be applied.
1713 If input doesn't have that channel the entry is ignored.
1716 Set central frequency for band.
1717 If input doesn't have that frequency the entry is ignored.
1720 Set band width in hertz.
1723 Set band gain in dB.
1726 Set filter type for band, optional, can be:
1730 Butterworth, this is default.
1741 With this option activated frequency response of anequalizer is displayed
1745 Set video stream size. Only useful if curves option is activated.
1748 Set max gain that will be displayed. Only useful if curves option is activated.
1749 Setting this to a reasonable value makes it possible to display gain which is derived from
1750 neighbour bands which are too close to each other and thus produce higher gain
1751 when both are activated.
1754 Set frequency scale used to draw frequency response in video output.
1755 Can be linear or logarithmic. Default is logarithmic.
1758 Set color for each channel curve which is going to be displayed in video stream.
1759 This is list of color names separated by space or by '|'.
1760 Unrecognised or missing colors will be replaced by white color.
1763 @subsection Examples
1767 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1768 for first 2 channels using Chebyshev type 1 filter:
1770 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1774 @subsection Commands
1776 This filter supports the following commands:
1779 Alter existing filter parameters.
1780 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1782 @var{fN} is existing filter number, starting from 0, if no such filter is available
1784 @var{freq} set new frequency parameter.
1785 @var{width} set new width parameter in herz.
1786 @var{gain} set new gain parameter in dB.
1788 Full filter invocation with asendcmd may look like this:
1789 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1794 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1796 Each sample is adjusted by looking for other samples with similar contexts. This
1797 context similarity is defined by comparing their surrounding patches of size
1798 @option{p}. Patches are searched in an area of @option{r} around the sample.
1800 The filter accepts the following options:
1804 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1807 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1808 Default value is 2 milliseconds.
1811 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1812 Default value is 6 milliseconds.
1815 Set the output mode.
1817 It accepts the following values:
1820 Pass input unchanged.
1823 Pass noise filtered out.
1828 Default value is @var{o}.
1832 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
1835 @subsection Commands
1837 This filter supports the following commands:
1840 Change denoise strength. Argument is single float number.
1841 Syntax for the command is : "@var{s}"
1845 Syntax for the command is : "i", "o" or "n" string.
1849 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
1851 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
1852 relate to producing the least mean square of the error signal (difference between the desired,
1853 2nd input audio stream and the actual signal, the 1st input audio stream).
1855 A description of the accepted options follows.
1868 Set the filter leakage.
1871 It accepts the following values:
1880 Pass filtered samples.
1883 Pass difference between desired and filtered samples.
1885 Default value is @var{o}.
1889 @subsection Examples
1893 One of many usages of this filter is noise reduction, input audio is filtered
1894 with same samples that are delayed by fixed amount, one such example for stereo audio is:
1896 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
1900 @subsection Commands
1902 This filter supports the same commands as options, excluding option @code{order}.
1906 Pass the audio source unchanged to the output.
1910 Pad the end of an audio stream with silence.
1912 This can be used together with @command{ffmpeg} @option{-shortest} to
1913 extend audio streams to the same length as the video stream.
1915 A description of the accepted options follows.
1919 Set silence packet size. Default value is 4096.
1922 Set the number of samples of silence to add to the end. After the
1923 value is reached, the stream is terminated. This option is mutually
1924 exclusive with @option{whole_len}.
1927 Set the minimum total number of samples in the output audio stream. If
1928 the value is longer than the input audio length, silence is added to
1929 the end, until the value is reached. This option is mutually exclusive
1930 with @option{pad_len}.
1933 Specify the duration of samples of silence to add. See
1934 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1935 for the accepted syntax. Used only if set to non-zero value.
1938 Specify the minimum total duration in the output audio stream. See
1939 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1940 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
1941 the input audio length, silence is added to the end, until the value is reached.
1942 This option is mutually exclusive with @option{pad_dur}
1945 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
1946 nor @option{whole_dur} option is set, the filter will add silence to the end of
1947 the input stream indefinitely.
1949 @subsection Examples
1953 Add 1024 samples of silence to the end of the input:
1959 Make sure the audio output will contain at least 10000 samples, pad
1960 the input with silence if required:
1962 apad=whole_len=10000
1966 Use @command{ffmpeg} to pad the audio input with silence, so that the
1967 video stream will always result the shortest and will be converted
1968 until the end in the output file when using the @option{shortest}
1971 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
1976 Add a phasing effect to the input audio.
1978 A phaser filter creates series of peaks and troughs in the frequency spectrum.
1979 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
1981 A description of the accepted parameters follows.
1985 Set input gain. Default is 0.4.
1988 Set output gain. Default is 0.74
1991 Set delay in milliseconds. Default is 3.0.
1994 Set decay. Default is 0.4.
1997 Set modulation speed in Hz. Default is 0.5.
2000 Set modulation type. Default is triangular.
2002 It accepts the following values:
2011 Audio pulsator is something between an autopanner and a tremolo.
2012 But it can produce funny stereo effects as well. Pulsator changes the volume
2013 of the left and right channel based on a LFO (low frequency oscillator) with
2014 different waveforms and shifted phases.
2015 This filter have the ability to define an offset between left and right
2016 channel. An offset of 0 means that both LFO shapes match each other.
2017 The left and right channel are altered equally - a conventional tremolo.
2018 An offset of 50% means that the shape of the right channel is exactly shifted
2019 in phase (or moved backwards about half of the frequency) - pulsator acts as
2020 an autopanner. At 1 both curves match again. Every setting in between moves the
2021 phase shift gapless between all stages and produces some "bypassing" sounds with
2022 sine and triangle waveforms. The more you set the offset near 1 (starting from
2023 the 0.5) the faster the signal passes from the left to the right speaker.
2025 The filter accepts the following options:
2029 Set input gain. By default it is 1. Range is [0.015625 - 64].
2032 Set output gain. By default it is 1. Range is [0.015625 - 64].
2035 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2036 sawup or sawdown. Default is sine.
2039 Set modulation. Define how much of original signal is affected by the LFO.
2042 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2045 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2048 Set pulse width. Default is 1. Allowed range is [0 - 2].
2051 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2054 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2058 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2062 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2063 if timing is set to hz.
2069 Resample the input audio to the specified parameters, using the
2070 libswresample library. If none are specified then the filter will
2071 automatically convert between its input and output.
2073 This filter is also able to stretch/squeeze the audio data to make it match
2074 the timestamps or to inject silence / cut out audio to make it match the
2075 timestamps, do a combination of both or do neither.
2077 The filter accepts the syntax
2078 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2079 expresses a sample rate and @var{resampler_options} is a list of
2080 @var{key}=@var{value} pairs, separated by ":". See the
2081 @ref{Resampler Options,,"Resampler Options" section in the
2082 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2083 for the complete list of supported options.
2085 @subsection Examples
2089 Resample the input audio to 44100Hz:
2095 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2096 samples per second compensation:
2098 aresample=async=1000
2104 Reverse an audio clip.
2106 Warning: This filter requires memory to buffer the entire clip, so trimming
2109 @subsection Examples
2113 Take the first 5 seconds of a clip, and reverse it.
2115 atrim=end=5,areverse
2121 Reduce noise from speech using Recurrent Neural Networks.
2123 This filter accepts the following options:
2127 Set train model file to load. This option is always required.
2130 @section asetnsamples
2132 Set the number of samples per each output audio frame.
2134 The last output packet may contain a different number of samples, as
2135 the filter will flush all the remaining samples when the input audio
2138 The filter accepts the following options:
2142 @item nb_out_samples, n
2143 Set the number of frames per each output audio frame. The number is
2144 intended as the number of samples @emph{per each channel}.
2145 Default value is 1024.
2148 If set to 1, the filter will pad the last audio frame with zeroes, so
2149 that the last frame will contain the same number of samples as the
2150 previous ones. Default value is 1.
2153 For example, to set the number of per-frame samples to 1234 and
2154 disable padding for the last frame, use:
2156 asetnsamples=n=1234:p=0
2161 Set the sample rate without altering the PCM data.
2162 This will result in a change of speed and pitch.
2164 The filter accepts the following options:
2167 @item sample_rate, r
2168 Set the output sample rate. Default is 44100 Hz.
2173 Show a line containing various information for each input audio frame.
2174 The input audio is not modified.
2176 The shown line contains a sequence of key/value pairs of the form
2177 @var{key}:@var{value}.
2179 The following values are shown in the output:
2183 The (sequential) number of the input frame, starting from 0.
2186 The presentation timestamp of the input frame, in time base units; the time base
2187 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2190 The presentation timestamp of the input frame in seconds.
2193 position of the frame in the input stream, -1 if this information in
2194 unavailable and/or meaningless (for example in case of synthetic audio)
2203 The sample rate for the audio frame.
2206 The number of samples (per channel) in the frame.
2209 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2210 audio, the data is treated as if all the planes were concatenated.
2212 @item plane_checksums
2213 A list of Adler-32 checksums for each data plane.
2217 Apply audio soft clipping.
2219 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2220 along a smooth curve, rather than the abrupt shape of hard-clipping.
2222 This filter accepts the following options:
2226 Set type of soft-clipping.
2228 It accepts the following values:
2240 Set additional parameter which controls sigmoid function.
2244 Automatic Speech Recognition
2246 This filter uses PocketSphinx for speech recognition. To enable
2247 compilation of this filter, you need to configure FFmpeg with
2248 @code{--enable-pocketsphinx}.
2250 It accepts the following options:
2254 Set sampling rate of input audio. Defaults is @code{16000}.
2255 This need to match speech models, otherwise one will get poor results.
2258 Set dictionary containing acoustic model files.
2261 Set pronunciation dictionary.
2264 Set language model file.
2267 Set language model set.
2270 Set which language model to use.
2273 Set output for log messages.
2276 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2281 Display time domain statistical information about the audio channels.
2282 Statistics are calculated and displayed for each audio channel and,
2283 where applicable, an overall figure is also given.
2285 It accepts the following option:
2288 Short window length in seconds, used for peak and trough RMS measurement.
2289 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2293 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2294 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2297 Available keys for each channel are:
2339 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2340 this @code{lavfi.astats.Overall.Peak_count}.
2342 For description what each key means read below.
2345 Set number of frame after which stats are going to be recalculated.
2346 Default is disabled.
2348 @item measure_perchannel
2349 Select the entries which need to be measured per channel. The metadata keys can
2350 be used as flags, default is @option{all} which measures everything.
2351 @option{none} disables all per channel measurement.
2353 @item measure_overall
2354 Select the entries which need to be measured overall. The metadata keys can
2355 be used as flags, default is @option{all} which measures everything.
2356 @option{none} disables all overall measurement.
2360 A description of each shown parameter follows:
2364 Mean amplitude displacement from zero.
2367 Minimal sample level.
2370 Maximal sample level.
2372 @item Min difference
2373 Minimal difference between two consecutive samples.
2375 @item Max difference
2376 Maximal difference between two consecutive samples.
2378 @item Mean difference
2379 Mean difference between two consecutive samples.
2380 The average of each difference between two consecutive samples.
2382 @item RMS difference
2383 Root Mean Square difference between two consecutive samples.
2387 Standard peak and RMS level measured in dBFS.
2391 Peak and trough values for RMS level measured over a short window.
2394 Standard ratio of peak to RMS level (note: not in dB).
2397 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2398 (i.e. either @var{Min level} or @var{Max level}).
2401 Number of occasions (not the number of samples) that the signal attained either
2402 @var{Min level} or @var{Max level}.
2405 Overall bit depth of audio. Number of bits used for each sample.
2408 Measured dynamic range of audio in dB.
2410 @item Zero crossings
2411 Number of points where the waveform crosses the zero level axis.
2413 @item Zero crossings rate
2414 Rate of Zero crossings and number of audio samples.
2421 The filter accepts exactly one parameter, the audio tempo. If not
2422 specified then the filter will assume nominal 1.0 tempo. Tempo must
2423 be in the [0.5, 100.0] range.
2425 Note that tempo greater than 2 will skip some samples rather than
2426 blend them in. If for any reason this is a concern it is always
2427 possible to daisy-chain several instances of atempo to achieve the
2428 desired product tempo.
2430 @subsection Examples
2434 Slow down audio to 80% tempo:
2440 To speed up audio to 300% tempo:
2446 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2448 atempo=sqrt(3),atempo=sqrt(3)
2452 @subsection Commands
2454 This filter supports the following commands:
2457 Change filter tempo scale factor.
2458 Syntax for the command is : "@var{tempo}"
2463 Trim the input so that the output contains one continuous subpart of the input.
2465 It accepts the following parameters:
2468 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2469 sample with the timestamp @var{start} will be the first sample in the output.
2472 Specify time of the first audio sample that will be dropped, i.e. the
2473 audio sample immediately preceding the one with the timestamp @var{end} will be
2474 the last sample in the output.
2477 Same as @var{start}, except this option sets the start timestamp in samples
2481 Same as @var{end}, except this option sets the end timestamp in samples instead
2485 The maximum duration of the output in seconds.
2488 The number of the first sample that should be output.
2491 The number of the first sample that should be dropped.
2494 @option{start}, @option{end}, and @option{duration} are expressed as time
2495 duration specifications; see
2496 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2498 Note that the first two sets of the start/end options and the @option{duration}
2499 option look at the frame timestamp, while the _sample options simply count the
2500 samples that pass through the filter. So start/end_pts and start/end_sample will
2501 give different results when the timestamps are wrong, inexact or do not start at
2502 zero. Also note that this filter does not modify the timestamps. If you wish
2503 to have the output timestamps start at zero, insert the asetpts filter after the
2506 If multiple start or end options are set, this filter tries to be greedy and
2507 keep all samples that match at least one of the specified constraints. To keep
2508 only the part that matches all the constraints at once, chain multiple atrim
2511 The defaults are such that all the input is kept. So it is possible to set e.g.
2512 just the end values to keep everything before the specified time.
2517 Drop everything except the second minute of input:
2519 ffmpeg -i INPUT -af atrim=60:120
2523 Keep only the first 1000 samples:
2525 ffmpeg -i INPUT -af atrim=end_sample=1000
2532 Apply a two-pole Butterworth band-pass filter with central
2533 frequency @var{frequency}, and (3dB-point) band-width width.
2534 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2535 instead of the default: constant 0dB peak gain.
2536 The filter roll off at 6dB per octave (20dB per decade).
2538 The filter accepts the following options:
2542 Set the filter's central frequency. Default is @code{3000}.
2545 Constant skirt gain if set to 1. Defaults to 0.
2548 Set method to specify band-width of filter.
2563 Specify the band-width of a filter in width_type units.
2566 How much to use filtered signal in output. Default is 1.
2567 Range is between 0 and 1.
2570 Specify which channels to filter, by default all available are filtered.
2573 @subsection Commands
2575 This filter supports the following commands:
2578 Change bandpass frequency.
2579 Syntax for the command is : "@var{frequency}"
2582 Change bandpass width_type.
2583 Syntax for the command is : "@var{width_type}"
2586 Change bandpass width.
2587 Syntax for the command is : "@var{width}"
2590 Change bandpass mix.
2591 Syntax for the command is : "@var{mix}"
2596 Apply a two-pole Butterworth band-reject filter with central
2597 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2598 The filter roll off at 6dB per octave (20dB per decade).
2600 The filter accepts the following options:
2604 Set the filter's central frequency. Default is @code{3000}.
2607 Set method to specify band-width of filter.
2622 Specify the band-width of a filter in width_type units.
2625 How much to use filtered signal in output. Default is 1.
2626 Range is between 0 and 1.
2629 Specify which channels to filter, by default all available are filtered.
2632 @subsection Commands
2634 This filter supports the following commands:
2637 Change bandreject frequency.
2638 Syntax for the command is : "@var{frequency}"
2641 Change bandreject width_type.
2642 Syntax for the command is : "@var{width_type}"
2645 Change bandreject width.
2646 Syntax for the command is : "@var{width}"
2649 Change bandreject mix.
2650 Syntax for the command is : "@var{mix}"
2653 @section bass, lowshelf
2655 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2656 shelving filter with a response similar to that of a standard
2657 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2659 The filter accepts the following options:
2663 Give the gain at 0 Hz. Its useful range is about -20
2664 (for a large cut) to +20 (for a large boost).
2665 Beware of clipping when using a positive gain.
2668 Set the filter's central frequency and so can be used
2669 to extend or reduce the frequency range to be boosted or cut.
2670 The default value is @code{100} Hz.
2673 Set method to specify band-width of filter.
2688 Determine how steep is the filter's shelf transition.
2691 How much to use filtered signal in output. Default is 1.
2692 Range is between 0 and 1.
2695 Specify which channels to filter, by default all available are filtered.
2698 @subsection Commands
2700 This filter supports the following commands:
2703 Change bass frequency.
2704 Syntax for the command is : "@var{frequency}"
2707 Change bass width_type.
2708 Syntax for the command is : "@var{width_type}"
2712 Syntax for the command is : "@var{width}"
2716 Syntax for the command is : "@var{gain}"
2720 Syntax for the command is : "@var{mix}"
2725 Apply a biquad IIR filter with the given coefficients.
2726 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
2727 are the numerator and denominator coefficients respectively.
2728 and @var{channels}, @var{c} specify which channels to filter, by default all
2729 available are filtered.
2731 @subsection Commands
2733 This filter supports the following commands:
2741 Change biquad parameter.
2742 Syntax for the command is : "@var{value}"
2745 How much to use filtered signal in output. Default is 1.
2746 Range is between 0 and 1.
2750 Bauer stereo to binaural transformation, which improves headphone listening of
2751 stereo audio records.
2753 To enable compilation of this filter you need to configure FFmpeg with
2754 @code{--enable-libbs2b}.
2756 It accepts the following parameters:
2760 Pre-defined crossfeed level.
2764 Default level (fcut=700, feed=50).
2767 Chu Moy circuit (fcut=700, feed=60).
2770 Jan Meier circuit (fcut=650, feed=95).
2775 Cut frequency (in Hz).
2784 Remap input channels to new locations.
2786 It accepts the following parameters:
2789 Map channels from input to output. The argument is a '|'-separated list of
2790 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
2791 @var{in_channel} form. @var{in_channel} can be either the name of the input
2792 channel (e.g. FL for front left) or its index in the input channel layout.
2793 @var{out_channel} is the name of the output channel or its index in the output
2794 channel layout. If @var{out_channel} is not given then it is implicitly an
2795 index, starting with zero and increasing by one for each mapping.
2797 @item channel_layout
2798 The channel layout of the output stream.
2801 If no mapping is present, the filter will implicitly map input channels to
2802 output channels, preserving indices.
2804 @subsection Examples
2808 For example, assuming a 5.1+downmix input MOV file,
2810 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
2812 will create an output WAV file tagged as stereo from the downmix channels of
2816 To fix a 5.1 WAV improperly encoded in AAC's native channel order
2818 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
2822 @section channelsplit
2824 Split each channel from an input audio stream into a separate output stream.
2826 It accepts the following parameters:
2828 @item channel_layout
2829 The channel layout of the input stream. The default is "stereo".
2831 A channel layout describing the channels to be extracted as separate output streams
2832 or "all" to extract each input channel as a separate stream. The default is "all".
2834 Choosing channels not present in channel layout in the input will result in an error.
2837 @subsection Examples
2841 For example, assuming a stereo input MP3 file,
2843 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
2845 will create an output Matroska file with two audio streams, one containing only
2846 the left channel and the other the right channel.
2849 Split a 5.1 WAV file into per-channel files:
2851 ffmpeg -i in.wav -filter_complex
2852 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
2853 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
2854 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
2859 Extract only LFE from a 5.1 WAV file:
2861 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
2862 -map '[LFE]' lfe.wav
2867 Add a chorus effect to the audio.
2869 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
2871 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
2872 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
2873 The modulation depth defines the range the modulated delay is played before or after
2874 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
2875 sound tuned around the original one, like in a chorus where some vocals are slightly
2878 It accepts the following parameters:
2881 Set input gain. Default is 0.4.
2884 Set output gain. Default is 0.4.
2887 Set delays. A typical delay is around 40ms to 60ms.
2899 @subsection Examples
2905 chorus=0.7:0.9:55:0.4:0.25:2
2911 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
2915 Fuller sounding chorus with three delays:
2917 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
2922 Compress or expand the audio's dynamic range.
2924 It accepts the following parameters:
2930 A list of times in seconds for each channel over which the instantaneous level
2931 of the input signal is averaged to determine its volume. @var{attacks} refers to
2932 increase of volume and @var{decays} refers to decrease of volume. For most
2933 situations, the attack time (response to the audio getting louder) should be
2934 shorter than the decay time, because the human ear is more sensitive to sudden
2935 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
2936 a typical value for decay is 0.8 seconds.
2937 If specified number of attacks & decays is lower than number of channels, the last
2938 set attack/decay will be used for all remaining channels.
2941 A list of points for the transfer function, specified in dB relative to the
2942 maximum possible signal amplitude. Each key points list must be defined using
2943 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
2944 @code{x0/y0 x1/y1 x2/y2 ....}
2946 The input values must be in strictly increasing order but the transfer function
2947 does not have to be monotonically rising. The point @code{0/0} is assumed but
2948 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
2949 function are @code{-70/-70|-60/-20|1/0}.
2952 Set the curve radius in dB for all joints. It defaults to 0.01.
2955 Set the additional gain in dB to be applied at all points on the transfer
2956 function. This allows for easy adjustment of the overall gain.
2960 Set an initial volume, in dB, to be assumed for each channel when filtering
2961 starts. This permits the user to supply a nominal level initially, so that, for
2962 example, a very large gain is not applied to initial signal levels before the
2963 companding has begun to operate. A typical value for audio which is initially
2964 quiet is -90 dB. It defaults to 0.
2967 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
2968 delayed before being fed to the volume adjuster. Specifying a delay
2969 approximately equal to the attack/decay times allows the filter to effectively
2970 operate in predictive rather than reactive mode. It defaults to 0.
2974 @subsection Examples
2978 Make music with both quiet and loud passages suitable for listening to in a
2981 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
2984 Another example for audio with whisper and explosion parts:
2986 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
2990 A noise gate for when the noise is at a lower level than the signal:
2992 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
2996 Here is another noise gate, this time for when the noise is at a higher level
2997 than the signal (making it, in some ways, similar to squelch):
2999 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3003 2:1 compression starting at -6dB:
3005 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3009 2:1 compression starting at -9dB:
3011 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3015 2:1 compression starting at -12dB:
3017 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3021 2:1 compression starting at -18dB:
3023 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3027 3:1 compression starting at -15dB:
3029 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3035 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3041 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
3045 Hard limiter at -6dB:
3047 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3051 Hard limiter at -12dB:
3053 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3057 Hard noise gate at -35 dB:
3059 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3065 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3069 @section compensationdelay
3071 Compensation Delay Line is a metric based delay to compensate differing
3072 positions of microphones or speakers.
3074 For example, you have recorded guitar with two microphones placed in
3075 different locations. Because the front of sound wave has fixed speed in
3076 normal conditions, the phasing of microphones can vary and depends on
3077 their location and interposition. The best sound mix can be achieved when
3078 these microphones are in phase (synchronized). Note that a distance of
3079 ~30 cm between microphones makes one microphone capture the signal in
3080 antiphase to the other microphone. That makes the final mix sound moody.
3081 This filter helps to solve phasing problems by adding different delays
3082 to each microphone track and make them synchronized.
3084 The best result can be reached when you take one track as base and
3085 synchronize other tracks one by one with it.
3086 Remember that synchronization/delay tolerance depends on sample rate, too.
3087 Higher sample rates will give more tolerance.
3089 The filter accepts the following parameters:
3093 Set millimeters distance. This is compensation distance for fine tuning.
3097 Set cm distance. This is compensation distance for tightening distance setup.
3101 Set meters distance. This is compensation distance for hard distance setup.
3105 Set dry amount. Amount of unprocessed (dry) signal.
3109 Set wet amount. Amount of processed (wet) signal.
3113 Set temperature in degrees Celsius. This is the temperature of the environment.
3118 Apply headphone crossfeed filter.
3120 Crossfeed is the process of blending the left and right channels of stereo
3122 It is mainly used to reduce extreme stereo separation of low frequencies.
3124 The intent is to produce more speaker like sound to the listener.
3126 The filter accepts the following options:
3130 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3131 This sets gain of low shelf filter for side part of stereo image.
3132 Default is -6dB. Max allowed is -30db when strength is set to 1.
3135 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3136 This sets cut off frequency of low shelf filter. Default is cut off near
3137 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3140 Set input gain. Default is 0.9.
3143 Set output gain. Default is 1.
3146 @section crystalizer
3147 Simple algorithm to expand audio dynamic range.
3149 The filter accepts the following options:
3153 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3154 (unchanged sound) to 10.0 (maximum effect).
3157 Enable clipping. By default is enabled.
3161 Apply a DC shift to the audio.
3163 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3164 in the recording chain) from the audio. The effect of a DC offset is reduced
3165 headroom and hence volume. The @ref{astats} filter can be used to determine if
3166 a signal has a DC offset.
3170 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3174 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3175 used to prevent clipping.
3180 Apply de-essing to the audio samples.
3184 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3188 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3192 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3196 Set the output mode.
3198 It accepts the following values:
3201 Pass input unchanged.
3204 Pass ess filtered out.
3209 Default value is @var{o}.
3215 Measure audio dynamic range.
3217 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3218 is found in transition material. And anything less that 8 have very poor dynamics
3219 and is very compressed.
3221 The filter accepts the following options:
3225 Set window length in seconds used to split audio into segments of equal length.
3226 Default is 3 seconds.
3230 Dynamic Audio Normalizer.
3232 This filter applies a certain amount of gain to the input audio in order
3233 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3234 contrast to more "simple" normalization algorithms, the Dynamic Audio
3235 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3236 This allows for applying extra gain to the "quiet" sections of the audio
3237 while avoiding distortions or clipping the "loud" sections. In other words:
3238 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3239 sections, in the sense that the volume of each section is brought to the
3240 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3241 this goal *without* applying "dynamic range compressing". It will retain 100%
3242 of the dynamic range *within* each section of the audio file.
3246 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3247 Default is 500 milliseconds.
3248 The Dynamic Audio Normalizer processes the input audio in small chunks,
3249 referred to as frames. This is required, because a peak magnitude has no
3250 meaning for just a single sample value. Instead, we need to determine the
3251 peak magnitude for a contiguous sequence of sample values. While a "standard"
3252 normalizer would simply use the peak magnitude of the complete file, the
3253 Dynamic Audio Normalizer determines the peak magnitude individually for each
3254 frame. The length of a frame is specified in milliseconds. By default, the
3255 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3256 been found to give good results with most files.
3257 Note that the exact frame length, in number of samples, will be determined
3258 automatically, based on the sampling rate of the individual input audio file.
3261 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3262 number. Default is 31.
3263 Probably the most important parameter of the Dynamic Audio Normalizer is the
3264 @code{window size} of the Gaussian smoothing filter. The filter's window size
3265 is specified in frames, centered around the current frame. For the sake of
3266 simplicity, this must be an odd number. Consequently, the default value of 31
3267 takes into account the current frame, as well as the 15 preceding frames and
3268 the 15 subsequent frames. Using a larger window results in a stronger
3269 smoothing effect and thus in less gain variation, i.e. slower gain
3270 adaptation. Conversely, using a smaller window results in a weaker smoothing
3271 effect and thus in more gain variation, i.e. faster gain adaptation.
3272 In other words, the more you increase this value, the more the Dynamic Audio
3273 Normalizer will behave like a "traditional" normalization filter. On the
3274 contrary, the more you decrease this value, the more the Dynamic Audio
3275 Normalizer will behave like a dynamic range compressor.
3278 Set the target peak value. This specifies the highest permissible magnitude
3279 level for the normalized audio input. This filter will try to approach the
3280 target peak magnitude as closely as possible, but at the same time it also
3281 makes sure that the normalized signal will never exceed the peak magnitude.
3282 A frame's maximum local gain factor is imposed directly by the target peak
3283 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3284 It is not recommended to go above this value.
3287 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3288 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3289 factor for each input frame, i.e. the maximum gain factor that does not
3290 result in clipping or distortion. The maximum gain factor is determined by
3291 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3292 additionally bounds the frame's maximum gain factor by a predetermined
3293 (global) maximum gain factor. This is done in order to avoid excessive gain
3294 factors in "silent" or almost silent frames. By default, the maximum gain
3295 factor is 10.0, For most inputs the default value should be sufficient and
3296 it usually is not recommended to increase this value. Though, for input
3297 with an extremely low overall volume level, it may be necessary to allow even
3298 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3299 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3300 Instead, a "sigmoid" threshold function will be applied. This way, the
3301 gain factors will smoothly approach the threshold value, but never exceed that
3305 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3306 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3307 This means that the maximum local gain factor for each frame is defined
3308 (only) by the frame's highest magnitude sample. This way, the samples can
3309 be amplified as much as possible without exceeding the maximum signal
3310 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3311 Normalizer can also take into account the frame's root mean square,
3312 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3313 determine the power of a time-varying signal. It is therefore considered
3314 that the RMS is a better approximation of the "perceived loudness" than
3315 just looking at the signal's peak magnitude. Consequently, by adjusting all
3316 frames to a constant RMS value, a uniform "perceived loudness" can be
3317 established. If a target RMS value has been specified, a frame's local gain
3318 factor is defined as the factor that would result in exactly that RMS value.
3319 Note, however, that the maximum local gain factor is still restricted by the
3320 frame's highest magnitude sample, in order to prevent clipping.
3323 Enable channels coupling. By default is enabled.
3324 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3325 amount. This means the same gain factor will be applied to all channels, i.e.
3326 the maximum possible gain factor is determined by the "loudest" channel.
3327 However, in some recordings, it may happen that the volume of the different
3328 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3329 In this case, this option can be used to disable the channel coupling. This way,
3330 the gain factor will be determined independently for each channel, depending
3331 only on the individual channel's highest magnitude sample. This allows for
3332 harmonizing the volume of the different channels.
3335 Enable DC bias correction. By default is disabled.
3336 An audio signal (in the time domain) is a sequence of sample values.
3337 In the Dynamic Audio Normalizer these sample values are represented in the
3338 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3339 audio signal, or "waveform", should be centered around the zero point.
3340 That means if we calculate the mean value of all samples in a file, or in a
3341 single frame, then the result should be 0.0 or at least very close to that
3342 value. If, however, there is a significant deviation of the mean value from
3343 0.0, in either positive or negative direction, this is referred to as a
3344 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3345 Audio Normalizer provides optional DC bias correction.
3346 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3347 the mean value, or "DC correction" offset, of each input frame and subtract
3348 that value from all of the frame's sample values which ensures those samples
3349 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3350 boundaries, the DC correction offset values will be interpolated smoothly
3351 between neighbouring frames.
3353 @item altboundary, b
3354 Enable alternative boundary mode. By default is disabled.
3355 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3356 around each frame. This includes the preceding frames as well as the
3357 subsequent frames. However, for the "boundary" frames, located at the very
3358 beginning and at the very end of the audio file, not all neighbouring
3359 frames are available. In particular, for the first few frames in the audio
3360 file, the preceding frames are not known. And, similarly, for the last few
3361 frames in the audio file, the subsequent frames are not known. Thus, the
3362 question arises which gain factors should be assumed for the missing frames
3363 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3364 to deal with this situation. The default boundary mode assumes a gain factor
3365 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3366 "fade out" at the beginning and at the end of the input, respectively.
3369 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3370 By default, the Dynamic Audio Normalizer does not apply "traditional"
3371 compression. This means that signal peaks will not be pruned and thus the
3372 full dynamic range will be retained within each local neighbourhood. However,
3373 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3374 normalization algorithm with a more "traditional" compression.
3375 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3376 (thresholding) function. If (and only if) the compression feature is enabled,
3377 all input frames will be processed by a soft knee thresholding function prior
3378 to the actual normalization process. Put simply, the thresholding function is
3379 going to prune all samples whose magnitude exceeds a certain threshold value.
3380 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3381 value. Instead, the threshold value will be adjusted for each individual
3383 In general, smaller parameters result in stronger compression, and vice versa.
3384 Values below 3.0 are not recommended, because audible distortion may appear.
3389 Make audio easier to listen to on headphones.
3391 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3392 so that when listened to on headphones the stereo image is moved from
3393 inside your head (standard for headphones) to outside and in front of
3394 the listener (standard for speakers).
3400 Apply a two-pole peaking equalisation (EQ) filter. With this
3401 filter, the signal-level at and around a selected frequency can
3402 be increased or decreased, whilst (unlike bandpass and bandreject
3403 filters) that at all other frequencies is unchanged.
3405 In order to produce complex equalisation curves, this filter can
3406 be given several times, each with a different central frequency.
3408 The filter accepts the following options:
3412 Set the filter's central frequency in Hz.
3415 Set method to specify band-width of filter.
3430 Specify the band-width of a filter in width_type units.
3433 Set the required gain or attenuation in dB.
3434 Beware of clipping when using a positive gain.
3437 How much to use filtered signal in output. Default is 1.
3438 Range is between 0 and 1.
3441 Specify which channels to filter, by default all available are filtered.
3444 @subsection Examples
3447 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3449 equalizer=f=1000:t=h:width=200:g=-10
3453 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3455 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3459 @subsection Commands
3461 This filter supports the following commands:
3464 Change equalizer frequency.
3465 Syntax for the command is : "@var{frequency}"
3468 Change equalizer width_type.
3469 Syntax for the command is : "@var{width_type}"
3472 Change equalizer width.
3473 Syntax for the command is : "@var{width}"
3476 Change equalizer gain.
3477 Syntax for the command is : "@var{gain}"
3480 Change equalizer mix.
3481 Syntax for the command is : "@var{mix}"
3484 @section extrastereo
3486 Linearly increases the difference between left and right channels which
3487 adds some sort of "live" effect to playback.
3489 The filter accepts the following options:
3493 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3494 (average of both channels), with 1.0 sound will be unchanged, with
3495 -1.0 left and right channels will be swapped.
3498 Enable clipping. By default is enabled.
3501 @section firequalizer
3502 Apply FIR Equalization using arbitrary frequency response.
3504 The filter accepts the following option:
3508 Set gain curve equation (in dB). The expression can contain variables:
3511 the evaluated frequency
3515 channel number, set to 0 when multichannels evaluation is disabled
3517 channel id, see libavutil/channel_layout.h, set to the first channel id when
3518 multichannels evaluation is disabled
3522 channel_layout, see libavutil/channel_layout.h
3527 @item gain_interpolate(f)
3528 interpolate gain on frequency f based on gain_entry
3529 @item cubic_interpolate(f)
3530 same as gain_interpolate, but smoother
3532 This option is also available as command. Default is @code{gain_interpolate(f)}.
3535 Set gain entry for gain_interpolate function. The expression can
3539 store gain entry at frequency f with value g
3541 This option is also available as command.
3544 Set filter delay in seconds. Higher value means more accurate.
3545 Default is @code{0.01}.
3548 Set filter accuracy in Hz. Lower value means more accurate.
3549 Default is @code{5}.
3552 Set window function. Acceptable values are:
3555 rectangular window, useful when gain curve is already smooth
3557 hann window (default)
3563 3-terms continuous 1st derivative nuttall window
3565 minimum 3-terms discontinuous nuttall window
3567 4-terms continuous 1st derivative nuttall window
3569 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3571 blackman-harris window
3577 If enabled, use fixed number of audio samples. This improves speed when
3578 filtering with large delay. Default is disabled.
3581 Enable multichannels evaluation on gain. Default is disabled.
3584 Enable zero phase mode by subtracting timestamp to compensate delay.
3585 Default is disabled.
3588 Set scale used by gain. Acceptable values are:
3591 linear frequency, linear gain
3593 linear frequency, logarithmic (in dB) gain (default)
3595 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3597 logarithmic frequency, logarithmic gain
3601 Set file for dumping, suitable for gnuplot.
3604 Set scale for dumpfile. Acceptable values are same with scale option.
3608 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3609 Default is disabled.
3612 Enable minimum phase impulse response. Default is disabled.
3615 @subsection Examples
3620 firequalizer=gain='if(lt(f,1000), 0, -INF)'
3623 lowpass at 1000 Hz with gain_entry:
3625 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
3628 custom equalization:
3630 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
3633 higher delay with zero phase to compensate delay:
3635 firequalizer=delay=0.1:fixed=on:zero_phase=on
3638 lowpass on left channel, highpass on right channel:
3640 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
3641 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
3646 Apply a flanging effect to the audio.
3648 The filter accepts the following options:
3652 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
3655 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
3658 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
3662 Set percentage of delayed signal mixed with original. Range from 0 to 100.
3663 Default value is 71.
3666 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
3669 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
3670 Default value is @var{sinusoidal}.
3673 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
3674 Default value is 25.
3677 Set delay-line interpolation, @var{linear} or @var{quadratic}.
3678 Default is @var{linear}.
3682 Apply Haas effect to audio.
3684 Note that this makes most sense to apply on mono signals.
3685 With this filter applied to mono signals it give some directionality and
3686 stretches its stereo image.
3688 The filter accepts the following options:
3692 Set input level. By default is @var{1}, or 0dB
3695 Set output level. By default is @var{1}, or 0dB.
3698 Set gain applied to side part of signal. By default is @var{1}.
3701 Set kind of middle source. Can be one of the following:
3711 Pick middle part signal of stereo image.
3714 Pick side part signal of stereo image.
3718 Change middle phase. By default is disabled.
3721 Set left channel delay. By default is @var{2.05} milliseconds.
3724 Set left channel balance. By default is @var{-1}.
3727 Set left channel gain. By default is @var{1}.
3730 Change left phase. By default is disabled.
3733 Set right channel delay. By defaults is @var{2.12} milliseconds.
3736 Set right channel balance. By default is @var{1}.
3739 Set right channel gain. By default is @var{1}.
3742 Change right phase. By default is enabled.
3747 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
3748 embedded HDCD codes is expanded into a 20-bit PCM stream.
3750 The filter supports the Peak Extend and Low-level Gain Adjustment features
3751 of HDCD, and detects the Transient Filter flag.
3754 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
3757 When using the filter with wav, note the default encoding for wav is 16-bit,
3758 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
3759 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
3761 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
3762 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
3765 The filter accepts the following options:
3768 @item disable_autoconvert
3769 Disable any automatic format conversion or resampling in the filter graph.
3771 @item process_stereo
3772 Process the stereo channels together. If target_gain does not match between
3773 channels, consider it invalid and use the last valid target_gain.
3776 Set the code detect timer period in ms.
3779 Always extend peaks above -3dBFS even if PE isn't signaled.
3782 Replace audio with a solid tone and adjust the amplitude to signal some
3783 specific aspect of the decoding process. The output file can be loaded in
3784 an audio editor alongside the original to aid analysis.
3786 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
3793 Gain adjustment level at each sample
3795 Samples where peak extend occurs
3797 Samples where the code detect timer is active
3799 Samples where the target gain does not match between channels
3805 Apply head-related transfer functions (HRTFs) to create virtual
3806 loudspeakers around the user for binaural listening via headphones.
3807 The HRIRs are provided via additional streams, for each channel
3808 one stereo input stream is needed.
3810 The filter accepts the following options:
3814 Set mapping of input streams for convolution.
3815 The argument is a '|'-separated list of channel names in order as they
3816 are given as additional stream inputs for filter.
3817 This also specify number of input streams. Number of input streams
3818 must be not less than number of channels in first stream plus one.
3821 Set gain applied to audio. Value is in dB. Default is 0.
3824 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
3825 processing audio in time domain which is slow.
3826 @var{freq} is processing audio in frequency domain which is fast.
3827 Default is @var{freq}.
3830 Set custom gain for LFE channels. Value is in dB. Default is 0.
3833 Set size of frame in number of samples which will be processed at once.
3834 Default value is @var{1024}. Allowed range is from 1024 to 96000.
3837 Set format of hrir stream.
3838 Default value is @var{stereo}. Alternative value is @var{multich}.
3839 If value is set to @var{stereo}, number of additional streams should
3840 be greater or equal to number of input channels in first input stream.
3841 Also each additional stream should have stereo number of channels.
3842 If value is set to @var{multich}, number of additional streams should
3843 be exactly one. Also number of input channels of additional stream
3844 should be equal or greater than twice number of channels of first input
3848 @subsection Examples
3852 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3853 each amovie filter use stereo file with IR coefficients as input.
3854 The files give coefficients for each position of virtual loudspeaker:
3857 -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"
3862 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3863 but now in @var{multich} @var{hrir} format.
3865 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"
3872 Apply a high-pass filter with 3dB point frequency.
3873 The filter can be either single-pole, or double-pole (the default).
3874 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
3876 The filter accepts the following options:
3880 Set frequency in Hz. Default is 3000.
3883 Set number of poles. Default is 2.
3886 Set method to specify band-width of filter.
3901 Specify the band-width of a filter in width_type units.
3902 Applies only to double-pole filter.
3903 The default is 0.707q and gives a Butterworth response.
3906 How much to use filtered signal in output. Default is 1.
3907 Range is between 0 and 1.
3910 Specify which channels to filter, by default all available are filtered.
3913 @subsection Commands
3915 This filter supports the following commands:
3918 Change highpass frequency.
3919 Syntax for the command is : "@var{frequency}"
3922 Change highpass width_type.
3923 Syntax for the command is : "@var{width_type}"
3926 Change highpass width.
3927 Syntax for the command is : "@var{width}"
3930 Change highpass mix.
3931 Syntax for the command is : "@var{mix}"
3936 Join multiple input streams into one multi-channel stream.
3938 It accepts the following parameters:
3942 The number of input streams. It defaults to 2.
3944 @item channel_layout
3945 The desired output channel layout. It defaults to stereo.
3948 Map channels from inputs to output. The argument is a '|'-separated list of
3949 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
3950 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
3951 can be either the name of the input channel (e.g. FL for front left) or its
3952 index in the specified input stream. @var{out_channel} is the name of the output
3956 The filter will attempt to guess the mappings when they are not specified
3957 explicitly. It does so by first trying to find an unused matching input channel
3958 and if that fails it picks the first unused input channel.
3960 Join 3 inputs (with properly set channel layouts):
3962 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
3965 Build a 5.1 output from 6 single-channel streams:
3967 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
3968 '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'
3974 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
3976 To enable compilation of this filter you need to configure FFmpeg with
3977 @code{--enable-ladspa}.
3981 Specifies the name of LADSPA plugin library to load. If the environment
3982 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
3983 each one of the directories specified by the colon separated list in
3984 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
3985 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
3986 @file{/usr/lib/ladspa/}.
3989 Specifies the plugin within the library. Some libraries contain only
3990 one plugin, but others contain many of them. If this is not set filter
3991 will list all available plugins within the specified library.
3994 Set the '|' separated list of controls which are zero or more floating point
3995 values that determine the behavior of the loaded plugin (for example delay,
3997 Controls need to be defined using the following syntax:
3998 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
3999 @var{valuei} is the value set on the @var{i}-th control.
4000 Alternatively they can be also defined using the following syntax:
4001 @var{value0}|@var{value1}|@var{value2}|..., where
4002 @var{valuei} is the value set on the @var{i}-th control.
4003 If @option{controls} is set to @code{help}, all available controls and
4004 their valid ranges are printed.
4006 @item sample_rate, s
4007 Specify the sample rate, default to 44100. Only used if plugin have
4011 Set the number of samples per channel per each output frame, default
4012 is 1024. Only used if plugin have zero inputs.
4015 Set the minimum duration of the sourced audio. See
4016 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4017 for the accepted syntax.
4018 Note that the resulting duration may be greater than the specified duration,
4019 as the generated audio is always cut at the end of a complete frame.
4020 If not specified, or the expressed duration is negative, the audio is
4021 supposed to be generated forever.
4022 Only used if plugin have zero inputs.
4026 @subsection Examples
4030 List all available plugins within amp (LADSPA example plugin) library:
4036 List all available controls and their valid ranges for @code{vcf_notch}
4037 plugin from @code{VCF} library:
4039 ladspa=f=vcf:p=vcf_notch:c=help
4043 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4046 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4050 Add reverberation to the audio using TAP-plugins
4051 (Tom's Audio Processing plugins):
4053 ladspa=file=tap_reverb:tap_reverb
4057 Generate white noise, with 0.2 amplitude:
4059 ladspa=file=cmt:noise_source_white:c=c0=.2
4063 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4064 @code{C* Audio Plugin Suite} (CAPS) library:
4066 ladspa=file=caps:Click:c=c1=20'
4070 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4072 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4076 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4077 @code{SWH Plugins} collection:
4079 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4083 Attenuate low frequencies using Multiband EQ from Steve Harris
4084 @code{SWH Plugins} collection:
4086 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4090 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4093 ladspa=caps:Narrower
4097 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4099 ladspa=caps:White:.2
4103 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4105 ladspa=caps:Fractal:c=c1=1
4109 Dynamic volume normalization using @code{VLevel} plugin:
4111 ladspa=vlevel-ladspa:vlevel_mono
4115 @subsection Commands
4117 This filter supports the following commands:
4120 Modify the @var{N}-th control value.
4122 If the specified value is not valid, it is ignored and prior one is kept.
4127 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4128 Support for both single pass (livestreams, files) and double pass (files) modes.
4129 This algorithm can target IL, LRA, and maximum true peak. To accurately detect true peaks,
4130 the audio stream will be upsampled to 192 kHz unless the normalization mode is linear.
4131 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4133 The filter accepts the following options:
4137 Set integrated loudness target.
4138 Range is -70.0 - -5.0. Default value is -24.0.
4141 Set loudness range target.
4142 Range is 1.0 - 20.0. Default value is 7.0.
4145 Set maximum true peak.
4146 Range is -9.0 - +0.0. Default value is -2.0.
4148 @item measured_I, measured_i
4149 Measured IL of input file.
4150 Range is -99.0 - +0.0.
4152 @item measured_LRA, measured_lra
4153 Measured LRA of input file.
4154 Range is 0.0 - 99.0.
4156 @item measured_TP, measured_tp
4157 Measured true peak of input file.
4158 Range is -99.0 - +99.0.
4160 @item measured_thresh
4161 Measured threshold of input file.
4162 Range is -99.0 - +0.0.
4165 Set offset gain. Gain is applied before the true-peak limiter.
4166 Range is -99.0 - +99.0. Default is +0.0.
4169 Normalize linearly if possible.
4170 measured_I, measured_LRA, measured_TP, and measured_thresh must also
4171 to be specified in order to use this mode.
4172 Options are true or false. Default is true.
4175 Treat mono input files as "dual-mono". If a mono file is intended for playback
4176 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4177 If set to @code{true}, this option will compensate for this effect.
4178 Multi-channel input files are not affected by this option.
4179 Options are true or false. Default is false.
4182 Set print format for stats. Options are summary, json, or none.
4183 Default value is none.
4188 Apply a low-pass filter with 3dB point frequency.
4189 The filter can be either single-pole or double-pole (the default).
4190 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4192 The filter accepts the following options:
4196 Set frequency in Hz. Default is 500.
4199 Set number of poles. Default is 2.
4202 Set method to specify band-width of filter.
4217 Specify the band-width of a filter in width_type units.
4218 Applies only to double-pole filter.
4219 The default is 0.707q and gives a Butterworth response.
4222 How much to use filtered signal in output. Default is 1.
4223 Range is between 0 and 1.
4226 Specify which channels to filter, by default all available are filtered.
4229 @subsection Examples
4232 Lowpass only LFE channel, it LFE is not present it does nothing:
4238 @subsection Commands
4240 This filter supports the following commands:
4243 Change lowpass frequency.
4244 Syntax for the command is : "@var{frequency}"
4247 Change lowpass width_type.
4248 Syntax for the command is : "@var{width_type}"
4251 Change lowpass width.
4252 Syntax for the command is : "@var{width}"
4256 Syntax for the command is : "@var{mix}"
4261 Load a LV2 (LADSPA Version 2) plugin.
4263 To enable compilation of this filter you need to configure FFmpeg with
4264 @code{--enable-lv2}.
4268 Specifies the plugin URI. You may need to escape ':'.
4271 Set the '|' separated list of controls which are zero or more floating point
4272 values that determine the behavior of the loaded plugin (for example delay,
4274 If @option{controls} is set to @code{help}, all available controls and
4275 their valid ranges are printed.
4277 @item sample_rate, s
4278 Specify the sample rate, default to 44100. Only used if plugin have
4282 Set the number of samples per channel per each output frame, default
4283 is 1024. Only used if plugin have zero inputs.
4286 Set the minimum duration of the sourced audio. See
4287 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4288 for the accepted syntax.
4289 Note that the resulting duration may be greater than the specified duration,
4290 as the generated audio is always cut at the end of a complete frame.
4291 If not specified, or the expressed duration is negative, the audio is
4292 supposed to be generated forever.
4293 Only used if plugin have zero inputs.
4296 @subsection Examples
4300 Apply bass enhancer plugin from Calf:
4302 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4306 Apply vinyl plugin from Calf:
4308 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4312 Apply bit crusher plugin from ArtyFX:
4314 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4319 Multiband Compress or expand the audio's dynamic range.
4321 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4322 This is akin to the crossover of a loudspeaker, and results in flat frequency
4323 response when absent compander action.
4325 It accepts the following parameters:
4329 This option syntax is:
4330 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4331 For explanation of each item refer to compand filter documentation.
4337 Mix channels with specific gain levels. The filter accepts the output
4338 channel layout followed by a set of channels definitions.
4340 This filter is also designed to efficiently remap the channels of an audio
4343 The filter accepts parameters of the form:
4344 "@var{l}|@var{outdef}|@var{outdef}|..."
4348 output channel layout or number of channels
4351 output channel specification, of the form:
4352 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4355 output channel to define, either a channel name (FL, FR, etc.) or a channel
4356 number (c0, c1, etc.)
4359 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4362 input channel to use, see out_name for details; it is not possible to mix
4363 named and numbered input channels
4366 If the `=' in a channel specification is replaced by `<', then the gains for
4367 that specification will be renormalized so that the total is 1, thus
4368 avoiding clipping noise.
4370 @subsection Mixing examples
4372 For example, if you want to down-mix from stereo to mono, but with a bigger
4373 factor for the left channel:
4375 pan=1c|c0=0.9*c0+0.1*c1
4378 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4379 7-channels surround:
4381 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4384 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4385 that should be preferred (see "-ac" option) unless you have very specific
4388 @subsection Remapping examples
4390 The channel remapping will be effective if, and only if:
4393 @item gain coefficients are zeroes or ones,
4394 @item only one input per channel output,
4397 If all these conditions are satisfied, the filter will notify the user ("Pure
4398 channel mapping detected"), and use an optimized and lossless method to do the
4401 For example, if you have a 5.1 source and want a stereo audio stream by
4402 dropping the extra channels:
4404 pan="stereo| c0=FL | c1=FR"
4407 Given the same source, you can also switch front left and front right channels
4408 and keep the input channel layout:
4410 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4413 If the input is a stereo audio stream, you can mute the front left channel (and
4414 still keep the stereo channel layout) with:
4419 Still with a stereo audio stream input, you can copy the right channel in both
4420 front left and right:
4422 pan="stereo| c0=FR | c1=FR"
4427 ReplayGain scanner filter. This filter takes an audio stream as an input and
4428 outputs it unchanged.
4429 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4433 Convert the audio sample format, sample rate and channel layout. It is
4434 not meant to be used directly.
4437 Apply time-stretching and pitch-shifting with librubberband.
4439 To enable compilation of this filter, you need to configure FFmpeg with
4440 @code{--enable-librubberband}.
4442 The filter accepts the following options:
4446 Set tempo scale factor.
4449 Set pitch scale factor.
4452 Set transients detector.
4453 Possible values are:
4462 Possible values are:
4471 Possible values are:
4478 Set processing window size.
4479 Possible values are:
4488 Possible values are:
4495 Enable formant preservation when shift pitching.
4496 Possible values are:
4504 Possible values are:
4513 Possible values are:
4520 @subsection Commands
4522 This filter supports the following commands:
4525 Change filter tempo scale factor.
4526 Syntax for the command is : "@var{tempo}"
4529 Change filter pitch scale factor.
4530 Syntax for the command is : "@var{pitch}"
4533 @section sidechaincompress
4535 This filter acts like normal compressor but has the ability to compress
4536 detected signal using second input signal.
4537 It needs two input streams and returns one output stream.
4538 First input stream will be processed depending on second stream signal.
4539 The filtered signal then can be filtered with other filters in later stages of
4540 processing. See @ref{pan} and @ref{amerge} filter.
4542 The filter accepts the following options:
4546 Set input gain. Default is 1. Range is between 0.015625 and 64.
4549 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4550 Default is @code{downward}.
4553 If a signal of second stream raises above this level it will affect the gain
4554 reduction of first stream.
4555 By default is 0.125. Range is between 0.00097563 and 1.
4558 Set a ratio about which the signal is reduced. 1:2 means that if the level
4559 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4560 Default is 2. Range is between 1 and 20.
4563 Amount of milliseconds the signal has to rise above the threshold before gain
4564 reduction starts. Default is 20. Range is between 0.01 and 2000.
4567 Amount of milliseconds the signal has to fall below the threshold before
4568 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4571 Set the amount by how much signal will be amplified after processing.
4572 Default is 1. Range is from 1 to 64.
4575 Curve the sharp knee around the threshold to enter gain reduction more softly.
4576 Default is 2.82843. Range is between 1 and 8.
4579 Choose if the @code{average} level between all channels of side-chain stream
4580 or the louder(@code{maximum}) channel of side-chain stream affects the
4581 reduction. Default is @code{average}.
4584 Should the exact signal be taken in case of @code{peak} or an RMS one in case
4585 of @code{rms}. Default is @code{rms} which is mainly smoother.
4588 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
4591 How much to use compressed signal in output. Default is 1.
4592 Range is between 0 and 1.
4595 @subsection Examples
4599 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
4600 depending on the signal of 2nd input and later compressed signal to be
4601 merged with 2nd input:
4603 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
4607 @section sidechaingate
4609 A sidechain gate acts like a normal (wideband) gate but has the ability to
4610 filter the detected signal before sending it to the gain reduction stage.
4611 Normally a gate uses the full range signal to detect a level above the
4613 For example: If you cut all lower frequencies from your sidechain signal
4614 the gate will decrease the volume of your track only if not enough highs
4615 appear. With this technique you are able to reduce the resonation of a
4616 natural drum or remove "rumbling" of muted strokes from a heavily distorted
4618 It needs two input streams and returns one output stream.
4619 First input stream will be processed depending on second stream signal.
4621 The filter accepts the following options:
4625 Set input level before filtering.
4626 Default is 1. Allowed range is from 0.015625 to 64.
4629 Set the mode of operation. Can be @code{upward} or @code{downward}.
4630 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
4631 will be amplified, expanding dynamic range in upward direction.
4632 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
4635 Set the level of gain reduction when the signal is below the threshold.
4636 Default is 0.06125. Allowed range is from 0 to 1.
4637 Setting this to 0 disables reduction and then filter behaves like expander.
4640 If a signal rises above this level the gain reduction is released.
4641 Default is 0.125. Allowed range is from 0 to 1.
4644 Set a ratio about which the signal is reduced.
4645 Default is 2. Allowed range is from 1 to 9000.
4648 Amount of milliseconds the signal has to rise above the threshold before gain
4650 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
4653 Amount of milliseconds the signal has to fall below the threshold before the
4654 reduction is increased again. Default is 250 milliseconds.
4655 Allowed range is from 0.01 to 9000.
4658 Set amount of amplification of signal after processing.
4659 Default is 1. Allowed range is from 1 to 64.
4662 Curve the sharp knee around the threshold to enter gain reduction more softly.
4663 Default is 2.828427125. Allowed range is from 1 to 8.
4666 Choose if exact signal should be taken for detection or an RMS like one.
4667 Default is rms. Can be peak or rms.
4670 Choose if the average level between all channels or the louder channel affects
4672 Default is average. Can be average or maximum.
4675 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
4678 @section silencedetect
4680 Detect silence in an audio stream.
4682 This filter logs a message when it detects that the input audio volume is less
4683 or equal to a noise tolerance value for a duration greater or equal to the
4684 minimum detected noise duration.
4686 The printed times and duration are expressed in seconds. The
4687 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
4688 is set on the first frame whose timestamp equals or exceeds the detection
4689 duration and it contains the timestamp of the first frame of the silence.
4691 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
4692 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
4693 keys are set on the first frame after the silence. If @option{mono} is
4694 enabled, and each channel is evaluated separately, the @code{.X}
4695 suffixed keys are used, and @code{X} corresponds to the channel number.
4697 The filter accepts the following options:
4701 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
4702 specified value) or amplitude ratio. Default is -60dB, or 0.001.
4705 Set silence duration until notification (default is 2 seconds). See
4706 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4707 for the accepted syntax.
4710 Process each channel separately, instead of combined. By default is disabled.
4713 @subsection Examples
4717 Detect 5 seconds of silence with -50dB noise tolerance:
4719 silencedetect=n=-50dB:d=5
4723 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
4724 tolerance in @file{silence.mp3}:
4726 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
4730 @section silenceremove
4732 Remove silence from the beginning, middle or end of the audio.
4734 The filter accepts the following options:
4738 This value is used to indicate if audio should be trimmed at beginning of
4739 the audio. A value of zero indicates no silence should be trimmed from the
4740 beginning. When specifying a non-zero value, it trims audio up until it
4741 finds non-silence. Normally, when trimming silence from beginning of audio
4742 the @var{start_periods} will be @code{1} but it can be increased to higher
4743 values to trim all audio up to specific count of non-silence periods.
4744 Default value is @code{0}.
4746 @item start_duration
4747 Specify the amount of time that non-silence must be detected before it stops
4748 trimming audio. By increasing the duration, bursts of noises can be treated
4749 as silence and trimmed off. Default value is @code{0}.
4751 @item start_threshold
4752 This indicates what sample value should be treated as silence. For digital
4753 audio, a value of @code{0} may be fine but for audio recorded from analog,
4754 you may wish to increase the value to account for background noise.
4755 Can be specified in dB (in case "dB" is appended to the specified value)
4756 or amplitude ratio. Default value is @code{0}.
4759 Specify max duration of silence at beginning that will be kept after
4760 trimming. Default is 0, which is equal to trimming all samples detected
4764 Specify mode of detection of silence end in start of multi-channel audio.
4765 Can be @var{any} or @var{all}. Default is @var{any}.
4766 With @var{any}, any sample that is detected as non-silence will cause
4767 stopped trimming of silence.
4768 With @var{all}, only if all channels are detected as non-silence will cause
4769 stopped trimming of silence.
4772 Set the count for trimming silence from the end of audio.
4773 To remove silence from the middle of a file, specify a @var{stop_periods}
4774 that is negative. This value is then treated as a positive value and is
4775 used to indicate the effect should restart processing as specified by
4776 @var{start_periods}, making it suitable for removing periods of silence
4777 in the middle of the audio.
4778 Default value is @code{0}.
4781 Specify a duration of silence that must exist before audio is not copied any
4782 more. By specifying a higher duration, silence that is wanted can be left in
4784 Default value is @code{0}.
4786 @item stop_threshold
4787 This is the same as @option{start_threshold} but for trimming silence from
4789 Can be specified in dB (in case "dB" is appended to the specified value)
4790 or amplitude ratio. Default value is @code{0}.
4793 Specify max duration of silence at end that will be kept after
4794 trimming. Default is 0, which is equal to trimming all samples detected
4798 Specify mode of detection of silence start in end of multi-channel audio.
4799 Can be @var{any} or @var{all}. Default is @var{any}.
4800 With @var{any}, any sample that is detected as non-silence will cause
4801 stopped trimming of silence.
4802 With @var{all}, only if all channels are detected as non-silence will cause
4803 stopped trimming of silence.
4806 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
4807 and works better with digital silence which is exactly 0.
4808 Default value is @code{rms}.
4811 Set duration in number of seconds used to calculate size of window in number
4812 of samples for detecting silence.
4813 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
4816 @subsection Examples
4820 The following example shows how this filter can be used to start a recording
4821 that does not contain the delay at the start which usually occurs between
4822 pressing the record button and the start of the performance:
4824 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
4828 Trim all silence encountered from beginning to end where there is more than 1
4829 second of silence in audio:
4831 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
4835 Trim all digital silence samples, using peak detection, from beginning to end
4836 where there is more than 0 samples of digital silence in audio and digital
4837 silence is detected in all channels at same positions in stream:
4839 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
4845 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
4846 loudspeakers around the user for binaural listening via headphones (audio
4847 formats up to 9 channels supported).
4848 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
4849 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
4850 Austrian Academy of Sciences.
4852 To enable compilation of this filter you need to configure FFmpeg with
4853 @code{--enable-libmysofa}.
4855 The filter accepts the following options:
4859 Set the SOFA file used for rendering.
4862 Set gain applied to audio. Value is in dB. Default is 0.
4865 Set rotation of virtual loudspeakers in deg. Default is 0.
4868 Set elevation of virtual speakers in deg. Default is 0.
4871 Set distance in meters between loudspeakers and the listener with near-field
4872 HRTFs. Default is 1.
4875 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4876 processing audio in time domain which is slow.
4877 @var{freq} is processing audio in frequency domain which is fast.
4878 Default is @var{freq}.
4881 Set custom positions of virtual loudspeakers. Syntax for this option is:
4882 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
4883 Each virtual loudspeaker is described with short channel name following with
4884 azimuth and elevation in degrees.
4885 Each virtual loudspeaker description is separated by '|'.
4886 For example to override front left and front right channel positions use:
4887 'speakers=FL 45 15|FR 345 15'.
4888 Descriptions with unrecognised channel names are ignored.
4891 Set custom gain for LFE channels. Value is in dB. Default is 0.
4894 Set custom frame size in number of samples. Default is 1024.
4895 Allowed range is from 1024 to 96000. Only used if option @samp{type}
4896 is set to @var{freq}.
4899 Should all IRs be normalized upon importing SOFA file.
4900 By default is enabled.
4903 Should nearest IRs be interpolated with neighbor IRs if exact position
4904 does not match. By default is disabled.
4907 Minphase all IRs upon loading of SOFA file. By default is disabled.
4910 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
4913 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
4916 @subsection Examples
4920 Using ClubFritz6 sofa file:
4922 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
4926 Using ClubFritz12 sofa file and bigger radius with small rotation:
4928 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
4932 Similar as above but with custom speaker positions for front left, front right, back left and back right
4933 and also with custom gain:
4935 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
4939 @section stereotools
4941 This filter has some handy utilities to manage stereo signals, for converting
4942 M/S stereo recordings to L/R signal while having control over the parameters
4943 or spreading the stereo image of master track.
4945 The filter accepts the following options:
4949 Set input level before filtering for both channels. Defaults is 1.
4950 Allowed range is from 0.015625 to 64.
4953 Set output level after filtering for both channels. Defaults is 1.
4954 Allowed range is from 0.015625 to 64.
4957 Set input balance between both channels. Default is 0.
4958 Allowed range is from -1 to 1.
4961 Set output balance between both channels. Default is 0.
4962 Allowed range is from -1 to 1.
4965 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
4966 clipping. Disabled by default.
4969 Mute the left channel. Disabled by default.
4972 Mute the right channel. Disabled by default.
4975 Change the phase of the left channel. Disabled by default.
4978 Change the phase of the right channel. Disabled by default.
4981 Set stereo mode. Available values are:
4985 Left/Right to Left/Right, this is default.
4988 Left/Right to Mid/Side.
4991 Mid/Side to Left/Right.
4994 Left/Right to Left/Left.
4997 Left/Right to Right/Right.
5000 Left/Right to Left + Right.
5003 Left/Right to Right/Left.
5006 Mid/Side to Left/Left.
5009 Mid/Side to Right/Right.
5013 Set level of side signal. Default is 1.
5014 Allowed range is from 0.015625 to 64.
5017 Set balance of side signal. Default is 0.
5018 Allowed range is from -1 to 1.
5021 Set level of the middle signal. Default is 1.
5022 Allowed range is from 0.015625 to 64.
5025 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5028 Set stereo base between mono and inversed channels. Default is 0.
5029 Allowed range is from -1 to 1.
5032 Set delay in milliseconds how much to delay left from right channel and
5033 vice versa. Default is 0. Allowed range is from -20 to 20.
5036 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5039 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5041 @item bmode_in, bmode_out
5042 Set balance mode for balance_in/balance_out option.
5044 Can be one of the following:
5048 Classic balance mode. Attenuate one channel at time.
5049 Gain is raised up to 1.
5052 Similar as classic mode above but gain is raised up to 2.
5055 Equal power distribution, from -6dB to +6dB range.
5059 @subsection Examples
5063 Apply karaoke like effect:
5065 stereotools=mlev=0.015625
5069 Convert M/S signal to L/R:
5071 "stereotools=mode=ms>lr"
5075 @section stereowiden
5077 This filter enhance the stereo effect by suppressing signal common to both
5078 channels and by delaying the signal of left into right and vice versa,
5079 thereby widening the stereo effect.
5081 The filter accepts the following options:
5085 Time in milliseconds of the delay of left signal into right and vice versa.
5086 Default is 20 milliseconds.
5089 Amount of gain in delayed signal into right and vice versa. Gives a delay
5090 effect of left signal in right output and vice versa which gives widening
5091 effect. Default is 0.3.
5094 Cross feed of left into right with inverted phase. This helps in suppressing
5095 the mono. If the value is 1 it will cancel all the signal common to both
5096 channels. Default is 0.3.
5099 Set level of input signal of original channel. Default is 0.8.
5102 @section superequalizer
5103 Apply 18 band equalizer.
5105 The filter accepts the following options:
5112 Set 131Hz band gain.
5114 Set 185Hz band gain.
5116 Set 262Hz band gain.
5118 Set 370Hz band gain.
5120 Set 523Hz band gain.
5122 Set 740Hz band gain.
5124 Set 1047Hz band gain.
5126 Set 1480Hz band gain.
5128 Set 2093Hz band gain.
5130 Set 2960Hz band gain.
5132 Set 4186Hz band gain.
5134 Set 5920Hz band gain.
5136 Set 8372Hz band gain.
5138 Set 11840Hz band gain.
5140 Set 16744Hz band gain.
5142 Set 20000Hz band gain.
5146 Apply audio surround upmix filter.
5148 This filter allows to produce multichannel output from audio stream.
5150 The filter accepts the following options:
5154 Set output channel layout. By default, this is @var{5.1}.
5156 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5157 for the required syntax.
5160 Set input channel layout. By default, this is @var{stereo}.
5162 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5163 for the required syntax.
5166 Set input volume level. By default, this is @var{1}.
5169 Set output volume level. By default, this is @var{1}.
5172 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5175 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5178 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5181 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5182 In @var{add} mode, LFE channel is created from input audio and added to output.
5183 In @var{sub} mode, LFE channel is created from input audio and added to output but
5184 also all non-LFE output channels are subtracted with output LFE channel.
5187 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5188 Default is @var{90}.
5191 Set front center input volume. By default, this is @var{1}.
5194 Set front center output volume. By default, this is @var{1}.
5197 Set front left input volume. By default, this is @var{1}.
5200 Set front left output volume. By default, this is @var{1}.
5203 Set front right input volume. By default, this is @var{1}.
5206 Set front right output volume. By default, this is @var{1}.
5209 Set side left input volume. By default, this is @var{1}.
5212 Set side left output volume. By default, this is @var{1}.
5215 Set side right input volume. By default, this is @var{1}.
5218 Set side right output volume. By default, this is @var{1}.
5221 Set back left input volume. By default, this is @var{1}.
5224 Set back left output volume. By default, this is @var{1}.
5227 Set back right input volume. By default, this is @var{1}.
5230 Set back right output volume. By default, this is @var{1}.
5233 Set back center input volume. By default, this is @var{1}.
5236 Set back center output volume. By default, this is @var{1}.
5239 Set LFE input volume. By default, this is @var{1}.
5242 Set LFE output volume. By default, this is @var{1}.
5245 Set spread usage of stereo image across X axis for all channels.
5248 Set spread usage of stereo image across Y axis for all channels.
5250 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5251 Set spread usage of stereo image across X axis for each channel.
5253 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5254 Set spread usage of stereo image across Y axis for each channel.
5257 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5260 Set window function.
5262 It accepts the following values:
5285 Default is @code{hann}.
5288 Set window overlap. If set to 1, the recommended overlap for selected
5289 window function will be picked. Default is @code{0.5}.
5292 @section treble, highshelf
5294 Boost or cut treble (upper) frequencies of the audio using a two-pole
5295 shelving filter with a response similar to that of a standard
5296 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5298 The filter accepts the following options:
5302 Give the gain at whichever is the lower of ~22 kHz and the
5303 Nyquist frequency. Its useful range is about -20 (for a large cut)
5304 to +20 (for a large boost). Beware of clipping when using a positive gain.
5307 Set the filter's central frequency and so can be used
5308 to extend or reduce the frequency range to be boosted or cut.
5309 The default value is @code{3000} Hz.
5312 Set method to specify band-width of filter.
5327 Determine how steep is the filter's shelf transition.
5330 How much to use filtered signal in output. Default is 1.
5331 Range is between 0 and 1.
5334 Specify which channels to filter, by default all available are filtered.
5337 @subsection Commands
5339 This filter supports the following commands:
5342 Change treble frequency.
5343 Syntax for the command is : "@var{frequency}"
5346 Change treble width_type.
5347 Syntax for the command is : "@var{width_type}"
5350 Change treble width.
5351 Syntax for the command is : "@var{width}"
5355 Syntax for the command is : "@var{gain}"
5359 Syntax for the command is : "@var{mix}"
5364 Sinusoidal amplitude modulation.
5366 The filter accepts the following options:
5370 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5371 (20 Hz or lower) will result in a tremolo effect.
5372 This filter may also be used as a ring modulator by specifying
5373 a modulation frequency higher than 20 Hz.
5374 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5377 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5378 Default value is 0.5.
5383 Sinusoidal phase modulation.
5385 The filter accepts the following options:
5389 Modulation frequency in Hertz.
5390 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5393 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5394 Default value is 0.5.
5399 Adjust the input audio volume.
5401 It accepts the following parameters:
5405 Set audio volume expression.
5407 Output values are clipped to the maximum value.
5409 The output audio volume is given by the relation:
5411 @var{output_volume} = @var{volume} * @var{input_volume}
5414 The default value for @var{volume} is "1.0".
5417 This parameter represents the mathematical precision.
5419 It determines which input sample formats will be allowed, which affects the
5420 precision of the volume scaling.
5424 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5426 32-bit floating-point; this limits input sample format to FLT. (default)
5428 64-bit floating-point; this limits input sample format to DBL.
5432 Choose the behaviour on encountering ReplayGain side data in input frames.
5436 Remove ReplayGain side data, ignoring its contents (the default).
5439 Ignore ReplayGain side data, but leave it in the frame.
5442 Prefer the track gain, if present.
5445 Prefer the album gain, if present.
5448 @item replaygain_preamp
5449 Pre-amplification gain in dB to apply to the selected replaygain gain.
5451 Default value for @var{replaygain_preamp} is 0.0.
5454 Set when the volume expression is evaluated.
5456 It accepts the following values:
5459 only evaluate expression once during the filter initialization, or
5460 when the @samp{volume} command is sent
5463 evaluate expression for each incoming frame
5466 Default value is @samp{once}.
5469 The volume expression can contain the following parameters.
5473 frame number (starting at zero)
5476 @item nb_consumed_samples
5477 number of samples consumed by the filter
5479 number of samples in the current frame
5481 original frame position in the file
5487 PTS at start of stream
5489 time at start of stream
5495 last set volume value
5498 Note that when @option{eval} is set to @samp{once} only the
5499 @var{sample_rate} and @var{tb} variables are available, all other
5500 variables will evaluate to NAN.
5502 @subsection Commands
5504 This filter supports the following commands:
5507 Modify the volume expression.
5508 The command accepts the same syntax of the corresponding option.
5510 If the specified expression is not valid, it is kept at its current
5512 @item replaygain_noclip
5513 Prevent clipping by limiting the gain applied.
5515 Default value for @var{replaygain_noclip} is 1.
5519 @subsection Examples
5523 Halve the input audio volume:
5527 volume=volume=-6.0206dB
5530 In all the above example the named key for @option{volume} can be
5531 omitted, for example like in:
5537 Increase input audio power by 6 decibels using fixed-point precision:
5539 volume=volume=6dB:precision=fixed
5543 Fade volume after time 10 with an annihilation period of 5 seconds:
5545 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
5549 @section volumedetect
5551 Detect the volume of the input video.
5553 The filter has no parameters. The input is not modified. Statistics about
5554 the volume will be printed in the log when the input stream end is reached.
5556 In particular it will show the mean volume (root mean square), maximum
5557 volume (on a per-sample basis), and the beginning of a histogram of the
5558 registered volume values (from the maximum value to a cumulated 1/1000 of
5561 All volumes are in decibels relative to the maximum PCM value.
5563 @subsection Examples
5565 Here is an excerpt of the output:
5567 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
5568 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
5569 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
5570 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
5571 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
5572 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
5573 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
5574 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
5575 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
5581 The mean square energy is approximately -27 dB, or 10^-2.7.
5583 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
5585 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
5588 In other words, raising the volume by +4 dB does not cause any clipping,
5589 raising it by +5 dB causes clipping for 6 samples, etc.
5591 @c man end AUDIO FILTERS
5593 @chapter Audio Sources
5594 @c man begin AUDIO SOURCES
5596 Below is a description of the currently available audio sources.
5600 Buffer audio frames, and make them available to the filter chain.
5602 This source is mainly intended for a programmatic use, in particular
5603 through the interface defined in @file{libavfilter/asrc_abuffer.h}.
5605 It accepts the following parameters:
5609 The timebase which will be used for timestamps of submitted frames. It must be
5610 either a floating-point number or in @var{numerator}/@var{denominator} form.
5613 The sample rate of the incoming audio buffers.
5616 The sample format of the incoming audio buffers.
5617 Either a sample format name or its corresponding integer representation from
5618 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
5620 @item channel_layout
5621 The channel layout of the incoming audio buffers.
5622 Either a channel layout name from channel_layout_map in
5623 @file{libavutil/channel_layout.c} or its corresponding integer representation
5624 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
5627 The number of channels of the incoming audio buffers.
5628 If both @var{channels} and @var{channel_layout} are specified, then they
5633 @subsection Examples
5636 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
5639 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
5640 Since the sample format with name "s16p" corresponds to the number
5641 6 and the "stereo" channel layout corresponds to the value 0x3, this is
5644 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
5649 Generate an audio signal specified by an expression.
5651 This source accepts in input one or more expressions (one for each
5652 channel), which are evaluated and used to generate a corresponding
5655 This source accepts the following options:
5659 Set the '|'-separated expressions list for each separate channel. In case the
5660 @option{channel_layout} option is not specified, the selected channel layout
5661 depends on the number of provided expressions. Otherwise the last
5662 specified expression is applied to the remaining output channels.
5664 @item channel_layout, c
5665 Set the channel layout. The number of channels in the specified layout
5666 must be equal to the number of specified expressions.
5669 Set the minimum duration of the sourced audio. See
5670 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5671 for the accepted syntax.
5672 Note that the resulting duration may be greater than the specified
5673 duration, as the generated audio is always cut at the end of a
5676 If not specified, or the expressed duration is negative, the audio is
5677 supposed to be generated forever.
5680 Set the number of samples per channel per each output frame,
5683 @item sample_rate, s
5684 Specify the sample rate, default to 44100.
5687 Each expression in @var{exprs} can contain the following constants:
5691 number of the evaluated sample, starting from 0
5694 time of the evaluated sample expressed in seconds, starting from 0
5701 @subsection Examples
5711 Generate a sin signal with frequency of 440 Hz, set sample rate to
5714 aevalsrc="sin(440*2*PI*t):s=8000"
5718 Generate a two channels signal, specify the channel layout (Front
5719 Center + Back Center) explicitly:
5721 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
5725 Generate white noise:
5727 aevalsrc="-2+random(0)"
5731 Generate an amplitude modulated signal:
5733 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
5737 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
5739 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
5746 The null audio source, return unprocessed audio frames. It is mainly useful
5747 as a template and to be employed in analysis / debugging tools, or as
5748 the source for filters which ignore the input data (for example the sox
5751 This source accepts the following options:
5755 @item channel_layout, cl
5757 Specifies the channel layout, and can be either an integer or a string
5758 representing a channel layout. The default value of @var{channel_layout}
5761 Check the channel_layout_map definition in
5762 @file{libavutil/channel_layout.c} for the mapping between strings and
5763 channel layout values.
5765 @item sample_rate, r
5766 Specifies the sample rate, and defaults to 44100.
5769 Set the number of samples per requested frames.
5773 @subsection Examples
5777 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
5779 anullsrc=r=48000:cl=4
5783 Do the same operation with a more obvious syntax:
5785 anullsrc=r=48000:cl=mono
5789 All the parameters need to be explicitly defined.
5793 Synthesize a voice utterance using the libflite library.
5795 To enable compilation of this filter you need to configure FFmpeg with
5796 @code{--enable-libflite}.
5798 Note that versions of the flite library prior to 2.0 are not thread-safe.
5800 The filter accepts the following options:
5805 If set to 1, list the names of the available voices and exit
5806 immediately. Default value is 0.
5809 Set the maximum number of samples per frame. Default value is 512.
5812 Set the filename containing the text to speak.
5815 Set the text to speak.
5818 Set the voice to use for the speech synthesis. Default value is
5819 @code{kal}. See also the @var{list_voices} option.
5822 @subsection Examples
5826 Read from file @file{speech.txt}, and synthesize the text using the
5827 standard flite voice:
5829 flite=textfile=speech.txt
5833 Read the specified text selecting the @code{slt} voice:
5835 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
5839 Input text to ffmpeg:
5841 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
5845 Make @file{ffplay} speak the specified text, using @code{flite} and
5846 the @code{lavfi} device:
5848 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
5852 For more information about libflite, check:
5853 @url{http://www.festvox.org/flite/}
5857 Generate a noise audio signal.
5859 The filter accepts the following options:
5862 @item sample_rate, r
5863 Specify the sample rate. Default value is 48000 Hz.
5866 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
5870 Specify the duration of the generated audio stream. Not specifying this option
5871 results in noise with an infinite length.
5873 @item color, colour, c
5874 Specify the color of noise. Available noise colors are white, pink, brown,
5875 blue and violet. Default color is white.
5878 Specify a value used to seed the PRNG.
5881 Set the number of samples per each output frame, default is 1024.
5884 @subsection Examples
5889 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
5891 anoisesrc=d=60:c=pink:r=44100:a=0.5
5897 Generate odd-tap Hilbert transform FIR coefficients.
5899 The resulting stream can be used with @ref{afir} filter for phase-shifting
5900 the signal by 90 degrees.
5902 This is used in many matrix coding schemes and for analytic signal generation.
5903 The process is often written as a multiplication by i (or j), the imaginary unit.
5905 The filter accepts the following options:
5909 @item sample_rate, s
5910 Set sample rate, default is 44100.
5913 Set length of FIR filter, default is 22051.
5916 Set number of samples per each frame.
5919 Set window function to be used when generating FIR coefficients.
5924 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
5926 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
5928 The filter accepts the following options:
5931 @item sample_rate, r
5932 Set sample rate, default is 44100.
5935 Set number of samples per each frame. Default is 1024.
5938 Set high-pass frequency. Default is 0.
5941 Set low-pass frequency. Default is 0.
5942 If high-pass frequency is lower than low-pass frequency and low-pass frequency
5943 is higher than 0 then filter will create band-pass filter coefficients,
5944 otherwise band-reject filter coefficients.
5947 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
5950 Set Kaiser window beta.
5953 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
5956 Enable rounding, by default is disabled.
5959 Set number of taps for high-pass filter.
5962 Set number of taps for low-pass filter.
5967 Generate an audio signal made of a sine wave with amplitude 1/8.
5969 The audio signal is bit-exact.
5971 The filter accepts the following options:
5976 Set the carrier frequency. Default is 440 Hz.
5978 @item beep_factor, b
5979 Enable a periodic beep every second with frequency @var{beep_factor} times
5980 the carrier frequency. Default is 0, meaning the beep is disabled.
5982 @item sample_rate, r
5983 Specify the sample rate, default is 44100.
5986 Specify the duration of the generated audio stream.
5988 @item samples_per_frame
5989 Set the number of samples per output frame.
5991 The expression can contain the following constants:
5995 The (sequential) number of the output audio frame, starting from 0.
5998 The PTS (Presentation TimeStamp) of the output audio frame,
5999 expressed in @var{TB} units.
6002 The PTS of the output audio frame, expressed in seconds.
6005 The timebase of the output audio frames.
6008 Default is @code{1024}.
6011 @subsection Examples
6016 Generate a simple 440 Hz sine wave:
6022 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6026 sine=frequency=220:beep_factor=4:duration=5
6030 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6033 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6037 @c man end AUDIO SOURCES
6039 @chapter Audio Sinks
6040 @c man begin AUDIO SINKS
6042 Below is a description of the currently available audio sinks.
6044 @section abuffersink
6046 Buffer audio frames, and make them available to the end of filter chain.
6048 This sink is mainly intended for programmatic use, in particular
6049 through the interface defined in @file{libavfilter/buffersink.h}
6050 or the options system.
6052 It accepts a pointer to an AVABufferSinkContext structure, which
6053 defines the incoming buffers' formats, to be passed as the opaque
6054 parameter to @code{avfilter_init_filter} for initialization.
6057 Null audio sink; do absolutely nothing with the input audio. It is
6058 mainly useful as a template and for use in analysis / debugging
6061 @c man end AUDIO SINKS
6063 @chapter Video Filters
6064 @c man begin VIDEO FILTERS
6066 When you configure your FFmpeg build, you can disable any of the
6067 existing filters using @code{--disable-filters}.
6068 The configure output will show the video filters included in your
6071 Below is a description of the currently available video filters.
6075 Mark a region of interest in a video frame.
6077 The frame data is passed through unchanged, but metadata is attached
6078 to the frame indicating regions of interest which can affect the
6079 behaviour of later encoding. Multiple regions can be marked by
6080 applying the filter multiple times.
6084 Region distance in pixels from the left edge of the frame.
6086 Region distance in pixels from the top edge of the frame.
6088 Region width in pixels.
6090 Region height in pixels.
6092 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6093 and may contain the following variables:
6096 Width of the input frame.
6098 Height of the input frame.
6102 Quantisation offset to apply within the region.
6104 This must be a real value in the range -1 to +1. A value of zero
6105 indicates no quality change. A negative value asks for better quality
6106 (less quantisation), while a positive value asks for worse quality
6107 (greater quantisation).
6109 The range is calibrated so that the extreme values indicate the
6110 largest possible offset - if the rest of the frame is encoded with the
6111 worst possible quality, an offset of -1 indicates that this region
6112 should be encoded with the best possible quality anyway. Intermediate
6113 values are then interpolated in some codec-dependent way.
6115 For example, in 10-bit H.264 the quantisation parameter varies between
6116 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6117 this region should be encoded with a QP around one-tenth of the full
6118 range better than the rest of the frame. So, if most of the frame
6119 were to be encoded with a QP of around 30, this region would get a QP
6120 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6121 An extreme value of -1 would indicate that this region should be
6122 encoded with the best possible quality regardless of the treatment of
6123 the rest of the frame - that is, should be encoded at a QP of -12.
6125 If set to true, remove any existing regions of interest marked on the
6126 frame before adding the new one.
6129 @subsection Examples
6133 Mark the centre quarter of the frame as interesting.
6135 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6138 Mark the 100-pixel-wide region on the left edge of the frame as very
6139 uninteresting (to be encoded at much lower quality than the rest of
6142 addroi=0:0:100:ih:+1/5
6146 @section alphaextract
6148 Extract the alpha component from the input as a grayscale video. This
6149 is especially useful with the @var{alphamerge} filter.
6153 Add or replace the alpha component of the primary input with the
6154 grayscale value of a second input. This is intended for use with
6155 @var{alphaextract} to allow the transmission or storage of frame
6156 sequences that have alpha in a format that doesn't support an alpha
6159 For example, to reconstruct full frames from a normal YUV-encoded video
6160 and a separate video created with @var{alphaextract}, you might use:
6162 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6165 Since this filter is designed for reconstruction, it operates on frame
6166 sequences without considering timestamps, and terminates when either
6167 input reaches end of stream. This will cause problems if your encoding
6168 pipeline drops frames. If you're trying to apply an image as an
6169 overlay to a video stream, consider the @var{overlay} filter instead.
6173 Amplify differences between current pixel and pixels of adjacent frames in
6174 same pixel location.
6176 This filter accepts the following options:
6180 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6181 For example radius of 3 will instruct filter to calculate average of 7 frames.
6184 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6187 Set threshold for difference amplification. Any difference greater or equal to
6188 this value will not alter source pixel. Default is 10.
6189 Allowed range is from 0 to 65535.
6192 Set tolerance for difference amplification. Any difference lower to
6193 this value will not alter source pixel. Default is 0.
6194 Allowed range is from 0 to 65535.
6197 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6198 This option controls maximum possible value that will decrease source pixel value.
6201 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6202 This option controls maximum possible value that will increase source pixel value.
6205 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6208 @subsection Commands
6210 This filter supports the following @ref{commands} that corresponds to option of same name:
6222 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6223 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6224 Substation Alpha) subtitles files.
6226 This filter accepts the following option in addition to the common options from
6227 the @ref{subtitles} filter:
6231 Set the shaping engine
6233 Available values are:
6236 The default libass shaping engine, which is the best available.
6238 Fast, font-agnostic shaper that can do only substitutions
6240 Slower shaper using OpenType for substitutions and positioning
6243 The default is @code{auto}.
6247 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6249 The filter accepts the following options:
6253 Set threshold A for 1st plane. Default is 0.02.
6254 Valid range is 0 to 0.3.
6257 Set threshold B for 1st plane. Default is 0.04.
6258 Valid range is 0 to 5.
6261 Set threshold A for 2nd plane. Default is 0.02.
6262 Valid range is 0 to 0.3.
6265 Set threshold B for 2nd plane. Default is 0.04.
6266 Valid range is 0 to 5.
6269 Set threshold A for 3rd plane. Default is 0.02.
6270 Valid range is 0 to 0.3.
6273 Set threshold B for 3rd plane. Default is 0.04.
6274 Valid range is 0 to 5.
6276 Threshold A is designed to react on abrupt changes in the input signal and
6277 threshold B is designed to react on continuous changes in the input signal.
6280 Set number of frames filter will use for averaging. Default is 9. Must be odd
6281 number in range [5, 129].
6284 Set what planes of frame filter will use for averaging. Default is all.
6287 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6288 Alternatively can be set to @code{s} serial.
6290 Parallel can be faster then serial, while other way around is never true.
6291 Parallel will abort early on first change being greater then thresholds, while serial
6292 will continue processing other side of frames if they are equal or bellow thresholds.
6295 @subsection Commands
6296 This filter supports same @ref{commands} as options except option @code{s}.
6297 The command accepts the same syntax of the corresponding option.
6301 Apply average blur filter.
6303 The filter accepts the following options:
6307 Set horizontal radius size.
6310 Set which planes to filter. By default all planes are filtered.
6313 Set vertical radius size, if zero it will be same as @code{sizeX}.
6314 Default is @code{0}.
6317 @subsection Commands
6318 This filter supports same commands as options.
6319 The command accepts the same syntax of the corresponding option.
6321 If the specified expression is not valid, it is kept at its current
6326 Compute the bounding box for the non-black pixels in the input frame
6329 This filter computes the bounding box containing all the pixels with a
6330 luminance value greater than the minimum allowed value.
6331 The parameters describing the bounding box are printed on the filter
6334 The filter accepts the following option:
6338 Set the minimal luminance value. Default is @code{16}.
6342 Apply bilateral filter, spatial smoothing while preserving edges.
6344 The filter accepts the following options:
6347 Set sigma of gaussian function to calculate spatial weight.
6348 Allowed range is 0 to 10. Default is 0.1.
6351 Set sigma of gaussian function to calculate range weight.
6352 Allowed range is 0 to 1. Default is 0.1.
6355 Set planes to filter. Default is first only.
6358 @section bitplanenoise
6360 Show and measure bit plane noise.
6362 The filter accepts the following options:
6366 Set which plane to analyze. Default is @code{1}.
6369 Filter out noisy pixels from @code{bitplane} set above.
6370 Default is disabled.
6373 @section blackdetect
6375 Detect video intervals that are (almost) completely black. Can be
6376 useful to detect chapter transitions, commercials, or invalid
6377 recordings. Output lines contains the time for the start, end and
6378 duration of the detected black interval expressed in seconds.
6380 In order to display the output lines, you need to set the loglevel at
6381 least to the AV_LOG_INFO value.
6383 The filter accepts the following options:
6386 @item black_min_duration, d
6387 Set the minimum detected black duration expressed in seconds. It must
6388 be a non-negative floating point number.
6390 Default value is 2.0.
6392 @item picture_black_ratio_th, pic_th
6393 Set the threshold for considering a picture "black".
6394 Express the minimum value for the ratio:
6396 @var{nb_black_pixels} / @var{nb_pixels}
6399 for which a picture is considered black.
6400 Default value is 0.98.
6402 @item pixel_black_th, pix_th
6403 Set the threshold for considering a pixel "black".
6405 The threshold expresses the maximum pixel luminance value for which a
6406 pixel is considered "black". The provided value is scaled according to
6407 the following equation:
6409 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6412 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6413 the input video format, the range is [0-255] for YUV full-range
6414 formats and [16-235] for YUV non full-range formats.
6416 Default value is 0.10.
6419 The following example sets the maximum pixel threshold to the minimum
6420 value, and detects only black intervals of 2 or more seconds:
6422 blackdetect=d=2:pix_th=0.00
6427 Detect frames that are (almost) completely black. Can be useful to
6428 detect chapter transitions or commercials. Output lines consist of
6429 the frame number of the detected frame, the percentage of blackness,
6430 the position in the file if known or -1 and the timestamp in seconds.
6432 In order to display the output lines, you need to set the loglevel at
6433 least to the AV_LOG_INFO value.
6435 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6436 The value represents the percentage of pixels in the picture that
6437 are below the threshold value.
6439 It accepts the following parameters:
6444 The percentage of the pixels that have to be below the threshold; it defaults to
6447 @item threshold, thresh
6448 The threshold below which a pixel value is considered black; it defaults to
6453 @section blend, tblend
6455 Blend two video frames into each other.
6457 The @code{blend} filter takes two input streams and outputs one
6458 stream, the first input is the "top" layer and second input is
6459 "bottom" layer. By default, the output terminates when the longest input terminates.
6461 The @code{tblend} (time blend) filter takes two consecutive frames
6462 from one single stream, and outputs the result obtained by blending
6463 the new frame on top of the old frame.
6465 A description of the accepted options follows.
6473 Set blend mode for specific pixel component or all pixel components in case
6474 of @var{all_mode}. Default value is @code{normal}.
6476 Available values for component modes are:
6518 Set blend opacity for specific pixel component or all pixel components in case
6519 of @var{all_opacity}. Only used in combination with pixel component blend modes.
6526 Set blend expression for specific pixel component or all pixel components in case
6527 of @var{all_expr}. Note that related mode options will be ignored if those are set.
6529 The expressions can use the following variables:
6533 The sequential number of the filtered frame, starting from @code{0}.
6537 the coordinates of the current sample
6541 the width and height of currently filtered plane
6545 Width and height scale for the plane being filtered. It is the
6546 ratio between the dimensions of the current plane to the luma plane,
6547 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
6548 the luma plane and @code{0.5,0.5} for the chroma planes.
6551 Time of the current frame, expressed in seconds.
6554 Value of pixel component at current location for first video frame (top layer).
6557 Value of pixel component at current location for second video frame (bottom layer).
6561 The @code{blend} filter also supports the @ref{framesync} options.
6563 @subsection Examples
6567 Apply transition from bottom layer to top layer in first 10 seconds:
6569 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
6573 Apply linear horizontal transition from top layer to bottom layer:
6575 blend=all_expr='A*(X/W)+B*(1-X/W)'
6579 Apply 1x1 checkerboard effect:
6581 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
6585 Apply uncover left effect:
6587 blend=all_expr='if(gte(N*SW+X,W),A,B)'
6591 Apply uncover down effect:
6593 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
6597 Apply uncover up-left effect:
6599 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
6603 Split diagonally video and shows top and bottom layer on each side:
6605 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
6609 Display differences between the current and the previous frame:
6611 tblend=all_mode=grainextract
6617 Denoise frames using Block-Matching 3D algorithm.
6619 The filter accepts the following options.
6623 Set denoising strength. Default value is 1.
6624 Allowed range is from 0 to 999.9.
6625 The denoising algorithm is very sensitive to sigma, so adjust it
6626 according to the source.
6629 Set local patch size. This sets dimensions in 2D.
6632 Set sliding step for processing blocks. Default value is 4.
6633 Allowed range is from 1 to 64.
6634 Smaller values allows processing more reference blocks and is slower.
6637 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
6638 When set to 1, no block matching is done. Larger values allows more blocks
6640 Allowed range is from 1 to 256.
6643 Set radius for search block matching. Default is 9.
6644 Allowed range is from 1 to INT32_MAX.
6647 Set step between two search locations for block matching. Default is 1.
6648 Allowed range is from 1 to 64. Smaller is slower.
6651 Set threshold of mean square error for block matching. Valid range is 0 to
6655 Set thresholding parameter for hard thresholding in 3D transformed domain.
6656 Larger values results in stronger hard-thresholding filtering in frequency
6660 Set filtering estimation mode. Can be @code{basic} or @code{final}.
6661 Default is @code{basic}.
6664 If enabled, filter will use 2nd stream for block matching.
6665 Default is disabled for @code{basic} value of @var{estim} option,
6666 and always enabled if value of @var{estim} is @code{final}.
6669 Set planes to filter. Default is all available except alpha.
6672 @subsection Examples
6676 Basic filtering with bm3d:
6678 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
6682 Same as above, but filtering only luma:
6684 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
6688 Same as above, but with both estimation modes:
6690 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
6694 Same as above, but prefilter with @ref{nlmeans} filter instead:
6696 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
6702 Apply a boxblur algorithm to the input video.
6704 It accepts the following parameters:
6708 @item luma_radius, lr
6709 @item luma_power, lp
6710 @item chroma_radius, cr
6711 @item chroma_power, cp
6712 @item alpha_radius, ar
6713 @item alpha_power, ap
6717 A description of the accepted options follows.
6720 @item luma_radius, lr
6721 @item chroma_radius, cr
6722 @item alpha_radius, ar
6723 Set an expression for the box radius in pixels used for blurring the
6724 corresponding input plane.
6726 The radius value must be a non-negative number, and must not be
6727 greater than the value of the expression @code{min(w,h)/2} for the
6728 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
6731 Default value for @option{luma_radius} is "2". If not specified,
6732 @option{chroma_radius} and @option{alpha_radius} default to the
6733 corresponding value set for @option{luma_radius}.
6735 The expressions can contain the following constants:
6739 The input width and height in pixels.
6743 The input chroma image width and height in pixels.
6747 The horizontal and vertical chroma subsample values. For example, for the
6748 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
6751 @item luma_power, lp
6752 @item chroma_power, cp
6753 @item alpha_power, ap
6754 Specify how many times the boxblur filter is applied to the
6755 corresponding plane.
6757 Default value for @option{luma_power} is 2. If not specified,
6758 @option{chroma_power} and @option{alpha_power} default to the
6759 corresponding value set for @option{luma_power}.
6761 A value of 0 will disable the effect.
6764 @subsection Examples
6768 Apply a boxblur filter with the luma, chroma, and alpha radii
6771 boxblur=luma_radius=2:luma_power=1
6776 Set the luma radius to 2, and alpha and chroma radius to 0:
6778 boxblur=2:1:cr=0:ar=0
6782 Set the luma and chroma radii to a fraction of the video dimension:
6784 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
6790 Deinterlace the input video ("bwdif" stands for "Bob Weaver
6791 Deinterlacing Filter").
6793 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
6794 interpolation algorithms.
6795 It accepts the following parameters:
6799 The interlacing mode to adopt. It accepts one of the following values:
6803 Output one frame for each frame.
6805 Output one frame for each field.
6808 The default value is @code{send_field}.
6811 The picture field parity assumed for the input interlaced video. It accepts one
6812 of the following values:
6816 Assume the top field is first.
6818 Assume the bottom field is first.
6820 Enable automatic detection of field parity.
6823 The default value is @code{auto}.
6824 If the interlacing is unknown or the decoder does not export this information,
6825 top field first will be assumed.
6828 Specify which frames to deinterlace. Accepts one of the following
6833 Deinterlace all frames.
6835 Only deinterlace frames marked as interlaced.
6838 The default value is @code{all}.
6842 Remove all color information for all colors except for certain one.
6844 The filter accepts the following options:
6848 The color which will not be replaced with neutral chroma.
6851 Similarity percentage with the above color.
6852 0.01 matches only the exact key color, while 1.0 matches everything.
6856 0.0 makes pixels either fully gray, or not gray at all.
6857 Higher values result in more preserved color.
6860 Signals that the color passed is already in YUV instead of RGB.
6862 Literal colors like "green" or "red" don't make sense with this enabled anymore.
6863 This can be used to pass exact YUV values as hexadecimal numbers.
6867 YUV colorspace color/chroma keying.
6869 The filter accepts the following options:
6873 The color which will be replaced with transparency.
6876 Similarity percentage with the key color.
6878 0.01 matches only the exact key color, while 1.0 matches everything.
6883 0.0 makes pixels either fully transparent, or not transparent at all.
6885 Higher values result in semi-transparent pixels, with a higher transparency
6886 the more similar the pixels color is to the key color.
6889 Signals that the color passed is already in YUV instead of RGB.
6891 Literal colors like "green" or "red" don't make sense with this enabled anymore.
6892 This can be used to pass exact YUV values as hexadecimal numbers.
6895 @subsection Examples
6899 Make every green pixel in the input image transparent:
6901 ffmpeg -i input.png -vf chromakey=green out.png
6905 Overlay a greenscreen-video on top of a static black background.
6907 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
6911 @section chromashift
6912 Shift chroma pixels horizontally and/or vertically.
6914 The filter accepts the following options:
6917 Set amount to shift chroma-blue horizontally.
6919 Set amount to shift chroma-blue vertically.
6921 Set amount to shift chroma-red horizontally.
6923 Set amount to shift chroma-red vertically.
6925 Set edge mode, can be @var{smear}, default, or @var{warp}.
6930 Display CIE color diagram with pixels overlaid onto it.
6932 The filter accepts the following options:
6947 @item uhdtv, rec2020
6961 Set what gamuts to draw.
6963 See @code{system} option for available values.
6966 Set ciescope size, by default set to 512.
6969 Set intensity used to map input pixel values to CIE diagram.
6972 Set contrast used to draw tongue colors that are out of active color system gamut.
6975 Correct gamma displayed on scope, by default enabled.
6978 Show white point on CIE diagram, by default disabled.
6981 Set input gamma. Used only with XYZ input color space.
6986 Visualize information exported by some codecs.
6988 Some codecs can export information through frames using side-data or other
6989 means. For example, some MPEG based codecs export motion vectors through the
6990 @var{export_mvs} flag in the codec @option{flags2} option.
6992 The filter accepts the following option:
6996 Set motion vectors to visualize.
6998 Available flags for @var{mv} are:
7002 forward predicted MVs of P-frames
7004 forward predicted MVs of B-frames
7006 backward predicted MVs of B-frames
7010 Display quantization parameters using the chroma planes.
7013 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7015 Available flags for @var{mv_type} are:
7019 forward predicted MVs
7021 backward predicted MVs
7024 @item frame_type, ft
7025 Set frame type to visualize motion vectors of.
7027 Available flags for @var{frame_type} are:
7031 intra-coded frames (I-frames)
7033 predicted frames (P-frames)
7035 bi-directionally predicted frames (B-frames)
7039 @subsection Examples
7043 Visualize forward predicted MVs of all frames using @command{ffplay}:
7045 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7049 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7051 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7055 @section colorbalance
7056 Modify intensity of primary colors (red, green and blue) of input frames.
7058 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7059 regions for the red-cyan, green-magenta or blue-yellow balance.
7061 A positive adjustment value shifts the balance towards the primary color, a negative
7062 value towards the complementary color.
7064 The filter accepts the following options:
7070 Adjust red, green and blue shadows (darkest pixels).
7075 Adjust red, green and blue midtones (medium pixels).
7080 Adjust red, green and blue highlights (brightest pixels).
7082 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7085 Preserve lightness when changing color balance. Default is disabled.
7088 @subsection Examples
7092 Add red color cast to shadows:
7098 @subsection Commands
7100 This filter supports the all above options as @ref{commands}.
7102 @section colorchannelmixer
7104 Adjust video input frames by re-mixing color channels.
7106 This filter modifies a color channel by adding the values associated to
7107 the other channels of the same pixels. For example if the value to
7108 modify is red, the output value will be:
7110 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7113 The filter accepts the following options:
7120 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7121 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7127 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7128 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7134 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7135 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7141 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7142 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7144 Allowed ranges for options are @code{[-2.0, 2.0]}.
7147 @subsection Examples
7151 Convert source to grayscale:
7153 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7156 Simulate sepia tones:
7158 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7162 @subsection Commands
7164 This filter supports the all above options as @ref{commands}.
7167 RGB colorspace color keying.
7169 The filter accepts the following options:
7173 The color which will be replaced with transparency.
7176 Similarity percentage with the key color.
7178 0.01 matches only the exact key color, while 1.0 matches everything.
7183 0.0 makes pixels either fully transparent, or not transparent at all.
7185 Higher values result in semi-transparent pixels, with a higher transparency
7186 the more similar the pixels color is to the key color.
7189 @subsection Examples
7193 Make every green pixel in the input image transparent:
7195 ffmpeg -i input.png -vf colorkey=green out.png
7199 Overlay a greenscreen-video on top of a static background image.
7201 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
7206 Remove all color information for all RGB colors except for certain one.
7208 The filter accepts the following options:
7212 The color which will not be replaced with neutral gray.
7215 Similarity percentage with the above color.
7216 0.01 matches only the exact key color, while 1.0 matches everything.
7219 Blend percentage. 0.0 makes pixels fully gray.
7220 Higher values result in more preserved color.
7223 @section colorlevels
7225 Adjust video input frames using levels.
7227 The filter accepts the following options:
7234 Adjust red, green, blue and alpha input black point.
7235 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7241 Adjust red, green, blue and alpha input white point.
7242 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7244 Input levels are used to lighten highlights (bright tones), darken shadows
7245 (dark tones), change the balance of bright and dark tones.
7251 Adjust red, green, blue and alpha output black point.
7252 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7258 Adjust red, green, blue and alpha output white point.
7259 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7261 Output levels allows manual selection of a constrained output level range.
7264 @subsection Examples
7268 Make video output darker:
7270 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7276 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7280 Make video output lighter:
7282 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7286 Increase brightness:
7288 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7292 @section colormatrix
7294 Convert color matrix.
7296 The filter accepts the following options:
7301 Specify the source and destination color matrix. Both values must be
7304 The accepted values are:
7332 For example to convert from BT.601 to SMPTE-240M, use the command:
7334 colormatrix=bt601:smpte240m
7339 Convert colorspace, transfer characteristics or color primaries.
7340 Input video needs to have an even size.
7342 The filter accepts the following options:
7347 Specify all color properties at once.
7349 The accepted values are:
7379 Specify output colorspace.
7381 The accepted values are:
7390 BT.470BG or BT.601-6 625
7393 SMPTE-170M or BT.601-6 525
7402 BT.2020 with non-constant luminance
7408 Specify output transfer characteristics.
7410 The accepted values are:
7422 Constant gamma of 2.2
7425 Constant gamma of 2.8
7428 SMPTE-170M, BT.601-6 625 or BT.601-6 525
7446 BT.2020 for 10-bits content
7449 BT.2020 for 12-bits content
7455 Specify output color primaries.
7457 The accepted values are:
7466 BT.470BG or BT.601-6 625
7469 SMPTE-170M or BT.601-6 525
7493 Specify output color range.
7495 The accepted values are:
7498 TV (restricted) range
7501 MPEG (restricted) range
7512 Specify output color format.
7514 The accepted values are:
7517 YUV 4:2:0 planar 8-bits
7520 YUV 4:2:0 planar 10-bits
7523 YUV 4:2:0 planar 12-bits
7526 YUV 4:2:2 planar 8-bits
7529 YUV 4:2:2 planar 10-bits
7532 YUV 4:2:2 planar 12-bits
7535 YUV 4:4:4 planar 8-bits
7538 YUV 4:4:4 planar 10-bits
7541 YUV 4:4:4 planar 12-bits
7546 Do a fast conversion, which skips gamma/primary correction. This will take
7547 significantly less CPU, but will be mathematically incorrect. To get output
7548 compatible with that produced by the colormatrix filter, use fast=1.
7551 Specify dithering mode.
7553 The accepted values are:
7559 Floyd-Steinberg dithering
7563 Whitepoint adaptation mode.
7565 The accepted values are:
7568 Bradford whitepoint adaptation
7571 von Kries whitepoint adaptation
7574 identity whitepoint adaptation (i.e. no whitepoint adaptation)
7578 Override all input properties at once. Same accepted values as @ref{all}.
7581 Override input colorspace. Same accepted values as @ref{space}.
7584 Override input color primaries. Same accepted values as @ref{primaries}.
7587 Override input transfer characteristics. Same accepted values as @ref{trc}.
7590 Override input color range. Same accepted values as @ref{range}.
7594 The filter converts the transfer characteristics, color space and color
7595 primaries to the specified user values. The output value, if not specified,
7596 is set to a default value based on the "all" property. If that property is
7597 also not specified, the filter will log an error. The output color range and
7598 format default to the same value as the input color range and format. The
7599 input transfer characteristics, color space, color primaries and color range
7600 should be set on the input data. If any of these are missing, the filter will
7601 log an error and no conversion will take place.
7603 For example to convert the input to SMPTE-240M, use the command:
7605 colorspace=smpte240m
7608 @section convolution
7610 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
7612 The filter accepts the following options:
7619 Set matrix for each plane.
7620 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
7621 and from 1 to 49 odd number of signed integers in @var{row} mode.
7627 Set multiplier for calculated value for each plane.
7628 If unset or 0, it will be sum of all matrix elements.
7634 Set bias for each plane. This value is added to the result of the multiplication.
7635 Useful for making the overall image brighter or darker. Default is 0.0.
7641 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
7642 Default is @var{square}.
7645 @subsection Examples
7651 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"
7657 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"
7663 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"
7669 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"
7673 Apply laplacian edge detector which includes diagonals:
7675 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"
7681 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"
7687 Apply 2D convolution of video stream in frequency domain using second stream
7690 The filter accepts the following options:
7694 Set which planes to process.
7697 Set which impulse video frames will be processed, can be @var{first}
7698 or @var{all}. Default is @var{all}.
7701 The @code{convolve} filter also supports the @ref{framesync} options.
7705 Copy the input video source unchanged to the output. This is mainly useful for
7710 Video filtering on GPU using Apple's CoreImage API on OSX.
7712 Hardware acceleration is based on an OpenGL context. Usually, this means it is
7713 processed by video hardware. However, software-based OpenGL implementations
7714 exist which means there is no guarantee for hardware processing. It depends on
7717 There are many filters and image generators provided by Apple that come with a
7718 large variety of options. The filter has to be referenced by its name along
7721 The coreimage filter accepts the following options:
7724 List all available filters and generators along with all their respective
7725 options as well as possible minimum and maximum values along with the default
7732 Specify all filters by their respective name and options.
7733 Use @var{list_filters} to determine all valid filter names and options.
7734 Numerical options are specified by a float value and are automatically clamped
7735 to their respective value range. Vector and color options have to be specified
7736 by a list of space separated float values. Character escaping has to be done.
7737 A special option name @code{default} is available to use default options for a
7740 It is required to specify either @code{default} or at least one of the filter options.
7741 All omitted options are used with their default values.
7742 The syntax of the filter string is as follows:
7744 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
7748 Specify a rectangle where the output of the filter chain is copied into the
7749 input image. It is given by a list of space separated float values:
7751 output_rect=x\ y\ width\ height
7753 If not given, the output rectangle equals the dimensions of the input image.
7754 The output rectangle is automatically cropped at the borders of the input
7755 image. Negative values are valid for each component.
7757 output_rect=25\ 25\ 100\ 100
7761 Several filters can be chained for successive processing without GPU-HOST
7762 transfers allowing for fast processing of complex filter chains.
7763 Currently, only filters with zero (generators) or exactly one (filters) input
7764 image and one output image are supported. Also, transition filters are not yet
7767 Some filters generate output images with additional padding depending on the
7768 respective filter kernel. The padding is automatically removed to ensure the
7769 filter output has the same size as the input image.
7771 For image generators, the size of the output image is determined by the
7772 previous output image of the filter chain or the input image of the whole
7773 filterchain, respectively. The generators do not use the pixel information of
7774 this image to generate their output. However, the generated output is
7775 blended onto this image, resulting in partial or complete coverage of the
7778 The @ref{coreimagesrc} video source can be used for generating input images
7779 which are directly fed into the filter chain. By using it, providing input
7780 images by another video source or an input video is not required.
7782 @subsection Examples
7787 List all filters available:
7789 coreimage=list_filters=true
7793 Use the CIBoxBlur filter with default options to blur an image:
7795 coreimage=filter=CIBoxBlur@@default
7799 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
7800 its center at 100x100 and a radius of 50 pixels:
7802 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
7806 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
7807 given as complete and escaped command-line for Apple's standard bash shell:
7809 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
7815 Cover a rectangular object
7817 It accepts the following options:
7821 Filepath of the optional cover image, needs to be in yuv420.
7826 It accepts the following values:
7829 cover it by the supplied image
7831 cover it by interpolating the surrounding pixels
7834 Default value is @var{blur}.
7837 @subsection Examples
7841 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
7843 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
7849 Crop the input video to given dimensions.
7851 It accepts the following parameters:
7855 The width of the output video. It defaults to @code{iw}.
7856 This expression is evaluated only once during the filter
7857 configuration, or when the @samp{w} or @samp{out_w} command is sent.
7860 The height of the output video. It defaults to @code{ih}.
7861 This expression is evaluated only once during the filter
7862 configuration, or when the @samp{h} or @samp{out_h} command is sent.
7865 The horizontal position, in the input video, of the left edge of the output
7866 video. It defaults to @code{(in_w-out_w)/2}.
7867 This expression is evaluated per-frame.
7870 The vertical position, in the input video, of the top edge of the output video.
7871 It defaults to @code{(in_h-out_h)/2}.
7872 This expression is evaluated per-frame.
7875 If set to 1 will force the output display aspect ratio
7876 to be the same of the input, by changing the output sample aspect
7877 ratio. It defaults to 0.
7880 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
7881 width/height/x/y as specified and will not be rounded to nearest smaller value.
7885 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
7886 expressions containing the following constants:
7891 The computed values for @var{x} and @var{y}. They are evaluated for
7896 The input width and height.
7900 These are the same as @var{in_w} and @var{in_h}.
7904 The output (cropped) width and height.
7908 These are the same as @var{out_w} and @var{out_h}.
7911 same as @var{iw} / @var{ih}
7914 input sample aspect ratio
7917 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
7921 horizontal and vertical chroma subsample values. For example for the
7922 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
7925 The number of the input frame, starting from 0.
7928 the position in the file of the input frame, NAN if unknown
7931 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
7935 The expression for @var{out_w} may depend on the value of @var{out_h},
7936 and the expression for @var{out_h} may depend on @var{out_w}, but they
7937 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
7938 evaluated after @var{out_w} and @var{out_h}.
7940 The @var{x} and @var{y} parameters specify the expressions for the
7941 position of the top-left corner of the output (non-cropped) area. They
7942 are evaluated for each frame. If the evaluated value is not valid, it
7943 is approximated to the nearest valid value.
7945 The expression for @var{x} may depend on @var{y}, and the expression
7946 for @var{y} may depend on @var{x}.
7948 @subsection Examples
7952 Crop area with size 100x100 at position (12,34).
7957 Using named options, the example above becomes:
7959 crop=w=100:h=100:x=12:y=34
7963 Crop the central input area with size 100x100:
7969 Crop the central input area with size 2/3 of the input video:
7971 crop=2/3*in_w:2/3*in_h
7975 Crop the input video central square:
7982 Delimit the rectangle with the top-left corner placed at position
7983 100:100 and the right-bottom corner corresponding to the right-bottom
7984 corner of the input image.
7986 crop=in_w-100:in_h-100:100:100
7990 Crop 10 pixels from the left and right borders, and 20 pixels from
7991 the top and bottom borders
7993 crop=in_w-2*10:in_h-2*20
7997 Keep only the bottom right quarter of the input image:
7999 crop=in_w/2:in_h/2:in_w/2:in_h/2
8003 Crop height for getting Greek harmony:
8005 crop=in_w:1/PHI*in_w
8009 Apply trembling effect:
8011 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)
8015 Apply erratic camera effect depending on timestamp:
8017 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)"
8021 Set x depending on the value of y:
8023 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8027 @subsection Commands
8029 This filter supports the following commands:
8035 Set width/height of the output video and the horizontal/vertical position
8037 The command accepts the same syntax of the corresponding option.
8039 If the specified expression is not valid, it is kept at its current
8045 Auto-detect the crop size.
8047 It calculates the necessary cropping parameters and prints the
8048 recommended parameters via the logging system. The detected dimensions
8049 correspond to the non-black area of the input video.
8051 It accepts the following parameters:
8056 Set higher black value threshold, which can be optionally specified
8057 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8058 value greater to the set value is considered non-black. It defaults to 24.
8059 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8060 on the bitdepth of the pixel format.
8063 The value which the width/height should be divisible by. It defaults to
8064 16. The offset is automatically adjusted to center the video. Use 2 to
8065 get only even dimensions (needed for 4:2:2 video). 16 is best when
8066 encoding to most video codecs.
8068 @item reset_count, reset
8069 Set the counter that determines after how many frames cropdetect will
8070 reset the previously detected largest video area and start over to
8071 detect the current optimal crop area. Default value is 0.
8073 This can be useful when channel logos distort the video area. 0
8074 indicates 'never reset', and returns the largest area encountered during
8081 Delay video filtering until a given wallclock timestamp. The filter first
8082 passes on @option{preroll} amount of frames, then it buffers at most
8083 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8084 it forwards the buffered frames and also any subsequent frames coming in its
8087 The filter can be used synchronize the output of multiple ffmpeg processes for
8088 realtime output devices like decklink. By putting the delay in the filtering
8089 chain and pre-buffering frames the process can pass on data to output almost
8090 immediately after the target wallclock timestamp is reached.
8092 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8098 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8101 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8104 The maximum duration of content to buffer before waiting for the cue expressed
8105 in seconds. Default is 0.
8112 Apply color adjustments using curves.
8114 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8115 component (red, green and blue) has its values defined by @var{N} key points
8116 tied from each other using a smooth curve. The x-axis represents the pixel
8117 values from the input frame, and the y-axis the new pixel values to be set for
8120 By default, a component curve is defined by the two points @var{(0;0)} and
8121 @var{(1;1)}. This creates a straight line where each original pixel value is
8122 "adjusted" to its own value, which means no change to the image.
8124 The filter allows you to redefine these two points and add some more. A new
8125 curve (using a natural cubic spline interpolation) will be define to pass
8126 smoothly through all these new coordinates. The new defined points needs to be
8127 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8128 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8129 the vector spaces, the values will be clipped accordingly.
8131 The filter accepts the following options:
8135 Select one of the available color presets. This option can be used in addition
8136 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8137 options takes priority on the preset values.
8138 Available presets are:
8141 @item color_negative
8144 @item increase_contrast
8146 @item linear_contrast
8147 @item medium_contrast
8149 @item strong_contrast
8152 Default is @code{none}.
8154 Set the master key points. These points will define a second pass mapping. It
8155 is sometimes called a "luminance" or "value" mapping. It can be used with
8156 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8157 post-processing LUT.
8159 Set the key points for the red component.
8161 Set the key points for the green component.
8163 Set the key points for the blue component.
8165 Set the key points for all components (not including master).
8166 Can be used in addition to the other key points component
8167 options. In this case, the unset component(s) will fallback on this
8168 @option{all} setting.
8170 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8172 Save Gnuplot script of the curves in specified file.
8175 To avoid some filtergraph syntax conflicts, each key points list need to be
8176 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8178 @subsection Examples
8182 Increase slightly the middle level of blue:
8184 curves=blue='0/0 0.5/0.58 1/1'
8190 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'
8192 Here we obtain the following coordinates for each components:
8195 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8197 @code{(0;0) (0.50;0.48) (1;1)}
8199 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8203 The previous example can also be achieved with the associated built-in preset:
8205 curves=preset=vintage
8215 Use a Photoshop preset and redefine the points of the green component:
8217 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8221 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8222 and @command{gnuplot}:
8224 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8225 gnuplot -p /tmp/curves.plt
8231 Video data analysis filter.
8233 This filter shows hexadecimal pixel values of part of video.
8235 The filter accepts the following options:
8239 Set output video size.
8242 Set x offset from where to pick pixels.
8245 Set y offset from where to pick pixels.
8248 Set scope mode, can be one of the following:
8251 Draw hexadecimal pixel values with white color on black background.
8254 Draw hexadecimal pixel values with input video pixel color on black
8258 Draw hexadecimal pixel values on color background picked from input video,
8259 the text color is picked in such way so its always visible.
8263 Draw rows and columns numbers on left and top of video.
8266 Set background opacity.
8271 Denoise frames using 2D DCT (frequency domain filtering).
8273 This filter is not designed for real time.
8275 The filter accepts the following options:
8279 Set the noise sigma constant.
8281 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8282 coefficient (absolute value) below this threshold with be dropped.
8284 If you need a more advanced filtering, see @option{expr}.
8286 Default is @code{0}.
8289 Set number overlapping pixels for each block. Since the filter can be slow, you
8290 may want to reduce this value, at the cost of a less effective filter and the
8291 risk of various artefacts.
8293 If the overlapping value doesn't permit processing the whole input width or
8294 height, a warning will be displayed and according borders won't be denoised.
8296 Default value is @var{blocksize}-1, which is the best possible setting.
8299 Set the coefficient factor expression.
8301 For each coefficient of a DCT block, this expression will be evaluated as a
8302 multiplier value for the coefficient.
8304 If this is option is set, the @option{sigma} option will be ignored.
8306 The absolute value of the coefficient can be accessed through the @var{c}
8310 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8311 @var{blocksize}, which is the width and height of the processed blocks.
8313 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8314 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8315 on the speed processing. Also, a larger block size does not necessarily means a
8319 @subsection Examples
8321 Apply a denoise with a @option{sigma} of @code{4.5}:
8326 The same operation can be achieved using the expression system:
8328 dctdnoiz=e='gte(c, 4.5*3)'
8331 Violent denoise using a block size of @code{16x16}:
8338 Remove banding artifacts from input video.
8339 It works by replacing banded pixels with average value of referenced pixels.
8341 The filter accepts the following options:
8348 Set banding detection threshold for each plane. Default is 0.02.
8349 Valid range is 0.00003 to 0.5.
8350 If difference between current pixel and reference pixel is less than threshold,
8351 it will be considered as banded.
8354 Banding detection range in pixels. Default is 16. If positive, random number
8355 in range 0 to set value will be used. If negative, exact absolute value
8357 The range defines square of four pixels around current pixel.
8360 Set direction in radians from which four pixel will be compared. If positive,
8361 random direction from 0 to set direction will be picked. If negative, exact of
8362 absolute value will be picked. For example direction 0, -PI or -2*PI radians
8363 will pick only pixels on same row and -PI/2 will pick only pixels on same
8367 If enabled, current pixel is compared with average value of all four
8368 surrounding pixels. The default is enabled. If disabled current pixel is
8369 compared with all four surrounding pixels. The pixel is considered banded
8370 if only all four differences with surrounding pixels are less than threshold.
8373 If enabled, current pixel is changed if and only if all pixel components are banded,
8374 e.g. banding detection threshold is triggered for all color components.
8375 The default is disabled.
8380 Remove blocking artifacts from input video.
8382 The filter accepts the following options:
8386 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
8387 This controls what kind of deblocking is applied.
8390 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
8396 Set blocking detection thresholds. Allowed range is 0 to 1.
8397 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
8398 Using higher threshold gives more deblocking strength.
8399 Setting @var{alpha} controls threshold detection at exact edge of block.
8400 Remaining options controls threshold detection near the edge. Each one for
8401 below/above or left/right. Setting any of those to @var{0} disables
8405 Set planes to filter. Default is to filter all available planes.
8408 @subsection Examples
8412 Deblock using weak filter and block size of 4 pixels.
8414 deblock=filter=weak:block=4
8418 Deblock using strong filter, block size of 4 pixels and custom thresholds for
8419 deblocking more edges.
8421 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
8425 Similar as above, but filter only first plane.
8427 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
8431 Similar as above, but filter only second and third plane.
8433 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
8440 Drop duplicated frames at regular intervals.
8442 The filter accepts the following options:
8446 Set the number of frames from which one will be dropped. Setting this to
8447 @var{N} means one frame in every batch of @var{N} frames will be dropped.
8448 Default is @code{5}.
8451 Set the threshold for duplicate detection. If the difference metric for a frame
8452 is less than or equal to this value, then it is declared as duplicate. Default
8456 Set scene change threshold. Default is @code{15}.
8460 Set the size of the x and y-axis blocks used during metric calculations.
8461 Larger blocks give better noise suppression, but also give worse detection of
8462 small movements. Must be a power of two. Default is @code{32}.
8465 Mark main input as a pre-processed input and activate clean source input
8466 stream. This allows the input to be pre-processed with various filters to help
8467 the metrics calculation while keeping the frame selection lossless. When set to
8468 @code{1}, the first stream is for the pre-processed input, and the second
8469 stream is the clean source from where the kept frames are chosen. Default is
8473 Set whether or not chroma is considered in the metric calculations. Default is
8479 Apply 2D deconvolution of video stream in frequency domain using second stream
8482 The filter accepts the following options:
8486 Set which planes to process.
8489 Set which impulse video frames will be processed, can be @var{first}
8490 or @var{all}. Default is @var{all}.
8493 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
8494 and height are not same and not power of 2 or if stream prior to convolving
8498 The @code{deconvolve} filter also supports the @ref{framesync} options.
8502 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
8504 It accepts the following options:
8508 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
8509 @var{rainbows} for cross-color reduction.
8512 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
8515 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
8518 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
8521 Set temporal chroma threshold. Lower values increases reduction of cross-color.
8526 Apply deflate effect to the video.
8528 This filter replaces the pixel by the local(3x3) average by taking into account
8529 only values lower than the pixel.
8531 It accepts the following options:
8538 Limit the maximum change for each plane, default is 65535.
8539 If 0, plane will remain unchanged.
8544 Remove temporal frame luminance variations.
8546 It accepts the following options:
8550 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
8553 Set averaging mode to smooth temporal luminance variations.
8555 Available values are:
8580 Do not actually modify frame. Useful when one only wants metadata.
8585 Remove judder produced by partially interlaced telecined content.
8587 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
8588 source was partially telecined content then the output of @code{pullup,dejudder}
8589 will have a variable frame rate. May change the recorded frame rate of the
8590 container. Aside from that change, this filter will not affect constant frame
8593 The option available in this filter is:
8597 Specify the length of the window over which the judder repeats.
8599 Accepts any integer greater than 1. Useful values are:
8603 If the original was telecined from 24 to 30 fps (Film to NTSC).
8606 If the original was telecined from 25 to 30 fps (PAL to NTSC).
8609 If a mixture of the two.
8612 The default is @samp{4}.
8617 Suppress a TV station logo by a simple interpolation of the surrounding
8618 pixels. Just set a rectangle covering the logo and watch it disappear
8619 (and sometimes something even uglier appear - your mileage may vary).
8621 It accepts the following parameters:
8626 Specify the top left corner coordinates of the logo. They must be
8631 Specify the width and height of the logo to clear. They must be
8635 Specify the thickness of the fuzzy edge of the rectangle (added to
8636 @var{w} and @var{h}). The default value is 1. This option is
8637 deprecated, setting higher values should no longer be necessary and
8641 When set to 1, a green rectangle is drawn on the screen to simplify
8642 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
8643 The default value is 0.
8645 The rectangle is drawn on the outermost pixels which will be (partly)
8646 replaced with interpolated values. The values of the next pixels
8647 immediately outside this rectangle in each direction will be used to
8648 compute the interpolated pixel values inside the rectangle.
8652 @subsection Examples
8656 Set a rectangle covering the area with top left corner coordinates 0,0
8657 and size 100x77, and a band of size 10:
8659 delogo=x=0:y=0:w=100:h=77:band=10
8666 Remove the rain in the input image/video by applying the derain methods based on
8667 convolutional neural networks. Supported models:
8671 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
8672 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
8675 Training as well as model generation scripts are provided in
8676 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
8678 Native model files (.model) can be generated from TensorFlow model
8679 files (.pb) by using tools/python/convert.py
8681 The filter accepts the following options:
8685 Specify which filter to use. This option accepts the following values:
8689 Derain filter. To conduct derain filter, you need to use a derain model.
8692 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
8694 Default value is @samp{derain}.
8697 Specify which DNN backend to use for model loading and execution. This option accepts
8698 the following values:
8702 Native implementation of DNN loading and execution.
8705 TensorFlow backend. To enable this backend you
8706 need to install the TensorFlow for C library (see
8707 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
8708 @code{--enable-libtensorflow}
8710 Default value is @samp{native}.
8713 Set path to model file specifying network architecture and its parameters.
8714 Note that different backends use different file formats. TensorFlow and native
8715 backend can load files for only its format.
8720 Attempt to fix small changes in horizontal and/or vertical shift. This
8721 filter helps remove camera shake from hand-holding a camera, bumping a
8722 tripod, moving on a vehicle, etc.
8724 The filter accepts the following options:
8732 Specify a rectangular area where to limit the search for motion
8734 If desired the search for motion vectors can be limited to a
8735 rectangular area of the frame defined by its top left corner, width
8736 and height. These parameters have the same meaning as the drawbox
8737 filter which can be used to visualise the position of the bounding
8740 This is useful when simultaneous movement of subjects within the frame
8741 might be confused for camera motion by the motion vector search.
8743 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
8744 then the full frame is used. This allows later options to be set
8745 without specifying the bounding box for the motion vector search.
8747 Default - search the whole frame.
8751 Specify the maximum extent of movement in x and y directions in the
8752 range 0-64 pixels. Default 16.
8755 Specify how to generate pixels to fill blanks at the edge of the
8756 frame. Available values are:
8759 Fill zeroes at blank locations
8761 Original image at blank locations
8763 Extruded edge value at blank locations
8765 Mirrored edge at blank locations
8767 Default value is @samp{mirror}.
8770 Specify the blocksize to use for motion search. Range 4-128 pixels,
8774 Specify the contrast threshold for blocks. Only blocks with more than
8775 the specified contrast (difference between darkest and lightest
8776 pixels) will be considered. Range 1-255, default 125.
8779 Specify the search strategy. Available values are:
8782 Set exhaustive search
8784 Set less exhaustive search.
8786 Default value is @samp{exhaustive}.
8789 If set then a detailed log of the motion search is written to the
8796 Remove unwanted contamination of foreground colors, caused by reflected color of
8797 greenscreen or bluescreen.
8799 This filter accepts the following options:
8803 Set what type of despill to use.
8806 Set how spillmap will be generated.
8809 Set how much to get rid of still remaining spill.
8812 Controls amount of red in spill area.
8815 Controls amount of green in spill area.
8816 Should be -1 for greenscreen.
8819 Controls amount of blue in spill area.
8820 Should be -1 for bluescreen.
8823 Controls brightness of spill area, preserving colors.
8826 Modify alpha from generated spillmap.
8831 Apply an exact inverse of the telecine operation. It requires a predefined
8832 pattern specified using the pattern option which must be the same as that passed
8833 to the telecine filter.
8835 This filter accepts the following options:
8844 The default value is @code{top}.
8848 A string of numbers representing the pulldown pattern you wish to apply.
8849 The default value is @code{23}.
8852 A number representing position of the first frame with respect to the telecine
8853 pattern. This is to be used if the stream is cut. The default value is @code{0}.
8858 Apply dilation effect to the video.
8860 This filter replaces the pixel by the local(3x3) maximum.
8862 It accepts the following options:
8869 Limit the maximum change for each plane, default is 65535.
8870 If 0, plane will remain unchanged.
8873 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
8876 Flags to local 3x3 coordinates maps like this:
8885 Displace pixels as indicated by second and third input stream.
8887 It takes three input streams and outputs one stream, the first input is the
8888 source, and second and third input are displacement maps.
8890 The second input specifies how much to displace pixels along the
8891 x-axis, while the third input specifies how much to displace pixels
8893 If one of displacement map streams terminates, last frame from that
8894 displacement map will be used.
8896 Note that once generated, displacements maps can be reused over and over again.
8898 A description of the accepted options follows.
8902 Set displace behavior for pixels that are out of range.
8904 Available values are:
8907 Missing pixels are replaced by black pixels.
8910 Adjacent pixels will spread out to replace missing pixels.
8913 Out of range pixels are wrapped so they point to pixels of other side.
8916 Out of range pixels will be replaced with mirrored pixels.
8918 Default is @samp{smear}.
8922 @subsection Examples
8926 Add ripple effect to rgb input of video size hd720:
8928 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
8932 Add wave effect to rgb input of video size hd720:
8934 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
8938 @section dnn_processing
8940 Do image processing with deep neural networks. Currently only AVFrame with RGB24
8941 and BGR24 are supported, more formats will be added later.
8943 The filter accepts the following options:
8947 Specify which DNN backend to use for model loading and execution. This option accepts
8948 the following values:
8952 Native implementation of DNN loading and execution.
8955 TensorFlow backend. To enable this backend you
8956 need to install the TensorFlow for C library (see
8957 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
8958 @code{--enable-libtensorflow}
8961 Default value is @samp{native}.
8964 Set path to model file specifying network architecture and its parameters.
8965 Note that different backends use different file formats. TensorFlow and native
8966 backend can load files for only its format.
8968 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
8971 Set the input name of the dnn network.
8974 Set the output name of the dnn network.
8977 Set the pixel format for the Frame. Allowed values are @code{AV_PIX_FMT_RGB24}, and @code{AV_PIX_FMT_BGR24}.
8978 Default value is @code{AV_PIX_FMT_RGB24}.
8984 Draw a colored box on the input image.
8986 It accepts the following parameters:
8991 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
8995 The expressions which specify the width and height of the box; if 0 they are interpreted as
8996 the input width and height. It defaults to 0.
8999 Specify the color of the box to write. For the general syntax of this option,
9000 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9001 value @code{invert} is used, the box edge color is the same as the
9002 video with inverted luma.
9005 The expression which sets the thickness of the box edge.
9006 A value of @code{fill} will create a filled box. Default value is @code{3}.
9008 See below for the list of accepted constants.
9011 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9012 will overwrite the video's color and alpha pixels.
9013 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9016 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9017 following constants:
9021 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9025 horizontal and vertical chroma subsample values. For example for the
9026 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9030 The input width and height.
9033 The input sample aspect ratio.
9037 The x and y offset coordinates where the box is drawn.
9041 The width and height of the drawn box.
9044 The thickness of the drawn box.
9046 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9047 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9051 @subsection Examples
9055 Draw a black box around the edge of the input image:
9061 Draw a box with color red and an opacity of 50%:
9063 drawbox=10:20:200:60:red@@0.5
9066 The previous example can be specified as:
9068 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9072 Fill the box with pink color:
9074 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9078 Draw a 2-pixel red 2.40:1 mask:
9080 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
9084 @subsection Commands
9085 This filter supports same commands as options.
9086 The command accepts the same syntax of the corresponding option.
9088 If the specified expression is not valid, it is kept at its current
9093 Draw a graph using input video metadata.
9095 It accepts the following parameters:
9099 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9102 Set 1st foreground color expression.
9105 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9108 Set 2nd foreground color expression.
9111 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9114 Set 3rd foreground color expression.
9117 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9120 Set 4th foreground color expression.
9123 Set minimal value of metadata value.
9126 Set maximal value of metadata value.
9129 Set graph background color. Default is white.
9134 Available values for mode is:
9141 Default is @code{line}.
9146 Available values for slide is:
9149 Draw new frame when right border is reached.
9152 Replace old columns with new ones.
9155 Scroll from right to left.
9158 Scroll from left to right.
9161 Draw single picture.
9164 Default is @code{frame}.
9167 Set size of graph video. For the syntax of this option, check the
9168 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9169 The default value is @code{900x256}.
9171 The foreground color expressions can use the following variables:
9174 Minimal value of metadata value.
9177 Maximal value of metadata value.
9180 Current metadata key value.
9183 The color is defined as 0xAABBGGRR.
9186 Example using metadata from @ref{signalstats} filter:
9188 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9191 Example using metadata from @ref{ebur128} filter:
9193 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9198 Draw a grid on the input image.
9200 It accepts the following parameters:
9205 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9209 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9210 input width and height, respectively, minus @code{thickness}, so image gets
9211 framed. Default to 0.
9214 Specify the color of the grid. For the general syntax of this option,
9215 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9216 value @code{invert} is used, the grid color is the same as the
9217 video with inverted luma.
9220 The expression which sets the thickness of the grid line. Default value is @code{1}.
9222 See below for the list of accepted constants.
9225 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9226 will overwrite the video's color and alpha pixels.
9227 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9230 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9231 following constants:
9235 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9239 horizontal and vertical chroma subsample values. For example for the
9240 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9244 The input grid cell width and height.
9247 The input sample aspect ratio.
9251 The x and y coordinates of some point of grid intersection (meant to configure offset).
9255 The width and height of the drawn cell.
9258 The thickness of the drawn cell.
9260 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9261 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9265 @subsection Examples
9269 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9271 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9275 Draw a white 3x3 grid with an opacity of 50%:
9277 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
9281 @subsection Commands
9282 This filter supports same commands as options.
9283 The command accepts the same syntax of the corresponding option.
9285 If the specified expression is not valid, it is kept at its current
9291 Draw a text string or text from a specified file on top of a video, using the
9292 libfreetype library.
9294 To enable compilation of this filter, you need to configure FFmpeg with
9295 @code{--enable-libfreetype}.
9296 To enable default font fallback and the @var{font} option you need to
9297 configure FFmpeg with @code{--enable-libfontconfig}.
9298 To enable the @var{text_shaping} option, you need to configure FFmpeg with
9299 @code{--enable-libfribidi}.
9303 It accepts the following parameters:
9308 Used to draw a box around text using the background color.
9309 The value must be either 1 (enable) or 0 (disable).
9310 The default value of @var{box} is 0.
9313 Set the width of the border to be drawn around the box using @var{boxcolor}.
9314 The default value of @var{boxborderw} is 0.
9317 The color to be used for drawing box around text. For the syntax of this
9318 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9320 The default value of @var{boxcolor} is "white".
9323 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
9324 The default value of @var{line_spacing} is 0.
9327 Set the width of the border to be drawn around the text using @var{bordercolor}.
9328 The default value of @var{borderw} is 0.
9331 Set the color to be used for drawing border around text. For the syntax of this
9332 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9334 The default value of @var{bordercolor} is "black".
9337 Select how the @var{text} is expanded. Can be either @code{none},
9338 @code{strftime} (deprecated) or
9339 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
9343 Set a start time for the count. Value is in microseconds. Only applied
9344 in the deprecated strftime expansion mode. To emulate in normal expansion
9345 mode use the @code{pts} function, supplying the start time (in seconds)
9346 as the second argument.
9349 If true, check and fix text coords to avoid clipping.
9352 The color to be used for drawing fonts. For the syntax of this option, check
9353 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9355 The default value of @var{fontcolor} is "black".
9357 @item fontcolor_expr
9358 String which is expanded the same way as @var{text} to obtain dynamic
9359 @var{fontcolor} value. By default this option has empty value and is not
9360 processed. When this option is set, it overrides @var{fontcolor} option.
9363 The font family to be used for drawing text. By default Sans.
9366 The font file to be used for drawing text. The path must be included.
9367 This parameter is mandatory if the fontconfig support is disabled.
9370 Draw the text applying alpha blending. The value can
9371 be a number between 0.0 and 1.0.
9372 The expression accepts the same variables @var{x, y} as well.
9373 The default value is 1.
9374 Please see @var{fontcolor_expr}.
9377 The font size to be used for drawing text.
9378 The default value of @var{fontsize} is 16.
9381 If set to 1, attempt to shape the text (for example, reverse the order of
9382 right-to-left text and join Arabic characters) before drawing it.
9383 Otherwise, just draw the text exactly as given.
9384 By default 1 (if supported).
9387 The flags to be used for loading the fonts.
9389 The flags map the corresponding flags supported by libfreetype, and are
9390 a combination of the following values:
9397 @item vertical_layout
9398 @item force_autohint
9401 @item ignore_global_advance_width
9403 @item ignore_transform
9409 Default value is "default".
9411 For more information consult the documentation for the FT_LOAD_*
9415 The color to be used for drawing a shadow behind the drawn text. For the
9416 syntax of this option, check the @ref{color syntax,,"Color" section in the
9417 ffmpeg-utils manual,ffmpeg-utils}.
9419 The default value of @var{shadowcolor} is "black".
9423 The x and y offsets for the text shadow position with respect to the
9424 position of the text. They can be either positive or negative
9425 values. The default value for both is "0".
9428 The starting frame number for the n/frame_num variable. The default value
9432 The size in number of spaces to use for rendering the tab.
9436 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
9437 format. It can be used with or without text parameter. @var{timecode_rate}
9438 option must be specified.
9440 @item timecode_rate, rate, r
9441 Set the timecode frame rate (timecode only). Value will be rounded to nearest
9442 integer. Minimum value is "1".
9443 Drop-frame timecode is supported for frame rates 30 & 60.
9446 If set to 1, the output of the timecode option will wrap around at 24 hours.
9447 Default is 0 (disabled).
9450 The text string to be drawn. The text must be a sequence of UTF-8
9452 This parameter is mandatory if no file is specified with the parameter
9456 A text file containing text to be drawn. The text must be a sequence
9457 of UTF-8 encoded characters.
9459 This parameter is mandatory if no text string is specified with the
9460 parameter @var{text}.
9462 If both @var{text} and @var{textfile} are specified, an error is thrown.
9465 If set to 1, the @var{textfile} will be reloaded before each frame.
9466 Be sure to update it atomically, or it may be read partially, or even fail.
9470 The expressions which specify the offsets where text will be drawn
9471 within the video frame. They are relative to the top/left border of the
9474 The default value of @var{x} and @var{y} is "0".
9476 See below for the list of accepted constants and functions.
9479 The parameters for @var{x} and @var{y} are expressions containing the
9480 following constants and functions:
9484 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
9488 horizontal and vertical chroma subsample values. For example for the
9489 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9492 the height of each text line
9500 @item max_glyph_a, ascent
9501 the maximum distance from the baseline to the highest/upper grid
9502 coordinate used to place a glyph outline point, for all the rendered
9504 It is a positive value, due to the grid's orientation with the Y axis
9507 @item max_glyph_d, descent
9508 the maximum distance from the baseline to the lowest grid coordinate
9509 used to place a glyph outline point, for all the rendered glyphs.
9510 This is a negative value, due to the grid's orientation, with the Y axis
9514 maximum glyph height, that is the maximum height for all the glyphs
9515 contained in the rendered text, it is equivalent to @var{ascent} -
9519 maximum glyph width, that is the maximum width for all the glyphs
9520 contained in the rendered text
9523 the number of input frame, starting from 0
9525 @item rand(min, max)
9526 return a random number included between @var{min} and @var{max}
9529 The input sample aspect ratio.
9532 timestamp expressed in seconds, NAN if the input timestamp is unknown
9535 the height of the rendered text
9538 the width of the rendered text
9542 the x and y offset coordinates where the text is drawn.
9544 These parameters allow the @var{x} and @var{y} expressions to refer
9545 to each other, so you can for example specify @code{y=x/dar}.
9548 A one character description of the current frame's picture type.
9551 The current packet's position in the input file or stream
9552 (in bytes, from the start of the input). A value of -1 indicates
9553 this info is not available.
9556 The current packet's duration, in seconds.
9559 The current packet's size (in bytes).
9562 @anchor{drawtext_expansion}
9563 @subsection Text expansion
9565 If @option{expansion} is set to @code{strftime},
9566 the filter recognizes strftime() sequences in the provided text and
9567 expands them accordingly. Check the documentation of strftime(). This
9568 feature is deprecated.
9570 If @option{expansion} is set to @code{none}, the text is printed verbatim.
9572 If @option{expansion} is set to @code{normal} (which is the default),
9573 the following expansion mechanism is used.
9575 The backslash character @samp{\}, followed by any character, always expands to
9576 the second character.
9578 Sequences of the form @code{%@{...@}} are expanded. The text between the
9579 braces is a function name, possibly followed by arguments separated by ':'.
9580 If the arguments contain special characters or delimiters (':' or '@}'),
9581 they should be escaped.
9583 Note that they probably must also be escaped as the value for the
9584 @option{text} option in the filter argument string and as the filter
9585 argument in the filtergraph description, and possibly also for the shell,
9586 that makes up to four levels of escaping; using a text file avoids these
9589 The following functions are available:
9594 The expression evaluation result.
9596 It must take one argument specifying the expression to be evaluated,
9597 which accepts the same constants and functions as the @var{x} and
9598 @var{y} values. Note that not all constants should be used, for
9599 example the text size is not known when evaluating the expression, so
9600 the constants @var{text_w} and @var{text_h} will have an undefined
9603 @item expr_int_format, eif
9604 Evaluate the expression's value and output as formatted integer.
9606 The first argument is the expression to be evaluated, just as for the @var{expr} function.
9607 The second argument specifies the output format. Allowed values are @samp{x},
9608 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
9609 @code{printf} function.
9610 The third parameter is optional and sets the number of positions taken by the output.
9611 It can be used to add padding with zeros from the left.
9614 The time at which the filter is running, expressed in UTC.
9615 It can accept an argument: a strftime() format string.
9618 The time at which the filter is running, expressed in the local time zone.
9619 It can accept an argument: a strftime() format string.
9622 Frame metadata. Takes one or two arguments.
9624 The first argument is mandatory and specifies the metadata key.
9626 The second argument is optional and specifies a default value, used when the
9627 metadata key is not found or empty.
9629 Available metadata can be identified by inspecting entries
9630 starting with TAG included within each frame section
9631 printed by running @code{ffprobe -show_frames}.
9633 String metadata generated in filters leading to
9634 the drawtext filter are also available.
9637 The frame number, starting from 0.
9640 A one character description of the current picture type.
9643 The timestamp of the current frame.
9644 It can take up to three arguments.
9646 The first argument is the format of the timestamp; it defaults to @code{flt}
9647 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
9648 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
9649 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
9650 @code{localtime} stands for the timestamp of the frame formatted as
9651 local time zone time.
9653 The second argument is an offset added to the timestamp.
9655 If the format is set to @code{hms}, a third argument @code{24HH} may be
9656 supplied to present the hour part of the formatted timestamp in 24h format
9659 If the format is set to @code{localtime} or @code{gmtime},
9660 a third argument may be supplied: a strftime() format string.
9661 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
9664 @subsection Commands
9666 This filter supports altering parameters via commands:
9669 Alter existing filter parameters.
9671 Syntax for the argument is the same as for filter invocation, e.g.
9674 fontsize=56:fontcolor=green:text='Hello World'
9677 Full filter invocation with sendcmd would look like this:
9680 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
9684 If the entire argument can't be parsed or applied as valid values then the filter will
9685 continue with its existing parameters.
9687 @subsection Examples
9691 Draw "Test Text" with font FreeSerif, using the default values for the
9692 optional parameters.
9695 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
9699 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
9700 and y=50 (counting from the top-left corner of the screen), text is
9701 yellow with a red box around it. Both the text and the box have an
9705 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
9706 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
9709 Note that the double quotes are not necessary if spaces are not used
9710 within the parameter list.
9713 Show the text at the center of the video frame:
9715 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
9719 Show the text at a random position, switching to a new position every 30 seconds:
9721 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)"
9725 Show a text line sliding from right to left in the last row of the video
9726 frame. The file @file{LONG_LINE} is assumed to contain a single line
9729 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
9733 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
9735 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
9739 Draw a single green letter "g", at the center of the input video.
9740 The glyph baseline is placed at half screen height.
9742 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
9746 Show text for 1 second every 3 seconds:
9748 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
9752 Use fontconfig to set the font. Note that the colons need to be escaped.
9754 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
9758 Print the date of a real-time encoding (see strftime(3)):
9760 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
9764 Show text fading in and out (appearing/disappearing):
9767 DS=1.0 # display start
9768 DE=10.0 # display end
9769 FID=1.5 # fade in duration
9770 FOD=5 # fade out duration
9771 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 @}"
9775 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
9776 and the @option{fontsize} value are included in the @option{y} offset.
9778 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
9779 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
9784 For more information about libfreetype, check:
9785 @url{http://www.freetype.org/}.
9787 For more information about fontconfig, check:
9788 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
9790 For more information about libfribidi, check:
9791 @url{http://fribidi.org/}.
9795 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
9797 The filter accepts the following options:
9802 Set low and high threshold values used by the Canny thresholding
9805 The high threshold selects the "strong" edge pixels, which are then
9806 connected through 8-connectivity with the "weak" edge pixels selected
9807 by the low threshold.
9809 @var{low} and @var{high} threshold values must be chosen in the range
9810 [0,1], and @var{low} should be lesser or equal to @var{high}.
9812 Default value for @var{low} is @code{20/255}, and default value for @var{high}
9816 Define the drawing mode.
9820 Draw white/gray wires on black background.
9823 Mix the colors to create a paint/cartoon effect.
9826 Apply Canny edge detector on all selected planes.
9828 Default value is @var{wires}.
9831 Select planes for filtering. By default all available planes are filtered.
9834 @subsection Examples
9838 Standard edge detection with custom values for the hysteresis thresholding:
9840 edgedetect=low=0.1:high=0.4
9844 Painting effect without thresholding:
9846 edgedetect=mode=colormix:high=0
9852 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
9854 For each input image, the filter will compute the optimal mapping from
9855 the input to the output given the codebook length, that is the number
9856 of distinct output colors.
9858 This filter accepts the following options.
9861 @item codebook_length, l
9862 Set codebook length. The value must be a positive integer, and
9863 represents the number of distinct output colors. Default value is 256.
9866 Set the maximum number of iterations to apply for computing the optimal
9867 mapping. The higher the value the better the result and the higher the
9868 computation time. Default value is 1.
9871 Set a random seed, must be an integer included between 0 and
9872 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
9873 will try to use a good random seed on a best effort basis.
9876 Set pal8 output pixel format. This option does not work with codebook
9877 length greater than 256.
9882 Measure graylevel entropy in histogram of color channels of video frames.
9884 It accepts the following parameters:
9888 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
9890 @var{diff} mode measures entropy of histogram delta values, absolute differences
9891 between neighbour histogram values.
9895 Set brightness, contrast, saturation and approximate gamma adjustment.
9897 The filter accepts the following options:
9901 Set the contrast expression. The value must be a float value in range
9902 @code{-1000.0} to @code{1000.0}. The default value is "1".
9905 Set the brightness expression. The value must be a float value in
9906 range @code{-1.0} to @code{1.0}. The default value is "0".
9909 Set the saturation expression. The value must be a float in
9910 range @code{0.0} to @code{3.0}. The default value is "1".
9913 Set the gamma expression. The value must be a float in range
9914 @code{0.1} to @code{10.0}. The default value is "1".
9917 Set the gamma expression for red. The value must be a float in
9918 range @code{0.1} to @code{10.0}. The default value is "1".
9921 Set the gamma expression for green. The value must be a float in range
9922 @code{0.1} to @code{10.0}. The default value is "1".
9925 Set the gamma expression for blue. The value must be a float in range
9926 @code{0.1} to @code{10.0}. The default value is "1".
9929 Set the gamma weight expression. It can be used to reduce the effect
9930 of a high gamma value on bright image areas, e.g. keep them from
9931 getting overamplified and just plain white. The value must be a float
9932 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
9933 gamma correction all the way down while @code{1.0} leaves it at its
9934 full strength. Default is "1".
9937 Set when the expressions for brightness, contrast, saturation and
9938 gamma expressions are evaluated.
9940 It accepts the following values:
9943 only evaluate expressions once during the filter initialization or
9944 when a command is processed
9947 evaluate expressions for each incoming frame
9950 Default value is @samp{init}.
9953 The expressions accept the following parameters:
9956 frame count of the input frame starting from 0
9959 byte position of the corresponding packet in the input file, NAN if
9963 frame rate of the input video, NAN if the input frame rate is unknown
9966 timestamp expressed in seconds, NAN if the input timestamp is unknown
9969 @subsection Commands
9970 The filter supports the following commands:
9974 Set the contrast expression.
9977 Set the brightness expression.
9980 Set the saturation expression.
9983 Set the gamma expression.
9986 Set the gamma_r expression.
9989 Set gamma_g expression.
9992 Set gamma_b expression.
9995 Set gamma_weight expression.
9997 The command accepts the same syntax of the corresponding option.
9999 If the specified expression is not valid, it is kept at its current
10006 Apply erosion effect to the video.
10008 This filter replaces the pixel by the local(3x3) minimum.
10010 It accepts the following options:
10017 Limit the maximum change for each plane, default is 65535.
10018 If 0, plane will remain unchanged.
10021 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10024 Flags to local 3x3 coordinates maps like this:
10031 @section extractplanes
10033 Extract color channel components from input video stream into
10034 separate grayscale video streams.
10036 The filter accepts the following option:
10040 Set plane(s) to extract.
10042 Available values for planes are:
10053 Choosing planes not available in the input will result in an error.
10054 That means you cannot select @code{r}, @code{g}, @code{b} planes
10055 with @code{y}, @code{u}, @code{v} planes at same time.
10058 @subsection Examples
10062 Extract luma, u and v color channel component from input video frame
10063 into 3 grayscale outputs:
10065 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
10071 Apply a fade-in/out effect to the input video.
10073 It accepts the following parameters:
10077 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10079 Default is @code{in}.
10081 @item start_frame, s
10082 Specify the number of the frame to start applying the fade
10083 effect at. Default is 0.
10086 The number of frames that the fade effect lasts. At the end of the
10087 fade-in effect, the output video will have the same intensity as the input video.
10088 At the end of the fade-out transition, the output video will be filled with the
10089 selected @option{color}.
10093 If set to 1, fade only alpha channel, if one exists on the input.
10094 Default value is 0.
10096 @item start_time, st
10097 Specify the timestamp (in seconds) of the frame to start to apply the fade
10098 effect. If both start_frame and start_time are specified, the fade will start at
10099 whichever comes last. Default is 0.
10102 The number of seconds for which the fade effect has to last. At the end of the
10103 fade-in effect the output video will have the same intensity as the input video,
10104 at the end of the fade-out transition the output video will be filled with the
10105 selected @option{color}.
10106 If both duration and nb_frames are specified, duration is used. Default is 0
10107 (nb_frames is used by default).
10110 Specify the color of the fade. Default is "black".
10113 @subsection Examples
10117 Fade in the first 30 frames of video:
10122 The command above is equivalent to:
10128 Fade out the last 45 frames of a 200-frame video:
10131 fade=type=out:start_frame=155:nb_frames=45
10135 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10137 fade=in:0:25, fade=out:975:25
10141 Make the first 5 frames yellow, then fade in from frame 5-24:
10143 fade=in:5:20:color=yellow
10147 Fade in alpha over first 25 frames of video:
10149 fade=in:0:25:alpha=1
10153 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10155 fade=t=in:st=5.5:d=0.5
10161 Denoise frames using 3D FFT (frequency domain filtering).
10163 The filter accepts the following options:
10167 Set the noise sigma constant. This sets denoising strength.
10168 Default value is 1. Allowed range is from 0 to 30.
10169 Using very high sigma with low overlap may give blocking artifacts.
10172 Set amount of denoising. By default all detected noise is reduced.
10173 Default value is 1. Allowed range is from 0 to 1.
10176 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10177 Actual size of block in pixels is 2 to power of @var{block}, so by default
10178 block size in pixels is 2^4 which is 16.
10181 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10184 Set number of previous frames to use for denoising. By default is set to 0.
10187 Set number of next frames to to use for denoising. By default is set to 0.
10190 Set planes which will be filtered, by default are all available filtered
10195 Apply arbitrary expressions to samples in frequency domain
10199 Adjust the dc value (gain) of the luma plane of the image. The filter
10200 accepts an integer value in range @code{0} to @code{1000}. The default
10201 value is set to @code{0}.
10204 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10205 filter accepts an integer value in range @code{0} to @code{1000}. The
10206 default value is set to @code{0}.
10209 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10210 filter accepts an integer value in range @code{0} to @code{1000}. The
10211 default value is set to @code{0}.
10214 Set the frequency domain weight expression for the luma plane.
10217 Set the frequency domain weight expression for the 1st chroma plane.
10220 Set the frequency domain weight expression for the 2nd chroma plane.
10223 Set when the expressions are evaluated.
10225 It accepts the following values:
10228 Only evaluate expressions once during the filter initialization.
10231 Evaluate expressions for each incoming frame.
10234 Default value is @samp{init}.
10236 The filter accepts the following variables:
10239 The coordinates of the current sample.
10243 The width and height of the image.
10246 The number of input frame, starting from 0.
10249 @subsection Examples
10255 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
10261 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
10267 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
10273 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
10280 Extract a single field from an interlaced image using stride
10281 arithmetic to avoid wasting CPU time. The output frames are marked as
10284 The filter accepts the following options:
10288 Specify whether to extract the top (if the value is @code{0} or
10289 @code{top}) or the bottom field (if the value is @code{1} or
10295 Create new frames by copying the top and bottom fields from surrounding frames
10296 supplied as numbers by the hint file.
10300 Set file containing hints: absolute/relative frame numbers.
10302 There must be one line for each frame in a clip. Each line must contain two
10303 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
10304 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
10305 is current frame number for @code{absolute} mode or out of [-1, 1] range
10306 for @code{relative} mode. First number tells from which frame to pick up top
10307 field and second number tells from which frame to pick up bottom field.
10309 If optionally followed by @code{+} output frame will be marked as interlaced,
10310 else if followed by @code{-} output frame will be marked as progressive, else
10311 it will be marked same as input frame.
10312 If optionally followed by @code{t} output frame will use only top field, or in
10313 case of @code{b} it will use only bottom field.
10314 If line starts with @code{#} or @code{;} that line is skipped.
10317 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
10320 Example of first several lines of @code{hint} file for @code{relative} mode:
10322 0,0 - # first frame
10323 1,0 - # second frame, use third's frame top field and second's frame bottom field
10324 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
10339 @section fieldmatch
10341 Field matching filter for inverse telecine. It is meant to reconstruct the
10342 progressive frames from a telecined stream. The filter does not drop duplicated
10343 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
10344 followed by a decimation filter such as @ref{decimate} in the filtergraph.
10346 The separation of the field matching and the decimation is notably motivated by
10347 the possibility of inserting a de-interlacing filter fallback between the two.
10348 If the source has mixed telecined and real interlaced content,
10349 @code{fieldmatch} will not be able to match fields for the interlaced parts.
10350 But these remaining combed frames will be marked as interlaced, and thus can be
10351 de-interlaced by a later filter such as @ref{yadif} before decimation.
10353 In addition to the various configuration options, @code{fieldmatch} can take an
10354 optional second stream, activated through the @option{ppsrc} option. If
10355 enabled, the frames reconstruction will be based on the fields and frames from
10356 this second stream. This allows the first input to be pre-processed in order to
10357 help the various algorithms of the filter, while keeping the output lossless
10358 (assuming the fields are matched properly). Typically, a field-aware denoiser,
10359 or brightness/contrast adjustments can help.
10361 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
10362 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
10363 which @code{fieldmatch} is based on. While the semantic and usage are very
10364 close, some behaviour and options names can differ.
10366 The @ref{decimate} filter currently only works for constant frame rate input.
10367 If your input has mixed telecined (30fps) and progressive content with a lower
10368 framerate like 24fps use the following filterchain to produce the necessary cfr
10369 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
10371 The filter accepts the following options:
10375 Specify the assumed field order of the input stream. Available values are:
10379 Auto detect parity (use FFmpeg's internal parity value).
10381 Assume bottom field first.
10383 Assume top field first.
10386 Note that it is sometimes recommended not to trust the parity announced by the
10389 Default value is @var{auto}.
10392 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
10393 sense that it won't risk creating jerkiness due to duplicate frames when
10394 possible, but if there are bad edits or blended fields it will end up
10395 outputting combed frames when a good match might actually exist. On the other
10396 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
10397 but will almost always find a good frame if there is one. The other values are
10398 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
10399 jerkiness and creating duplicate frames versus finding good matches in sections
10400 with bad edits, orphaned fields, blended fields, etc.
10402 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
10404 Available values are:
10408 2-way matching (p/c)
10410 2-way matching, and trying 3rd match if still combed (p/c + n)
10412 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
10414 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
10415 still combed (p/c + n + u/b)
10417 3-way matching (p/c/n)
10419 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
10420 detected as combed (p/c/n + u/b)
10423 The parenthesis at the end indicate the matches that would be used for that
10424 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
10427 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
10430 Default value is @var{pc_n}.
10433 Mark the main input stream as a pre-processed input, and enable the secondary
10434 input stream as the clean source to pick the fields from. See the filter
10435 introduction for more details. It is similar to the @option{clip2} feature from
10438 Default value is @code{0} (disabled).
10441 Set the field to match from. It is recommended to set this to the same value as
10442 @option{order} unless you experience matching failures with that setting. In
10443 certain circumstances changing the field that is used to match from can have a
10444 large impact on matching performance. Available values are:
10448 Automatic (same value as @option{order}).
10450 Match from the bottom field.
10452 Match from the top field.
10455 Default value is @var{auto}.
10458 Set whether or not chroma is included during the match comparisons. In most
10459 cases it is recommended to leave this enabled. You should set this to @code{0}
10460 only if your clip has bad chroma problems such as heavy rainbowing or other
10461 artifacts. Setting this to @code{0} could also be used to speed things up at
10462 the cost of some accuracy.
10464 Default value is @code{1}.
10468 These define an exclusion band which excludes the lines between @option{y0} and
10469 @option{y1} from being included in the field matching decision. An exclusion
10470 band can be used to ignore subtitles, a logo, or other things that may
10471 interfere with the matching. @option{y0} sets the starting scan line and
10472 @option{y1} sets the ending line; all lines in between @option{y0} and
10473 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
10474 @option{y0} and @option{y1} to the same value will disable the feature.
10475 @option{y0} and @option{y1} defaults to @code{0}.
10478 Set the scene change detection threshold as a percentage of maximum change on
10479 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
10480 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
10481 @option{scthresh} is @code{[0.0, 100.0]}.
10483 Default value is @code{12.0}.
10486 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
10487 account the combed scores of matches when deciding what match to use as the
10488 final match. Available values are:
10492 No final matching based on combed scores.
10494 Combed scores are only used when a scene change is detected.
10496 Use combed scores all the time.
10499 Default is @var{sc}.
10502 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
10503 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
10504 Available values are:
10508 No forced calculation.
10510 Force p/c/n calculations.
10512 Force p/c/n/u/b calculations.
10515 Default value is @var{none}.
10518 This is the area combing threshold used for combed frame detection. This
10519 essentially controls how "strong" or "visible" combing must be to be detected.
10520 Larger values mean combing must be more visible and smaller values mean combing
10521 can be less visible or strong and still be detected. Valid settings are from
10522 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
10523 be detected as combed). This is basically a pixel difference value. A good
10524 range is @code{[8, 12]}.
10526 Default value is @code{9}.
10529 Sets whether or not chroma is considered in the combed frame decision. Only
10530 disable this if your source has chroma problems (rainbowing, etc.) that are
10531 causing problems for the combed frame detection with chroma enabled. Actually,
10532 using @option{chroma}=@var{0} is usually more reliable, except for the case
10533 where there is chroma only combing in the source.
10535 Default value is @code{0}.
10539 Respectively set the x-axis and y-axis size of the window used during combed
10540 frame detection. This has to do with the size of the area in which
10541 @option{combpel} pixels are required to be detected as combed for a frame to be
10542 declared combed. See the @option{combpel} parameter description for more info.
10543 Possible values are any number that is a power of 2 starting at 4 and going up
10546 Default value is @code{16}.
10549 The number of combed pixels inside any of the @option{blocky} by
10550 @option{blockx} size blocks on the frame for the frame to be detected as
10551 combed. While @option{cthresh} controls how "visible" the combing must be, this
10552 setting controls "how much" combing there must be in any localized area (a
10553 window defined by the @option{blockx} and @option{blocky} settings) on the
10554 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
10555 which point no frames will ever be detected as combed). This setting is known
10556 as @option{MI} in TFM/VFM vocabulary.
10558 Default value is @code{80}.
10561 @anchor{p/c/n/u/b meaning}
10562 @subsection p/c/n/u/b meaning
10564 @subsubsection p/c/n
10566 We assume the following telecined stream:
10569 Top fields: 1 2 2 3 4
10570 Bottom fields: 1 2 3 4 4
10573 The numbers correspond to the progressive frame the fields relate to. Here, the
10574 first two frames are progressive, the 3rd and 4th are combed, and so on.
10576 When @code{fieldmatch} is configured to run a matching from bottom
10577 (@option{field}=@var{bottom}) this is how this input stream get transformed:
10582 B 1 2 3 4 4 <-- matching reference
10591 As a result of the field matching, we can see that some frames get duplicated.
10592 To perform a complete inverse telecine, you need to rely on a decimation filter
10593 after this operation. See for instance the @ref{decimate} filter.
10595 The same operation now matching from top fields (@option{field}=@var{top})
10600 T 1 2 2 3 4 <-- matching reference
10610 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
10611 basically, they refer to the frame and field of the opposite parity:
10614 @item @var{p} matches the field of the opposite parity in the previous frame
10615 @item @var{c} matches the field of the opposite parity in the current frame
10616 @item @var{n} matches the field of the opposite parity in the next frame
10621 The @var{u} and @var{b} matching are a bit special in the sense that they match
10622 from the opposite parity flag. In the following examples, we assume that we are
10623 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
10624 'x' is placed above and below each matched fields.
10626 With bottom matching (@option{field}=@var{bottom}):
10631 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10632 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10640 With top matching (@option{field}=@var{top}):
10645 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10646 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10654 @subsection Examples
10656 Simple IVTC of a top field first telecined stream:
10658 fieldmatch=order=tff:combmatch=none, decimate
10661 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
10663 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
10666 @section fieldorder
10668 Transform the field order of the input video.
10670 It accepts the following parameters:
10675 The output field order. Valid values are @var{tff} for top field first or @var{bff}
10676 for bottom field first.
10679 The default value is @samp{tff}.
10681 The transformation is done by shifting the picture content up or down
10682 by one line, and filling the remaining line with appropriate picture content.
10683 This method is consistent with most broadcast field order converters.
10685 If the input video is not flagged as being interlaced, or it is already
10686 flagged as being of the required output field order, then this filter does
10687 not alter the incoming video.
10689 It is very useful when converting to or from PAL DV material,
10690 which is bottom field first.
10694 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
10697 @section fifo, afifo
10699 Buffer input images and send them when they are requested.
10701 It is mainly useful when auto-inserted by the libavfilter
10704 It does not take parameters.
10706 @section fillborders
10708 Fill borders of the input video, without changing video stream dimensions.
10709 Sometimes video can have garbage at the four edges and you may not want to
10710 crop video input to keep size multiple of some number.
10712 This filter accepts the following options:
10716 Number of pixels to fill from left border.
10719 Number of pixels to fill from right border.
10722 Number of pixels to fill from top border.
10725 Number of pixels to fill from bottom border.
10730 It accepts the following values:
10733 fill pixels using outermost pixels
10736 fill pixels using mirroring
10739 fill pixels with constant value
10742 Default is @var{smear}.
10745 Set color for pixels in fixed mode. Default is @var{black}.
10750 Find a rectangular object
10752 It accepts the following options:
10756 Filepath of the object image, needs to be in gray8.
10759 Detection threshold, default is 0.5.
10762 Number of mipmaps, default is 3.
10764 @item xmin, ymin, xmax, ymax
10765 Specifies the rectangle in which to search.
10768 @subsection Examples
10772 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
10774 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
10780 Flood area with values of same pixel components with another values.
10782 It accepts the following options:
10785 Set pixel x coordinate.
10788 Set pixel y coordinate.
10791 Set source #0 component value.
10794 Set source #1 component value.
10797 Set source #2 component value.
10800 Set source #3 component value.
10803 Set destination #0 component value.
10806 Set destination #1 component value.
10809 Set destination #2 component value.
10812 Set destination #3 component value.
10818 Convert the input video to one of the specified pixel formats.
10819 Libavfilter will try to pick one that is suitable as input to
10822 It accepts the following parameters:
10826 A '|'-separated list of pixel format names, such as
10827 "pix_fmts=yuv420p|monow|rgb24".
10831 @subsection Examples
10835 Convert the input video to the @var{yuv420p} format
10837 format=pix_fmts=yuv420p
10840 Convert the input video to any of the formats in the list
10842 format=pix_fmts=yuv420p|yuv444p|yuv410p
10849 Convert the video to specified constant frame rate by duplicating or dropping
10850 frames as necessary.
10852 It accepts the following parameters:
10856 The desired output frame rate. The default is @code{25}.
10859 Assume the first PTS should be the given value, in seconds. This allows for
10860 padding/trimming at the start of stream. By default, no assumption is made
10861 about the first frame's expected PTS, so no padding or trimming is done.
10862 For example, this could be set to 0 to pad the beginning with duplicates of
10863 the first frame if a video stream starts after the audio stream or to trim any
10864 frames with a negative PTS.
10867 Timestamp (PTS) rounding method.
10869 Possible values are:
10876 round towards -infinity
10878 round towards +infinity
10882 The default is @code{near}.
10885 Action performed when reading the last frame.
10887 Possible values are:
10890 Use same timestamp rounding method as used for other frames.
10892 Pass through last frame if input duration has not been reached yet.
10894 The default is @code{round}.
10898 Alternatively, the options can be specified as a flat string:
10899 @var{fps}[:@var{start_time}[:@var{round}]].
10901 See also the @ref{setpts} filter.
10903 @subsection Examples
10907 A typical usage in order to set the fps to 25:
10913 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
10915 fps=fps=film:round=near
10921 Pack two different video streams into a stereoscopic video, setting proper
10922 metadata on supported codecs. The two views should have the same size and
10923 framerate and processing will stop when the shorter video ends. Please note
10924 that you may conveniently adjust view properties with the @ref{scale} and
10927 It accepts the following parameters:
10931 The desired packing format. Supported values are:
10936 The views are next to each other (default).
10939 The views are on top of each other.
10942 The views are packed by line.
10945 The views are packed by column.
10948 The views are temporally interleaved.
10957 # Convert left and right views into a frame-sequential video
10958 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
10960 # Convert views into a side-by-side video with the same output resolution as the input
10961 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
10966 Change the frame rate by interpolating new video output frames from the source
10969 This filter is not designed to function correctly with interlaced media. If
10970 you wish to change the frame rate of interlaced media then you are required
10971 to deinterlace before this filter and re-interlace after this filter.
10973 A description of the accepted options follows.
10977 Specify the output frames per second. This option can also be specified
10978 as a value alone. The default is @code{50}.
10981 Specify the start of a range where the output frame will be created as a
10982 linear interpolation of two frames. The range is [@code{0}-@code{255}],
10983 the default is @code{15}.
10986 Specify the end of a range where the output frame will be created as a
10987 linear interpolation of two frames. The range is [@code{0}-@code{255}],
10988 the default is @code{240}.
10991 Specify the level at which a scene change is detected as a value between
10992 0 and 100 to indicate a new scene; a low value reflects a low
10993 probability for the current frame to introduce a new scene, while a higher
10994 value means the current frame is more likely to be one.
10995 The default is @code{8.2}.
10998 Specify flags influencing the filter process.
11000 Available value for @var{flags} is:
11003 @item scene_change_detect, scd
11004 Enable scene change detection using the value of the option @var{scene}.
11005 This flag is enabled by default.
11011 Select one frame every N-th frame.
11013 This filter accepts the following option:
11016 Select frame after every @code{step} frames.
11017 Allowed values are positive integers higher than 0. Default value is @code{1}.
11020 @section freezedetect
11022 Detect frozen video.
11024 This filter logs a message and sets frame metadata when it detects that the
11025 input video has no significant change in content during a specified duration.
11026 Video freeze detection calculates the mean average absolute difference of all
11027 the components of video frames and compares it to a noise floor.
11029 The printed times and duration are expressed in seconds. The
11030 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11031 whose timestamp equals or exceeds the detection duration and it contains the
11032 timestamp of the first frame of the freeze. The
11033 @code{lavfi.freezedetect.freeze_duration} and
11034 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11037 The filter accepts the following options:
11041 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11042 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11046 Set freeze duration until notification (default is 2 seconds).
11052 Apply a frei0r effect to the input video.
11054 To enable the compilation of this filter, you need to install the frei0r
11055 header and configure FFmpeg with @code{--enable-frei0r}.
11057 It accepts the following parameters:
11062 The name of the frei0r effect to load. If the environment variable
11063 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11064 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11065 Otherwise, the standard frei0r paths are searched, in this order:
11066 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11067 @file{/usr/lib/frei0r-1/}.
11069 @item filter_params
11070 A '|'-separated list of parameters to pass to the frei0r effect.
11074 A frei0r effect parameter can be a boolean (its value is either
11075 "y" or "n"), a double, a color (specified as
11076 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11077 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11078 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11079 a position (specified as @var{X}/@var{Y}, where
11080 @var{X} and @var{Y} are floating point numbers) and/or a string.
11082 The number and types of parameters depend on the loaded effect. If an
11083 effect parameter is not specified, the default value is set.
11085 @subsection Examples
11089 Apply the distort0r effect, setting the first two double parameters:
11091 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11095 Apply the colordistance effect, taking a color as the first parameter:
11097 frei0r=colordistance:0.2/0.3/0.4
11098 frei0r=colordistance:violet
11099 frei0r=colordistance:0x112233
11103 Apply the perspective effect, specifying the top left and top right image
11106 frei0r=perspective:0.2/0.2|0.8/0.2
11110 For more information, see
11111 @url{http://frei0r.dyne.org}
11115 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11117 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11118 processing filter, one of them is performed once per block, not per pixel.
11119 This allows for much higher speed.
11121 The filter accepts the following options:
11125 Set quality. This option defines the number of levels for averaging. It accepts
11126 an integer in the range 4-5. Default value is @code{4}.
11129 Force a constant quantization parameter. It accepts an integer in range 0-63.
11130 If not set, the filter will use the QP from the video stream (if available).
11133 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11134 more details but also more artifacts, while higher values make the image smoother
11135 but also blurrier. Default value is @code{0} − PSNR optimal.
11137 @item use_bframe_qp
11138 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11139 option may cause flicker since the B-Frames have often larger QP. Default is
11140 @code{0} (not enabled).
11146 Apply Gaussian blur filter.
11148 The filter accepts the following options:
11152 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11155 Set number of steps for Gaussian approximation. Default is @code{1}.
11158 Set which planes to filter. By default all planes are filtered.
11161 Set vertical sigma, if negative it will be same as @code{sigma}.
11162 Default is @code{-1}.
11165 @subsection Commands
11166 This filter supports same commands as options.
11167 The command accepts the same syntax of the corresponding option.
11169 If the specified expression is not valid, it is kept at its current
11174 Apply generic equation to each pixel.
11176 The filter accepts the following options:
11179 @item lum_expr, lum
11180 Set the luminance expression.
11182 Set the chrominance blue expression.
11184 Set the chrominance red expression.
11185 @item alpha_expr, a
11186 Set the alpha expression.
11188 Set the red expression.
11189 @item green_expr, g
11190 Set the green expression.
11192 Set the blue expression.
11195 The colorspace is selected according to the specified options. If one
11196 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11197 options is specified, the filter will automatically select a YCbCr
11198 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11199 @option{blue_expr} options is specified, it will select an RGB
11202 If one of the chrominance expression is not defined, it falls back on the other
11203 one. If no alpha expression is specified it will evaluate to opaque value.
11204 If none of chrominance expressions are specified, they will evaluate
11205 to the luminance expression.
11207 The expressions can use the following variables and functions:
11211 The sequential number of the filtered frame, starting from @code{0}.
11215 The coordinates of the current sample.
11219 The width and height of the image.
11223 Width and height scale depending on the currently filtered plane. It is the
11224 ratio between the corresponding luma plane number of pixels and the current
11225 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11226 @code{0.5,0.5} for chroma planes.
11229 Time of the current frame, expressed in seconds.
11232 Return the value of the pixel at location (@var{x},@var{y}) of the current
11236 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
11240 Return the value of the pixel at location (@var{x},@var{y}) of the
11241 blue-difference chroma plane. Return 0 if there is no such plane.
11244 Return the value of the pixel at location (@var{x},@var{y}) of the
11245 red-difference chroma plane. Return 0 if there is no such plane.
11250 Return the value of the pixel at location (@var{x},@var{y}) of the
11251 red/green/blue component. Return 0 if there is no such component.
11254 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
11255 plane. Return 0 if there is no such plane.
11257 @item interpolation
11258 Set one of interpolation methods:
11263 Default is bilinear.
11266 For functions, if @var{x} and @var{y} are outside the area, the value will be
11267 automatically clipped to the closer edge.
11269 @subsection Examples
11273 Flip the image horizontally:
11279 Generate a bidimensional sine wave, with angle @code{PI/3} and a
11280 wavelength of 100 pixels:
11282 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
11286 Generate a fancy enigmatic moving light:
11288 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
11292 Generate a quick emboss effect:
11294 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
11298 Modify RGB components depending on pixel position:
11300 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
11304 Create a radial gradient that is the same size as the input (also see
11305 the @ref{vignette} filter):
11307 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
11313 Fix the banding artifacts that are sometimes introduced into nearly flat
11314 regions by truncation to 8-bit color depth.
11315 Interpolate the gradients that should go where the bands are, and
11318 It is designed for playback only. Do not use it prior to
11319 lossy compression, because compression tends to lose the dither and
11320 bring back the bands.
11322 It accepts the following parameters:
11327 The maximum amount by which the filter will change any one pixel. This is also
11328 the threshold for detecting nearly flat regions. Acceptable values range from
11329 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
11333 The neighborhood to fit the gradient to. A larger radius makes for smoother
11334 gradients, but also prevents the filter from modifying the pixels near detailed
11335 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
11336 values will be clipped to the valid range.
11340 Alternatively, the options can be specified as a flat string:
11341 @var{strength}[:@var{radius}]
11343 @subsection Examples
11347 Apply the filter with a @code{3.5} strength and radius of @code{8}:
11353 Specify radius, omitting the strength (which will fall-back to the default
11361 @anchor{graphmonitor}
11362 @section graphmonitor
11363 Show various filtergraph stats.
11365 With this filter one can debug complete filtergraph.
11366 Especially issues with links filling with queued frames.
11368 The filter accepts the following options:
11372 Set video output size. Default is @var{hd720}.
11375 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
11378 Set output mode, can be @var{fulll} or @var{compact}.
11379 In @var{compact} mode only filters with some queued frames have displayed stats.
11382 Set flags which enable which stats are shown in video.
11384 Available values for flags are:
11387 Display number of queued frames in each link.
11389 @item frame_count_in
11390 Display number of frames taken from filter.
11392 @item frame_count_out
11393 Display number of frames given out from filter.
11396 Display current filtered frame pts.
11399 Display current filtered frame time.
11402 Display time base for filter link.
11405 Display used format for filter link.
11408 Display video size or number of audio channels in case of audio used by filter link.
11411 Display video frame rate or sample rate in case of audio used by filter link.
11415 Set upper limit for video rate of output stream, Default value is @var{25}.
11416 This guarantee that output video frame rate will not be higher than this value.
11420 A color constancy variation filter which estimates scene illumination via grey edge algorithm
11421 and corrects the scene colors accordingly.
11423 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
11425 The filter accepts the following options:
11429 The order of differentiation to be applied on the scene. Must be chosen in the range
11430 [0,2] and default value is 1.
11433 The Minkowski parameter to be used for calculating the Minkowski distance. Must
11434 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
11435 max value instead of calculating Minkowski distance.
11438 The standard deviation of Gaussian blur to be applied on the scene. Must be
11439 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
11440 can't be equal to 0 if @var{difford} is greater than 0.
11443 @subsection Examples
11449 greyedge=difford=1:minknorm=5:sigma=2
11455 greyedge=difford=1:minknorm=0:sigma=2
11463 Apply a Hald CLUT to a video stream.
11465 First input is the video stream to process, and second one is the Hald CLUT.
11466 The Hald CLUT input can be a simple picture or a complete video stream.
11468 The filter accepts the following options:
11472 Force termination when the shortest input terminates. Default is @code{0}.
11474 Continue applying the last CLUT after the end of the stream. A value of
11475 @code{0} disable the filter after the last frame of the CLUT is reached.
11476 Default is @code{1}.
11479 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
11480 filters share the same internals).
11482 This filter also supports the @ref{framesync} options.
11484 More information about the Hald CLUT can be found on Eskil Steenberg's website
11485 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
11487 @subsection Workflow examples
11489 @subsubsection Hald CLUT video stream
11491 Generate an identity Hald CLUT stream altered with various effects:
11493 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
11496 Note: make sure you use a lossless codec.
11498 Then use it with @code{haldclut} to apply it on some random stream:
11500 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
11503 The Hald CLUT will be applied to the 10 first seconds (duration of
11504 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
11505 to the remaining frames of the @code{mandelbrot} stream.
11507 @subsubsection Hald CLUT with preview
11509 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
11510 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
11511 biggest possible square starting at the top left of the picture. The remaining
11512 padding pixels (bottom or right) will be ignored. This area can be used to add
11513 a preview of the Hald CLUT.
11515 Typically, the following generated Hald CLUT will be supported by the
11516 @code{haldclut} filter:
11519 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
11520 pad=iw+320 [padded_clut];
11521 smptebars=s=320x256, split [a][b];
11522 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
11523 [main][b] overlay=W-320" -frames:v 1 clut.png
11526 It contains the original and a preview of the effect of the CLUT: SMPTE color
11527 bars are displayed on the right-top, and below the same color bars processed by
11530 Then, the effect of this Hald CLUT can be visualized with:
11532 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
11537 Flip the input video horizontally.
11539 For example, to horizontally flip the input video with @command{ffmpeg}:
11541 ffmpeg -i in.avi -vf "hflip" out.avi
11545 This filter applies a global color histogram equalization on a
11548 It can be used to correct video that has a compressed range of pixel
11549 intensities. The filter redistributes the pixel intensities to
11550 equalize their distribution across the intensity range. It may be
11551 viewed as an "automatically adjusting contrast filter". This filter is
11552 useful only for correcting degraded or poorly captured source
11555 The filter accepts the following options:
11559 Determine the amount of equalization to be applied. As the strength
11560 is reduced, the distribution of pixel intensities more-and-more
11561 approaches that of the input frame. The value must be a float number
11562 in the range [0,1] and defaults to 0.200.
11565 Set the maximum intensity that can generated and scale the output
11566 values appropriately. The strength should be set as desired and then
11567 the intensity can be limited if needed to avoid washing-out. The value
11568 must be a float number in the range [0,1] and defaults to 0.210.
11571 Set the antibanding level. If enabled the filter will randomly vary
11572 the luminance of output pixels by a small amount to avoid banding of
11573 the histogram. Possible values are @code{none}, @code{weak} or
11574 @code{strong}. It defaults to @code{none}.
11579 Compute and draw a color distribution histogram for the input video.
11581 The computed histogram is a representation of the color component
11582 distribution in an image.
11584 Standard histogram displays the color components distribution in an image.
11585 Displays color graph for each color component. Shows distribution of
11586 the Y, U, V, A or R, G, B components, depending on input format, in the
11587 current frame. Below each graph a color component scale meter is shown.
11589 The filter accepts the following options:
11593 Set height of level. Default value is @code{200}.
11594 Allowed range is [50, 2048].
11597 Set height of color scale. Default value is @code{12}.
11598 Allowed range is [0, 40].
11602 It accepts the following values:
11605 Per color component graphs are placed below each other.
11608 Per color component graphs are placed side by side.
11611 Presents information identical to that in the @code{parade}, except
11612 that the graphs representing color components are superimposed directly
11615 Default is @code{stack}.
11618 Set mode. Can be either @code{linear}, or @code{logarithmic}.
11619 Default is @code{linear}.
11622 Set what color components to display.
11623 Default is @code{7}.
11626 Set foreground opacity. Default is @code{0.7}.
11629 Set background opacity. Default is @code{0.5}.
11632 @subsection Examples
11637 Calculate and draw histogram:
11639 ffplay -i input -vf histogram
11647 This is a high precision/quality 3d denoise filter. It aims to reduce
11648 image noise, producing smooth images and making still images really
11649 still. It should enhance compressibility.
11651 It accepts the following optional parameters:
11655 A non-negative floating point number which specifies spatial luma strength.
11656 It defaults to 4.0.
11658 @item chroma_spatial
11659 A non-negative floating point number which specifies spatial chroma strength.
11660 It defaults to 3.0*@var{luma_spatial}/4.0.
11663 A floating point number which specifies luma temporal strength. It defaults to
11664 6.0*@var{luma_spatial}/4.0.
11667 A floating point number which specifies chroma temporal strength. It defaults to
11668 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
11671 @anchor{hwdownload}
11672 @section hwdownload
11674 Download hardware frames to system memory.
11676 The input must be in hardware frames, and the output a non-hardware format.
11677 Not all formats will be supported on the output - it may be necessary to insert
11678 an additional @option{format} filter immediately following in the graph to get
11679 the output in a supported format.
11683 Map hardware frames to system memory or to another device.
11685 This filter has several different modes of operation; which one is used depends
11686 on the input and output formats:
11689 Hardware frame input, normal frame output
11691 Map the input frames to system memory and pass them to the output. If the
11692 original hardware frame is later required (for example, after overlaying
11693 something else on part of it), the @option{hwmap} filter can be used again
11694 in the next mode to retrieve it.
11696 Normal frame input, hardware frame output
11698 If the input is actually a software-mapped hardware frame, then unmap it -
11699 that is, return the original hardware frame.
11701 Otherwise, a device must be provided. Create new hardware surfaces on that
11702 device for the output, then map them back to the software format at the input
11703 and give those frames to the preceding filter. This will then act like the
11704 @option{hwupload} filter, but may be able to avoid an additional copy when
11705 the input is already in a compatible format.
11707 Hardware frame input and output
11709 A device must be supplied for the output, either directly or with the
11710 @option{derive_device} option. The input and output devices must be of
11711 different types and compatible - the exact meaning of this is
11712 system-dependent, but typically it means that they must refer to the same
11713 underlying hardware context (for example, refer to the same graphics card).
11715 If the input frames were originally created on the output device, then unmap
11716 to retrieve the original frames.
11718 Otherwise, map the frames to the output device - create new hardware frames
11719 on the output corresponding to the frames on the input.
11722 The following additional parameters are accepted:
11726 Set the frame mapping mode. Some combination of:
11729 The mapped frame should be readable.
11731 The mapped frame should be writeable.
11733 The mapping will always overwrite the entire frame.
11735 This may improve performance in some cases, as the original contents of the
11736 frame need not be loaded.
11738 The mapping must not involve any copying.
11740 Indirect mappings to copies of frames are created in some cases where either
11741 direct mapping is not possible or it would have unexpected properties.
11742 Setting this flag ensures that the mapping is direct and will fail if that is
11745 Defaults to @var{read+write} if not specified.
11747 @item derive_device @var{type}
11748 Rather than using the device supplied at initialisation, instead derive a new
11749 device of type @var{type} from the device the input frames exist on.
11752 In a hardware to hardware mapping, map in reverse - create frames in the sink
11753 and map them back to the source. This may be necessary in some cases where
11754 a mapping in one direction is required but only the opposite direction is
11755 supported by the devices being used.
11757 This option is dangerous - it may break the preceding filter in undefined
11758 ways if there are any additional constraints on that filter's output.
11759 Do not use it without fully understanding the implications of its use.
11765 Upload system memory frames to hardware surfaces.
11767 The device to upload to must be supplied when the filter is initialised. If
11768 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
11771 @anchor{hwupload_cuda}
11772 @section hwupload_cuda
11774 Upload system memory frames to a CUDA device.
11776 It accepts the following optional parameters:
11780 The number of the CUDA device to use
11785 Apply a high-quality magnification filter designed for pixel art. This filter
11786 was originally created by Maxim Stepin.
11788 It accepts the following option:
11792 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
11793 @code{hq3x} and @code{4} for @code{hq4x}.
11794 Default is @code{3}.
11798 Stack input videos horizontally.
11800 All streams must be of same pixel format and of same height.
11802 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
11803 to create same output.
11805 The filter accepts the following option:
11809 Set number of input streams. Default is 2.
11812 If set to 1, force the output to terminate when the shortest input
11813 terminates. Default value is 0.
11818 Modify the hue and/or the saturation of the input.
11820 It accepts the following parameters:
11824 Specify the hue angle as a number of degrees. It accepts an expression,
11825 and defaults to "0".
11828 Specify the saturation in the [-10,10] range. It accepts an expression and
11832 Specify the hue angle as a number of radians. It accepts an
11833 expression, and defaults to "0".
11836 Specify the brightness in the [-10,10] range. It accepts an expression and
11840 @option{h} and @option{H} are mutually exclusive, and can't be
11841 specified at the same time.
11843 The @option{b}, @option{h}, @option{H} and @option{s} option values are
11844 expressions containing the following constants:
11848 frame count of the input frame starting from 0
11851 presentation timestamp of the input frame expressed in time base units
11854 frame rate of the input video, NAN if the input frame rate is unknown
11857 timestamp expressed in seconds, NAN if the input timestamp is unknown
11860 time base of the input video
11863 @subsection Examples
11867 Set the hue to 90 degrees and the saturation to 1.0:
11873 Same command but expressing the hue in radians:
11879 Rotate hue and make the saturation swing between 0
11880 and 2 over a period of 1 second:
11882 hue="H=2*PI*t: s=sin(2*PI*t)+1"
11886 Apply a 3 seconds saturation fade-in effect starting at 0:
11888 hue="s=min(t/3\,1)"
11891 The general fade-in expression can be written as:
11893 hue="s=min(0\, max((t-START)/DURATION\, 1))"
11897 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
11899 hue="s=max(0\, min(1\, (8-t)/3))"
11902 The general fade-out expression can be written as:
11904 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
11909 @subsection Commands
11911 This filter supports the following commands:
11917 Modify the hue and/or the saturation and/or brightness of the input video.
11918 The command accepts the same syntax of the corresponding option.
11920 If the specified expression is not valid, it is kept at its current
11924 @section hysteresis
11926 Grow first stream into second stream by connecting components.
11927 This makes it possible to build more robust edge masks.
11929 This filter accepts the following options:
11933 Set which planes will be processed as bitmap, unprocessed planes will be
11934 copied from first stream.
11935 By default value 0xf, all planes will be processed.
11938 Set threshold which is used in filtering. If pixel component value is higher than
11939 this value filter algorithm for connecting components is activated.
11940 By default value is 0.
11945 Detect video interlacing type.
11947 This filter tries to detect if the input frames are interlaced, progressive,
11948 top or bottom field first. It will also try to detect fields that are
11949 repeated between adjacent frames (a sign of telecine).
11951 Single frame detection considers only immediately adjacent frames when classifying each frame.
11952 Multiple frame detection incorporates the classification history of previous frames.
11954 The filter will log these metadata values:
11957 @item single.current_frame
11958 Detected type of current frame using single-frame detection. One of:
11959 ``tff'' (top field first), ``bff'' (bottom field first),
11960 ``progressive'', or ``undetermined''
11963 Cumulative number of frames detected as top field first using single-frame detection.
11966 Cumulative number of frames detected as top field first using multiple-frame detection.
11969 Cumulative number of frames detected as bottom field first using single-frame detection.
11971 @item multiple.current_frame
11972 Detected type of current frame using multiple-frame detection. One of:
11973 ``tff'' (top field first), ``bff'' (bottom field first),
11974 ``progressive'', or ``undetermined''
11977 Cumulative number of frames detected as bottom field first using multiple-frame detection.
11979 @item single.progressive
11980 Cumulative number of frames detected as progressive using single-frame detection.
11982 @item multiple.progressive
11983 Cumulative number of frames detected as progressive using multiple-frame detection.
11985 @item single.undetermined
11986 Cumulative number of frames that could not be classified using single-frame detection.
11988 @item multiple.undetermined
11989 Cumulative number of frames that could not be classified using multiple-frame detection.
11991 @item repeated.current_frame
11992 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
11994 @item repeated.neither
11995 Cumulative number of frames with no repeated field.
11998 Cumulative number of frames with the top field repeated from the previous frame's top field.
12000 @item repeated.bottom
12001 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12004 The filter accepts the following options:
12008 Set interlacing threshold.
12010 Set progressive threshold.
12012 Threshold for repeated field detection.
12014 Number of frames after which a given frame's contribution to the
12015 statistics is halved (i.e., it contributes only 0.5 to its
12016 classification). The default of 0 means that all frames seen are given
12017 full weight of 1.0 forever.
12018 @item analyze_interlaced_flag
12019 When this is not 0 then idet will use the specified number of frames to determine
12020 if the interlaced flag is accurate, it will not count undetermined frames.
12021 If the flag is found to be accurate it will be used without any further
12022 computations, if it is found to be inaccurate it will be cleared without any
12023 further computations. This allows inserting the idet filter as a low computational
12024 method to clean up the interlaced flag
12029 Deinterleave or interleave fields.
12031 This filter allows one to process interlaced images fields without
12032 deinterlacing them. Deinterleaving splits the input frame into 2
12033 fields (so called half pictures). Odd lines are moved to the top
12034 half of the output image, even lines to the bottom half.
12035 You can process (filter) them independently and then re-interleave them.
12037 The filter accepts the following options:
12041 @item chroma_mode, c
12042 @item alpha_mode, a
12043 Available values for @var{luma_mode}, @var{chroma_mode} and
12044 @var{alpha_mode} are:
12050 @item deinterleave, d
12051 Deinterleave fields, placing one above the other.
12053 @item interleave, i
12054 Interleave fields. Reverse the effect of deinterleaving.
12056 Default value is @code{none}.
12058 @item luma_swap, ls
12059 @item chroma_swap, cs
12060 @item alpha_swap, as
12061 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12066 Apply inflate effect to the video.
12068 This filter replaces the pixel by the local(3x3) average by taking into account
12069 only values higher than the pixel.
12071 It accepts the following options:
12078 Limit the maximum change for each plane, default is 65535.
12079 If 0, plane will remain unchanged.
12084 Simple interlacing filter from progressive contents. This interleaves upper (or
12085 lower) lines from odd frames with lower (or upper) lines from even frames,
12086 halving the frame rate and preserving image height.
12089 Original Original New Frame
12090 Frame 'j' Frame 'j+1' (tff)
12091 ========== =========== ==================
12092 Line 0 --------------------> Frame 'j' Line 0
12093 Line 1 Line 1 ----> Frame 'j+1' Line 1
12094 Line 2 ---------------------> Frame 'j' Line 2
12095 Line 3 Line 3 ----> Frame 'j+1' Line 3
12097 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12100 It accepts the following optional parameters:
12104 This determines whether the interlaced frame is taken from the even
12105 (tff - default) or odd (bff) lines of the progressive frame.
12108 Vertical lowpass filter to avoid twitter interlacing and
12109 reduce moire patterns.
12113 Disable vertical lowpass filter
12116 Enable linear filter (default)
12119 Enable complex filter. This will slightly less reduce twitter and moire
12120 but better retain detail and subjective sharpness impression.
12127 Deinterlace input video by applying Donald Graft's adaptive kernel
12128 deinterling. Work on interlaced parts of a video to produce
12129 progressive frames.
12131 The description of the accepted parameters follows.
12135 Set the threshold which affects the filter's tolerance when
12136 determining if a pixel line must be processed. It must be an integer
12137 in the range [0,255] and defaults to 10. A value of 0 will result in
12138 applying the process on every pixels.
12141 Paint pixels exceeding the threshold value to white if set to 1.
12145 Set the fields order. Swap fields if set to 1, leave fields alone if
12149 Enable additional sharpening if set to 1. Default is 0.
12152 Enable twoway sharpening if set to 1. Default is 0.
12155 @subsection Examples
12159 Apply default values:
12161 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12165 Enable additional sharpening:
12171 Paint processed pixels in white:
12179 Slowly update darker pixels.
12181 This filter makes short flashes of light appear longer.
12182 This filter accepts the following options:
12186 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
12189 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
12192 @section lenscorrection
12194 Correct radial lens distortion
12196 This filter can be used to correct for radial distortion as can result from the use
12197 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
12198 one can use tools available for example as part of opencv or simply trial-and-error.
12199 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
12200 and extract the k1 and k2 coefficients from the resulting matrix.
12202 Note that effectively the same filter is available in the open-source tools Krita and
12203 Digikam from the KDE project.
12205 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
12206 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
12207 brightness distribution, so you may want to use both filters together in certain
12208 cases, though you will have to take care of ordering, i.e. whether vignetting should
12209 be applied before or after lens correction.
12211 @subsection Options
12213 The filter accepts the following options:
12217 Relative x-coordinate of the focal point of the image, and thereby the center of the
12218 distortion. This value has a range [0,1] and is expressed as fractions of the image
12219 width. Default is 0.5.
12221 Relative y-coordinate of the focal point of the image, and thereby the center of the
12222 distortion. This value has a range [0,1] and is expressed as fractions of the image
12223 height. Default is 0.5.
12225 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
12226 no correction. Default is 0.
12228 Coefficient of the double quadratic correction term. This value has a range [-1,1].
12229 0 means no correction. Default is 0.
12232 The formula that generates the correction is:
12234 @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)
12236 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
12237 distances from the focal point in the source and target images, respectively.
12241 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
12243 The @code{lensfun} filter requires the camera make, camera model, and lens model
12244 to apply the lens correction. The filter will load the lensfun database and
12245 query it to find the corresponding camera and lens entries in the database. As
12246 long as these entries can be found with the given options, the filter can
12247 perform corrections on frames. Note that incomplete strings will result in the
12248 filter choosing the best match with the given options, and the filter will
12249 output the chosen camera and lens models (logged with level "info"). You must
12250 provide the make, camera model, and lens model as they are required.
12252 The filter accepts the following options:
12256 The make of the camera (for example, "Canon"). This option is required.
12259 The model of the camera (for example, "Canon EOS 100D"). This option is
12263 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
12264 option is required.
12267 The type of correction to apply. The following values are valid options:
12271 Enables fixing lens vignetting.
12274 Enables fixing lens geometry. This is the default.
12277 Enables fixing chromatic aberrations.
12280 Enables fixing lens vignetting and lens geometry.
12283 Enables fixing lens vignetting and chromatic aberrations.
12286 Enables fixing both lens geometry and chromatic aberrations.
12289 Enables all possible corrections.
12293 The focal length of the image/video (zoom; expected constant for video). For
12294 example, a 18--55mm lens has focal length range of [18--55], so a value in that
12295 range should be chosen when using that lens. Default 18.
12298 The aperture of the image/video (expected constant for video). Note that
12299 aperture is only used for vignetting correction. Default 3.5.
12301 @item focus_distance
12302 The focus distance of the image/video (expected constant for video). Note that
12303 focus distance is only used for vignetting and only slightly affects the
12304 vignetting correction process. If unknown, leave it at the default value (which
12308 The scale factor which is applied after transformation. After correction the
12309 video is no longer necessarily rectangular. This parameter controls how much of
12310 the resulting image is visible. The value 0 means that a value will be chosen
12311 automatically such that there is little or no unmapped area in the output
12312 image. 1.0 means that no additional scaling is done. Lower values may result
12313 in more of the corrected image being visible, while higher values may avoid
12314 unmapped areas in the output.
12316 @item target_geometry
12317 The target geometry of the output image/video. The following values are valid
12321 @item rectilinear (default)
12324 @item equirectangular
12325 @item fisheye_orthographic
12326 @item fisheye_stereographic
12327 @item fisheye_equisolid
12328 @item fisheye_thoby
12331 Apply the reverse of image correction (instead of correcting distortion, apply
12334 @item interpolation
12335 The type of interpolation used when correcting distortion. The following values
12340 @item linear (default)
12345 @subsection Examples
12349 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
12350 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
12354 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
12358 Apply the same as before, but only for the first 5 seconds of video.
12361 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
12368 Obtain the VMAF (Video Multi-Method Assessment Fusion)
12369 score between two input videos.
12371 The obtained VMAF score is printed through the logging system.
12373 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
12374 After installing the library it can be enabled using:
12375 @code{./configure --enable-libvmaf --enable-version3}.
12376 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
12378 The filter has following options:
12382 Set the model path which is to be used for SVM.
12383 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
12386 Set the file path to be used to store logs.
12389 Set the format of the log file (xml or json).
12391 @item enable_transform
12392 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
12393 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
12394 Default value: @code{false}
12397 Invokes the phone model which will generate VMAF scores higher than in the
12398 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
12399 Default value: @code{false}
12402 Enables computing psnr along with vmaf.
12403 Default value: @code{false}
12406 Enables computing ssim along with vmaf.
12407 Default value: @code{false}
12410 Enables computing ms_ssim along with vmaf.
12411 Default value: @code{false}
12414 Set the pool method to be used for computing vmaf.
12415 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
12418 Set number of threads to be used when computing vmaf.
12419 Default value: @code{0}, which makes use of all available logical processors.
12422 Set interval for frame subsampling used when computing vmaf.
12423 Default value: @code{1}
12425 @item enable_conf_interval
12426 Enables confidence interval.
12427 Default value: @code{false}
12430 This filter also supports the @ref{framesync} options.
12432 @subsection Examples
12435 On the below examples the input file @file{main.mpg} being processed is
12436 compared with the reference file @file{ref.mpg}.
12439 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
12443 Example with options:
12445 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
12449 Example with options and different containers:
12451 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=1/AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=1/AVTB,setpts=PTS-STARTPTS[ref];[main][ref]libvmaf=psnr=1:log_fmt=json" -f null -
12457 Limits the pixel components values to the specified range [min, max].
12459 The filter accepts the following options:
12463 Lower bound. Defaults to the lowest allowed value for the input.
12466 Upper bound. Defaults to the highest allowed value for the input.
12469 Specify which planes will be processed. Defaults to all available.
12476 The filter accepts the following options:
12480 Set the number of loops. Setting this value to -1 will result in infinite loops.
12484 Set maximal size in number of frames. Default is 0.
12487 Set first frame of loop. Default is 0.
12490 @subsection Examples
12494 Loop single first frame infinitely:
12496 loop=loop=-1:size=1:start=0
12500 Loop single first frame 10 times:
12502 loop=loop=10:size=1:start=0
12506 Loop 10 first frames 5 times:
12508 loop=loop=5:size=10:start=0
12514 Apply a 1D LUT to an input video.
12516 The filter accepts the following options:
12520 Set the 1D LUT file name.
12522 Currently supported formats:
12531 Select interpolation mode.
12533 Available values are:
12537 Use values from the nearest defined point.
12539 Interpolate values using the linear interpolation.
12541 Interpolate values using the cosine interpolation.
12543 Interpolate values using the cubic interpolation.
12545 Interpolate values using the spline interpolation.
12552 Apply a 3D LUT to an input video.
12554 The filter accepts the following options:
12558 Set the 3D LUT file name.
12560 Currently supported formats:
12574 Select interpolation mode.
12576 Available values are:
12580 Use values from the nearest defined point.
12582 Interpolate values using the 8 points defining a cube.
12584 Interpolate values using a tetrahedron.
12590 Turn certain luma values into transparency.
12592 The filter accepts the following options:
12596 Set the luma which will be used as base for transparency.
12597 Default value is @code{0}.
12600 Set the range of luma values to be keyed out.
12601 Default value is @code{0}.
12604 Set the range of softness. Default value is @code{0}.
12605 Use this to control gradual transition from zero to full transparency.
12608 @section lut, lutrgb, lutyuv
12610 Compute a look-up table for binding each pixel component input value
12611 to an output value, and apply it to the input video.
12613 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
12614 to an RGB input video.
12616 These filters accept the following parameters:
12619 set first pixel component expression
12621 set second pixel component expression
12623 set third pixel component expression
12625 set fourth pixel component expression, corresponds to the alpha component
12628 set red component expression
12630 set green component expression
12632 set blue component expression
12634 alpha component expression
12637 set Y/luminance component expression
12639 set U/Cb component expression
12641 set V/Cr component expression
12644 Each of them specifies the expression to use for computing the lookup table for
12645 the corresponding pixel component values.
12647 The exact component associated to each of the @var{c*} options depends on the
12650 The @var{lut} filter requires either YUV or RGB pixel formats in input,
12651 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
12653 The expressions can contain the following constants and functions:
12658 The input width and height.
12661 The input value for the pixel component.
12664 The input value, clipped to the @var{minval}-@var{maxval} range.
12667 The maximum value for the pixel component.
12670 The minimum value for the pixel component.
12673 The negated value for the pixel component value, clipped to the
12674 @var{minval}-@var{maxval} range; it corresponds to the expression
12675 "maxval-clipval+minval".
12678 The computed value in @var{val}, clipped to the
12679 @var{minval}-@var{maxval} range.
12681 @item gammaval(gamma)
12682 The computed gamma correction value of the pixel component value,
12683 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
12685 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
12689 All expressions default to "val".
12691 @subsection Examples
12695 Negate input video:
12697 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
12698 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
12701 The above is the same as:
12703 lutrgb="r=negval:g=negval:b=negval"
12704 lutyuv="y=negval:u=negval:v=negval"
12714 Remove chroma components, turning the video into a graytone image:
12716 lutyuv="u=128:v=128"
12720 Apply a luma burning effect:
12726 Remove green and blue components:
12732 Set a constant alpha channel value on input:
12734 format=rgba,lutrgb=a="maxval-minval/2"
12738 Correct luminance gamma by a factor of 0.5:
12740 lutyuv=y=gammaval(0.5)
12744 Discard least significant bits of luma:
12746 lutyuv=y='bitand(val, 128+64+32)'
12750 Technicolor like effect:
12752 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
12756 @section lut2, tlut2
12758 The @code{lut2} filter takes two input streams and outputs one
12761 The @code{tlut2} (time lut2) filter takes two consecutive frames
12762 from one single stream.
12764 This filter accepts the following parameters:
12767 set first pixel component expression
12769 set second pixel component expression
12771 set third pixel component expression
12773 set fourth pixel component expression, corresponds to the alpha component
12776 set output bit depth, only available for @code{lut2} filter. By default is 0,
12777 which means bit depth is automatically picked from first input format.
12780 Each of them specifies the expression to use for computing the lookup table for
12781 the corresponding pixel component values.
12783 The exact component associated to each of the @var{c*} options depends on the
12786 The expressions can contain the following constants:
12791 The input width and height.
12794 The first input value for the pixel component.
12797 The second input value for the pixel component.
12800 The first input video bit depth.
12803 The second input video bit depth.
12806 All expressions default to "x".
12808 @subsection Examples
12812 Highlight differences between two RGB video streams:
12814 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)'
12818 Highlight differences between two YUV video streams:
12820 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)'
12824 Show max difference between two video streams:
12826 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)))'
12830 @section maskedclamp
12832 Clamp the first input stream with the second input and third input stream.
12834 Returns the value of first stream to be between second input
12835 stream - @code{undershoot} and third input stream + @code{overshoot}.
12837 This filter accepts the following options:
12840 Default value is @code{0}.
12843 Default value is @code{0}.
12846 Set which planes will be processed as bitmap, unprocessed planes will be
12847 copied from first stream.
12848 By default value 0xf, all planes will be processed.
12853 Merge the second and third input stream into output stream using absolute differences
12854 between second input stream and first input stream and absolute difference between
12855 third input stream and first input stream. The picked value will be from second input
12856 stream if second absolute difference is greater than first one or from third input stream
12859 This filter accepts the following options:
12862 Set which planes will be processed as bitmap, unprocessed planes will be
12863 copied from first stream.
12864 By default value 0xf, all planes will be processed.
12867 @section maskedmerge
12869 Merge the first input stream with the second input stream using per pixel
12870 weights in the third input stream.
12872 A value of 0 in the third stream pixel component means that pixel component
12873 from first stream is returned unchanged, while maximum value (eg. 255 for
12874 8-bit videos) means that pixel component from second stream is returned
12875 unchanged. Intermediate values define the amount of merging between both
12876 input stream's pixel components.
12878 This filter accepts the following options:
12881 Set which planes will be processed as bitmap, unprocessed planes will be
12882 copied from first stream.
12883 By default value 0xf, all planes will be processed.
12888 Merge the second and third input stream into output stream using absolute differences
12889 between second input stream and first input stream and absolute difference between
12890 third input stream and first input stream. The picked value will be from second input
12891 stream if second absolute difference is less than first one or from third input stream
12894 This filter accepts the following options:
12897 Set which planes will be processed as bitmap, unprocessed planes will be
12898 copied from first stream.
12899 By default value 0xf, all planes will be processed.
12903 Create mask from input video.
12905 For example it is useful to create motion masks after @code{tblend} filter.
12907 This filter accepts the following options:
12911 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
12914 Set high threshold. Any pixel component higher than this value will be set to max value
12915 allowed for current pixel format.
12918 Set planes to filter, by default all available planes are filtered.
12921 Fill all frame pixels with this value.
12924 Set max average pixel value for frame. If sum of all pixel components is higher that this
12925 average, output frame will be completely filled with value set by @var{fill} option.
12926 Typically useful for scene changes when used in combination with @code{tblend} filter.
12931 Apply motion-compensation deinterlacing.
12933 It needs one field per frame as input and must thus be used together
12934 with yadif=1/3 or equivalent.
12936 This filter accepts the following options:
12939 Set the deinterlacing mode.
12941 It accepts one of the following values:
12946 use iterative motion estimation
12948 like @samp{slow}, but use multiple reference frames.
12950 Default value is @samp{fast}.
12953 Set the picture field parity assumed for the input video. It must be
12954 one of the following values:
12958 assume top field first
12960 assume bottom field first
12963 Default value is @samp{bff}.
12966 Set per-block quantization parameter (QP) used by the internal
12969 Higher values should result in a smoother motion vector field but less
12970 optimal individual vectors. Default value is 1.
12975 Pick median pixel from certain rectangle defined by radius.
12977 This filter accepts the following options:
12981 Set horizontal radius size. Default value is @code{1}.
12982 Allowed range is integer from 1 to 127.
12985 Set which planes to process. Default is @code{15}, which is all available planes.
12988 Set vertical radius size. Default value is @code{0}.
12989 Allowed range is integer from 0 to 127.
12990 If it is 0, value will be picked from horizontal @code{radius} option.
12993 @section mergeplanes
12995 Merge color channel components from several video streams.
12997 The filter accepts up to 4 input streams, and merge selected input
12998 planes to the output video.
13000 This filter accepts the following options:
13003 Set input to output plane mapping. Default is @code{0}.
13005 The mappings is specified as a bitmap. It should be specified as a
13006 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13007 mapping for the first plane of the output stream. 'A' sets the number of
13008 the input stream to use (from 0 to 3), and 'a' the plane number of the
13009 corresponding input to use (from 0 to 3). The rest of the mappings is
13010 similar, 'Bb' describes the mapping for the output stream second
13011 plane, 'Cc' describes the mapping for the output stream third plane and
13012 'Dd' describes the mapping for the output stream fourth plane.
13015 Set output pixel format. Default is @code{yuva444p}.
13018 @subsection Examples
13022 Merge three gray video streams of same width and height into single video stream:
13024 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13028 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13030 [a0][a1]mergeplanes=0x00010210:yuva444p
13034 Swap Y and A plane in yuva444p stream:
13036 format=yuva444p,mergeplanes=0x03010200:yuva444p
13040 Swap U and V plane in yuv420p stream:
13042 format=yuv420p,mergeplanes=0x000201:yuv420p
13046 Cast a rgb24 clip to yuv444p:
13048 format=rgb24,mergeplanes=0x000102:yuv444p
13054 Estimate and export motion vectors using block matching algorithms.
13055 Motion vectors are stored in frame side data to be used by other filters.
13057 This filter accepts the following options:
13060 Specify the motion estimation method. Accepts one of the following values:
13064 Exhaustive search algorithm.
13066 Three step search algorithm.
13068 Two dimensional logarithmic search algorithm.
13070 New three step search algorithm.
13072 Four step search algorithm.
13074 Diamond search algorithm.
13076 Hexagon-based search algorithm.
13078 Enhanced predictive zonal search algorithm.
13080 Uneven multi-hexagon search algorithm.
13082 Default value is @samp{esa}.
13085 Macroblock size. Default @code{16}.
13088 Search parameter. Default @code{7}.
13091 @section midequalizer
13093 Apply Midway Image Equalization effect using two video streams.
13095 Midway Image Equalization adjusts a pair of images to have the same
13096 histogram, while maintaining their dynamics as much as possible. It's
13097 useful for e.g. matching exposures from a pair of stereo cameras.
13099 This filter has two inputs and one output, which must be of same pixel format, but
13100 may be of different sizes. The output of filter is first input adjusted with
13101 midway histogram of both inputs.
13103 This filter accepts the following option:
13107 Set which planes to process. Default is @code{15}, which is all available planes.
13110 @section minterpolate
13112 Convert the video to specified frame rate using motion interpolation.
13114 This filter accepts the following options:
13117 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}.
13120 Motion interpolation mode. Following values are accepted:
13123 Duplicate previous or next frame for interpolating new ones.
13125 Blend source frames. Interpolated frame is mean of previous and next frames.
13127 Motion compensated interpolation. Following options are effective when this mode is selected:
13131 Motion compensation mode. Following values are accepted:
13134 Overlapped block motion compensation.
13136 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13138 Default mode is @samp{obmc}.
13141 Motion estimation mode. Following values are accepted:
13144 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
13146 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
13148 Default mode is @samp{bilat}.
13151 The algorithm to be used for motion estimation. Following values are accepted:
13154 Exhaustive search algorithm.
13156 Three step search algorithm.
13158 Two dimensional logarithmic search algorithm.
13160 New three step search algorithm.
13162 Four step search algorithm.
13164 Diamond search algorithm.
13166 Hexagon-based search algorithm.
13168 Enhanced predictive zonal search algorithm.
13170 Uneven multi-hexagon search algorithm.
13172 Default algorithm is @samp{epzs}.
13175 Macroblock size. Default @code{16}.
13178 Motion estimation search parameter. Default @code{32}.
13181 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).
13186 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:
13189 Disable scene change detection.
13191 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
13193 Default method is @samp{fdiff}.
13195 @item scd_threshold
13196 Scene change detection threshold. Default is @code{5.0}.
13201 Mix several video input streams into one video stream.
13203 A description of the accepted options follows.
13207 The number of inputs. If unspecified, it defaults to 2.
13210 Specify weight of each input video stream as sequence.
13211 Each weight is separated by space. If number of weights
13212 is smaller than number of @var{frames} last specified
13213 weight will be used for all remaining unset weights.
13216 Specify scale, if it is set it will be multiplied with sum
13217 of each weight multiplied with pixel values to give final destination
13218 pixel value. By default @var{scale} is auto scaled to sum of weights.
13221 Specify how end of stream is determined.
13224 The duration of the longest input. (default)
13227 The duration of the shortest input.
13230 The duration of the first input.
13234 @section mpdecimate
13236 Drop frames that do not differ greatly from the previous frame in
13237 order to reduce frame rate.
13239 The main use of this filter is for very-low-bitrate encoding
13240 (e.g. streaming over dialup modem), but it could in theory be used for
13241 fixing movies that were inverse-telecined incorrectly.
13243 A description of the accepted options follows.
13247 Set the maximum number of consecutive frames which can be dropped (if
13248 positive), or the minimum interval between dropped frames (if
13249 negative). If the value is 0, the frame is dropped disregarding the
13250 number of previous sequentially dropped frames.
13252 Default value is 0.
13257 Set the dropping threshold values.
13259 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
13260 represent actual pixel value differences, so a threshold of 64
13261 corresponds to 1 unit of difference for each pixel, or the same spread
13262 out differently over the block.
13264 A frame is a candidate for dropping if no 8x8 blocks differ by more
13265 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
13266 meaning the whole image) differ by more than a threshold of @option{lo}.
13268 Default value for @option{hi} is 64*12, default value for @option{lo} is
13269 64*5, and default value for @option{frac} is 0.33.
13275 Negate (invert) the input video.
13277 It accepts the following option:
13282 With value 1, it negates the alpha component, if present. Default value is 0.
13288 Denoise frames using Non-Local Means algorithm.
13290 Each pixel is adjusted by looking for other pixels with similar contexts. This
13291 context similarity is defined by comparing their surrounding patches of size
13292 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
13295 Note that the research area defines centers for patches, which means some
13296 patches will be made of pixels outside that research area.
13298 The filter accepts the following options.
13302 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
13305 Set patch size. Default is 7. Must be odd number in range [0, 99].
13308 Same as @option{p} but for chroma planes.
13310 The default value is @var{0} and means automatic.
13313 Set research size. Default is 15. Must be odd number in range [0, 99].
13316 Same as @option{r} but for chroma planes.
13318 The default value is @var{0} and means automatic.
13323 Deinterlace video using neural network edge directed interpolation.
13325 This filter accepts the following options:
13329 Mandatory option, without binary file filter can not work.
13330 Currently file can be found here:
13331 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
13334 Set which frames to deinterlace, by default it is @code{all}.
13335 Can be @code{all} or @code{interlaced}.
13338 Set mode of operation.
13340 Can be one of the following:
13344 Use frame flags, both fields.
13346 Use frame flags, single field.
13348 Use top field only.
13350 Use bottom field only.
13352 Use both fields, top first.
13354 Use both fields, bottom first.
13358 Set which planes to process, by default filter process all frames.
13361 Set size of local neighborhood around each pixel, used by the predictor neural
13364 Can be one of the following:
13377 Set the number of neurons in predictor neural network.
13378 Can be one of the following:
13389 Controls the number of different neural network predictions that are blended
13390 together to compute the final output value. Can be @code{fast}, default or
13394 Set which set of weights to use in the predictor.
13395 Can be one of the following:
13399 weights trained to minimize absolute error
13401 weights trained to minimize squared error
13405 Controls whether or not the prescreener neural network is used to decide
13406 which pixels should be processed by the predictor neural network and which
13407 can be handled by simple cubic interpolation.
13408 The prescreener is trained to know whether cubic interpolation will be
13409 sufficient for a pixel or whether it should be predicted by the predictor nn.
13410 The computational complexity of the prescreener nn is much less than that of
13411 the predictor nn. Since most pixels can be handled by cubic interpolation,
13412 using the prescreener generally results in much faster processing.
13413 The prescreener is pretty accurate, so the difference between using it and not
13414 using it is almost always unnoticeable.
13416 Can be one of the following:
13424 Default is @code{new}.
13427 Set various debugging flags.
13432 Force libavfilter not to use any of the specified pixel formats for the
13433 input to the next filter.
13435 It accepts the following parameters:
13439 A '|'-separated list of pixel format names, such as
13440 pix_fmts=yuv420p|monow|rgb24".
13444 @subsection Examples
13448 Force libavfilter to use a format different from @var{yuv420p} for the
13449 input to the vflip filter:
13451 noformat=pix_fmts=yuv420p,vflip
13455 Convert the input video to any of the formats not contained in the list:
13457 noformat=yuv420p|yuv444p|yuv410p
13463 Add noise on video input frame.
13465 The filter accepts the following options:
13473 Set noise seed for specific pixel component or all pixel components in case
13474 of @var{all_seed}. Default value is @code{123457}.
13476 @item all_strength, alls
13477 @item c0_strength, c0s
13478 @item c1_strength, c1s
13479 @item c2_strength, c2s
13480 @item c3_strength, c3s
13481 Set noise strength for specific pixel component or all pixel components in case
13482 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
13484 @item all_flags, allf
13485 @item c0_flags, c0f
13486 @item c1_flags, c1f
13487 @item c2_flags, c2f
13488 @item c3_flags, c3f
13489 Set pixel component flags or set flags for all components if @var{all_flags}.
13490 Available values for component flags are:
13493 averaged temporal noise (smoother)
13495 mix random noise with a (semi)regular pattern
13497 temporal noise (noise pattern changes between frames)
13499 uniform noise (gaussian otherwise)
13503 @subsection Examples
13505 Add temporal and uniform noise to input video:
13507 noise=alls=20:allf=t+u
13512 Normalize RGB video (aka histogram stretching, contrast stretching).
13513 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
13515 For each channel of each frame, the filter computes the input range and maps
13516 it linearly to the user-specified output range. The output range defaults
13517 to the full dynamic range from pure black to pure white.
13519 Temporal smoothing can be used on the input range to reduce flickering (rapid
13520 changes in brightness) caused when small dark or bright objects enter or leave
13521 the scene. This is similar to the auto-exposure (automatic gain control) on a
13522 video camera, and, like a video camera, it may cause a period of over- or
13523 under-exposure of the video.
13525 The R,G,B channels can be normalized independently, which may cause some
13526 color shifting, or linked together as a single channel, which prevents
13527 color shifting. Linked normalization preserves hue. Independent normalization
13528 does not, so it can be used to remove some color casts. Independent and linked
13529 normalization can be combined in any ratio.
13531 The normalize filter accepts the following options:
13536 Colors which define the output range. The minimum input value is mapped to
13537 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
13538 The defaults are black and white respectively. Specifying white for
13539 @var{blackpt} and black for @var{whitept} will give color-inverted,
13540 normalized video. Shades of grey can be used to reduce the dynamic range
13541 (contrast). Specifying saturated colors here can create some interesting
13545 The number of previous frames to use for temporal smoothing. The input range
13546 of each channel is smoothed using a rolling average over the current frame
13547 and the @var{smoothing} previous frames. The default is 0 (no temporal
13551 Controls the ratio of independent (color shifting) channel normalization to
13552 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
13553 independent. Defaults to 1.0 (fully independent).
13556 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
13557 expensive no-op. Defaults to 1.0 (full strength).
13561 @subsection Examples
13563 Stretch video contrast to use the full dynamic range, with no temporal
13564 smoothing; may flicker depending on the source content:
13566 normalize=blackpt=black:whitept=white:smoothing=0
13569 As above, but with 50 frames of temporal smoothing; flicker should be
13570 reduced, depending on the source content:
13572 normalize=blackpt=black:whitept=white:smoothing=50
13575 As above, but with hue-preserving linked channel normalization:
13577 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
13580 As above, but with half strength:
13582 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
13585 Map the darkest input color to red, the brightest input color to cyan:
13587 normalize=blackpt=red:whitept=cyan
13592 Pass the video source unchanged to the output.
13595 Optical Character Recognition
13597 This filter uses Tesseract for optical character recognition. To enable
13598 compilation of this filter, you need to configure FFmpeg with
13599 @code{--enable-libtesseract}.
13601 It accepts the following options:
13605 Set datapath to tesseract data. Default is to use whatever was
13606 set at installation.
13609 Set language, default is "eng".
13612 Set character whitelist.
13615 Set character blacklist.
13618 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
13619 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
13623 Apply a video transform using libopencv.
13625 To enable this filter, install the libopencv library and headers and
13626 configure FFmpeg with @code{--enable-libopencv}.
13628 It accepts the following parameters:
13633 The name of the libopencv filter to apply.
13635 @item filter_params
13636 The parameters to pass to the libopencv filter. If not specified, the default
13637 values are assumed.
13641 Refer to the official libopencv documentation for more precise
13643 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
13645 Several libopencv filters are supported; see the following subsections.
13650 Dilate an image by using a specific structuring element.
13651 It corresponds to the libopencv function @code{cvDilate}.
13653 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
13655 @var{struct_el} represents a structuring element, and has the syntax:
13656 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
13658 @var{cols} and @var{rows} represent the number of columns and rows of
13659 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
13660 point, and @var{shape} the shape for the structuring element. @var{shape}
13661 must be "rect", "cross", "ellipse", or "custom".
13663 If the value for @var{shape} is "custom", it must be followed by a
13664 string of the form "=@var{filename}". The file with name
13665 @var{filename} is assumed to represent a binary image, with each
13666 printable character corresponding to a bright pixel. When a custom
13667 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
13668 or columns and rows of the read file are assumed instead.
13670 The default value for @var{struct_el} is "3x3+0x0/rect".
13672 @var{nb_iterations} specifies the number of times the transform is
13673 applied to the image, and defaults to 1.
13677 # Use the default values
13680 # Dilate using a structuring element with a 5x5 cross, iterating two times
13681 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
13683 # Read the shape from the file diamond.shape, iterating two times.
13684 # The file diamond.shape may contain a pattern of characters like this
13690 # The specified columns and rows are ignored
13691 # but the anchor point coordinates are not
13692 ocv=dilate:0x0+2x2/custom=diamond.shape|2
13697 Erode an image by using a specific structuring element.
13698 It corresponds to the libopencv function @code{cvErode}.
13700 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
13701 with the same syntax and semantics as the @ref{dilate} filter.
13705 Smooth the input video.
13707 The filter takes the following parameters:
13708 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
13710 @var{type} is the type of smooth filter to apply, and must be one of
13711 the following values: "blur", "blur_no_scale", "median", "gaussian",
13712 or "bilateral". The default value is "gaussian".
13714 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
13715 depends on the smooth type. @var{param1} and
13716 @var{param2} accept integer positive values or 0. @var{param3} and
13717 @var{param4} accept floating point values.
13719 The default value for @var{param1} is 3. The default value for the
13720 other parameters is 0.
13722 These parameters correspond to the parameters assigned to the
13723 libopencv function @code{cvSmooth}.
13725 @section oscilloscope
13727 2D Video Oscilloscope.
13729 Useful to measure spatial impulse, step responses, chroma delays, etc.
13731 It accepts the following parameters:
13735 Set scope center x position.
13738 Set scope center y position.
13741 Set scope size, relative to frame diagonal.
13744 Set scope tilt/rotation.
13750 Set trace center x position.
13753 Set trace center y position.
13756 Set trace width, relative to width of frame.
13759 Set trace height, relative to height of frame.
13762 Set which components to trace. By default it traces first three components.
13765 Draw trace grid. By default is enabled.
13768 Draw some statistics. By default is enabled.
13771 Draw scope. By default is enabled.
13774 @subsection Examples
13778 Inspect full first row of video frame.
13780 oscilloscope=x=0.5:y=0:s=1
13784 Inspect full last row of video frame.
13786 oscilloscope=x=0.5:y=1:s=1
13790 Inspect full 5th line of video frame of height 1080.
13792 oscilloscope=x=0.5:y=5/1080:s=1
13796 Inspect full last column of video frame.
13798 oscilloscope=x=1:y=0.5:s=1:t=1
13806 Overlay one video on top of another.
13808 It takes two inputs and has one output. The first input is the "main"
13809 video on which the second input is overlaid.
13811 It accepts the following parameters:
13813 A description of the accepted options follows.
13818 Set the expression for the x and y coordinates of the overlaid video
13819 on the main video. Default value is "0" for both expressions. In case
13820 the expression is invalid, it is set to a huge value (meaning that the
13821 overlay will not be displayed within the output visible area).
13824 See @ref{framesync}.
13827 Set when the expressions for @option{x}, and @option{y} are evaluated.
13829 It accepts the following values:
13832 only evaluate expressions once during the filter initialization or
13833 when a command is processed
13836 evaluate expressions for each incoming frame
13839 Default value is @samp{frame}.
13842 See @ref{framesync}.
13845 Set the format for the output video.
13847 It accepts the following values:
13850 force YUV420 output
13853 force YUV422 output
13856 force YUV444 output
13859 force packed RGB output
13862 force planar RGB output
13865 automatically pick format
13868 Default value is @samp{yuv420}.
13871 See @ref{framesync}.
13874 Set format of alpha of the overlaid video, it can be @var{straight} or
13875 @var{premultiplied}. Default is @var{straight}.
13878 The @option{x}, and @option{y} expressions can contain the following
13884 The main input width and height.
13888 The overlay input width and height.
13892 The computed values for @var{x} and @var{y}. They are evaluated for
13897 horizontal and vertical chroma subsample values of the output
13898 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
13902 the number of input frame, starting from 0
13905 the position in the file of the input frame, NAN if unknown
13908 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
13912 This filter also supports the @ref{framesync} options.
13914 Note that the @var{n}, @var{pos}, @var{t} variables are available only
13915 when evaluation is done @emph{per frame}, and will evaluate to NAN
13916 when @option{eval} is set to @samp{init}.
13918 Be aware that frames are taken from each input video in timestamp
13919 order, hence, if their initial timestamps differ, it is a good idea
13920 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
13921 have them begin in the same zero timestamp, as the example for
13922 the @var{movie} filter does.
13924 You can chain together more overlays but you should test the
13925 efficiency of such approach.
13927 @subsection Commands
13929 This filter supports the following commands:
13933 Modify the x and y of the overlay input.
13934 The command accepts the same syntax of the corresponding option.
13936 If the specified expression is not valid, it is kept at its current
13940 @subsection Examples
13944 Draw the overlay at 10 pixels from the bottom right corner of the main
13947 overlay=main_w-overlay_w-10:main_h-overlay_h-10
13950 Using named options the example above becomes:
13952 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
13956 Insert a transparent PNG logo in the bottom left corner of the input,
13957 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
13959 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
13963 Insert 2 different transparent PNG logos (second logo on bottom
13964 right corner) using the @command{ffmpeg} tool:
13966 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
13970 Add a transparent color layer on top of the main video; @code{WxH}
13971 must specify the size of the main input to the overlay filter:
13973 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
13977 Play an original video and a filtered version (here with the deshake
13978 filter) side by side using the @command{ffplay} tool:
13980 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
13983 The above command is the same as:
13985 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
13989 Make a sliding overlay appearing from the left to the right top part of the
13990 screen starting since time 2:
13992 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
13996 Compose output by putting two input videos side to side:
13998 ffmpeg -i left.avi -i right.avi -filter_complex "
13999 nullsrc=size=200x100 [background];
14000 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14001 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14002 [background][left] overlay=shortest=1 [background+left];
14003 [background+left][right] overlay=shortest=1:x=100 [left+right]
14008 Mask 10-20 seconds of a video by applying the delogo filter to a section
14010 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14011 -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]'
14016 Chain several overlays in cascade:
14018 nullsrc=s=200x200 [bg];
14019 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14020 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14021 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14022 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14023 [in3] null, [mid2] overlay=100:100 [out0]
14030 Apply Overcomplete Wavelet denoiser.
14032 The filter accepts the following options:
14038 Larger depth values will denoise lower frequency components more, but
14039 slow down filtering.
14041 Must be an int in the range 8-16, default is @code{8}.
14043 @item luma_strength, ls
14046 Must be a double value in the range 0-1000, default is @code{1.0}.
14048 @item chroma_strength, cs
14049 Set chroma strength.
14051 Must be a double value in the range 0-1000, default is @code{1.0}.
14057 Add paddings to the input image, and place the original input at the
14058 provided @var{x}, @var{y} coordinates.
14060 It accepts the following parameters:
14065 Specify an expression for the size of the output image with the
14066 paddings added. If the value for @var{width} or @var{height} is 0, the
14067 corresponding input size is used for the output.
14069 The @var{width} expression can reference the value set by the
14070 @var{height} expression, and vice versa.
14072 The default value of @var{width} and @var{height} is 0.
14076 Specify the offsets to place the input image at within the padded area,
14077 with respect to the top/left border of the output image.
14079 The @var{x} expression can reference the value set by the @var{y}
14080 expression, and vice versa.
14082 The default value of @var{x} and @var{y} is 0.
14084 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14085 so the input image is centered on the padded area.
14088 Specify the color of the padded area. For the syntax of this option,
14089 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14090 manual,ffmpeg-utils}.
14092 The default value of @var{color} is "black".
14095 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
14097 It accepts the following values:
14101 Only evaluate expressions once during the filter initialization or when
14102 a command is processed.
14105 Evaluate expressions for each incoming frame.
14109 Default value is @samp{init}.
14112 Pad to aspect instead to a resolution.
14116 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
14117 options are expressions containing the following constants:
14122 The input video width and height.
14126 These are the same as @var{in_w} and @var{in_h}.
14130 The output width and height (the size of the padded area), as
14131 specified by the @var{width} and @var{height} expressions.
14135 These are the same as @var{out_w} and @var{out_h}.
14139 The x and y offsets as specified by the @var{x} and @var{y}
14140 expressions, or NAN if not yet specified.
14143 same as @var{iw} / @var{ih}
14146 input sample aspect ratio
14149 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
14153 The horizontal and vertical chroma subsample values. For example for the
14154 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
14157 @subsection Examples
14161 Add paddings with the color "violet" to the input video. The output video
14162 size is 640x480, and the top-left corner of the input video is placed at
14165 pad=640:480:0:40:violet
14168 The example above is equivalent to the following command:
14170 pad=width=640:height=480:x=0:y=40:color=violet
14174 Pad the input to get an output with dimensions increased by 3/2,
14175 and put the input video at the center of the padded area:
14177 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
14181 Pad the input to get a squared output with size equal to the maximum
14182 value between the input width and height, and put the input video at
14183 the center of the padded area:
14185 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
14189 Pad the input to get a final w/h ratio of 16:9:
14191 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
14195 In case of anamorphic video, in order to set the output display aspect
14196 correctly, it is necessary to use @var{sar} in the expression,
14197 according to the relation:
14199 (ih * X / ih) * sar = output_dar
14200 X = output_dar / sar
14203 Thus the previous example needs to be modified to:
14205 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
14209 Double the output size and put the input video in the bottom-right
14210 corner of the output padded area:
14212 pad="2*iw:2*ih:ow-iw:oh-ih"
14216 @anchor{palettegen}
14217 @section palettegen
14219 Generate one palette for a whole video stream.
14221 It accepts the following options:
14225 Set the maximum number of colors to quantize in the palette.
14226 Note: the palette will still contain 256 colors; the unused palette entries
14229 @item reserve_transparent
14230 Create a palette of 255 colors maximum and reserve the last one for
14231 transparency. Reserving the transparency color is useful for GIF optimization.
14232 If not set, the maximum of colors in the palette will be 256. You probably want
14233 to disable this option for a standalone image.
14236 @item transparency_color
14237 Set the color that will be used as background for transparency.
14240 Set statistics mode.
14242 It accepts the following values:
14245 Compute full frame histograms.
14247 Compute histograms only for the part that differs from previous frame. This
14248 might be relevant to give more importance to the moving part of your input if
14249 the background is static.
14251 Compute new histogram for each frame.
14254 Default value is @var{full}.
14257 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
14258 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
14259 color quantization of the palette. This information is also visible at
14260 @var{info} logging level.
14262 @subsection Examples
14266 Generate a representative palette of a given video using @command{ffmpeg}:
14268 ffmpeg -i input.mkv -vf palettegen palette.png
14272 @section paletteuse
14274 Use a palette to downsample an input video stream.
14276 The filter takes two inputs: one video stream and a palette. The palette must
14277 be a 256 pixels image.
14279 It accepts the following options:
14283 Select dithering mode. Available algorithms are:
14286 Ordered 8x8 bayer dithering (deterministic)
14288 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
14289 Note: this dithering is sometimes considered "wrong" and is included as a
14291 @item floyd_steinberg
14292 Floyd and Steingberg dithering (error diffusion)
14294 Frankie Sierra dithering v2 (error diffusion)
14296 Frankie Sierra dithering v2 "Lite" (error diffusion)
14299 Default is @var{sierra2_4a}.
14302 When @var{bayer} dithering is selected, this option defines the scale of the
14303 pattern (how much the crosshatch pattern is visible). A low value means more
14304 visible pattern for less banding, and higher value means less visible pattern
14305 at the cost of more banding.
14307 The option must be an integer value in the range [0,5]. Default is @var{2}.
14310 If set, define the zone to process
14314 Only the changing rectangle will be reprocessed. This is similar to GIF
14315 cropping/offsetting compression mechanism. This option can be useful for speed
14316 if only a part of the image is changing, and has use cases such as limiting the
14317 scope of the error diffusal @option{dither} to the rectangle that bounds the
14318 moving scene (it leads to more deterministic output if the scene doesn't change
14319 much, and as a result less moving noise and better GIF compression).
14322 Default is @var{none}.
14325 Take new palette for each output frame.
14327 @item alpha_threshold
14328 Sets the alpha threshold for transparency. Alpha values above this threshold
14329 will be treated as completely opaque, and values below this threshold will be
14330 treated as completely transparent.
14332 The option must be an integer value in the range [0,255]. Default is @var{128}.
14335 @subsection Examples
14339 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
14340 using @command{ffmpeg}:
14342 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
14346 @section perspective
14348 Correct perspective of video not recorded perpendicular to the screen.
14350 A description of the accepted parameters follows.
14361 Set coordinates expression for top left, top right, bottom left and bottom right corners.
14362 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
14363 If the @code{sense} option is set to @code{source}, then the specified points will be sent
14364 to the corners of the destination. If the @code{sense} option is set to @code{destination},
14365 then the corners of the source will be sent to the specified coordinates.
14367 The expressions can use the following variables:
14372 the width and height of video frame.
14376 Output frame count.
14379 @item interpolation
14380 Set interpolation for perspective correction.
14382 It accepts the following values:
14388 Default value is @samp{linear}.
14391 Set interpretation of coordinate options.
14393 It accepts the following values:
14397 Send point in the source specified by the given coordinates to
14398 the corners of the destination.
14400 @item 1, destination
14402 Send the corners of the source to the point in the destination specified
14403 by the given coordinates.
14405 Default value is @samp{source}.
14409 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
14411 It accepts the following values:
14414 only evaluate expressions once during the filter initialization or
14415 when a command is processed
14418 evaluate expressions for each incoming frame
14421 Default value is @samp{init}.
14426 Delay interlaced video by one field time so that the field order changes.
14428 The intended use is to fix PAL movies that have been captured with the
14429 opposite field order to the film-to-video transfer.
14431 A description of the accepted parameters follows.
14437 It accepts the following values:
14440 Capture field order top-first, transfer bottom-first.
14441 Filter will delay the bottom field.
14444 Capture field order bottom-first, transfer top-first.
14445 Filter will delay the top field.
14448 Capture and transfer with the same field order. This mode only exists
14449 for the documentation of the other options to refer to, but if you
14450 actually select it, the filter will faithfully do nothing.
14453 Capture field order determined automatically by field flags, transfer
14455 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
14456 basis using field flags. If no field information is available,
14457 then this works just like @samp{u}.
14460 Capture unknown or varying, transfer opposite.
14461 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
14462 analyzing the images and selecting the alternative that produces best
14463 match between the fields.
14466 Capture top-first, transfer unknown or varying.
14467 Filter selects among @samp{t} and @samp{p} using image analysis.
14470 Capture bottom-first, transfer unknown or varying.
14471 Filter selects among @samp{b} and @samp{p} using image analysis.
14474 Capture determined by field flags, transfer unknown or varying.
14475 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
14476 image analysis. If no field information is available, then this works just
14477 like @samp{U}. This is the default mode.
14480 Both capture and transfer unknown or varying.
14481 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
14485 @section photosensitivity
14486 Reduce various flashes in video, so to help users with epilepsy.
14488 It accepts the following options:
14491 Set how many frames to use when filtering. Default is 30.
14494 Set detection threshold factor. Default is 1.
14498 Set how many pixels to skip when sampling frames. Default is 1.
14499 Allowed range is from 1 to 1024.
14502 Leave frames unchanged. Default is disabled.
14505 @section pixdesctest
14507 Pixel format descriptor test filter, mainly useful for internal
14508 testing. The output video should be equal to the input video.
14512 format=monow, pixdesctest
14515 can be used to test the monowhite pixel format descriptor definition.
14519 Display sample values of color channels. Mainly useful for checking color
14520 and levels. Minimum supported resolution is 640x480.
14522 The filters accept the following options:
14526 Set scope X position, relative offset on X axis.
14529 Set scope Y position, relative offset on Y axis.
14538 Set window opacity. This window also holds statistics about pixel area.
14541 Set window X position, relative offset on X axis.
14544 Set window Y position, relative offset on Y axis.
14549 Enable the specified chain of postprocessing subfilters using libpostproc. This
14550 library should be automatically selected with a GPL build (@code{--enable-gpl}).
14551 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
14552 Each subfilter and some options have a short and a long name that can be used
14553 interchangeably, i.e. dr/dering are the same.
14555 The filters accept the following options:
14559 Set postprocessing subfilters string.
14562 All subfilters share common options to determine their scope:
14566 Honor the quality commands for this subfilter.
14569 Do chrominance filtering, too (default).
14572 Do luminance filtering only (no chrominance).
14575 Do chrominance filtering only (no luminance).
14578 These options can be appended after the subfilter name, separated by a '|'.
14580 Available subfilters are:
14583 @item hb/hdeblock[|difference[|flatness]]
14584 Horizontal deblocking filter
14587 Difference factor where higher values mean more deblocking (default: @code{32}).
14589 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14592 @item vb/vdeblock[|difference[|flatness]]
14593 Vertical deblocking filter
14596 Difference factor where higher values mean more deblocking (default: @code{32}).
14598 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14601 @item ha/hadeblock[|difference[|flatness]]
14602 Accurate horizontal deblocking filter
14605 Difference factor where higher values mean more deblocking (default: @code{32}).
14607 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14610 @item va/vadeblock[|difference[|flatness]]
14611 Accurate vertical deblocking filter
14614 Difference factor where higher values mean more deblocking (default: @code{32}).
14616 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14620 The horizontal and vertical deblocking filters share the difference and
14621 flatness values so you cannot set different horizontal and vertical
14625 @item h1/x1hdeblock
14626 Experimental horizontal deblocking filter
14628 @item v1/x1vdeblock
14629 Experimental vertical deblocking filter
14634 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
14637 larger -> stronger filtering
14639 larger -> stronger filtering
14641 larger -> stronger filtering
14644 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
14647 Stretch luminance to @code{0-255}.
14650 @item lb/linblenddeint
14651 Linear blend deinterlacing filter that deinterlaces the given block by
14652 filtering all lines with a @code{(1 2 1)} filter.
14654 @item li/linipoldeint
14655 Linear interpolating deinterlacing filter that deinterlaces the given block by
14656 linearly interpolating every second line.
14658 @item ci/cubicipoldeint
14659 Cubic interpolating deinterlacing filter deinterlaces the given block by
14660 cubically interpolating every second line.
14662 @item md/mediandeint
14663 Median deinterlacing filter that deinterlaces the given block by applying a
14664 median filter to every second line.
14666 @item fd/ffmpegdeint
14667 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
14668 second line with a @code{(-1 4 2 4 -1)} filter.
14671 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
14672 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
14674 @item fq/forceQuant[|quantizer]
14675 Overrides the quantizer table from the input with the constant quantizer you
14683 Default pp filter combination (@code{hb|a,vb|a,dr|a})
14686 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
14689 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
14692 @subsection Examples
14696 Apply horizontal and vertical deblocking, deringing and automatic
14697 brightness/contrast:
14703 Apply default filters without brightness/contrast correction:
14709 Apply default filters and temporal denoiser:
14711 pp=default/tmpnoise|1|2|3
14715 Apply deblocking on luminance only, and switch vertical deblocking on or off
14716 automatically depending on available CPU time:
14723 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
14724 similar to spp = 6 with 7 point DCT, where only the center sample is
14727 The filter accepts the following options:
14731 Force a constant quantization parameter. It accepts an integer in range
14732 0 to 63. If not set, the filter will use the QP from the video stream
14736 Set thresholding mode. Available modes are:
14740 Set hard thresholding.
14742 Set soft thresholding (better de-ringing effect, but likely blurrier).
14744 Set medium thresholding (good results, default).
14748 @section premultiply
14749 Apply alpha premultiply effect to input video stream using first plane
14750 of second stream as alpha.
14752 Both streams must have same dimensions and same pixel format.
14754 The filter accepts the following option:
14758 Set which planes will be processed, unprocessed planes will be copied.
14759 By default value 0xf, all planes will be processed.
14762 Do not require 2nd input for processing, instead use alpha plane from input stream.
14766 Apply prewitt operator to input video stream.
14768 The filter accepts the following option:
14772 Set which planes will be processed, unprocessed planes will be copied.
14773 By default value 0xf, all planes will be processed.
14776 Set value which will be multiplied with filtered result.
14779 Set value which will be added to filtered result.
14782 @anchor{program_opencl}
14783 @section program_opencl
14785 Filter video using an OpenCL program.
14790 OpenCL program source file.
14793 Kernel name in program.
14796 Number of inputs to the filter. Defaults to 1.
14799 Size of output frames. Defaults to the same as the first input.
14803 The program source file must contain a kernel function with the given name,
14804 which will be run once for each plane of the output. Each run on a plane
14805 gets enqueued as a separate 2D global NDRange with one work-item for each
14806 pixel to be generated. The global ID offset for each work-item is therefore
14807 the coordinates of a pixel in the destination image.
14809 The kernel function needs to take the following arguments:
14812 Destination image, @var{__write_only image2d_t}.
14814 This image will become the output; the kernel should write all of it.
14816 Frame index, @var{unsigned int}.
14818 This is a counter starting from zero and increasing by one for each frame.
14820 Source images, @var{__read_only image2d_t}.
14822 These are the most recent images on each input. The kernel may read from
14823 them to generate the output, but they can't be written to.
14830 Copy the input to the output (output must be the same size as the input).
14832 __kernel void copy(__write_only image2d_t destination,
14833 unsigned int index,
14834 __read_only image2d_t source)
14836 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
14838 int2 location = (int2)(get_global_id(0), get_global_id(1));
14840 float4 value = read_imagef(source, sampler, location);
14842 write_imagef(destination, location, value);
14847 Apply a simple transformation, rotating the input by an amount increasing
14848 with the index counter. Pixel values are linearly interpolated by the
14849 sampler, and the output need not have the same dimensions as the input.
14851 __kernel void rotate_image(__write_only image2d_t dst,
14852 unsigned int index,
14853 __read_only image2d_t src)
14855 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
14856 CLK_FILTER_LINEAR);
14858 float angle = (float)index / 100.0f;
14860 float2 dst_dim = convert_float2(get_image_dim(dst));
14861 float2 src_dim = convert_float2(get_image_dim(src));
14863 float2 dst_cen = dst_dim / 2.0f;
14864 float2 src_cen = src_dim / 2.0f;
14866 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
14868 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
14870 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
14871 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
14873 src_pos = src_pos * src_dim / dst_dim;
14875 float2 src_loc = src_pos + src_cen;
14877 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
14878 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
14879 write_imagef(dst, dst_loc, 0.5f);
14881 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
14886 Blend two inputs together, with the amount of each input used varying
14887 with the index counter.
14889 __kernel void blend_images(__write_only image2d_t dst,
14890 unsigned int index,
14891 __read_only image2d_t src1,
14892 __read_only image2d_t src2)
14894 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
14895 CLK_FILTER_LINEAR);
14897 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
14899 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
14900 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
14901 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
14903 float4 val1 = read_imagef(src1, sampler, src1_loc);
14904 float4 val2 = read_imagef(src2, sampler, src2_loc);
14906 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
14912 @section pseudocolor
14914 Alter frame colors in video with pseudocolors.
14916 This filter accepts the following options:
14920 set pixel first component expression
14923 set pixel second component expression
14926 set pixel third component expression
14929 set pixel fourth component expression, corresponds to the alpha component
14932 set component to use as base for altering colors
14935 Each of them specifies the expression to use for computing the lookup table for
14936 the corresponding pixel component values.
14938 The expressions can contain the following constants and functions:
14943 The input width and height.
14946 The input value for the pixel component.
14948 @item ymin, umin, vmin, amin
14949 The minimum allowed component value.
14951 @item ymax, umax, vmax, amax
14952 The maximum allowed component value.
14955 All expressions default to "val".
14957 @subsection Examples
14961 Change too high luma values to gradient:
14963 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'"
14969 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
14970 Ratio) between two input videos.
14972 This filter takes in input two input videos, the first input is
14973 considered the "main" source and is passed unchanged to the
14974 output. The second input is used as a "reference" video for computing
14977 Both video inputs must have the same resolution and pixel format for
14978 this filter to work correctly. Also it assumes that both inputs
14979 have the same number of frames, which are compared one by one.
14981 The obtained average PSNR is printed through the logging system.
14983 The filter stores the accumulated MSE (mean squared error) of each
14984 frame, and at the end of the processing it is averaged across all frames
14985 equally, and the following formula is applied to obtain the PSNR:
14988 PSNR = 10*log10(MAX^2/MSE)
14991 Where MAX is the average of the maximum values of each component of the
14994 The description of the accepted parameters follows.
14997 @item stats_file, f
14998 If specified the filter will use the named file to save the PSNR of
14999 each individual frame. When filename equals "-" the data is sent to
15002 @item stats_version
15003 Specifies which version of the stats file format to use. Details of
15004 each format are written below.
15005 Default value is 1.
15007 @item stats_add_max
15008 Determines whether the max value is output to the stats log.
15009 Default value is 0.
15010 Requires stats_version >= 2. If this is set and stats_version < 2,
15011 the filter will return an error.
15014 This filter also supports the @ref{framesync} options.
15016 The file printed if @var{stats_file} is selected, contains a sequence of
15017 key/value pairs of the form @var{key}:@var{value} for each compared
15020 If a @var{stats_version} greater than 1 is specified, a header line precedes
15021 the list of per-frame-pair stats, with key value pairs following the frame
15022 format with the following parameters:
15025 @item psnr_log_version
15026 The version of the log file format. Will match @var{stats_version}.
15029 A comma separated list of the per-frame-pair parameters included in
15033 A description of each shown per-frame-pair parameter follows:
15037 sequential number of the input frame, starting from 1
15040 Mean Square Error pixel-by-pixel average difference of the compared
15041 frames, averaged over all the image components.
15043 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15044 Mean Square Error pixel-by-pixel average difference of the compared
15045 frames for the component specified by the suffix.
15047 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15048 Peak Signal to Noise ratio of the compared frames for the component
15049 specified by the suffix.
15051 @item max_avg, max_y, max_u, max_v
15052 Maximum allowed value for each channel, and average over all
15056 @subsection Examples
15061 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15062 [main][ref] psnr="stats_file=stats.log" [out]
15065 On this example the input file being processed is compared with the
15066 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15067 is stored in @file{stats.log}.
15070 Another example with different containers:
15072 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 -
15079 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15080 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15083 The pullup filter is designed to take advantage of future context in making
15084 its decisions. This filter is stateless in the sense that it does not lock
15085 onto a pattern to follow, but it instead looks forward to the following
15086 fields in order to identify matches and rebuild progressive frames.
15088 To produce content with an even framerate, insert the fps filter after
15089 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15090 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15092 The filter accepts the following options:
15099 These options set the amount of "junk" to ignore at the left, right, top, and
15100 bottom of the image, respectively. Left and right are in units of 8 pixels,
15101 while top and bottom are in units of 2 lines.
15102 The default is 8 pixels on each side.
15105 Set the strict breaks. Setting this option to 1 will reduce the chances of
15106 filter generating an occasional mismatched frame, but it may also cause an
15107 excessive number of frames to be dropped during high motion sequences.
15108 Conversely, setting it to -1 will make filter match fields more easily.
15109 This may help processing of video where there is slight blurring between
15110 the fields, but may also cause there to be interlaced frames in the output.
15111 Default value is @code{0}.
15114 Set the metric plane to use. It accepts the following values:
15120 Use chroma blue plane.
15123 Use chroma red plane.
15126 This option may be set to use chroma plane instead of the default luma plane
15127 for doing filter's computations. This may improve accuracy on very clean
15128 source material, but more likely will decrease accuracy, especially if there
15129 is chroma noise (rainbow effect) or any grayscale video.
15130 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15131 load and make pullup usable in realtime on slow machines.
15134 For best results (without duplicated frames in the output file) it is
15135 necessary to change the output frame rate. For example, to inverse
15136 telecine NTSC input:
15138 ffmpeg -i input -vf pullup -r 24000/1001 ...
15143 Change video quantization parameters (QP).
15145 The filter accepts the following option:
15149 Set expression for quantization parameter.
15152 The expression is evaluated through the eval API and can contain, among others,
15153 the following constants:
15157 1 if index is not 129, 0 otherwise.
15160 Sequential index starting from -129 to 128.
15163 @subsection Examples
15167 Some equation like:
15175 Flush video frames from internal cache of frames into a random order.
15176 No frame is discarded.
15177 Inspired by @ref{frei0r} nervous filter.
15181 Set size in number of frames of internal cache, in range from @code{2} to
15182 @code{512}. Default is @code{30}.
15185 Set seed for random number generator, must be an integer included between
15186 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15187 less than @code{0}, the filter will try to use a good random seed on a
15191 @section readeia608
15193 Read closed captioning (EIA-608) information from the top lines of a video frame.
15195 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15196 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15197 with EIA-608 data (starting from 0). A description of each metadata value follows:
15200 @item lavfi.readeia608.X.cc
15201 The two bytes stored as EIA-608 data (printed in hexadecimal).
15203 @item lavfi.readeia608.X.line
15204 The number of the line on which the EIA-608 data was identified and read.
15207 This filter accepts the following options:
15211 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15214 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15217 Set minimal acceptable amplitude change for sync codes detection.
15218 Default is @code{0.2}. Allowed range is @code{[0.001 - 1]}.
15221 Set the ratio of width reserved for sync code detection.
15222 Default is @code{0.27}. Allowed range is @code{[0.01 - 0.7]}.
15225 Set the max peaks height difference for sync code detection.
15226 Default is @code{0.1}. Allowed range is @code{[0.0 - 0.5]}.
15229 Set max peaks period difference for sync code detection.
15230 Default is @code{0.1}. Allowed range is @code{[0.0 - 0.5]}.
15233 Set the first two max start code bits differences.
15234 Default is @code{0.02}. Allowed range is @code{[0.0 - 0.5]}.
15237 Set the minimum ratio of bits height compared to 3rd start code bit.
15238 Default is @code{0.75}. Allowed range is @code{[0.01 - 1]}.
15241 Set the white color threshold. Default is @code{0.35}. Allowed range is @code{[0.1 - 1]}.
15244 Set the black color threshold. Default is @code{0.15}. Allowed range is @code{[0.0 - 0.5]}.
15247 Enable checking the parity bit. In the event of a parity error, the filter will output
15248 @code{0x00} for that character. Default is false.
15251 Lowpass lines prior to further processing. Default is disabled.
15254 @subsection Examples
15258 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
15260 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
15266 Read vertical interval timecode (VITC) information from the top lines of a
15269 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
15270 timecode value, if a valid timecode has been detected. Further metadata key
15271 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
15272 timecode data has been found or not.
15274 This filter accepts the following options:
15278 Set the maximum number of lines to scan for VITC data. If the value is set to
15279 @code{-1} the full video frame is scanned. Default is @code{45}.
15282 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
15283 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
15286 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
15287 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
15290 @subsection Examples
15294 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
15295 draw @code{--:--:--:--} as a placeholder:
15297 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
15303 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
15305 Destination pixel at position (X, Y) will be picked from source (x, y) position
15306 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
15307 value for pixel will be used for destination pixel.
15309 Xmap and Ymap input video streams must be of same dimensions. Output video stream
15310 will have Xmap/Ymap video stream dimensions.
15311 Xmap and Ymap input video streams are 16bit depth, single channel.
15315 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
15316 Default is @code{color}.
15319 @section removegrain
15321 The removegrain filter is a spatial denoiser for progressive video.
15325 Set mode for the first plane.
15328 Set mode for the second plane.
15331 Set mode for the third plane.
15334 Set mode for the fourth plane.
15337 Range of mode is from 0 to 24. Description of each mode follows:
15341 Leave input plane unchanged. Default.
15344 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
15347 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
15350 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
15353 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
15354 This is equivalent to a median filter.
15357 Line-sensitive clipping giving the minimal change.
15360 Line-sensitive clipping, intermediate.
15363 Line-sensitive clipping, intermediate.
15366 Line-sensitive clipping, intermediate.
15369 Line-sensitive clipping on a line where the neighbours pixels are the closest.
15372 Replaces the target pixel with the closest neighbour.
15375 [1 2 1] horizontal and vertical kernel blur.
15381 Bob mode, interpolates top field from the line where the neighbours
15382 pixels are the closest.
15385 Bob mode, interpolates bottom field from the line where the neighbours
15386 pixels are the closest.
15389 Bob mode, interpolates top field. Same as 13 but with a more complicated
15390 interpolation formula.
15393 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
15394 interpolation formula.
15397 Clips the pixel with the minimum and maximum of respectively the maximum and
15398 minimum of each pair of opposite neighbour pixels.
15401 Line-sensitive clipping using opposite neighbours whose greatest distance from
15402 the current pixel is minimal.
15405 Replaces the pixel with the average of its 8 neighbours.
15408 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
15411 Clips pixels using the averages of opposite neighbour.
15414 Same as mode 21 but simpler and faster.
15417 Small edge and halo removal, but reputed useless.
15423 @section removelogo
15425 Suppress a TV station logo, using an image file to determine which
15426 pixels comprise the logo. It works by filling in the pixels that
15427 comprise the logo with neighboring pixels.
15429 The filter accepts the following options:
15433 Set the filter bitmap file, which can be any image format supported by
15434 libavformat. The width and height of the image file must match those of the
15435 video stream being processed.
15438 Pixels in the provided bitmap image with a value of zero are not
15439 considered part of the logo, non-zero pixels are considered part of
15440 the logo. If you use white (255) for the logo and black (0) for the
15441 rest, you will be safe. For making the filter bitmap, it is
15442 recommended to take a screen capture of a black frame with the logo
15443 visible, and then using a threshold filter followed by the erode
15444 filter once or twice.
15446 If needed, little splotches can be fixed manually. Remember that if
15447 logo pixels are not covered, the filter quality will be much
15448 reduced. Marking too many pixels as part of the logo does not hurt as
15449 much, but it will increase the amount of blurring needed to cover over
15450 the image and will destroy more information than necessary, and extra
15451 pixels will slow things down on a large logo.
15453 @section repeatfields
15455 This filter uses the repeat_field flag from the Video ES headers and hard repeats
15456 fields based on its value.
15460 Reverse a video clip.
15462 Warning: This filter requires memory to buffer the entire clip, so trimming
15465 @subsection Examples
15469 Take the first 5 seconds of a clip, and reverse it.
15476 Shift R/G/B/A pixels horizontally and/or vertically.
15478 The filter accepts the following options:
15481 Set amount to shift red horizontally.
15483 Set amount to shift red vertically.
15485 Set amount to shift green horizontally.
15487 Set amount to shift green vertically.
15489 Set amount to shift blue horizontally.
15491 Set amount to shift blue vertically.
15493 Set amount to shift alpha horizontally.
15495 Set amount to shift alpha vertically.
15497 Set edge mode, can be @var{smear}, default, or @var{warp}.
15501 Apply roberts cross operator to input video stream.
15503 The filter accepts the following option:
15507 Set which planes will be processed, unprocessed planes will be copied.
15508 By default value 0xf, all planes will be processed.
15511 Set value which will be multiplied with filtered result.
15514 Set value which will be added to filtered result.
15519 Rotate video by an arbitrary angle expressed in radians.
15521 The filter accepts the following options:
15523 A description of the optional parameters follows.
15526 Set an expression for the angle by which to rotate the input video
15527 clockwise, expressed as a number of radians. A negative value will
15528 result in a counter-clockwise rotation. By default it is set to "0".
15530 This expression is evaluated for each frame.
15533 Set the output width expression, default value is "iw".
15534 This expression is evaluated just once during configuration.
15537 Set the output height expression, default value is "ih".
15538 This expression is evaluated just once during configuration.
15541 Enable bilinear interpolation if set to 1, a value of 0 disables
15542 it. Default value is 1.
15545 Set the color used to fill the output area not covered by the rotated
15546 image. For the general syntax of this option, check the
15547 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
15548 If the special value "none" is selected then no
15549 background is printed (useful for example if the background is never shown).
15551 Default value is "black".
15554 The expressions for the angle and the output size can contain the
15555 following constants and functions:
15559 sequential number of the input frame, starting from 0. It is always NAN
15560 before the first frame is filtered.
15563 time in seconds of the input frame, it is set to 0 when the filter is
15564 configured. It is always NAN before the first frame is filtered.
15568 horizontal and vertical chroma subsample values. For example for the
15569 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15573 the input video width and height
15577 the output width and height, that is the size of the padded area as
15578 specified by the @var{width} and @var{height} expressions
15582 the minimal width/height required for completely containing the input
15583 video rotated by @var{a} radians.
15585 These are only available when computing the @option{out_w} and
15586 @option{out_h} expressions.
15589 @subsection Examples
15593 Rotate the input by PI/6 radians clockwise:
15599 Rotate the input by PI/6 radians counter-clockwise:
15605 Rotate the input by 45 degrees clockwise:
15611 Apply a constant rotation with period T, starting from an angle of PI/3:
15613 rotate=PI/3+2*PI*t/T
15617 Make the input video rotation oscillating with a period of T
15618 seconds and an amplitude of A radians:
15620 rotate=A*sin(2*PI/T*t)
15624 Rotate the video, output size is chosen so that the whole rotating
15625 input video is always completely contained in the output:
15627 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
15631 Rotate the video, reduce the output size so that no background is ever
15634 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
15638 @subsection Commands
15640 The filter supports the following commands:
15644 Set the angle expression.
15645 The command accepts the same syntax of the corresponding option.
15647 If the specified expression is not valid, it is kept at its current
15653 Apply Shape Adaptive Blur.
15655 The filter accepts the following options:
15658 @item luma_radius, lr
15659 Set luma blur filter strength, must be a value in range 0.1-4.0, default
15660 value is 1.0. A greater value will result in a more blurred image, and
15661 in slower processing.
15663 @item luma_pre_filter_radius, lpfr
15664 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
15667 @item luma_strength, ls
15668 Set luma maximum difference between pixels to still be considered, must
15669 be a value in the 0.1-100.0 range, default value is 1.0.
15671 @item chroma_radius, cr
15672 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
15673 greater value will result in a more blurred image, and in slower
15676 @item chroma_pre_filter_radius, cpfr
15677 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
15679 @item chroma_strength, cs
15680 Set chroma maximum difference between pixels to still be considered,
15681 must be a value in the -0.9-100.0 range.
15684 Each chroma option value, if not explicitly specified, is set to the
15685 corresponding luma option value.
15690 Scale (resize) the input video, using the libswscale library.
15692 The scale filter forces the output display aspect ratio to be the same
15693 of the input, by changing the output sample aspect ratio.
15695 If the input image format is different from the format requested by
15696 the next filter, the scale filter will convert the input to the
15699 @subsection Options
15700 The filter accepts the following options, or any of the options
15701 supported by the libswscale scaler.
15703 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
15704 the complete list of scaler options.
15709 Set the output video dimension expression. Default value is the input
15712 If the @var{width} or @var{w} value is 0, the input width is used for
15713 the output. If the @var{height} or @var{h} value is 0, the input height
15714 is used for the output.
15716 If one and only one of the values is -n with n >= 1, the scale filter
15717 will use a value that maintains the aspect ratio of the input image,
15718 calculated from the other specified dimension. After that it will,
15719 however, make sure that the calculated dimension is divisible by n and
15720 adjust the value if necessary.
15722 If both values are -n with n >= 1, the behavior will be identical to
15723 both values being set to 0 as previously detailed.
15725 See below for the list of accepted constants for use in the dimension
15729 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
15733 Only evaluate expressions once during the filter initialization or when a command is processed.
15736 Evaluate expressions for each incoming frame.
15740 Default value is @samp{init}.
15744 Set the interlacing mode. It accepts the following values:
15748 Force interlaced aware scaling.
15751 Do not apply interlaced scaling.
15754 Select interlaced aware scaling depending on whether the source frames
15755 are flagged as interlaced or not.
15758 Default value is @samp{0}.
15761 Set libswscale scaling flags. See
15762 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15763 complete list of values. If not explicitly specified the filter applies
15767 @item param0, param1
15768 Set libswscale input parameters for scaling algorithms that need them. See
15769 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15770 complete documentation. If not explicitly specified the filter applies
15776 Set the video size. For the syntax of this option, check the
15777 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
15779 @item in_color_matrix
15780 @item out_color_matrix
15781 Set in/output YCbCr color space type.
15783 This allows the autodetected value to be overridden as well as allows forcing
15784 a specific value used for the output and encoder.
15786 If not specified, the color space type depends on the pixel format.
15792 Choose automatically.
15795 Format conforming to International Telecommunication Union (ITU)
15796 Recommendation BT.709.
15799 Set color space conforming to the United States Federal Communications
15800 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
15805 Set color space conforming to:
15809 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
15812 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
15815 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
15820 Set color space conforming to SMPTE ST 240:1999.
15823 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
15828 Set in/output YCbCr sample range.
15830 This allows the autodetected value to be overridden as well as allows forcing
15831 a specific value used for the output and encoder. If not specified, the
15832 range depends on the pixel format. Possible values:
15836 Choose automatically.
15839 Set full range (0-255 in case of 8-bit luma).
15841 @item mpeg/limited/tv
15842 Set "MPEG" range (16-235 in case of 8-bit luma).
15845 @item force_original_aspect_ratio
15846 Enable decreasing or increasing output video width or height if necessary to
15847 keep the original aspect ratio. Possible values:
15851 Scale the video as specified and disable this feature.
15854 The output video dimensions will automatically be decreased if needed.
15857 The output video dimensions will automatically be increased if needed.
15861 One useful instance of this option is that when you know a specific device's
15862 maximum allowed resolution, you can use this to limit the output video to
15863 that, while retaining the aspect ratio. For example, device A allows
15864 1280x720 playback, and your video is 1920x800. Using this option (set it to
15865 decrease) and specifying 1280x720 to the command line makes the output
15868 Please note that this is a different thing than specifying -1 for @option{w}
15869 or @option{h}, you still need to specify the output resolution for this option
15872 @item force_divisible_by
15873 Ensures that both the output dimensions, width and height, are divisible by the
15874 given integer when used together with @option{force_original_aspect_ratio}. This
15875 works similar to using @code{-n} in the @option{w} and @option{h} options.
15877 This option respects the value set for @option{force_original_aspect_ratio},
15878 increasing or decreasing the resolution accordingly. The video's aspect ratio
15879 may be slightly modified.
15881 This option can be handy if you need to have a video fit within or exceed
15882 a defined resolution using @option{force_original_aspect_ratio} but also have
15883 encoder restrictions on width or height divisibility.
15887 The values of the @option{w} and @option{h} options are expressions
15888 containing the following constants:
15893 The input width and height
15897 These are the same as @var{in_w} and @var{in_h}.
15901 The output (scaled) width and height
15905 These are the same as @var{out_w} and @var{out_h}
15908 The same as @var{iw} / @var{ih}
15911 input sample aspect ratio
15914 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
15918 horizontal and vertical input chroma subsample values. For example for the
15919 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15923 horizontal and vertical output chroma subsample values. For example for the
15924 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15927 @subsection Examples
15931 Scale the input video to a size of 200x100
15936 This is equivalent to:
15947 Specify a size abbreviation for the output size:
15952 which can also be written as:
15958 Scale the input to 2x:
15960 scale=w=2*iw:h=2*ih
15964 The above is the same as:
15966 scale=2*in_w:2*in_h
15970 Scale the input to 2x with forced interlaced scaling:
15972 scale=2*iw:2*ih:interl=1
15976 Scale the input to half size:
15978 scale=w=iw/2:h=ih/2
15982 Increase the width, and set the height to the same size:
15988 Seek Greek harmony:
15995 Increase the height, and set the width to 3/2 of the height:
15997 scale=w=3/2*oh:h=3/5*ih
16001 Increase the size, making the size a multiple of the chroma
16004 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16008 Increase the width to a maximum of 500 pixels,
16009 keeping the same aspect ratio as the input:
16011 scale=w='min(500\, iw*3/2):h=-1'
16015 Make pixels square by combining scale and setsar:
16017 scale='trunc(ih*dar):ih',setsar=1/1
16021 Make pixels square by combining scale and setsar,
16022 making sure the resulting resolution is even (required by some codecs):
16024 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16028 @subsection Commands
16030 This filter supports the following commands:
16034 Set the output video dimension expression.
16035 The command accepts the same syntax of the corresponding option.
16037 If the specified expression is not valid, it is kept at its current
16043 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16044 format conversion on CUDA video frames. Setting the output width and height
16045 works in the same way as for the @var{scale} filter.
16047 The following additional options are accepted:
16050 The pixel format of the output CUDA frames. If set to the string "same" (the
16051 default), the input format will be kept. Note that automatic format negotiation
16052 and conversion is not yet supported for hardware frames
16055 The interpolation algorithm used for resizing. One of the following:
16062 @item cubic2p_bspline
16063 2-parameter cubic (B=1, C=0)
16065 @item cubic2p_catmullrom
16066 2-parameter cubic (B=0, C=1/2)
16068 @item cubic2p_b05c03
16069 2-parameter cubic (B=1/2, C=3/10)
16081 Scale (resize) the input video, based on a reference video.
16083 See the scale filter for available options, scale2ref supports the same but
16084 uses the reference video instead of the main input as basis. scale2ref also
16085 supports the following additional constants for the @option{w} and
16086 @option{h} options:
16091 The main input video's width and height
16094 The same as @var{main_w} / @var{main_h}
16097 The main input video's sample aspect ratio
16099 @item main_dar, mdar
16100 The main input video's display aspect ratio. Calculated from
16101 @code{(main_w / main_h) * main_sar}.
16105 The main input video's horizontal and vertical chroma subsample values.
16106 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16110 @subsection Examples
16114 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16116 'scale2ref[b][a];[a][b]overlay'
16120 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16122 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16127 Scroll input video horizontally and/or vertically by constant speed.
16129 The filter accepts the following options:
16131 @item horizontal, h
16132 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16133 Negative values changes scrolling direction.
16136 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16137 Negative values changes scrolling direction.
16140 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16143 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16146 @subsection Commands
16148 This filter supports the following @ref{commands}:
16150 @item horizontal, h
16151 Set the horizontal scrolling speed.
16153 Set the vertical scrolling speed.
16156 @anchor{selectivecolor}
16157 @section selectivecolor
16159 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
16160 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
16161 by the "purity" of the color (that is, how saturated it already is).
16163 This filter is similar to the Adobe Photoshop Selective Color tool.
16165 The filter accepts the following options:
16168 @item correction_method
16169 Select color correction method.
16171 Available values are:
16174 Specified adjustments are applied "as-is" (added/subtracted to original pixel
16177 Specified adjustments are relative to the original component value.
16179 Default is @code{absolute}.
16181 Adjustments for red pixels (pixels where the red component is the maximum)
16183 Adjustments for yellow pixels (pixels where the blue component is the minimum)
16185 Adjustments for green pixels (pixels where the green component is the maximum)
16187 Adjustments for cyan pixels (pixels where the red component is the minimum)
16189 Adjustments for blue pixels (pixels where the blue component is the maximum)
16191 Adjustments for magenta pixels (pixels where the green component is the minimum)
16193 Adjustments for white pixels (pixels where all components are greater than 128)
16195 Adjustments for all pixels except pure black and pure white
16197 Adjustments for black pixels (pixels where all components are lesser than 128)
16199 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
16202 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
16203 4 space separated floating point adjustment values in the [-1,1] range,
16204 respectively to adjust the amount of cyan, magenta, yellow and black for the
16205 pixels of its range.
16207 @subsection Examples
16211 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
16212 increase magenta by 27% in blue areas:
16214 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
16218 Use a Photoshop selective color preset:
16220 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
16224 @anchor{separatefields}
16225 @section separatefields
16227 The @code{separatefields} takes a frame-based video input and splits
16228 each frame into its components fields, producing a new half height clip
16229 with twice the frame rate and twice the frame count.
16231 This filter use field-dominance information in frame to decide which
16232 of each pair of fields to place first in the output.
16233 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
16235 @section setdar, setsar
16237 The @code{setdar} filter sets the Display Aspect Ratio for the filter
16240 This is done by changing the specified Sample (aka Pixel) Aspect
16241 Ratio, according to the following equation:
16243 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
16246 Keep in mind that the @code{setdar} filter does not modify the pixel
16247 dimensions of the video frame. Also, the display aspect ratio set by
16248 this filter may be changed by later filters in the filterchain,
16249 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
16252 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
16253 the filter output video.
16255 Note that as a consequence of the application of this filter, the
16256 output display aspect ratio will change according to the equation
16259 Keep in mind that the sample aspect ratio set by the @code{setsar}
16260 filter may be changed by later filters in the filterchain, e.g. if
16261 another "setsar" or a "setdar" filter is applied.
16263 It accepts the following parameters:
16266 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
16267 Set the aspect ratio used by the filter.
16269 The parameter can be a floating point number string, an expression, or
16270 a string of the form @var{num}:@var{den}, where @var{num} and
16271 @var{den} are the numerator and denominator of the aspect ratio. If
16272 the parameter is not specified, it is assumed the value "0".
16273 In case the form "@var{num}:@var{den}" is used, the @code{:} character
16277 Set the maximum integer value to use for expressing numerator and
16278 denominator when reducing the expressed aspect ratio to a rational.
16279 Default value is @code{100}.
16283 The parameter @var{sar} is an expression containing
16284 the following constants:
16288 These are approximated values for the mathematical constants e
16289 (Euler's number), pi (Greek pi), and phi (the golden ratio).
16292 The input width and height.
16295 These are the same as @var{w} / @var{h}.
16298 The input sample aspect ratio.
16301 The input display aspect ratio. It is the same as
16302 (@var{w} / @var{h}) * @var{sar}.
16305 Horizontal and vertical chroma subsample values. For example, for the
16306 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16309 @subsection Examples
16314 To change the display aspect ratio to 16:9, specify one of the following:
16321 To change the sample aspect ratio to 10:11, specify:
16327 To set a display aspect ratio of 16:9, and specify a maximum integer value of
16328 1000 in the aspect ratio reduction, use the command:
16330 setdar=ratio=16/9:max=1000
16338 Force field for the output video frame.
16340 The @code{setfield} filter marks the interlace type field for the
16341 output frames. It does not change the input frame, but only sets the
16342 corresponding property, which affects how the frame is treated by
16343 following filters (e.g. @code{fieldorder} or @code{yadif}).
16345 The filter accepts the following options:
16350 Available values are:
16354 Keep the same field property.
16357 Mark the frame as bottom-field-first.
16360 Mark the frame as top-field-first.
16363 Mark the frame as progressive.
16370 Force frame parameter for the output video frame.
16372 The @code{setparams} filter marks interlace and color range for the
16373 output frames. It does not change the input frame, but only sets the
16374 corresponding property, which affects how the frame is treated by
16379 Available values are:
16383 Keep the same field property (default).
16386 Mark the frame as bottom-field-first.
16389 Mark the frame as top-field-first.
16392 Mark the frame as progressive.
16396 Available values are:
16400 Keep the same color range property (default).
16402 @item unspecified, unknown
16403 Mark the frame as unspecified color range.
16405 @item limited, tv, mpeg
16406 Mark the frame as limited range.
16408 @item full, pc, jpeg
16409 Mark the frame as full range.
16412 @item color_primaries
16413 Set the color primaries.
16414 Available values are:
16418 Keep the same color primaries property (default).
16435 Set the color transfer.
16436 Available values are:
16440 Keep the same color trc property (default).
16462 Set the colorspace.
16463 Available values are:
16467 Keep the same colorspace property (default).
16480 @item chroma-derived-nc
16481 @item chroma-derived-c
16488 Show a line containing various information for each input video frame.
16489 The input video is not modified.
16491 This filter supports the following options:
16495 Calculate checksums of each plane. By default enabled.
16498 The shown line contains a sequence of key/value pairs of the form
16499 @var{key}:@var{value}.
16501 The following values are shown in the output:
16505 The (sequential) number of the input frame, starting from 0.
16508 The Presentation TimeStamp of the input frame, expressed as a number of
16509 time base units. The time base unit depends on the filter input pad.
16512 The Presentation TimeStamp of the input frame, expressed as a number of
16516 The position of the frame in the input stream, or -1 if this information is
16517 unavailable and/or meaningless (for example in case of synthetic video).
16520 The pixel format name.
16523 The sample aspect ratio of the input frame, expressed in the form
16524 @var{num}/@var{den}.
16527 The size of the input frame. For the syntax of this option, check the
16528 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16531 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
16532 for bottom field first).
16535 This is 1 if the frame is a key frame, 0 otherwise.
16538 The picture type of the input frame ("I" for an I-frame, "P" for a
16539 P-frame, "B" for a B-frame, or "?" for an unknown type).
16540 Also refer to the documentation of the @code{AVPictureType} enum and of
16541 the @code{av_get_picture_type_char} function defined in
16542 @file{libavutil/avutil.h}.
16545 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
16547 @item plane_checksum
16548 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
16549 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
16552 @section showpalette
16554 Displays the 256 colors palette of each frame. This filter is only relevant for
16555 @var{pal8} pixel format frames.
16557 It accepts the following option:
16561 Set the size of the box used to represent one palette color entry. Default is
16562 @code{30} (for a @code{30x30} pixel box).
16565 @section shuffleframes
16567 Reorder and/or duplicate and/or drop video frames.
16569 It accepts the following parameters:
16573 Set the destination indexes of input frames.
16574 This is space or '|' separated list of indexes that maps input frames to output
16575 frames. Number of indexes also sets maximal value that each index may have.
16576 '-1' index have special meaning and that is to drop frame.
16579 The first frame has the index 0. The default is to keep the input unchanged.
16581 @subsection Examples
16585 Swap second and third frame of every three frames of the input:
16587 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
16591 Swap 10th and 1st frame of every ten frames of the input:
16593 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
16597 @section shuffleplanes
16599 Reorder and/or duplicate video planes.
16601 It accepts the following parameters:
16606 The index of the input plane to be used as the first output plane.
16609 The index of the input plane to be used as the second output plane.
16612 The index of the input plane to be used as the third output plane.
16615 The index of the input plane to be used as the fourth output plane.
16619 The first plane has the index 0. The default is to keep the input unchanged.
16621 @subsection Examples
16625 Swap the second and third planes of the input:
16627 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
16631 @anchor{signalstats}
16632 @section signalstats
16633 Evaluate various visual metrics that assist in determining issues associated
16634 with the digitization of analog video media.
16636 By default the filter will log these metadata values:
16640 Display the minimal Y value contained within the input frame. Expressed in
16644 Display the Y value at the 10% percentile within the input frame. Expressed in
16648 Display the average Y value within the input frame. Expressed in range of
16652 Display the Y value at the 90% percentile within the input frame. Expressed in
16656 Display the maximum Y value contained within the input frame. Expressed in
16660 Display the minimal U value contained within the input frame. Expressed in
16664 Display the U value at the 10% percentile within the input frame. Expressed in
16668 Display the average U value within the input frame. Expressed in range of
16672 Display the U value at the 90% percentile within the input frame. Expressed in
16676 Display the maximum U value contained within the input frame. Expressed in
16680 Display the minimal V value contained within the input frame. Expressed in
16684 Display the V value at the 10% percentile within the input frame. Expressed in
16688 Display the average V value within the input frame. Expressed in range of
16692 Display the V value at the 90% percentile within the input frame. Expressed in
16696 Display the maximum V value contained within the input frame. Expressed in
16700 Display the minimal saturation value contained within the input frame.
16701 Expressed in range of [0-~181.02].
16704 Display the saturation value at the 10% percentile within the input frame.
16705 Expressed in range of [0-~181.02].
16708 Display the average saturation value within the input frame. Expressed in range
16712 Display the saturation value at the 90% percentile within the input frame.
16713 Expressed in range of [0-~181.02].
16716 Display the maximum saturation value contained within the input frame.
16717 Expressed in range of [0-~181.02].
16720 Display the median value for hue within the input frame. Expressed in range of
16724 Display the average value for hue within the input frame. Expressed in range of
16728 Display the average of sample value difference between all values of the Y
16729 plane in the current frame and corresponding values of the previous input frame.
16730 Expressed in range of [0-255].
16733 Display the average of sample value difference between all values of the U
16734 plane in the current frame and corresponding values of the previous input frame.
16735 Expressed in range of [0-255].
16738 Display the average of sample value difference between all values of the V
16739 plane in the current frame and corresponding values of the previous input frame.
16740 Expressed in range of [0-255].
16743 Display bit depth of Y plane in current frame.
16744 Expressed in range of [0-16].
16747 Display bit depth of U plane in current frame.
16748 Expressed in range of [0-16].
16751 Display bit depth of V plane in current frame.
16752 Expressed in range of [0-16].
16755 The filter accepts the following options:
16761 @option{stat} specify an additional form of image analysis.
16762 @option{out} output video with the specified type of pixel highlighted.
16764 Both options accept the following values:
16768 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
16769 unlike the neighboring pixels of the same field. Examples of temporal outliers
16770 include the results of video dropouts, head clogs, or tape tracking issues.
16773 Identify @var{vertical line repetition}. Vertical line repetition includes
16774 similar rows of pixels within a frame. In born-digital video vertical line
16775 repetition is common, but this pattern is uncommon in video digitized from an
16776 analog source. When it occurs in video that results from the digitization of an
16777 analog source it can indicate concealment from a dropout compensator.
16780 Identify pixels that fall outside of legal broadcast range.
16784 Set the highlight color for the @option{out} option. The default color is
16788 @subsection Examples
16792 Output data of various video metrics:
16794 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
16798 Output specific data about the minimum and maximum values of the Y plane per frame:
16800 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
16804 Playback video while highlighting pixels that are outside of broadcast range in red.
16806 ffplay example.mov -vf signalstats="out=brng:color=red"
16810 Playback video with signalstats metadata drawn over the frame.
16812 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
16815 The contents of signalstat_drawtext.txt used in the command are:
16818 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
16819 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
16820 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
16821 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
16829 Calculates the MPEG-7 Video Signature. The filter can handle more than one
16830 input. In this case the matching between the inputs can be calculated additionally.
16831 The filter always passes through the first input. The signature of each stream can
16832 be written into a file.
16834 It accepts the following options:
16838 Enable or disable the matching process.
16840 Available values are:
16844 Disable the calculation of a matching (default).
16846 Calculate the matching for the whole video and output whether the whole video
16847 matches or only parts.
16849 Calculate only until a matching is found or the video ends. Should be faster in
16854 Set the number of inputs. The option value must be a non negative integer.
16855 Default value is 1.
16858 Set the path to which the output is written. If there is more than one input,
16859 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
16860 integer), that will be replaced with the input number. If no filename is
16861 specified, no output will be written. This is the default.
16864 Choose the output format.
16866 Available values are:
16870 Use the specified binary representation (default).
16872 Use the specified xml representation.
16876 Set threshold to detect one word as similar. The option value must be an integer
16877 greater than zero. The default value is 9000.
16880 Set threshold to detect all words as similar. The option value must be an integer
16881 greater than zero. The default value is 60000.
16884 Set threshold to detect frames as similar. The option value must be an integer
16885 greater than zero. The default value is 116.
16888 Set the minimum length of a sequence in frames to recognize it as matching
16889 sequence. The option value must be a non negative integer value.
16890 The default value is 0.
16893 Set the minimum relation, that matching frames to all frames must have.
16894 The option value must be a double value between 0 and 1. The default value is 0.5.
16897 @subsection Examples
16901 To calculate the signature of an input video and store it in signature.bin:
16903 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
16907 To detect whether two videos match and store the signatures in XML format in
16908 signature0.xml and signature1.xml:
16910 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 -
16918 Blur the input video without impacting the outlines.
16920 It accepts the following options:
16923 @item luma_radius, lr
16924 Set the luma radius. The option value must be a float number in
16925 the range [0.1,5.0] that specifies the variance of the gaussian filter
16926 used to blur the image (slower if larger). Default value is 1.0.
16928 @item luma_strength, ls
16929 Set the luma strength. The option value must be a float number
16930 in the range [-1.0,1.0] that configures the blurring. A value included
16931 in [0.0,1.0] will blur the image whereas a value included in
16932 [-1.0,0.0] will sharpen the image. Default value is 1.0.
16934 @item luma_threshold, lt
16935 Set the luma threshold used as a coefficient to determine
16936 whether a pixel should be blurred or not. The option value must be an
16937 integer in the range [-30,30]. A value of 0 will filter all the image,
16938 a value included in [0,30] will filter flat areas and a value included
16939 in [-30,0] will filter edges. Default value is 0.
16941 @item chroma_radius, cr
16942 Set the chroma radius. The option value must be a float number in
16943 the range [0.1,5.0] that specifies the variance of the gaussian filter
16944 used to blur the image (slower if larger). Default value is @option{luma_radius}.
16946 @item chroma_strength, cs
16947 Set the chroma strength. The option value must be a float number
16948 in the range [-1.0,1.0] that configures the blurring. A value included
16949 in [0.0,1.0] will blur the image whereas a value included in
16950 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
16952 @item chroma_threshold, ct
16953 Set the chroma threshold used as a coefficient to determine
16954 whether a pixel should be blurred or not. The option value must be an
16955 integer in the range [-30,30]. A value of 0 will filter all the image,
16956 a value included in [0,30] will filter flat areas and a value included
16957 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
16960 If a chroma option is not explicitly set, the corresponding luma value
16964 Apply sobel operator to input video stream.
16966 The filter accepts the following option:
16970 Set which planes will be processed, unprocessed planes will be copied.
16971 By default value 0xf, all planes will be processed.
16974 Set value which will be multiplied with filtered result.
16977 Set value which will be added to filtered result.
16983 Apply a simple postprocessing filter that compresses and decompresses the image
16984 at several (or - in the case of @option{quality} level @code{6} - all) shifts
16985 and average the results.
16987 The filter accepts the following options:
16991 Set quality. This option defines the number of levels for averaging. It accepts
16992 an integer in the range 0-6. If set to @code{0}, the filter will have no
16993 effect. A value of @code{6} means the higher quality. For each increment of
16994 that value the speed drops by a factor of approximately 2. Default value is
16998 Force a constant quantization parameter. If not set, the filter will use the QP
16999 from the video stream (if available).
17002 Set thresholding mode. Available modes are:
17006 Set hard thresholding (default).
17008 Set soft thresholding (better de-ringing effect, but likely blurrier).
17011 @item use_bframe_qp
17012 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17013 option may cause flicker since the B-Frames have often larger QP. Default is
17014 @code{0} (not enabled).
17019 Scale the input by applying one of the super-resolution methods based on
17020 convolutional neural networks. Supported models:
17024 Super-Resolution Convolutional Neural Network model (SRCNN).
17025 See @url{https://arxiv.org/abs/1501.00092}.
17028 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17029 See @url{https://arxiv.org/abs/1609.05158}.
17032 Training scripts as well as scripts for model file (.pb) saving can be found at
17033 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17034 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17036 Native model files (.model) can be generated from TensorFlow model
17037 files (.pb) by using tools/python/convert.py
17039 The filter accepts the following options:
17043 Specify which DNN backend to use for model loading and execution. This option accepts
17044 the following values:
17048 Native implementation of DNN loading and execution.
17051 TensorFlow backend. To enable this backend you
17052 need to install the TensorFlow for C library (see
17053 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17054 @code{--enable-libtensorflow}
17057 Default value is @samp{native}.
17060 Set path to model file specifying network architecture and its parameters.
17061 Note that different backends use different file formats. TensorFlow backend
17062 can load files for both formats, while native backend can load files for only
17066 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17067 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17068 input upscaled using bicubic upscaling with proper scale factor.
17073 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17075 This filter takes in input two input videos, the first input is
17076 considered the "main" source and is passed unchanged to the
17077 output. The second input is used as a "reference" video for computing
17080 Both video inputs must have the same resolution and pixel format for
17081 this filter to work correctly. Also it assumes that both inputs
17082 have the same number of frames, which are compared one by one.
17084 The filter stores the calculated SSIM of each frame.
17086 The description of the accepted parameters follows.
17089 @item stats_file, f
17090 If specified the filter will use the named file to save the SSIM of
17091 each individual frame. When filename equals "-" the data is sent to
17095 The file printed if @var{stats_file} is selected, contains a sequence of
17096 key/value pairs of the form @var{key}:@var{value} for each compared
17099 A description of each shown parameter follows:
17103 sequential number of the input frame, starting from 1
17105 @item Y, U, V, R, G, B
17106 SSIM of the compared frames for the component specified by the suffix.
17109 SSIM of the compared frames for the whole frame.
17112 Same as above but in dB representation.
17115 This filter also supports the @ref{framesync} options.
17117 @subsection Examples
17122 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
17123 [main][ref] ssim="stats_file=stats.log" [out]
17126 On this example the input file being processed is compared with the
17127 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
17128 is stored in @file{stats.log}.
17131 Another example with both psnr and ssim at same time:
17133 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
17137 Another example with different containers:
17139 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=1/AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=1/AVTB,setpts=PTS-STARTPTS[ref];[main][ref]ssim" -f null -
17145 Convert between different stereoscopic image formats.
17147 The filters accept the following options:
17151 Set stereoscopic image format of input.
17153 Available values for input image formats are:
17156 side by side parallel (left eye left, right eye right)
17159 side by side crosseye (right eye left, left eye right)
17162 side by side parallel with half width resolution
17163 (left eye left, right eye right)
17166 side by side crosseye with half width resolution
17167 (right eye left, left eye right)
17171 above-below (left eye above, right eye below)
17175 above-below (right eye above, left eye below)
17179 above-below with half height resolution
17180 (left eye above, right eye below)
17184 above-below with half height resolution
17185 (right eye above, left eye below)
17188 alternating frames (left eye first, right eye second)
17191 alternating frames (right eye first, left eye second)
17194 interleaved rows (left eye has top row, right eye starts on next row)
17197 interleaved rows (right eye has top row, left eye starts on next row)
17200 interleaved columns, left eye first
17203 interleaved columns, right eye first
17205 Default value is @samp{sbsl}.
17209 Set stereoscopic image format of output.
17213 side by side parallel (left eye left, right eye right)
17216 side by side crosseye (right eye left, left eye right)
17219 side by side parallel with half width resolution
17220 (left eye left, right eye right)
17223 side by side crosseye with half width resolution
17224 (right eye left, left eye right)
17228 above-below (left eye above, right eye below)
17232 above-below (right eye above, left eye below)
17236 above-below with half height resolution
17237 (left eye above, right eye below)
17241 above-below with half height resolution
17242 (right eye above, left eye below)
17245 alternating frames (left eye first, right eye second)
17248 alternating frames (right eye first, left eye second)
17251 interleaved rows (left eye has top row, right eye starts on next row)
17254 interleaved rows (right eye has top row, left eye starts on next row)
17257 anaglyph red/blue gray
17258 (red filter on left eye, blue filter on right eye)
17261 anaglyph red/green gray
17262 (red filter on left eye, green filter on right eye)
17265 anaglyph red/cyan gray
17266 (red filter on left eye, cyan filter on right eye)
17269 anaglyph red/cyan half colored
17270 (red filter on left eye, cyan filter on right eye)
17273 anaglyph red/cyan color
17274 (red filter on left eye, cyan filter on right eye)
17277 anaglyph red/cyan color optimized with the least squares projection of dubois
17278 (red filter on left eye, cyan filter on right eye)
17281 anaglyph green/magenta gray
17282 (green filter on left eye, magenta filter on right eye)
17285 anaglyph green/magenta half colored
17286 (green filter on left eye, magenta filter on right eye)
17289 anaglyph green/magenta colored
17290 (green filter on left eye, magenta filter on right eye)
17293 anaglyph green/magenta color optimized with the least squares projection of dubois
17294 (green filter on left eye, magenta filter on right eye)
17297 anaglyph yellow/blue gray
17298 (yellow filter on left eye, blue filter on right eye)
17301 anaglyph yellow/blue half colored
17302 (yellow filter on left eye, blue filter on right eye)
17305 anaglyph yellow/blue colored
17306 (yellow filter on left eye, blue filter on right eye)
17309 anaglyph yellow/blue color optimized with the least squares projection of dubois
17310 (yellow filter on left eye, blue filter on right eye)
17313 mono output (left eye only)
17316 mono output (right eye only)
17319 checkerboard, left eye first
17322 checkerboard, right eye first
17325 interleaved columns, left eye first
17328 interleaved columns, right eye first
17334 Default value is @samp{arcd}.
17337 @subsection Examples
17341 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
17347 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
17353 @section streamselect, astreamselect
17354 Select video or audio streams.
17356 The filter accepts the following options:
17360 Set number of inputs. Default is 2.
17363 Set input indexes to remap to outputs.
17366 @subsection Commands
17368 The @code{streamselect} and @code{astreamselect} filter supports the following
17373 Set input indexes to remap to outputs.
17376 @subsection Examples
17380 Select first 5 seconds 1st stream and rest of time 2nd stream:
17382 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
17386 Same as above, but for audio:
17388 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
17395 Draw subtitles on top of input video using the libass library.
17397 To enable compilation of this filter you need to configure FFmpeg with
17398 @code{--enable-libass}. This filter also requires a build with libavcodec and
17399 libavformat to convert the passed subtitles file to ASS (Advanced Substation
17400 Alpha) subtitles format.
17402 The filter accepts the following options:
17406 Set the filename of the subtitle file to read. It must be specified.
17408 @item original_size
17409 Specify the size of the original video, the video for which the ASS file
17410 was composed. For the syntax of this option, check the
17411 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17412 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
17413 correctly scale the fonts if the aspect ratio has been changed.
17416 Set a directory path containing fonts that can be used by the filter.
17417 These fonts will be used in addition to whatever the font provider uses.
17420 Process alpha channel, by default alpha channel is untouched.
17423 Set subtitles input character encoding. @code{subtitles} filter only. Only
17424 useful if not UTF-8.
17426 @item stream_index, si
17427 Set subtitles stream index. @code{subtitles} filter only.
17430 Override default style or script info parameters of the subtitles. It accepts a
17431 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
17434 If the first key is not specified, it is assumed that the first value
17435 specifies the @option{filename}.
17437 For example, to render the file @file{sub.srt} on top of the input
17438 video, use the command:
17443 which is equivalent to:
17445 subtitles=filename=sub.srt
17448 To render the default subtitles stream from file @file{video.mkv}, use:
17450 subtitles=video.mkv
17453 To render the second subtitles stream from that file, use:
17455 subtitles=video.mkv:si=1
17458 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
17459 @code{DejaVu Serif}, use:
17461 subtitles=sub.srt:force_style='FontName=DejaVu Serif,PrimaryColour=&HCCFF0000'
17464 @section super2xsai
17466 Scale the input by 2x and smooth using the Super2xSaI (Scale and
17467 Interpolate) pixel art scaling algorithm.
17469 Useful for enlarging pixel art images without reducing sharpness.
17473 Swap two rectangular objects in video.
17475 This filter accepts the following options:
17485 Set 1st rect x coordinate.
17488 Set 1st rect y coordinate.
17491 Set 2nd rect x coordinate.
17494 Set 2nd rect y coordinate.
17496 All expressions are evaluated once for each frame.
17499 The all options are expressions containing the following constants:
17504 The input width and height.
17507 same as @var{w} / @var{h}
17510 input sample aspect ratio
17513 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
17516 The number of the input frame, starting from 0.
17519 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
17522 the position in the file of the input frame, NAN if unknown
17530 Apply telecine process to the video.
17532 This filter accepts the following options:
17541 The default value is @code{top}.
17545 A string of numbers representing the pulldown pattern you wish to apply.
17546 The default value is @code{23}.
17550 Some typical patterns:
17555 24p: 2332 (preferred)
17562 24p: 222222222223 ("Euro pulldown")
17569 Apply threshold effect to video stream.
17571 This filter needs four video streams to perform thresholding.
17572 First stream is stream we are filtering.
17573 Second stream is holding threshold values, third stream is holding min values,
17574 and last, fourth stream is holding max values.
17576 The filter accepts the following option:
17580 Set which planes will be processed, unprocessed planes will be copied.
17581 By default value 0xf, all planes will be processed.
17584 For example if first stream pixel's component value is less then threshold value
17585 of pixel component from 2nd threshold stream, third stream value will picked,
17586 otherwise fourth stream pixel component value will be picked.
17588 Using color source filter one can perform various types of thresholding:
17590 @subsection Examples
17594 Binary threshold, using gray color as threshold:
17596 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
17600 Inverted binary threshold, using gray color as threshold:
17602 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
17606 Truncate binary threshold, using gray color as threshold:
17608 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
17612 Threshold to zero, using gray color as threshold:
17614 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
17618 Inverted threshold to zero, using gray color as threshold:
17620 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
17625 Select the most representative frame in a given sequence of consecutive frames.
17627 The filter accepts the following options:
17631 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
17632 will pick one of them, and then handle the next batch of @var{n} frames until
17633 the end. Default is @code{100}.
17636 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
17637 value will result in a higher memory usage, so a high value is not recommended.
17639 @subsection Examples
17643 Extract one picture each 50 frames:
17649 Complete example of a thumbnail creation with @command{ffmpeg}:
17651 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
17657 Tile several successive frames together.
17659 The filter accepts the following options:
17664 Set the grid size (i.e. the number of lines and columns). For the syntax of
17665 this option, check the
17666 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17669 Set the maximum number of frames to render in the given area. It must be less
17670 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
17671 the area will be used.
17674 Set the outer border margin in pixels.
17677 Set the inner border thickness (i.e. the number of pixels between frames). For
17678 more advanced padding options (such as having different values for the edges),
17679 refer to the pad video filter.
17682 Specify the color of the unused area. For the syntax of this option, check the
17683 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
17684 The default value of @var{color} is "black".
17687 Set the number of frames to overlap when tiling several successive frames together.
17688 The value must be between @code{0} and @var{nb_frames - 1}.
17691 Set the number of frames to initially be empty before displaying first output frame.
17692 This controls how soon will one get first output frame.
17693 The value must be between @code{0} and @var{nb_frames - 1}.
17696 @subsection Examples
17700 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
17702 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
17704 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
17705 duplicating each output frame to accommodate the originally detected frame
17709 Display @code{5} pictures in an area of @code{3x2} frames,
17710 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
17711 mixed flat and named options:
17713 tile=3x2:nb_frames=5:padding=7:margin=2
17717 @section tinterlace
17719 Perform various types of temporal field interlacing.
17721 Frames are counted starting from 1, so the first input frame is
17724 The filter accepts the following options:
17729 Specify the mode of the interlacing. This option can also be specified
17730 as a value alone. See below for a list of values for this option.
17732 Available values are:
17736 Move odd frames into the upper field, even into the lower field,
17737 generating a double height frame at half frame rate.
17741 Frame 1 Frame 2 Frame 3 Frame 4
17743 11111 22222 33333 44444
17744 11111 22222 33333 44444
17745 11111 22222 33333 44444
17746 11111 22222 33333 44444
17760 Only output odd frames, even frames are dropped, generating a frame with
17761 unchanged height at half frame rate.
17766 Frame 1 Frame 2 Frame 3 Frame 4
17768 11111 22222 33333 44444
17769 11111 22222 33333 44444
17770 11111 22222 33333 44444
17771 11111 22222 33333 44444
17781 Only output even frames, odd frames are dropped, generating a frame with
17782 unchanged height at half frame rate.
17787 Frame 1 Frame 2 Frame 3 Frame 4
17789 11111 22222 33333 44444
17790 11111 22222 33333 44444
17791 11111 22222 33333 44444
17792 11111 22222 33333 44444
17802 Expand each frame to full height, but pad alternate lines with black,
17803 generating a frame with double height at the same input frame rate.
17808 Frame 1 Frame 2 Frame 3 Frame 4
17810 11111 22222 33333 44444
17811 11111 22222 33333 44444
17812 11111 22222 33333 44444
17813 11111 22222 33333 44444
17816 11111 ..... 33333 .....
17817 ..... 22222 ..... 44444
17818 11111 ..... 33333 .....
17819 ..... 22222 ..... 44444
17820 11111 ..... 33333 .....
17821 ..... 22222 ..... 44444
17822 11111 ..... 33333 .....
17823 ..... 22222 ..... 44444
17827 @item interleave_top, 4
17828 Interleave the upper field from odd frames with the lower field from
17829 even frames, generating a frame with unchanged height at half frame rate.
17834 Frame 1 Frame 2 Frame 3 Frame 4
17836 11111<- 22222 33333<- 44444
17837 11111 22222<- 33333 44444<-
17838 11111<- 22222 33333<- 44444
17839 11111 22222<- 33333 44444<-
17849 @item interleave_bottom, 5
17850 Interleave the lower field from odd frames with the upper field from
17851 even frames, generating a frame with unchanged height at half frame rate.
17856 Frame 1 Frame 2 Frame 3 Frame 4
17858 11111 22222<- 33333 44444<-
17859 11111<- 22222 33333<- 44444
17860 11111 22222<- 33333 44444<-
17861 11111<- 22222 33333<- 44444
17871 @item interlacex2, 6
17872 Double frame rate with unchanged height. Frames are inserted each
17873 containing the second temporal field from the previous input frame and
17874 the first temporal field from the next input frame. This mode relies on
17875 the top_field_first flag. Useful for interlaced video displays with no
17876 field synchronisation.
17881 Frame 1 Frame 2 Frame 3 Frame 4
17883 11111 22222 33333 44444
17884 11111 22222 33333 44444
17885 11111 22222 33333 44444
17886 11111 22222 33333 44444
17889 11111 22222 22222 33333 33333 44444 44444
17890 11111 11111 22222 22222 33333 33333 44444
17891 11111 22222 22222 33333 33333 44444 44444
17892 11111 11111 22222 22222 33333 33333 44444
17897 Move odd frames into the upper field, even into the lower field,
17898 generating a double height frame at same frame rate.
17903 Frame 1 Frame 2 Frame 3 Frame 4
17905 11111 22222 33333 44444
17906 11111 22222 33333 44444
17907 11111 22222 33333 44444
17908 11111 22222 33333 44444
17911 11111 33333 33333 55555
17912 22222 22222 44444 44444
17913 11111 33333 33333 55555
17914 22222 22222 44444 44444
17915 11111 33333 33333 55555
17916 22222 22222 44444 44444
17917 11111 33333 33333 55555
17918 22222 22222 44444 44444
17923 Numeric values are deprecated but are accepted for backward
17924 compatibility reasons.
17926 Default mode is @code{merge}.
17929 Specify flags influencing the filter process.
17931 Available value for @var{flags} is:
17934 @item low_pass_filter, vlpf
17935 Enable linear vertical low-pass filtering in the filter.
17936 Vertical low-pass filtering is required when creating an interlaced
17937 destination from a progressive source which contains high-frequency
17938 vertical detail. Filtering will reduce interlace 'twitter' and Moire
17941 @item complex_filter, cvlpf
17942 Enable complex vertical low-pass filtering.
17943 This will slightly less reduce interlace 'twitter' and Moire
17944 patterning but better retain detail and subjective sharpness impression.
17948 Vertical low-pass filtering can only be enabled for @option{mode}
17949 @var{interleave_top} and @var{interleave_bottom}.
17955 Mix successive video frames.
17957 A description of the accepted options follows.
17961 The number of successive frames to mix. If unspecified, it defaults to 3.
17964 Specify weight of each input video frame.
17965 Each weight is separated by space. If number of weights is smaller than
17966 number of @var{frames} last specified weight will be used for all remaining
17970 Specify scale, if it is set it will be multiplied with sum
17971 of each weight multiplied with pixel values to give final destination
17972 pixel value. By default @var{scale} is auto scaled to sum of weights.
17975 @subsection Examples
17979 Average 7 successive frames:
17981 tmix=frames=7:weights="1 1 1 1 1 1 1"
17985 Apply simple temporal convolution:
17987 tmix=frames=3:weights="-1 3 -1"
17991 Similar as above but only showing temporal differences:
17993 tmix=frames=3:weights="-1 2 -1":scale=1
17999 Tone map colors from different dynamic ranges.
18001 This filter expects data in single precision floating point, as it needs to
18002 operate on (and can output) out-of-range values. Another filter, such as
18003 @ref{zscale}, is needed to convert the resulting frame to a usable format.
18005 The tonemapping algorithms implemented only work on linear light, so input
18006 data should be linearized beforehand (and possibly correctly tagged).
18009 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
18012 @subsection Options
18013 The filter accepts the following options.
18017 Set the tone map algorithm to use.
18019 Possible values are:
18022 Do not apply any tone map, only desaturate overbright pixels.
18025 Hard-clip any out-of-range values. Use it for perfect color accuracy for
18026 in-range values, while distorting out-of-range values.
18029 Stretch the entire reference gamut to a linear multiple of the display.
18032 Fit a logarithmic transfer between the tone curves.
18035 Preserve overall image brightness with a simple curve, using nonlinear
18036 contrast, which results in flattening details and degrading color accuracy.
18039 Preserve both dark and bright details better than @var{reinhard}, at the cost
18040 of slightly darkening everything. Use it when detail preservation is more
18041 important than color and brightness accuracy.
18044 Smoothly map out-of-range values, while retaining contrast and colors for
18045 in-range material as much as possible. Use it when color accuracy is more
18046 important than detail preservation.
18052 Tune the tone mapping algorithm.
18054 This affects the following algorithms:
18060 Specifies the scale factor to use while stretching.
18064 Specifies the exponent of the function.
18068 Specify an extra linear coefficient to multiply into the signal before clipping.
18072 Specify the local contrast coefficient at the display peak.
18073 Default to 0.5, which means that in-gamut values will be about half as bright
18080 Specify the transition point from linear to mobius transform. Every value
18081 below this point is guaranteed to be mapped 1:1. The higher the value, the
18082 more accurate the result will be, at the cost of losing bright details.
18083 Default to 0.3, which due to the steep initial slope still preserves in-range
18084 colors fairly accurately.
18088 Apply desaturation for highlights that exceed this level of brightness. The
18089 higher the parameter, the more color information will be preserved. This
18090 setting helps prevent unnaturally blown-out colors for super-highlights, by
18091 (smoothly) turning into white instead. This makes images feel more natural,
18092 at the cost of reducing information about out-of-range colors.
18094 The default of 2.0 is somewhat conservative and will mostly just apply to
18095 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
18097 This option works only if the input frame has a supported color tag.
18100 Override signal/nominal/reference peak with this value. Useful when the
18101 embedded peak information in display metadata is not reliable or when tone
18102 mapping from a lower range to a higher range.
18107 Temporarily pad video frames.
18109 The filter accepts the following options:
18113 Specify number of delay frames before input video stream.
18116 Specify number of padding frames after input video stream.
18117 Set to -1 to pad indefinitely.
18120 Set kind of frames added to beginning of stream.
18121 Can be either @var{add} or @var{clone}.
18122 With @var{add} frames of solid-color are added.
18123 With @var{clone} frames are clones of first frame.
18126 Set kind of frames added to end of stream.
18127 Can be either @var{add} or @var{clone}.
18128 With @var{add} frames of solid-color are added.
18129 With @var{clone} frames are clones of last frame.
18131 @item start_duration, stop_duration
18132 Specify the duration of the start/stop delay. See
18133 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18134 for the accepted syntax.
18135 These options override @var{start} and @var{stop}.
18138 Specify the color of the padded area. For the syntax of this option,
18139 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
18140 manual,ffmpeg-utils}.
18142 The default value of @var{color} is "black".
18148 Transpose rows with columns in the input video and optionally flip it.
18150 It accepts the following parameters:
18155 Specify the transposition direction.
18157 Can assume the following values:
18159 @item 0, 4, cclock_flip
18160 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
18168 Rotate by 90 degrees clockwise, that is:
18176 Rotate by 90 degrees counterclockwise, that is:
18183 @item 3, 7, clock_flip
18184 Rotate by 90 degrees clockwise and vertically flip, that is:
18192 For values between 4-7, the transposition is only done if the input
18193 video geometry is portrait and not landscape. These values are
18194 deprecated, the @code{passthrough} option should be used instead.
18196 Numerical values are deprecated, and should be dropped in favor of
18197 symbolic constants.
18200 Do not apply the transposition if the input geometry matches the one
18201 specified by the specified value. It accepts the following values:
18204 Always apply transposition.
18206 Preserve portrait geometry (when @var{height} >= @var{width}).
18208 Preserve landscape geometry (when @var{width} >= @var{height}).
18211 Default value is @code{none}.
18214 For example to rotate by 90 degrees clockwise and preserve portrait
18217 transpose=dir=1:passthrough=portrait
18220 The command above can also be specified as:
18222 transpose=1:portrait
18225 @section transpose_npp
18227 Transpose rows with columns in the input video and optionally flip it.
18228 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
18230 It accepts the following parameters:
18235 Specify the transposition direction.
18237 Can assume the following values:
18240 Rotate by 90 degrees counterclockwise and vertically flip. (default)
18243 Rotate by 90 degrees clockwise.
18246 Rotate by 90 degrees counterclockwise.
18249 Rotate by 90 degrees clockwise and vertically flip.
18253 Do not apply the transposition if the input geometry matches the one
18254 specified by the specified value. It accepts the following values:
18257 Always apply transposition. (default)
18259 Preserve portrait geometry (when @var{height} >= @var{width}).
18261 Preserve landscape geometry (when @var{width} >= @var{height}).
18267 Trim the input so that the output contains one continuous subpart of the input.
18269 It accepts the following parameters:
18272 Specify the time of the start of the kept section, i.e. the frame with the
18273 timestamp @var{start} will be the first frame in the output.
18276 Specify the time of the first frame that will be dropped, i.e. the frame
18277 immediately preceding the one with the timestamp @var{end} will be the last
18278 frame in the output.
18281 This is the same as @var{start}, except this option sets the start timestamp
18282 in timebase units instead of seconds.
18285 This is the same as @var{end}, except this option sets the end timestamp
18286 in timebase units instead of seconds.
18289 The maximum duration of the output in seconds.
18292 The number of the first frame that should be passed to the output.
18295 The number of the first frame that should be dropped.
18298 @option{start}, @option{end}, and @option{duration} are expressed as time
18299 duration specifications; see
18300 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18301 for the accepted syntax.
18303 Note that the first two sets of the start/end options and the @option{duration}
18304 option look at the frame timestamp, while the _frame variants simply count the
18305 frames that pass through the filter. Also note that this filter does not modify
18306 the timestamps. If you wish for the output timestamps to start at zero, insert a
18307 setpts filter after the trim filter.
18309 If multiple start or end options are set, this filter tries to be greedy and
18310 keep all the frames that match at least one of the specified constraints. To keep
18311 only the part that matches all the constraints at once, chain multiple trim
18314 The defaults are such that all the input is kept. So it is possible to set e.g.
18315 just the end values to keep everything before the specified time.
18320 Drop everything except the second minute of input:
18322 ffmpeg -i INPUT -vf trim=60:120
18326 Keep only the first second:
18328 ffmpeg -i INPUT -vf trim=duration=1
18333 @section unpremultiply
18334 Apply alpha unpremultiply effect to input video stream using first plane
18335 of second stream as alpha.
18337 Both streams must have same dimensions and same pixel format.
18339 The filter accepts the following option:
18343 Set which planes will be processed, unprocessed planes will be copied.
18344 By default value 0xf, all planes will be processed.
18346 If the format has 1 or 2 components, then luma is bit 0.
18347 If the format has 3 or 4 components:
18348 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
18349 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
18350 If present, the alpha channel is always the last bit.
18353 Do not require 2nd input for processing, instead use alpha plane from input stream.
18359 Sharpen or blur the input video.
18361 It accepts the following parameters:
18364 @item luma_msize_x, lx
18365 Set the luma matrix horizontal size. It must be an odd integer between
18366 3 and 23. The default value is 5.
18368 @item luma_msize_y, ly
18369 Set the luma matrix vertical size. It must be an odd integer between 3
18370 and 23. The default value is 5.
18372 @item luma_amount, la
18373 Set the luma effect strength. It must be a floating point number, reasonable
18374 values lay between -1.5 and 1.5.
18376 Negative values will blur the input video, while positive values will
18377 sharpen it, a value of zero will disable the effect.
18379 Default value is 1.0.
18381 @item chroma_msize_x, cx
18382 Set the chroma matrix horizontal size. It must be an odd integer
18383 between 3 and 23. The default value is 5.
18385 @item chroma_msize_y, cy
18386 Set the chroma matrix vertical size. It must be an odd integer
18387 between 3 and 23. The default value is 5.
18389 @item chroma_amount, ca
18390 Set the chroma effect strength. It must be a floating point number, reasonable
18391 values lay between -1.5 and 1.5.
18393 Negative values will blur the input video, while positive values will
18394 sharpen it, a value of zero will disable the effect.
18396 Default value is 0.0.
18400 All parameters are optional and default to the equivalent of the
18401 string '5:5:1.0:5:5:0.0'.
18403 @subsection Examples
18407 Apply strong luma sharpen effect:
18409 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
18413 Apply a strong blur of both luma and chroma parameters:
18415 unsharp=7:7:-2:7:7:-2
18421 Apply ultra slow/simple postprocessing filter that compresses and decompresses
18422 the image at several (or - in the case of @option{quality} level @code{8} - all)
18423 shifts and average the results.
18425 The way this differs from the behavior of spp is that uspp actually encodes &
18426 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
18427 DCT similar to MJPEG.
18429 The filter accepts the following options:
18433 Set quality. This option defines the number of levels for averaging. It accepts
18434 an integer in the range 0-8. If set to @code{0}, the filter will have no
18435 effect. A value of @code{8} means the higher quality. For each increment of
18436 that value the speed drops by a factor of approximately 2. Default value is
18440 Force a constant quantization parameter. If not set, the filter will use the QP
18441 from the video stream (if available).
18446 Convert 360 videos between various formats.
18448 The filter accepts the following options:
18454 Set format of the input/output video.
18462 Equirectangular projection.
18467 Cubemap with 3x2/6x1/1x6 layout.
18469 Format specific options:
18474 Set padding proportion for the input/output cubemap. Values in decimals.
18481 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)
18484 Default value is @b{@samp{0}}.
18488 Set fixed padding for the input/output cubemap. Values in pixels.
18490 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
18494 Set order of faces for the input/output cubemap. Choose one direction for each position.
18496 Designation of directions:
18512 Default value is @b{@samp{rludfb}}.
18516 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
18518 Designation of angles:
18521 0 degrees clockwise
18523 90 degrees clockwise
18525 180 degrees clockwise
18527 270 degrees clockwise
18530 Default value is @b{@samp{000000}}.
18534 Equi-Angular Cubemap.
18539 Regular video. @i{(output only)}
18541 Format specific options:
18546 Set horizontal/vertical/diagonal field of view. Values in degrees.
18548 If diagonal field of view is set it overrides horizontal and vertical field of view.
18554 Format specific options:
18558 Set padding proportion. Values in decimals.
18568 Default value is @b{@samp{0}}.
18573 Facebook's 360 format.
18576 Stereographic format.
18578 Format specific options:
18583 Set horizontal/vertical/diagonal field of view. Values in degrees.
18585 If diagonal field of view is set it overrides horizontal and vertical field of view.
18592 Ball format, gives significant distortion toward the back.
18595 Hammer-Aitoff map projection format.
18598 Sinusoidal map projection format.
18603 Set interpolation method.@*
18604 @i{Note: more complex interpolation methods require much more memory to run.}
18614 Bilinear interpolation.
18617 Bicubic interpolation.
18620 Lanczos interpolation.
18623 Default value is @b{@samp{line}}.
18627 Set the output video resolution.
18629 Default resolution depends on formats.
18633 Set the input/output stereo format.
18644 Default value is @b{@samp{2d}} for input and output format.
18649 Set rotation for the output video. Values in degrees.
18652 Set rotation order for the output video. Choose one item for each position.
18663 Default value is @b{@samp{ypr}}.
18668 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
18672 Set if input video is flipped horizontally/vertically. Boolean values.
18675 Set if input video is transposed. Boolean value, by default disabled.
18678 Set if output video needs to be transposed. Boolean value, by default disabled.
18682 @subsection Examples
18686 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
18688 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
18691 Extract back view of Equi-Angular Cubemap:
18693 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
18696 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
18698 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
18702 @section vaguedenoiser
18704 Apply a wavelet based denoiser.
18706 It transforms each frame from the video input into the wavelet domain,
18707 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
18708 the obtained coefficients. It does an inverse wavelet transform after.
18709 Due to wavelet properties, it should give a nice smoothed result, and
18710 reduced noise, without blurring picture features.
18712 This filter accepts the following options:
18716 The filtering strength. The higher, the more filtered the video will be.
18717 Hard thresholding can use a higher threshold than soft thresholding
18718 before the video looks overfiltered. Default value is 2.
18721 The filtering method the filter will use.
18723 It accepts the following values:
18726 All values under the threshold will be zeroed.
18729 All values under the threshold will be zeroed. All values above will be
18730 reduced by the threshold.
18733 Scales or nullifies coefficients - intermediary between (more) soft and
18734 (less) hard thresholding.
18737 Default is garrote.
18740 Number of times, the wavelet will decompose the picture. Picture can't
18741 be decomposed beyond a particular point (typically, 8 for a 640x480
18742 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
18745 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
18748 A list of the planes to process. By default all planes are processed.
18751 @section vectorscope
18753 Display 2 color component values in the two dimensional graph (which is called
18756 This filter accepts the following options:
18760 Set vectorscope mode.
18762 It accepts the following values:
18765 Gray values are displayed on graph, higher brightness means more pixels have
18766 same component color value on location in graph. This is the default mode.
18769 Gray values are displayed on graph. Surrounding pixels values which are not
18770 present in video frame are drawn in gradient of 2 color components which are
18771 set by option @code{x} and @code{y}. The 3rd color component is static.
18774 Actual color components values present in video frame are displayed on graph.
18777 Similar as color2 but higher frequency of same values @code{x} and @code{y}
18778 on graph increases value of another color component, which is luminance by
18779 default values of @code{x} and @code{y}.
18782 Actual colors present in video frame are displayed on graph. If two different
18783 colors map to same position on graph then color with higher value of component
18784 not present in graph is picked.
18787 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
18788 component picked from radial gradient.
18792 Set which color component will be represented on X-axis. Default is @code{1}.
18795 Set which color component will be represented on Y-axis. Default is @code{2}.
18798 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
18799 of color component which represents frequency of (X, Y) location in graph.
18804 No envelope, this is default.
18807 Instant envelope, even darkest single pixel will be clearly highlighted.
18810 Hold maximum and minimum values presented in graph over time. This way you
18811 can still spot out of range values without constantly looking at vectorscope.
18814 Peak and instant envelope combined together.
18818 Set what kind of graticule to draw.
18826 Set graticule opacity.
18829 Set graticule flags.
18833 Draw graticule for white point.
18836 Draw graticule for black point.
18839 Draw color points short names.
18843 Set background opacity.
18845 @item lthreshold, l
18846 Set low threshold for color component not represented on X or Y axis.
18847 Values lower than this value will be ignored. Default is 0.
18848 Note this value is multiplied with actual max possible value one pixel component
18849 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
18852 @item hthreshold, h
18853 Set high threshold for color component not represented on X or Y axis.
18854 Values higher than this value will be ignored. Default is 1.
18855 Note this value is multiplied with actual max possible value one pixel component
18856 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
18857 is 0.9 * 255 = 230.
18859 @item colorspace, c
18860 Set what kind of colorspace to use when drawing graticule.
18869 @anchor{vidstabdetect}
18870 @section vidstabdetect
18872 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
18873 @ref{vidstabtransform} for pass 2.
18875 This filter generates a file with relative translation and rotation
18876 transform information about subsequent frames, which is then used by
18877 the @ref{vidstabtransform} filter.
18879 To enable compilation of this filter you need to configure FFmpeg with
18880 @code{--enable-libvidstab}.
18882 This filter accepts the following options:
18886 Set the path to the file used to write the transforms information.
18887 Default value is @file{transforms.trf}.
18890 Set how shaky the video is and how quick the camera is. It accepts an
18891 integer in the range 1-10, a value of 1 means little shakiness, a
18892 value of 10 means strong shakiness. Default value is 5.
18895 Set the accuracy of the detection process. It must be a value in the
18896 range 1-15. A value of 1 means low accuracy, a value of 15 means high
18897 accuracy. Default value is 15.
18900 Set stepsize of the search process. The region around minimum is
18901 scanned with 1 pixel resolution. Default value is 6.
18904 Set minimum contrast. Below this value a local measurement field is
18905 discarded. Must be a floating point value in the range 0-1. Default
18909 Set reference frame number for tripod mode.
18911 If enabled, the motion of the frames is compared to a reference frame
18912 in the filtered stream, identified by the specified number. The idea
18913 is to compensate all movements in a more-or-less static scene and keep
18914 the camera view absolutely still.
18916 If set to 0, it is disabled. The frames are counted starting from 1.
18919 Show fields and transforms in the resulting frames. It accepts an
18920 integer in the range 0-2. Default value is 0, which disables any
18924 @subsection Examples
18928 Use default values:
18934 Analyze strongly shaky movie and put the results in file
18935 @file{mytransforms.trf}:
18937 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
18941 Visualize the result of internal transformations in the resulting
18944 vidstabdetect=show=1
18948 Analyze a video with medium shakiness using @command{ffmpeg}:
18950 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
18954 @anchor{vidstabtransform}
18955 @section vidstabtransform
18957 Video stabilization/deshaking: pass 2 of 2,
18958 see @ref{vidstabdetect} for pass 1.
18960 Read a file with transform information for each frame and
18961 apply/compensate them. Together with the @ref{vidstabdetect}
18962 filter this can be used to deshake videos. See also
18963 @url{http://public.hronopik.de/vid.stab}. It is important to also use
18964 the @ref{unsharp} filter, see below.
18966 To enable compilation of this filter you need to configure FFmpeg with
18967 @code{--enable-libvidstab}.
18969 @subsection Options
18973 Set path to the file used to read the transforms. Default value is
18974 @file{transforms.trf}.
18977 Set the number of frames (value*2 + 1) used for lowpass filtering the
18978 camera movements. Default value is 10.
18980 For example a number of 10 means that 21 frames are used (10 in the
18981 past and 10 in the future) to smoothen the motion in the video. A
18982 larger value leads to a smoother video, but limits the acceleration of
18983 the camera (pan/tilt movements). 0 is a special case where a static
18984 camera is simulated.
18987 Set the camera path optimization algorithm.
18989 Accepted values are:
18992 gaussian kernel low-pass filter on camera motion (default)
18994 averaging on transformations
18998 Set maximal number of pixels to translate frames. Default value is -1,
19002 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
19003 value is -1, meaning no limit.
19006 Specify how to deal with borders that may be visible due to movement
19009 Available values are:
19012 keep image information from previous frame (default)
19014 fill the border black
19018 Invert transforms if set to 1. Default value is 0.
19021 Consider transforms as relative to previous frame if set to 1,
19022 absolute if set to 0. Default value is 0.
19025 Set percentage to zoom. A positive value will result in a zoom-in
19026 effect, a negative value in a zoom-out effect. Default value is 0 (no
19030 Set optimal zooming to avoid borders.
19032 Accepted values are:
19037 optimal static zoom value is determined (only very strong movements
19038 will lead to visible borders) (default)
19040 optimal adaptive zoom value is determined (no borders will be
19041 visible), see @option{zoomspeed}
19044 Note that the value given at zoom is added to the one calculated here.
19047 Set percent to zoom maximally each frame (enabled when
19048 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
19052 Specify type of interpolation.
19054 Available values are:
19059 linear only horizontal
19061 linear in both directions (default)
19063 cubic in both directions (slow)
19067 Enable virtual tripod mode if set to 1, which is equivalent to
19068 @code{relative=0:smoothing=0}. Default value is 0.
19070 Use also @code{tripod} option of @ref{vidstabdetect}.
19073 Increase log verbosity if set to 1. Also the detected global motions
19074 are written to the temporary file @file{global_motions.trf}. Default
19078 @subsection Examples
19082 Use @command{ffmpeg} for a typical stabilization with default values:
19084 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
19087 Note the use of the @ref{unsharp} filter which is always recommended.
19090 Zoom in a bit more and load transform data from a given file:
19092 vidstabtransform=zoom=5:input="mytransforms.trf"
19096 Smoothen the video even more:
19098 vidstabtransform=smoothing=30
19104 Flip the input video vertically.
19106 For example, to vertically flip a video with @command{ffmpeg}:
19108 ffmpeg -i in.avi -vf "vflip" out.avi
19113 Detect variable frame rate video.
19115 This filter tries to detect if the input is variable or constant frame rate.
19117 At end it will output number of frames detected as having variable delta pts,
19118 and ones with constant delta pts.
19119 If there was frames with variable delta, than it will also show min, max and
19120 average delta encountered.
19124 Boost or alter saturation.
19126 The filter accepts the following options:
19129 Set strength of boost if positive value or strength of alter if negative value.
19130 Default is 0. Allowed range is from -2 to 2.
19133 Set the red balance. Default is 1. Allowed range is from -10 to 10.
19136 Set the green balance. Default is 1. Allowed range is from -10 to 10.
19139 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
19142 Set the red luma coefficient.
19145 Set the green luma coefficient.
19148 Set the blue luma coefficient.
19151 If @code{intensity} is negative and this is set to 1, colors will change,
19152 otherwise colors will be less saturated, more towards gray.
19158 Make or reverse a natural vignetting effect.
19160 The filter accepts the following options:
19164 Set lens angle expression as a number of radians.
19166 The value is clipped in the @code{[0,PI/2]} range.
19168 Default value: @code{"PI/5"}
19172 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
19176 Set forward/backward mode.
19178 Available modes are:
19181 The larger the distance from the central point, the darker the image becomes.
19184 The larger the distance from the central point, the brighter the image becomes.
19185 This can be used to reverse a vignette effect, though there is no automatic
19186 detection to extract the lens @option{angle} and other settings (yet). It can
19187 also be used to create a burning effect.
19190 Default value is @samp{forward}.
19193 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
19195 It accepts the following values:
19198 Evaluate expressions only once during the filter initialization.
19201 Evaluate expressions for each incoming frame. This is way slower than the
19202 @samp{init} mode since it requires all the scalers to be re-computed, but it
19203 allows advanced dynamic expressions.
19206 Default value is @samp{init}.
19209 Set dithering to reduce the circular banding effects. Default is @code{1}
19213 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
19214 Setting this value to the SAR of the input will make a rectangular vignetting
19215 following the dimensions of the video.
19217 Default is @code{1/1}.
19220 @subsection Expressions
19222 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
19223 following parameters.
19228 input width and height
19231 the number of input frame, starting from 0
19234 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
19235 @var{TB} units, NAN if undefined
19238 frame rate of the input video, NAN if the input frame rate is unknown
19241 the PTS (Presentation TimeStamp) of the filtered video frame,
19242 expressed in seconds, NAN if undefined
19245 time base of the input video
19249 @subsection Examples
19253 Apply simple strong vignetting effect:
19259 Make a flickering vignetting:
19261 vignette='PI/4+random(1)*PI/50':eval=frame
19266 @section vmafmotion
19268 Obtain the average vmaf motion score of a video.
19269 It is one of the component filters of VMAF.
19271 The obtained average motion score is printed through the logging system.
19273 In the below example the input file @file{ref.mpg} is being processed and score
19277 ffmpeg -i ref.mpg -lavfi vmafmotion -f null -
19281 Stack input videos vertically.
19283 All streams must be of same pixel format and of same width.
19285 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
19286 to create same output.
19288 The filter accepts the following options:
19292 Set number of input streams. Default is 2.
19295 If set to 1, force the output to terminate when the shortest input
19296 terminates. Default value is 0.
19301 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
19302 Deinterlacing Filter").
19304 Based on the process described by Martin Weston for BBC R&D, and
19305 implemented based on the de-interlace algorithm written by Jim
19306 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
19307 uses filter coefficients calculated by BBC R&D.
19309 This filter uses field-dominance information in frame to decide which
19310 of each pair of fields to place first in the output.
19311 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
19313 There are two sets of filter coefficients, so called "simple"
19314 and "complex". Which set of filter coefficients is used can
19315 be set by passing an optional parameter:
19319 Set the interlacing filter coefficients. Accepts one of the following values:
19323 Simple filter coefficient set.
19325 More-complex filter coefficient set.
19327 Default value is @samp{complex}.
19330 Specify which frames to deinterlace. Accepts one of the following values:
19334 Deinterlace all frames,
19336 Only deinterlace frames marked as interlaced.
19339 Default value is @samp{all}.
19343 Video waveform monitor.
19345 The waveform monitor plots color component intensity. By default luminance
19346 only. Each column of the waveform corresponds to a column of pixels in the
19349 It accepts the following options:
19353 Can be either @code{row}, or @code{column}. Default is @code{column}.
19354 In row mode, the graph on the left side represents color component value 0 and
19355 the right side represents value = 255. In column mode, the top side represents
19356 color component value = 0 and bottom side represents value = 255.
19359 Set intensity. Smaller values are useful to find out how many values of the same
19360 luminance are distributed across input rows/columns.
19361 Default value is @code{0.04}. Allowed range is [0, 1].
19364 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
19365 In mirrored mode, higher values will be represented on the left
19366 side for @code{row} mode and at the top for @code{column} mode. Default is
19367 @code{1} (mirrored).
19371 It accepts the following values:
19374 Presents information identical to that in the @code{parade}, except
19375 that the graphs representing color components are superimposed directly
19378 This display mode makes it easier to spot relative differences or similarities
19379 in overlapping areas of the color components that are supposed to be identical,
19380 such as neutral whites, grays, or blacks.
19383 Display separate graph for the color components side by side in
19384 @code{row} mode or one below the other in @code{column} mode.
19387 Display separate graph for the color components side by side in
19388 @code{column} mode or one below the other in @code{row} mode.
19390 Using this display mode makes it easy to spot color casts in the highlights
19391 and shadows of an image, by comparing the contours of the top and the bottom
19392 graphs of each waveform. Since whites, grays, and blacks are characterized
19393 by exactly equal amounts of red, green, and blue, neutral areas of the picture
19394 should display three waveforms of roughly equal width/height. If not, the
19395 correction is easy to perform by making level adjustments the three waveforms.
19397 Default is @code{stack}.
19399 @item components, c
19400 Set which color components to display. Default is 1, which means only luminance
19401 or red color component if input is in RGB colorspace. If is set for example to
19402 7 it will display all 3 (if) available color components.
19407 No envelope, this is default.
19410 Instant envelope, minimum and maximum values presented in graph will be easily
19411 visible even with small @code{step} value.
19414 Hold minimum and maximum values presented in graph across time. This way you
19415 can still spot out of range values without constantly looking at waveforms.
19418 Peak and instant envelope combined together.
19424 No filtering, this is default.
19427 Luma and chroma combined together.
19430 Similar as above, but shows difference between blue and red chroma.
19433 Similar as above, but use different colors.
19436 Similar as above, but again with different colors.
19439 Displays only chroma.
19442 Displays actual color value on waveform.
19445 Similar as above, but with luma showing frequency of chroma values.
19449 Set which graticule to display.
19453 Do not display graticule.
19456 Display green graticule showing legal broadcast ranges.
19459 Display orange graticule showing legal broadcast ranges.
19462 Display invert graticule showing legal broadcast ranges.
19466 Set graticule opacity.
19469 Set graticule flags.
19473 Draw numbers above lines. By default enabled.
19476 Draw dots instead of lines.
19480 Set scale used for displaying graticule.
19487 Default is digital.
19490 Set background opacity.
19493 @section weave, doubleweave
19495 The @code{weave} takes a field-based video input and join
19496 each two sequential fields into single frame, producing a new double
19497 height clip with half the frame rate and half the frame count.
19499 The @code{doubleweave} works same as @code{weave} but without
19500 halving frame rate and frame count.
19502 It accepts the following option:
19506 Set first field. Available values are:
19510 Set the frame as top-field-first.
19513 Set the frame as bottom-field-first.
19517 @subsection Examples
19521 Interlace video using @ref{select} and @ref{separatefields} filter:
19523 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
19528 Apply the xBR high-quality magnification filter which is designed for pixel
19529 art. It follows a set of edge-detection rules, see
19530 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
19532 It accepts the following option:
19536 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
19537 @code{3xBR} and @code{4} for @code{4xBR}.
19538 Default is @code{3}.
19542 Pick median pixels from several input videos.
19544 The filter accepts the following options:
19548 Set number of inputs.
19549 Default is 3. Allowed range is from 3 to 255.
19550 If number of inputs is even number, than result will be mean value between two median values.
19553 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19557 Stack video inputs into custom layout.
19559 All streams must be of same pixel format.
19561 The filter accepts the following options:
19565 Set number of input streams. Default is 2.
19568 Specify layout of inputs.
19569 This option requires the desired layout configuration to be explicitly set by the user.
19570 This sets position of each video input in output. Each input
19571 is separated by '|'.
19572 The first number represents the column, and the second number represents the row.
19573 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
19574 where X is video input from which to take width or height.
19575 Multiple values can be used when separated by '+'. In such
19576 case values are summed together.
19578 Note that if inputs are of different sizes gaps may appear, as not all of
19579 the output video frame will be filled. Similarly, videos can overlap each
19580 other if their position doesn't leave enough space for the full frame of
19583 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
19584 a layout must be set by the user.
19587 If set to 1, force the output to terminate when the shortest input
19588 terminates. Default value is 0.
19591 @subsection Examples
19595 Display 4 inputs into 2x2 grid.
19599 input1(0, 0) | input3(w0, 0)
19600 input2(0, h0) | input4(w0, h0)
19604 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
19607 Note that if inputs are of different sizes, gaps or overlaps may occur.
19610 Display 4 inputs into 1x4 grid.
19617 input4(0, h0+h1+h2)
19621 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
19624 Note that if inputs are of different widths, unused space will appear.
19627 Display 9 inputs into 3x3 grid.
19631 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
19632 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
19633 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
19637 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
19640 Note that if inputs are of different sizes, gaps or overlaps may occur.
19643 Display 16 inputs into 4x4 grid.
19647 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
19648 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
19649 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
19650 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
19654 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|
19655 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
19658 Note that if inputs are of different sizes, gaps or overlaps may occur.
19665 Deinterlace the input video ("yadif" means "yet another deinterlacing
19668 It accepts the following parameters:
19674 The interlacing mode to adopt. It accepts one of the following values:
19677 @item 0, send_frame
19678 Output one frame for each frame.
19679 @item 1, send_field
19680 Output one frame for each field.
19681 @item 2, send_frame_nospatial
19682 Like @code{send_frame}, but it skips the spatial interlacing check.
19683 @item 3, send_field_nospatial
19684 Like @code{send_field}, but it skips the spatial interlacing check.
19687 The default value is @code{send_frame}.
19690 The picture field parity assumed for the input interlaced video. It accepts one
19691 of the following values:
19695 Assume the top field is first.
19697 Assume the bottom field is first.
19699 Enable automatic detection of field parity.
19702 The default value is @code{auto}.
19703 If the interlacing is unknown or the decoder does not export this information,
19704 top field first will be assumed.
19707 Specify which frames to deinterlace. Accepts one of the following
19712 Deinterlace all frames.
19713 @item 1, interlaced
19714 Only deinterlace frames marked as interlaced.
19717 The default value is @code{all}.
19720 @section yadif_cuda
19722 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
19723 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
19726 It accepts the following parameters:
19732 The interlacing mode to adopt. It accepts one of the following values:
19735 @item 0, send_frame
19736 Output one frame for each frame.
19737 @item 1, send_field
19738 Output one frame for each field.
19739 @item 2, send_frame_nospatial
19740 Like @code{send_frame}, but it skips the spatial interlacing check.
19741 @item 3, send_field_nospatial
19742 Like @code{send_field}, but it skips the spatial interlacing check.
19745 The default value is @code{send_frame}.
19748 The picture field parity assumed for the input interlaced video. It accepts one
19749 of the following values:
19753 Assume the top field is first.
19755 Assume the bottom field is first.
19757 Enable automatic detection of field parity.
19760 The default value is @code{auto}.
19761 If the interlacing is unknown or the decoder does not export this information,
19762 top field first will be assumed.
19765 Specify which frames to deinterlace. Accepts one of the following
19770 Deinterlace all frames.
19771 @item 1, interlaced
19772 Only deinterlace frames marked as interlaced.
19775 The default value is @code{all}.
19780 Apply Zoom & Pan effect.
19782 This filter accepts the following options:
19786 Set the zoom expression. Range is 1-10. Default is 1.
19790 Set the x and y expression. Default is 0.
19793 Set the duration expression in number of frames.
19794 This sets for how many number of frames effect will last for
19795 single input image.
19798 Set the output image size, default is 'hd720'.
19801 Set the output frame rate, default is '25'.
19804 Each expression can contain the following constants:
19823 Output frame count.
19827 Last calculated 'x' and 'y' position from 'x' and 'y' expression
19828 for current input frame.
19832 'x' and 'y' of last output frame of previous input frame or 0 when there was
19833 not yet such frame (first input frame).
19836 Last calculated zoom from 'z' expression for current input frame.
19839 Last calculated zoom of last output frame of previous input frame.
19842 Number of output frames for current input frame. Calculated from 'd' expression
19843 for each input frame.
19846 number of output frames created for previous input frame
19849 Rational number: input width / input height
19852 sample aspect ratio
19855 display aspect ratio
19859 @subsection Examples
19863 Zoom-in up to 1.5 and pan at same time to some spot near center of picture:
19865 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
19869 Zoom-in up to 1.5 and pan always at center of picture:
19871 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
19875 Same as above but without pausing:
19877 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
19883 Scale (resize) the input video, using the z.lib library:
19884 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
19885 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
19887 The zscale filter forces the output display aspect ratio to be the same
19888 as the input, by changing the output sample aspect ratio.
19890 If the input image format is different from the format requested by
19891 the next filter, the zscale filter will convert the input to the
19894 @subsection Options
19895 The filter accepts the following options.
19900 Set the output video dimension expression. Default value is the input
19903 If the @var{width} or @var{w} value is 0, the input width is used for
19904 the output. If the @var{height} or @var{h} value is 0, the input height
19905 is used for the output.
19907 If one and only one of the values is -n with n >= 1, the zscale filter
19908 will use a value that maintains the aspect ratio of the input image,
19909 calculated from the other specified dimension. After that it will,
19910 however, make sure that the calculated dimension is divisible by n and
19911 adjust the value if necessary.
19913 If both values are -n with n >= 1, the behavior will be identical to
19914 both values being set to 0 as previously detailed.
19916 See below for the list of accepted constants for use in the dimension
19920 Set the video size. For the syntax of this option, check the
19921 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19924 Set the dither type.
19926 Possible values are:
19931 @item error_diffusion
19937 Set the resize filter type.
19939 Possible values are:
19949 Default is bilinear.
19952 Set the color range.
19954 Possible values are:
19961 Default is same as input.
19964 Set the color primaries.
19966 Possible values are:
19976 Default is same as input.
19979 Set the transfer characteristics.
19981 Possible values are:
19995 Default is same as input.
19998 Set the colorspace matrix.
20000 Possible value are:
20011 Default is same as input.
20014 Set the input color range.
20016 Possible values are:
20023 Default is same as input.
20025 @item primariesin, pin
20026 Set the input color primaries.
20028 Possible values are:
20038 Default is same as input.
20040 @item transferin, tin
20041 Set the input transfer characteristics.
20043 Possible values are:
20054 Default is same as input.
20056 @item matrixin, min
20057 Set the input colorspace matrix.
20059 Possible value are:
20071 Set the output chroma location.
20073 Possible values are:
20084 @item chromalin, cin
20085 Set the input chroma location.
20087 Possible values are:
20099 Set the nominal peak luminance.
20102 The values of the @option{w} and @option{h} options are expressions
20103 containing the following constants:
20108 The input width and height
20112 These are the same as @var{in_w} and @var{in_h}.
20116 The output (scaled) width and height
20120 These are the same as @var{out_w} and @var{out_h}
20123 The same as @var{iw} / @var{ih}
20126 input sample aspect ratio
20129 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
20133 horizontal and vertical input chroma subsample values. For example for the
20134 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20138 horizontal and vertical output chroma subsample values. For example for the
20139 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20145 @c man end VIDEO FILTERS
20147 @chapter OpenCL Video Filters
20148 @c man begin OPENCL VIDEO FILTERS
20150 Below is a description of the currently available OpenCL video filters.
20152 To enable compilation of these filters you need to configure FFmpeg with
20153 @code{--enable-opencl}.
20155 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
20158 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
20159 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
20160 given device parameters.
20162 @item -filter_hw_device @var{name}
20163 Pass the hardware device called @var{name} to all filters in any filter graph.
20167 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
20171 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
20173 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
20177 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.
20179 @section avgblur_opencl
20181 Apply average blur filter.
20183 The filter accepts the following options:
20187 Set horizontal radius size.
20188 Range is @code{[1, 1024]} and default value is @code{1}.
20191 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20194 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
20197 @subsection Example
20201 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.
20203 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
20207 @section boxblur_opencl
20209 Apply a boxblur algorithm to the input video.
20211 It accepts the following parameters:
20215 @item luma_radius, lr
20216 @item luma_power, lp
20217 @item chroma_radius, cr
20218 @item chroma_power, cp
20219 @item alpha_radius, ar
20220 @item alpha_power, ap
20224 A description of the accepted options follows.
20227 @item luma_radius, lr
20228 @item chroma_radius, cr
20229 @item alpha_radius, ar
20230 Set an expression for the box radius in pixels used for blurring the
20231 corresponding input plane.
20233 The radius value must be a non-negative number, and must not be
20234 greater than the value of the expression @code{min(w,h)/2} for the
20235 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
20238 Default value for @option{luma_radius} is "2". If not specified,
20239 @option{chroma_radius} and @option{alpha_radius} default to the
20240 corresponding value set for @option{luma_radius}.
20242 The expressions can contain the following constants:
20246 The input width and height in pixels.
20250 The input chroma image width and height in pixels.
20254 The horizontal and vertical chroma subsample values. For example, for the
20255 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
20258 @item luma_power, lp
20259 @item chroma_power, cp
20260 @item alpha_power, ap
20261 Specify how many times the boxblur filter is applied to the
20262 corresponding plane.
20264 Default value for @option{luma_power} is 2. If not specified,
20265 @option{chroma_power} and @option{alpha_power} default to the
20266 corresponding value set for @option{luma_power}.
20268 A value of 0 will disable the effect.
20271 @subsection Examples
20273 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.
20277 Apply a boxblur filter with the luma, chroma, and alpha radius
20278 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.
20280 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
20281 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
20285 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.
20287 For the luma plane, a 2x2 box radius will be run once.
20289 For the chroma plane, a 4x4 box radius will be run 5 times.
20291 For the alpha plane, a 3x3 box radius will be run 7 times.
20293 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
20297 @section convolution_opencl
20299 Apply convolution of 3x3, 5x5, 7x7 matrix.
20301 The filter accepts the following options:
20308 Set matrix for each plane.
20309 Matrix is sequence of 9, 25 or 49 signed numbers.
20310 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
20316 Set multiplier for calculated value for each plane.
20317 If unset or 0, it will be sum of all matrix elements.
20318 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
20324 Set bias for each plane. This value is added to the result of the multiplication.
20325 Useful for making the overall image brighter or darker.
20326 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
20330 @subsection Examples
20336 -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
20342 -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
20346 Apply edge enhance:
20348 -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
20354 -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
20358 Apply laplacian edge detector which includes diagonals:
20360 -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
20366 -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
20370 @section dilation_opencl
20372 Apply dilation effect to the video.
20374 This filter replaces the pixel by the local(3x3) maximum.
20376 It accepts the following options:
20383 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
20384 If @code{0}, plane will remain unchanged.
20387 Flag which specifies the pixel to refer to.
20388 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
20390 Flags to local 3x3 coordinates region centered on @code{x}:
20399 @subsection Example
20403 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.
20405 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
20409 @section erosion_opencl
20411 Apply erosion effect to the video.
20413 This filter replaces the pixel by the local(3x3) minimum.
20415 It accepts the following options:
20422 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
20423 If @code{0}, plane will remain unchanged.
20426 Flag which specifies the pixel to refer to.
20427 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
20429 Flags to local 3x3 coordinates region centered on @code{x}:
20438 @subsection Example
20442 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.
20444 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
20448 @section colorkey_opencl
20449 RGB colorspace color keying.
20451 The filter accepts the following options:
20455 The color which will be replaced with transparency.
20458 Similarity percentage with the key color.
20460 0.01 matches only the exact key color, while 1.0 matches everything.
20465 0.0 makes pixels either fully transparent, or not transparent at all.
20467 Higher values result in semi-transparent pixels, with a higher transparency
20468 the more similar the pixels color is to the key color.
20471 @subsection Examples
20475 Make every semi-green pixel in the input transparent with some slight blending:
20477 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
20481 @section deshake_opencl
20482 Feature-point based video stabilization filter.
20484 The filter accepts the following options:
20488 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
20491 Whether or not additional debug info should be displayed, both in the processed output and in the console.
20493 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
20495 Viewing point matches in the output video is only supported for RGB input.
20497 Defaults to @code{0}.
20499 @item adaptive_crop
20500 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
20502 Defaults to @code{1}.
20504 @item refine_features
20505 Whether or not feature points should be refined at a sub-pixel level.
20507 This can be turned off for a slight performance gain at the cost of precision.
20509 Defaults to @code{1}.
20511 @item smooth_strength
20512 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
20514 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
20516 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
20518 Defaults to @code{0.0}.
20520 @item smooth_window_multiplier
20521 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
20523 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
20525 Acceptable values range from @code{0.1} to @code{10.0}.
20527 Larger values increase the amount of motion data available for determining how to smooth the camera path,
20528 potentially improving smoothness, but also increase latency and memory usage.
20530 Defaults to @code{2.0}.
20534 @subsection Examples
20538 Stabilize a video with a fixed, medium smoothing strength:
20540 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
20544 Stabilize a video with debugging (both in console and in rendered video):
20546 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
20550 @section nlmeans_opencl
20552 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
20554 @section overlay_opencl
20556 Overlay one video on top of another.
20558 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
20559 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
20561 The filter accepts the following options:
20566 Set the x coordinate of the overlaid video on the main video.
20567 Default value is @code{0}.
20570 Set the y coordinate of the overlaid video on the main video.
20571 Default value is @code{0}.
20575 @subsection Examples
20579 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
20581 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
20584 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
20586 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
20591 @section prewitt_opencl
20593 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
20595 The filter accepts the following option:
20599 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20602 Set value which will be multiplied with filtered result.
20603 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20606 Set value which will be added to filtered result.
20607 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20610 @subsection Example
20614 Apply the Prewitt operator with scale set to 2 and delta set to 10.
20616 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
20620 @section roberts_opencl
20621 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
20623 The filter accepts the following option:
20627 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20630 Set value which will be multiplied with filtered result.
20631 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20634 Set value which will be added to filtered result.
20635 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20638 @subsection Example
20642 Apply the Roberts cross operator with scale set to 2 and delta set to 10
20644 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
20648 @section sobel_opencl
20650 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
20652 The filter accepts the following option:
20656 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20659 Set value which will be multiplied with filtered result.
20660 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20663 Set value which will be added to filtered result.
20664 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20667 @subsection Example
20671 Apply sobel operator with scale set to 2 and delta set to 10
20673 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
20677 @section tonemap_opencl
20679 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
20681 It accepts the following parameters:
20685 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
20688 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
20691 Apply desaturation for highlights that exceed this level of brightness. The
20692 higher the parameter, the more color information will be preserved. This
20693 setting helps prevent unnaturally blown-out colors for super-highlights, by
20694 (smoothly) turning into white instead. This makes images feel more natural,
20695 at the cost of reducing information about out-of-range colors.
20697 The default value is 0.5, and the algorithm here is a little different from
20698 the cpu version tonemap currently. A setting of 0.0 disables this option.
20701 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
20702 is used to detect whether the scene has changed or not. If the distance between
20703 the current frame average brightness and the current running average exceeds
20704 a threshold value, we would re-calculate scene average and peak brightness.
20705 The default value is 0.2.
20708 Specify the output pixel format.
20710 Currently supported formats are:
20717 Set the output color range.
20719 Possible values are:
20725 Default is same as input.
20728 Set the output color primaries.
20730 Possible values are:
20736 Default is same as input.
20739 Set the output transfer characteristics.
20741 Possible values are:
20750 Set the output colorspace matrix.
20752 Possible value are:
20758 Default is same as input.
20762 @subsection Example
20766 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
20768 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
20772 @section unsharp_opencl
20774 Sharpen or blur the input video.
20776 It accepts the following parameters:
20779 @item luma_msize_x, lx
20780 Set the luma matrix horizontal size.
20781 Range is @code{[1, 23]} and default value is @code{5}.
20783 @item luma_msize_y, ly
20784 Set the luma matrix vertical size.
20785 Range is @code{[1, 23]} and default value is @code{5}.
20787 @item luma_amount, la
20788 Set the luma effect strength.
20789 Range is @code{[-10, 10]} and default value is @code{1.0}.
20791 Negative values will blur the input video, while positive values will
20792 sharpen it, a value of zero will disable the effect.
20794 @item chroma_msize_x, cx
20795 Set the chroma matrix horizontal size.
20796 Range is @code{[1, 23]} and default value is @code{5}.
20798 @item chroma_msize_y, cy
20799 Set the chroma matrix vertical size.
20800 Range is @code{[1, 23]} and default value is @code{5}.
20802 @item chroma_amount, ca
20803 Set the chroma effect strength.
20804 Range is @code{[-10, 10]} and default value is @code{0.0}.
20806 Negative values will blur the input video, while positive values will
20807 sharpen it, a value of zero will disable the effect.
20811 All parameters are optional and default to the equivalent of the
20812 string '5:5:1.0:5:5:0.0'.
20814 @subsection Examples
20818 Apply strong luma sharpen effect:
20820 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
20824 Apply a strong blur of both luma and chroma parameters:
20826 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
20830 @c man end OPENCL VIDEO FILTERS
20832 @chapter Video Sources
20833 @c man begin VIDEO SOURCES
20835 Below is a description of the currently available video sources.
20839 Buffer video frames, and make them available to the filter chain.
20841 This source is mainly intended for a programmatic use, in particular
20842 through the interface defined in @file{libavfilter/vsrc_buffer.h}.
20844 It accepts the following parameters:
20849 Specify the size (width and height) of the buffered video frames. For the
20850 syntax of this option, check the
20851 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
20854 The input video width.
20857 The input video height.
20860 A string representing the pixel format of the buffered video frames.
20861 It may be a number corresponding to a pixel format, or a pixel format
20865 Specify the timebase assumed by the timestamps of the buffered frames.
20868 Specify the frame rate expected for the video stream.
20870 @item pixel_aspect, sar
20871 The sample (pixel) aspect ratio of the input video.
20874 Specify the optional parameters to be used for the scale filter which
20875 is automatically inserted when an input change is detected in the
20876 input size or format.
20878 @item hw_frames_ctx
20879 When using a hardware pixel format, this should be a reference to an
20880 AVHWFramesContext describing input frames.
20885 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
20888 will instruct the source to accept video frames with size 320x240 and
20889 with format "yuv410p", assuming 1/24 as the timestamps timebase and
20890 square pixels (1:1 sample aspect ratio).
20891 Since the pixel format with name "yuv410p" corresponds to the number 6
20892 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
20893 this example corresponds to:
20895 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
20898 Alternatively, the options can be specified as a flat string, but this
20899 syntax is deprecated:
20901 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}[:@var{sws_param}]
20905 Create a pattern generated by an elementary cellular automaton.
20907 The initial state of the cellular automaton can be defined through the
20908 @option{filename} and @option{pattern} options. If such options are
20909 not specified an initial state is created randomly.
20911 At each new frame a new row in the video is filled with the result of
20912 the cellular automaton next generation. The behavior when the whole
20913 frame is filled is defined by the @option{scroll} option.
20915 This source accepts the following options:
20919 Read the initial cellular automaton state, i.e. the starting row, from
20920 the specified file.
20921 In the file, each non-whitespace character is considered an alive
20922 cell, a newline will terminate the row, and further characters in the
20923 file will be ignored.
20926 Read the initial cellular automaton state, i.e. the starting row, from
20927 the specified string.
20929 Each non-whitespace character in the string is considered an alive
20930 cell, a newline will terminate the row, and further characters in the
20931 string will be ignored.
20934 Set the video rate, that is the number of frames generated per second.
20937 @item random_fill_ratio, ratio
20938 Set the random fill ratio for the initial cellular automaton row. It
20939 is a floating point number value ranging from 0 to 1, defaults to
20942 This option is ignored when a file or a pattern is specified.
20944 @item random_seed, seed
20945 Set the seed for filling randomly the initial row, must be an integer
20946 included between 0 and UINT32_MAX. If not specified, or if explicitly
20947 set to -1, the filter will try to use a good random seed on a best
20951 Set the cellular automaton rule, it is a number ranging from 0 to 255.
20952 Default value is 110.
20955 Set the size of the output video. For the syntax of this option, check the
20956 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
20958 If @option{filename} or @option{pattern} is specified, the size is set
20959 by default to the width of the specified initial state row, and the
20960 height is set to @var{width} * PHI.
20962 If @option{size} is set, it must contain the width of the specified
20963 pattern string, and the specified pattern will be centered in the
20966 If a filename or a pattern string is not specified, the size value
20967 defaults to "320x518" (used for a randomly generated initial state).
20970 If set to 1, scroll the output upward when all the rows in the output
20971 have been already filled. If set to 0, the new generated row will be
20972 written over the top row just after the bottom row is filled.
20975 @item start_full, full
20976 If set to 1, completely fill the output with generated rows before
20977 outputting the first frame.
20978 This is the default behavior, for disabling set the value to 0.
20981 If set to 1, stitch the left and right row edges together.
20982 This is the default behavior, for disabling set the value to 0.
20985 @subsection Examples
20989 Read the initial state from @file{pattern}, and specify an output of
20992 cellauto=f=pattern:s=200x400
20996 Generate a random initial row with a width of 200 cells, with a fill
20999 cellauto=ratio=2/3:s=200x200
21003 Create a pattern generated by rule 18 starting by a single alive cell
21004 centered on an initial row with width 100:
21006 cellauto=p=@@:s=100x400:full=0:rule=18
21010 Specify a more elaborated initial pattern:
21012 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
21017 @anchor{coreimagesrc}
21018 @section coreimagesrc
21019 Video source generated on GPU using Apple's CoreImage API on OSX.
21021 This video source is a specialized version of the @ref{coreimage} video filter.
21022 Use a core image generator at the beginning of the applied filterchain to
21023 generate the content.
21025 The coreimagesrc video source accepts the following options:
21027 @item list_generators
21028 List all available generators along with all their respective options as well as
21029 possible minimum and maximum values along with the default values.
21031 list_generators=true
21035 Specify the size of the sourced video. For the syntax of this option, check the
21036 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21037 The default value is @code{320x240}.
21040 Specify the frame rate of the sourced video, as the number of frames
21041 generated per second. It has to be a string in the format
21042 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21043 number or a valid video frame rate abbreviation. The default value is
21047 Set the sample aspect ratio of the sourced video.
21050 Set the duration of the sourced video. See
21051 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21052 for the accepted syntax.
21054 If not specified, or the expressed duration is negative, the video is
21055 supposed to be generated forever.
21058 Additionally, all options of the @ref{coreimage} video filter are accepted.
21059 A complete filterchain can be used for further processing of the
21060 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
21061 and examples for details.
21063 @subsection Examples
21068 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
21069 given as complete and escaped command-line for Apple's standard bash shell:
21071 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
21073 This example is equivalent to the QRCode example of @ref{coreimage} without the
21074 need for a nullsrc video source.
21078 @section mandelbrot
21080 Generate a Mandelbrot set fractal, and progressively zoom towards the
21081 point specified with @var{start_x} and @var{start_y}.
21083 This source accepts the following options:
21088 Set the terminal pts value. Default value is 400.
21091 Set the terminal scale value.
21092 Must be a floating point value. Default value is 0.3.
21095 Set the inner coloring mode, that is the algorithm used to draw the
21096 Mandelbrot fractal internal region.
21098 It shall assume one of the following values:
21103 Show time until convergence.
21105 Set color based on point closest to the origin of the iterations.
21110 Default value is @var{mincol}.
21113 Set the bailout value. Default value is 10.0.
21116 Set the maximum of iterations performed by the rendering
21117 algorithm. Default value is 7189.
21120 Set outer coloring mode.
21121 It shall assume one of following values:
21123 @item iteration_count
21124 Set iteration count mode.
21125 @item normalized_iteration_count
21126 set normalized iteration count mode.
21128 Default value is @var{normalized_iteration_count}.
21131 Set frame rate, expressed as number of frames per second. Default
21135 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
21136 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
21139 Set the initial scale value. Default value is 3.0.
21142 Set the initial x position. Must be a floating point value between
21143 -100 and 100. Default value is -0.743643887037158704752191506114774.
21146 Set the initial y position. Must be a floating point value between
21147 -100 and 100. Default value is -0.131825904205311970493132056385139.
21152 Generate various test patterns, as generated by the MPlayer test filter.
21154 The size of the generated video is fixed, and is 256x256.
21155 This source is useful in particular for testing encoding features.
21157 This source accepts the following options:
21162 Specify the frame rate of the sourced video, as the number of frames
21163 generated per second. It has to be a string in the format
21164 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21165 number or a valid video frame rate abbreviation. The default value is
21169 Set the duration of the sourced video. See
21170 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21171 for the accepted syntax.
21173 If not specified, or the expressed duration is negative, the video is
21174 supposed to be generated forever.
21178 Set the number or the name of the test to perform. Supported tests are:
21192 @item max_frames, m
21193 Set the maximum number of frames generated for each test, default value is 30.
21197 Default value is "all", which will cycle through the list of all tests.
21202 mptestsrc=t=dc_luma
21205 will generate a "dc_luma" test pattern.
21207 @section frei0r_src
21209 Provide a frei0r source.
21211 To enable compilation of this filter you need to install the frei0r
21212 header and configure FFmpeg with @code{--enable-frei0r}.
21214 This source accepts the following parameters:
21219 The size of the video to generate. For the syntax of this option, check the
21220 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21223 The framerate of the generated video. It may be a string of the form
21224 @var{num}/@var{den} or a frame rate abbreviation.
21227 The name to the frei0r source to load. For more information regarding frei0r and
21228 how to set the parameters, read the @ref{frei0r} section in the video filters
21231 @item filter_params
21232 A '|'-separated list of parameters to pass to the frei0r source.
21236 For example, to generate a frei0r partik0l source with size 200x200
21237 and frame rate 10 which is overlaid on the overlay filter main input:
21239 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
21244 Generate a life pattern.
21246 This source is based on a generalization of John Conway's life game.
21248 The sourced input represents a life grid, each pixel represents a cell
21249 which can be in one of two possible states, alive or dead. Every cell
21250 interacts with its eight neighbours, which are the cells that are
21251 horizontally, vertically, or diagonally adjacent.
21253 At each interaction the grid evolves according to the adopted rule,
21254 which specifies the number of neighbor alive cells which will make a
21255 cell stay alive or born. The @option{rule} option allows one to specify
21258 This source accepts the following options:
21262 Set the file from which to read the initial grid state. In the file,
21263 each non-whitespace character is considered an alive cell, and newline
21264 is used to delimit the end of each row.
21266 If this option is not specified, the initial grid is generated
21270 Set the video rate, that is the number of frames generated per second.
21273 @item random_fill_ratio, ratio
21274 Set the random fill ratio for the initial random grid. It is a
21275 floating point number value ranging from 0 to 1, defaults to 1/PHI.
21276 It is ignored when a file is specified.
21278 @item random_seed, seed
21279 Set the seed for filling the initial random grid, must be an integer
21280 included between 0 and UINT32_MAX. If not specified, or if explicitly
21281 set to -1, the filter will try to use a good random seed on a best
21287 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
21288 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
21289 @var{NS} specifies the number of alive neighbor cells which make a
21290 live cell stay alive, and @var{NB} the number of alive neighbor cells
21291 which make a dead cell to become alive (i.e. to "born").
21292 "s" and "b" can be used in place of "S" and "B", respectively.
21294 Alternatively a rule can be specified by an 18-bits integer. The 9
21295 high order bits are used to encode the next cell state if it is alive
21296 for each number of neighbor alive cells, the low order bits specify
21297 the rule for "borning" new cells. Higher order bits encode for an
21298 higher number of neighbor cells.
21299 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
21300 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
21302 Default value is "S23/B3", which is the original Conway's game of life
21303 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
21304 cells, and will born a new cell if there are three alive cells around
21308 Set the size of the output video. For the syntax of this option, check the
21309 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21311 If @option{filename} is specified, the size is set by default to the
21312 same size of the input file. If @option{size} is set, it must contain
21313 the size specified in the input file, and the initial grid defined in
21314 that file is centered in the larger resulting area.
21316 If a filename is not specified, the size value defaults to "320x240"
21317 (used for a randomly generated initial grid).
21320 If set to 1, stitch the left and right grid edges together, and the
21321 top and bottom edges also. Defaults to 1.
21324 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
21325 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
21326 value from 0 to 255.
21329 Set the color of living (or new born) cells.
21332 Set the color of dead cells. If @option{mold} is set, this is the first color
21333 used to represent a dead cell.
21336 Set mold color, for definitely dead and moldy cells.
21338 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
21339 ffmpeg-utils manual,ffmpeg-utils}.
21342 @subsection Examples
21346 Read a grid from @file{pattern}, and center it on a grid of size
21349 life=f=pattern:s=300x300
21353 Generate a random grid of size 200x200, with a fill ratio of 2/3:
21355 life=ratio=2/3:s=200x200
21359 Specify a custom rule for evolving a randomly generated grid:
21365 Full example with slow death effect (mold) using @command{ffplay}:
21367 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
21374 @anchor{haldclutsrc}
21377 @anchor{pal100bars}
21378 @anchor{rgbtestsrc}
21380 @anchor{smptehdbars}
21383 @anchor{yuvtestsrc}
21384 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
21386 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
21388 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
21390 The @code{color} source provides an uniformly colored input.
21392 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
21393 @ref{haldclut} filter.
21395 The @code{nullsrc} source returns unprocessed video frames. It is
21396 mainly useful to be employed in analysis / debugging tools, or as the
21397 source for filters which ignore the input data.
21399 The @code{pal75bars} source generates a color bars pattern, based on
21400 EBU PAL recommendations with 75% color levels.
21402 The @code{pal100bars} source generates a color bars pattern, based on
21403 EBU PAL recommendations with 100% color levels.
21405 The @code{rgbtestsrc} source generates an RGB test pattern useful for
21406 detecting RGB vs BGR issues. You should see a red, green and blue
21407 stripe from top to bottom.
21409 The @code{smptebars} source generates a color bars pattern, based on
21410 the SMPTE Engineering Guideline EG 1-1990.
21412 The @code{smptehdbars} source generates a color bars pattern, based on
21413 the SMPTE RP 219-2002.
21415 The @code{testsrc} source generates a test video pattern, showing a
21416 color pattern, a scrolling gradient and a timestamp. This is mainly
21417 intended for testing purposes.
21419 The @code{testsrc2} source is similar to testsrc, but supports more
21420 pixel formats instead of just @code{rgb24}. This allows using it as an
21421 input for other tests without requiring a format conversion.
21423 The @code{yuvtestsrc} source generates an YUV test pattern. You should
21424 see a y, cb and cr stripe from top to bottom.
21426 The sources accept the following parameters:
21431 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
21432 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
21433 pixels to be used as identity matrix for 3D lookup tables. Each component is
21434 coded on a @code{1/(N*N)} scale.
21437 Specify the color of the source, only available in the @code{color}
21438 source. For the syntax of this option, check the
21439 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
21442 Specify the size of the sourced video. For the syntax of this option, check the
21443 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21444 The default value is @code{320x240}.
21446 This option is not available with the @code{allrgb}, @code{allyuv}, and
21447 @code{haldclutsrc} filters.
21450 Specify the frame rate of the sourced video, as the number of frames
21451 generated per second. It has to be a string in the format
21452 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21453 number or a valid video frame rate abbreviation. The default value is
21457 Set the duration of the sourced video. See
21458 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21459 for the accepted syntax.
21461 If not specified, or the expressed duration is negative, the video is
21462 supposed to be generated forever.
21465 Set the sample aspect ratio of the sourced video.
21468 Specify the alpha (opacity) of the background, only available in the
21469 @code{testsrc2} source. The value must be between 0 (fully transparent) and
21470 255 (fully opaque, the default).
21473 Set the number of decimals to show in the timestamp, only available in the
21474 @code{testsrc} source.
21476 The displayed timestamp value will correspond to the original
21477 timestamp value multiplied by the power of 10 of the specified
21478 value. Default value is 0.
21481 @subsection Examples
21485 Generate a video with a duration of 5.3 seconds, with size
21486 176x144 and a frame rate of 10 frames per second:
21488 testsrc=duration=5.3:size=qcif:rate=10
21492 The following graph description will generate a red source
21493 with an opacity of 0.2, with size "qcif" and a frame rate of 10
21496 color=c=red@@0.2:s=qcif:r=10
21500 If the input content is to be ignored, @code{nullsrc} can be used. The
21501 following command generates noise in the luminance plane by employing
21502 the @code{geq} filter:
21504 nullsrc=s=256x256, geq=random(1)*255:128:128
21508 @subsection Commands
21510 The @code{color} source supports the following commands:
21514 Set the color of the created image. Accepts the same syntax of the
21515 corresponding @option{color} option.
21520 Generate video using an OpenCL program.
21525 OpenCL program source file.
21528 Kernel name in program.
21531 Size of frames to generate. This must be set.
21534 Pixel format to use for the generated frames. This must be set.
21537 Number of frames generated every second. Default value is '25'.
21541 For details of how the program loading works, see the @ref{program_opencl}
21548 Generate a colour ramp by setting pixel values from the position of the pixel
21549 in the output image. (Note that this will work with all pixel formats, but
21550 the generated output will not be the same.)
21552 __kernel void ramp(__write_only image2d_t dst,
21553 unsigned int index)
21555 int2 loc = (int2)(get_global_id(0), get_global_id(1));
21558 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
21560 write_imagef(dst, loc, val);
21565 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
21567 __kernel void sierpinski_carpet(__write_only image2d_t dst,
21568 unsigned int index)
21570 int2 loc = (int2)(get_global_id(0), get_global_id(1));
21572 float4 value = 0.0f;
21573 int x = loc.x + index;
21574 int y = loc.y + index;
21575 while (x > 0 || y > 0) {
21576 if (x % 3 == 1 && y % 3 == 1) {
21584 write_imagef(dst, loc, value);
21590 @section sierpinski
21592 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
21594 This source accepts the following options:
21598 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
21599 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
21602 Set frame rate, expressed as number of frames per second. Default
21606 Set seed which is used for random panning.
21609 Set max jump for single pan destination. Allowed range is from 1 to 10000.
21612 Set fractal type, can be default @code{carpet} or @code{triangle}.
21615 @c man end VIDEO SOURCES
21617 @chapter Video Sinks
21618 @c man begin VIDEO SINKS
21620 Below is a description of the currently available video sinks.
21622 @section buffersink
21624 Buffer video frames, and make them available to the end of the filter
21627 This sink is mainly intended for programmatic use, in particular
21628 through the interface defined in @file{libavfilter/buffersink.h}
21629 or the options system.
21631 It accepts a pointer to an AVBufferSinkContext structure, which
21632 defines the incoming buffers' formats, to be passed as the opaque
21633 parameter to @code{avfilter_init_filter} for initialization.
21637 Null video sink: do absolutely nothing with the input video. It is
21638 mainly useful as a template and for use in analysis / debugging
21641 @c man end VIDEO SINKS
21643 @chapter Multimedia Filters
21644 @c man begin MULTIMEDIA FILTERS
21646 Below is a description of the currently available multimedia filters.
21650 Convert input audio to a video output, displaying the audio bit scope.
21652 The filter accepts the following options:
21656 Set frame rate, expressed as number of frames per second. Default
21660 Specify the video size for the output. For the syntax of this option, check the
21661 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21662 Default value is @code{1024x256}.
21665 Specify list of colors separated by space or by '|' which will be used to
21666 draw channels. Unrecognized or missing colors will be replaced
21670 @section adrawgraph
21671 Draw a graph using input audio metadata.
21673 See @ref{drawgraph}
21675 @section agraphmonitor
21677 See @ref{graphmonitor}.
21679 @section ahistogram
21681 Convert input audio to a video output, displaying the volume histogram.
21683 The filter accepts the following options:
21687 Specify how histogram is calculated.
21689 It accepts the following values:
21692 Use single histogram for all channels.
21694 Use separate histogram for each channel.
21696 Default is @code{single}.
21699 Set frame rate, expressed as number of frames per second. Default
21703 Specify the video size for the output. For the syntax of this option, check the
21704 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21705 Default value is @code{hd720}.
21710 It accepts the following values:
21721 reverse logarithmic
21723 Default is @code{log}.
21726 Set amplitude scale.
21728 It accepts the following values:
21735 Default is @code{log}.
21738 Set how much frames to accumulate in histogram.
21739 Default is 1. Setting this to -1 accumulates all frames.
21742 Set histogram ratio of window height.
21745 Set sonogram sliding.
21747 It accepts the following values:
21750 replace old rows with new ones.
21752 scroll from top to bottom.
21754 Default is @code{replace}.
21757 @section aphasemeter
21759 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
21760 representing mean phase of current audio frame. A video output can also be produced and is
21761 enabled by default. The audio is passed through as first output.
21763 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
21764 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
21765 and @code{1} means channels are in phase.
21767 The filter accepts the following options, all related to its video output:
21771 Set the output frame rate. Default value is @code{25}.
21774 Set the video size for the output. For the syntax of this option, check the
21775 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21776 Default value is @code{800x400}.
21781 Specify the red, green, blue contrast. Default values are @code{2},
21782 @code{7} and @code{1}.
21783 Allowed range is @code{[0, 255]}.
21786 Set color which will be used for drawing median phase. If color is
21787 @code{none} which is default, no median phase value will be drawn.
21790 Enable video output. Default is enabled.
21793 @section avectorscope
21795 Convert input audio to a video output, representing the audio vector
21798 The filter is used to measure the difference between channels of stereo
21799 audio stream. A monaural signal, consisting of identical left and right
21800 signal, results in straight vertical line. Any stereo separation is visible
21801 as a deviation from this line, creating a Lissajous figure.
21802 If the straight (or deviation from it) but horizontal line appears this
21803 indicates that the left and right channels are out of phase.
21805 The filter accepts the following options:
21809 Set the vectorscope mode.
21811 Available values are:
21814 Lissajous rotated by 45 degrees.
21817 Same as above but not rotated.
21820 Shape resembling half of circle.
21823 Default value is @samp{lissajous}.
21826 Set the video size for the output. For the syntax of this option, check the
21827 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21828 Default value is @code{400x400}.
21831 Set the output frame rate. Default value is @code{25}.
21837 Specify the red, green, blue and alpha contrast. Default values are @code{40},
21838 @code{160}, @code{80} and @code{255}.
21839 Allowed range is @code{[0, 255]}.
21845 Specify the red, green, blue and alpha fade. Default values are @code{15},
21846 @code{10}, @code{5} and @code{5}.
21847 Allowed range is @code{[0, 255]}.
21850 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
21851 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
21854 Set the vectorscope drawing mode.
21856 Available values are:
21859 Draw dot for each sample.
21862 Draw line between previous and current sample.
21865 Default value is @samp{dot}.
21868 Specify amplitude scale of audio samples.
21870 Available values are:
21886 Swap left channel axis with right channel axis.
21896 Mirror only x axis.
21899 Mirror only y axis.
21907 @subsection Examples
21911 Complete example using @command{ffplay}:
21913 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
21914 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
21918 @section bench, abench
21920 Benchmark part of a filtergraph.
21922 The filter accepts the following options:
21926 Start or stop a timer.
21928 Available values are:
21931 Get the current time, set it as frame metadata (using the key
21932 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
21935 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
21936 the input frame metadata to get the time difference. Time difference, average,
21937 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
21938 @code{min}) are then printed. The timestamps are expressed in seconds.
21942 @subsection Examples
21946 Benchmark @ref{selectivecolor} filter:
21948 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
21954 Concatenate audio and video streams, joining them together one after the
21957 The filter works on segments of synchronized video and audio streams. All
21958 segments must have the same number of streams of each type, and that will
21959 also be the number of streams at output.
21961 The filter accepts the following options:
21966 Set the number of segments. Default is 2.
21969 Set the number of output video streams, that is also the number of video
21970 streams in each segment. Default is 1.
21973 Set the number of output audio streams, that is also the number of audio
21974 streams in each segment. Default is 0.
21977 Activate unsafe mode: do not fail if segments have a different format.
21981 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
21982 @var{a} audio outputs.
21984 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
21985 segment, in the same order as the outputs, then the inputs for the second
21988 Related streams do not always have exactly the same duration, for various
21989 reasons including codec frame size or sloppy authoring. For that reason,
21990 related synchronized streams (e.g. a video and its audio track) should be
21991 concatenated at once. The concat filter will use the duration of the longest
21992 stream in each segment (except the last one), and if necessary pad shorter
21993 audio streams with silence.
21995 For this filter to work correctly, all segments must start at timestamp 0.
21997 All corresponding streams must have the same parameters in all segments; the
21998 filtering system will automatically select a common pixel format for video
21999 streams, and a common sample format, sample rate and channel layout for
22000 audio streams, but other settings, such as resolution, must be converted
22001 explicitly by the user.
22003 Different frame rates are acceptable but will result in variable frame rate
22004 at output; be sure to configure the output file to handle it.
22006 @subsection Examples
22010 Concatenate an opening, an episode and an ending, all in bilingual version
22011 (video in stream 0, audio in streams 1 and 2):
22013 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
22014 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
22015 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
22016 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
22020 Concatenate two parts, handling audio and video separately, using the
22021 (a)movie sources, and adjusting the resolution:
22023 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
22024 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
22025 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
22027 Note that a desync will happen at the stitch if the audio and video streams
22028 do not have exactly the same duration in the first file.
22032 @subsection Commands
22034 This filter supports the following commands:
22037 Close the current segment and step to the next one
22043 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
22044 level. By default, it logs a message at a frequency of 10Hz with the
22045 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
22046 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
22048 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
22049 sample format is double-precision floating point. The input stream will be converted to
22050 this specification, if needed. Users may need to insert aformat and/or aresample filters
22051 after this filter to obtain the original parameters.
22053 The filter also has a video output (see the @var{video} option) with a real
22054 time graph to observe the loudness evolution. The graphic contains the logged
22055 message mentioned above, so it is not printed anymore when this option is set,
22056 unless the verbose logging is set. The main graphing area contains the
22057 short-term loudness (3 seconds of analysis), and the gauge on the right is for
22058 the momentary loudness (400 milliseconds), but can optionally be configured
22059 to instead display short-term loudness (see @var{gauge}).
22061 The green area marks a +/- 1LU target range around the target loudness
22062 (-23LUFS by default, unless modified through @var{target}).
22064 More information about the Loudness Recommendation EBU R128 on
22065 @url{http://tech.ebu.ch/loudness}.
22067 The filter accepts the following options:
22072 Activate the video output. The audio stream is passed unchanged whether this
22073 option is set or no. The video stream will be the first output stream if
22074 activated. Default is @code{0}.
22077 Set the video size. This option is for video only. For the syntax of this
22079 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22080 Default and minimum resolution is @code{640x480}.
22083 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
22084 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
22085 other integer value between this range is allowed.
22088 Set metadata injection. If set to @code{1}, the audio input will be segmented
22089 into 100ms output frames, each of them containing various loudness information
22090 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
22092 Default is @code{0}.
22095 Force the frame logging level.
22097 Available values are:
22100 information logging level
22102 verbose logging level
22105 By default, the logging level is set to @var{info}. If the @option{video} or
22106 the @option{metadata} options are set, it switches to @var{verbose}.
22111 Available modes can be cumulated (the option is a @code{flag} type). Possible
22115 Disable any peak mode (default).
22117 Enable sample-peak mode.
22119 Simple peak mode looking for the higher sample value. It logs a message
22120 for sample-peak (identified by @code{SPK}).
22122 Enable true-peak mode.
22124 If enabled, the peak lookup is done on an over-sampled version of the input
22125 stream for better peak accuracy. It logs a message for true-peak.
22126 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
22127 This mode requires a build with @code{libswresample}.
22131 Treat mono input files as "dual mono". If a mono file is intended for playback
22132 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
22133 If set to @code{true}, this option will compensate for this effect.
22134 Multi-channel input files are not affected by this option.
22137 Set a specific pan law to be used for the measurement of dual mono files.
22138 This parameter is optional, and has a default value of -3.01dB.
22141 Set a specific target level (in LUFS) used as relative zero in the visualization.
22142 This parameter is optional and has a default value of -23LUFS as specified
22143 by EBU R128. However, material published online may prefer a level of -16LUFS
22144 (e.g. for use with podcasts or video platforms).
22147 Set the value displayed by the gauge. Valid values are @code{momentary} and s
22148 @code{shortterm}. By default the momentary value will be used, but in certain
22149 scenarios it may be more useful to observe the short term value instead (e.g.
22153 Sets the display scale for the loudness. Valid parameters are @code{absolute}
22154 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
22155 video output, not the summary or continuous log output.
22158 @subsection Examples
22162 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
22164 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
22168 Run an analysis with @command{ffmpeg}:
22170 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
22174 @section interleave, ainterleave
22176 Temporally interleave frames from several inputs.
22178 @code{interleave} works with video inputs, @code{ainterleave} with audio.
22180 These filters read frames from several inputs and send the oldest
22181 queued frame to the output.
22183 Input streams must have well defined, monotonically increasing frame
22186 In order to submit one frame to output, these filters need to enqueue
22187 at least one frame for each input, so they cannot work in case one
22188 input is not yet terminated and will not receive incoming frames.
22190 For example consider the case when one input is a @code{select} filter
22191 which always drops input frames. The @code{interleave} filter will keep
22192 reading from that input, but it will never be able to send new frames
22193 to output until the input sends an end-of-stream signal.
22195 Also, depending on inputs synchronization, the filters will drop
22196 frames in case one input receives more frames than the other ones, and
22197 the queue is already filled.
22199 These filters accept the following options:
22203 Set the number of different inputs, it is 2 by default.
22206 @subsection Examples
22210 Interleave frames belonging to different streams using @command{ffmpeg}:
22212 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
22216 Add flickering blur effect:
22218 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
22222 @section metadata, ametadata
22224 Manipulate frame metadata.
22226 This filter accepts the following options:
22230 Set mode of operation of the filter.
22232 Can be one of the following:
22236 If both @code{value} and @code{key} is set, select frames
22237 which have such metadata. If only @code{key} is set, select
22238 every frame that has such key in metadata.
22241 Add new metadata @code{key} and @code{value}. If key is already available
22245 Modify value of already present key.
22248 If @code{value} is set, delete only keys that have such value.
22249 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
22253 Print key and its value if metadata was found. If @code{key} is not set print all
22254 metadata values available in frame.
22258 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
22261 Set metadata value which will be used. This option is mandatory for
22262 @code{modify} and @code{add} mode.
22265 Which function to use when comparing metadata value and @code{value}.
22267 Can be one of following:
22271 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
22274 Values are interpreted as strings, returns true if metadata value starts with
22275 the @code{value} option string.
22278 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
22281 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
22284 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
22287 Values are interpreted as floats, returns true if expression from option @code{expr}
22291 Values are interpreted as strings, returns true if metadata value ends with
22292 the @code{value} option string.
22296 Set expression which is used when @code{function} is set to @code{expr}.
22297 The expression is evaluated through the eval API and can contain the following
22302 Float representation of @code{value} from metadata key.
22305 Float representation of @code{value} as supplied by user in @code{value} option.
22309 If specified in @code{print} mode, output is written to the named file. Instead of
22310 plain filename any writable url can be specified. Filename ``-'' is a shorthand
22311 for standard output. If @code{file} option is not set, output is written to the log
22312 with AV_LOG_INFO loglevel.
22316 @subsection Examples
22320 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
22323 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
22326 Print silencedetect output to file @file{metadata.txt}.
22328 silencedetect,ametadata=mode=print:file=metadata.txt
22331 Direct all metadata to a pipe with file descriptor 4.
22333 metadata=mode=print:file='pipe\:4'
22337 @section perms, aperms
22339 Set read/write permissions for the output frames.
22341 These filters are mainly aimed at developers to test direct path in the
22342 following filter in the filtergraph.
22344 The filters accept the following options:
22348 Select the permissions mode.
22350 It accepts the following values:
22353 Do nothing. This is the default.
22355 Set all the output frames read-only.
22357 Set all the output frames directly writable.
22359 Make the frame read-only if writable, and writable if read-only.
22361 Set each output frame read-only or writable randomly.
22365 Set the seed for the @var{random} mode, must be an integer included between
22366 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
22367 @code{-1}, the filter will try to use a good random seed on a best effort
22371 Note: in case of auto-inserted filter between the permission filter and the
22372 following one, the permission might not be received as expected in that
22373 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
22374 perms/aperms filter can avoid this problem.
22376 @section realtime, arealtime
22378 Slow down filtering to match real time approximately.
22380 These filters will pause the filtering for a variable amount of time to
22381 match the output rate with the input timestamps.
22382 They are similar to the @option{re} option to @code{ffmpeg}.
22384 They accept the following options:
22388 Time limit for the pauses. Any pause longer than that will be considered
22389 a timestamp discontinuity and reset the timer. Default is 2 seconds.
22391 Speed factor for processing. The value must be a float larger than zero.
22392 Values larger than 1.0 will result in faster than realtime processing,
22393 smaller will slow processing down. The @var{limit} is automatically adapted
22394 accordingly. Default is 1.0.
22396 A processing speed faster than what is possible without these filters cannot
22401 @section select, aselect
22403 Select frames to pass in output.
22405 This filter accepts the following options:
22410 Set expression, which is evaluated for each input frame.
22412 If the expression is evaluated to zero, the frame is discarded.
22414 If the evaluation result is negative or NaN, the frame is sent to the
22415 first output; otherwise it is sent to the output with index
22416 @code{ceil(val)-1}, assuming that the input index starts from 0.
22418 For example a value of @code{1.2} corresponds to the output with index
22419 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
22422 Set the number of outputs. The output to which to send the selected
22423 frame is based on the result of the evaluation. Default value is 1.
22426 The expression can contain the following constants:
22430 The (sequential) number of the filtered frame, starting from 0.
22433 The (sequential) number of the selected frame, starting from 0.
22435 @item prev_selected_n
22436 The sequential number of the last selected frame. It's NAN if undefined.
22439 The timebase of the input timestamps.
22442 The PTS (Presentation TimeStamp) of the filtered video frame,
22443 expressed in @var{TB} units. It's NAN if undefined.
22446 The PTS of the filtered video frame,
22447 expressed in seconds. It's NAN if undefined.
22450 The PTS of the previously filtered video frame. It's NAN if undefined.
22452 @item prev_selected_pts
22453 The PTS of the last previously filtered video frame. It's NAN if undefined.
22455 @item prev_selected_t
22456 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
22459 The PTS of the first video frame in the video. It's NAN if undefined.
22462 The time of the first video frame in the video. It's NAN if undefined.
22464 @item pict_type @emph{(video only)}
22465 The type of the filtered frame. It can assume one of the following
22477 @item interlace_type @emph{(video only)}
22478 The frame interlace type. It can assume one of the following values:
22481 The frame is progressive (not interlaced).
22483 The frame is top-field-first.
22485 The frame is bottom-field-first.
22488 @item consumed_sample_n @emph{(audio only)}
22489 the number of selected samples before the current frame
22491 @item samples_n @emph{(audio only)}
22492 the number of samples in the current frame
22494 @item sample_rate @emph{(audio only)}
22495 the input sample rate
22498 This is 1 if the filtered frame is a key-frame, 0 otherwise.
22501 the position in the file of the filtered frame, -1 if the information
22502 is not available (e.g. for synthetic video)
22504 @item scene @emph{(video only)}
22505 value between 0 and 1 to indicate a new scene; a low value reflects a low
22506 probability for the current frame to introduce a new scene, while a higher
22507 value means the current frame is more likely to be one (see the example below)
22509 @item concatdec_select
22510 The concat demuxer can select only part of a concat input file by setting an
22511 inpoint and an outpoint, but the output packets may not be entirely contained
22512 in the selected interval. By using this variable, it is possible to skip frames
22513 generated by the concat demuxer which are not exactly contained in the selected
22516 This works by comparing the frame pts against the @var{lavf.concat.start_time}
22517 and the @var{lavf.concat.duration} packet metadata values which are also
22518 present in the decoded frames.
22520 The @var{concatdec_select} variable is -1 if the frame pts is at least
22521 start_time and either the duration metadata is missing or the frame pts is less
22522 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
22525 That basically means that an input frame is selected if its pts is within the
22526 interval set by the concat demuxer.
22530 The default value of the select expression is "1".
22532 @subsection Examples
22536 Select all frames in input:
22541 The example above is the same as:
22553 Select only I-frames:
22555 select='eq(pict_type\,I)'
22559 Select one frame every 100:
22561 select='not(mod(n\,100))'
22565 Select only frames contained in the 10-20 time interval:
22567 select=between(t\,10\,20)
22571 Select only I-frames contained in the 10-20 time interval:
22573 select=between(t\,10\,20)*eq(pict_type\,I)
22577 Select frames with a minimum distance of 10 seconds:
22579 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
22583 Use aselect to select only audio frames with samples number > 100:
22585 aselect='gt(samples_n\,100)'
22589 Create a mosaic of the first scenes:
22591 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
22594 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
22598 Send even and odd frames to separate outputs, and compose them:
22600 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
22604 Select useful frames from an ffconcat file which is using inpoints and
22605 outpoints but where the source files are not intra frame only.
22607 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
22611 @section sendcmd, asendcmd
22613 Send commands to filters in the filtergraph.
22615 These filters read commands to be sent to other filters in the
22618 @code{sendcmd} must be inserted between two video filters,
22619 @code{asendcmd} must be inserted between two audio filters, but apart
22620 from that they act the same way.
22622 The specification of commands can be provided in the filter arguments
22623 with the @var{commands} option, or in a file specified by the
22624 @var{filename} option.
22626 These filters accept the following options:
22629 Set the commands to be read and sent to the other filters.
22631 Set the filename of the commands to be read and sent to the other
22635 @subsection Commands syntax
22637 A commands description consists of a sequence of interval
22638 specifications, comprising a list of commands to be executed when a
22639 particular event related to that interval occurs. The occurring event
22640 is typically the current frame time entering or leaving a given time
22643 An interval is specified by the following syntax:
22645 @var{START}[-@var{END}] @var{COMMANDS};
22648 The time interval is specified by the @var{START} and @var{END} times.
22649 @var{END} is optional and defaults to the maximum time.
22651 The current frame time is considered within the specified interval if
22652 it is included in the interval [@var{START}, @var{END}), that is when
22653 the time is greater or equal to @var{START} and is lesser than
22656 @var{COMMANDS} consists of a sequence of one or more command
22657 specifications, separated by ",", relating to that interval. The
22658 syntax of a command specification is given by:
22660 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
22663 @var{FLAGS} is optional and specifies the type of events relating to
22664 the time interval which enable sending the specified command, and must
22665 be a non-null sequence of identifier flags separated by "+" or "|" and
22666 enclosed between "[" and "]".
22668 The following flags are recognized:
22671 The command is sent when the current frame timestamp enters the
22672 specified interval. In other words, the command is sent when the
22673 previous frame timestamp was not in the given interval, and the
22677 The command is sent when the current frame timestamp leaves the
22678 specified interval. In other words, the command is sent when the
22679 previous frame timestamp was in the given interval, and the
22683 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
22686 @var{TARGET} specifies the target of the command, usually the name of
22687 the filter class or a specific filter instance name.
22689 @var{COMMAND} specifies the name of the command for the target filter.
22691 @var{ARG} is optional and specifies the optional list of argument for
22692 the given @var{COMMAND}.
22694 Between one interval specification and another, whitespaces, or
22695 sequences of characters starting with @code{#} until the end of line,
22696 are ignored and can be used to annotate comments.
22698 A simplified BNF description of the commands specification syntax
22701 @var{COMMAND_FLAG} ::= "enter" | "leave"
22702 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
22703 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
22704 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
22705 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
22706 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
22709 @subsection Examples
22713 Specify audio tempo change at second 4:
22715 asendcmd=c='4.0 atempo tempo 1.5',atempo
22719 Target a specific filter instance:
22721 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
22725 Specify a list of drawtext and hue commands in a file.
22727 # show text in the interval 5-10
22728 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
22729 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
22731 # desaturate the image in the interval 15-20
22732 15.0-20.0 [enter] hue s 0,
22733 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
22735 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
22737 # apply an exponential saturation fade-out effect, starting from time 25
22738 25 [enter] hue s exp(25-t)
22741 A filtergraph allowing to read and process the above command list
22742 stored in a file @file{test.cmd}, can be specified with:
22744 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
22749 @section setpts, asetpts
22751 Change the PTS (presentation timestamp) of the input frames.
22753 @code{setpts} works on video frames, @code{asetpts} on audio frames.
22755 This filter accepts the following options:
22760 The expression which is evaluated for each frame to construct its timestamp.
22764 The expression is evaluated through the eval API and can contain the following
22768 @item FRAME_RATE, FR
22769 frame rate, only defined for constant frame-rate video
22772 The presentation timestamp in input
22775 The count of the input frame for video or the number of consumed samples,
22776 not including the current frame for audio, starting from 0.
22778 @item NB_CONSUMED_SAMPLES
22779 The number of consumed samples, not including the current frame (only
22782 @item NB_SAMPLES, S
22783 The number of samples in the current frame (only audio)
22785 @item SAMPLE_RATE, SR
22786 The audio sample rate.
22789 The PTS of the first frame.
22792 the time in seconds of the first frame
22795 State whether the current frame is interlaced.
22798 the time in seconds of the current frame
22801 original position in the file of the frame, or undefined if undefined
22802 for the current frame
22805 The previous input PTS.
22808 previous input time in seconds
22811 The previous output PTS.
22814 previous output time in seconds
22817 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
22821 The wallclock (RTC) time at the start of the movie in microseconds.
22824 The timebase of the input timestamps.
22828 @subsection Examples
22832 Start counting PTS from zero
22834 setpts=PTS-STARTPTS
22838 Apply fast motion effect:
22844 Apply slow motion effect:
22850 Set fixed rate of 25 frames per second:
22856 Set fixed rate 25 fps with some jitter:
22858 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
22862 Apply an offset of 10 seconds to the input PTS:
22868 Generate timestamps from a "live source" and rebase onto the current timebase:
22870 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
22874 Generate timestamps by counting samples:
22883 Force color range for the output video frame.
22885 The @code{setrange} filter marks the color range property for the
22886 output frames. It does not change the input frame, but only sets the
22887 corresponding property, which affects how the frame is treated by
22890 The filter accepts the following options:
22895 Available values are:
22899 Keep the same color range property.
22901 @item unspecified, unknown
22902 Set the color range as unspecified.
22904 @item limited, tv, mpeg
22905 Set the color range as limited.
22907 @item full, pc, jpeg
22908 Set the color range as full.
22912 @section settb, asettb
22914 Set the timebase to use for the output frames timestamps.
22915 It is mainly useful for testing timebase configuration.
22917 It accepts the following parameters:
22922 The expression which is evaluated into the output timebase.
22926 The value for @option{tb} is an arithmetic expression representing a
22927 rational. The expression can contain the constants "AVTB" (the default
22928 timebase), "intb" (the input timebase) and "sr" (the sample rate,
22929 audio only). Default value is "intb".
22931 @subsection Examples
22935 Set the timebase to 1/25:
22941 Set the timebase to 1/10:
22947 Set the timebase to 1001/1000:
22953 Set the timebase to 2*intb:
22959 Set the default timebase value:
22966 Convert input audio to a video output representing frequency spectrum
22967 logarithmically using Brown-Puckette constant Q transform algorithm with
22968 direct frequency domain coefficient calculation (but the transform itself
22969 is not really constant Q, instead the Q factor is actually variable/clamped),
22970 with musical tone scale, from E0 to D#10.
22972 The filter accepts the following options:
22976 Specify the video size for the output. It must be even. For the syntax of this option,
22977 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22978 Default value is @code{1920x1080}.
22981 Set the output frame rate. Default value is @code{25}.
22984 Set the bargraph height. It must be even. Default value is @code{-1} which
22985 computes the bargraph height automatically.
22988 Set the axis height. It must be even. Default value is @code{-1} which computes
22989 the axis height automatically.
22992 Set the sonogram height. It must be even. Default value is @code{-1} which
22993 computes the sonogram height automatically.
22996 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
22997 instead. Default value is @code{1}.
22999 @item sono_v, volume
23000 Specify the sonogram volume expression. It can contain variables:
23003 the @var{bar_v} evaluated expression
23004 @item frequency, freq, f
23005 the frequency where it is evaluated
23006 @item timeclamp, tc
23007 the value of @var{timeclamp} option
23011 @item a_weighting(f)
23012 A-weighting of equal loudness
23013 @item b_weighting(f)
23014 B-weighting of equal loudness
23015 @item c_weighting(f)
23016 C-weighting of equal loudness.
23018 Default value is @code{16}.
23020 @item bar_v, volume2
23021 Specify the bargraph volume expression. It can contain variables:
23024 the @var{sono_v} evaluated expression
23025 @item frequency, freq, f
23026 the frequency where it is evaluated
23027 @item timeclamp, tc
23028 the value of @var{timeclamp} option
23032 @item a_weighting(f)
23033 A-weighting of equal loudness
23034 @item b_weighting(f)
23035 B-weighting of equal loudness
23036 @item c_weighting(f)
23037 C-weighting of equal loudness.
23039 Default value is @code{sono_v}.
23041 @item sono_g, gamma
23042 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
23043 higher gamma makes the spectrum having more range. Default value is @code{3}.
23044 Acceptable range is @code{[1, 7]}.
23046 @item bar_g, gamma2
23047 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
23051 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
23052 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
23054 @item timeclamp, tc
23055 Specify the transform timeclamp. At low frequency, there is trade-off between
23056 accuracy in time domain and frequency domain. If timeclamp is lower,
23057 event in time domain is represented more accurately (such as fast bass drum),
23058 otherwise event in frequency domain is represented more accurately
23059 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
23062 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
23063 limits future samples by applying asymmetric windowing in time domain, useful
23064 when low latency is required. Accepted range is @code{[0, 1]}.
23067 Specify the transform base frequency. Default value is @code{20.01523126408007475},
23068 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
23071 Specify the transform end frequency. Default value is @code{20495.59681441799654},
23072 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
23075 This option is deprecated and ignored.
23078 Specify the transform length in time domain. Use this option to control accuracy
23079 trade-off between time domain and frequency domain at every frequency sample.
23080 It can contain variables:
23082 @item frequency, freq, f
23083 the frequency where it is evaluated
23084 @item timeclamp, tc
23085 the value of @var{timeclamp} option.
23087 Default value is @code{384*tc/(384+tc*f)}.
23090 Specify the transform count for every video frame. Default value is @code{6}.
23091 Acceptable range is @code{[1, 30]}.
23094 Specify the transform count for every single pixel. Default value is @code{0},
23095 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
23098 Specify font file for use with freetype to draw the axis. If not specified,
23099 use embedded font. Note that drawing with font file or embedded font is not
23100 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
23104 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
23105 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
23109 Specify font color expression. This is arithmetic expression that should return
23110 integer value 0xRRGGBB. It can contain variables:
23112 @item frequency, freq, f
23113 the frequency where it is evaluated
23114 @item timeclamp, tc
23115 the value of @var{timeclamp} option
23120 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
23121 @item r(x), g(x), b(x)
23122 red, green, and blue value of intensity x.
23124 Default value is @code{st(0, (midi(f)-59.5)/12);
23125 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
23126 r(1-ld(1)) + b(ld(1))}.
23129 Specify image file to draw the axis. This option override @var{fontfile} and
23130 @var{fontcolor} option.
23133 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
23134 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
23135 Default value is @code{1}.
23138 Set colorspace. The accepted values are:
23141 Unspecified (default)
23150 BT.470BG or BT.601-6 625
23153 SMPTE-170M or BT.601-6 525
23159 BT.2020 with non-constant luminance
23164 Set spectrogram color scheme. This is list of floating point values with format
23165 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
23166 The default is @code{1|0.5|0|0|0.5|1}.
23170 @subsection Examples
23174 Playing audio while showing the spectrum:
23176 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
23180 Same as above, but with frame rate 30 fps:
23182 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
23186 Playing at 1280x720:
23188 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
23192 Disable sonogram display:
23198 A1 and its harmonics: A1, A2, (near)E3, A3:
23200 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),
23201 asplit[a][out1]; [a] showcqt [out0]'
23205 Same as above, but with more accuracy in frequency domain:
23207 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),
23208 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
23214 bar_v=10:sono_v=bar_v*a_weighting(f)
23218 Custom gamma, now spectrum is linear to the amplitude.
23224 Custom tlength equation:
23226 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)))'
23230 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
23232 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
23236 Custom font using fontconfig:
23238 font='Courier New,Monospace,mono|bold'
23242 Custom frequency range with custom axis using image file:
23244 axisfile=myaxis.png:basefreq=40:endfreq=10000
23250 Convert input audio to video output representing the audio power spectrum.
23251 Audio amplitude is on Y-axis while frequency is on X-axis.
23253 The filter accepts the following options:
23257 Specify size of video. For the syntax of this option, check the
23258 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23259 Default is @code{1024x512}.
23263 This set how each frequency bin will be represented.
23265 It accepts the following values:
23271 Default is @code{bar}.
23274 Set amplitude scale.
23276 It accepts the following values:
23290 Default is @code{log}.
23293 Set frequency scale.
23295 It accepts the following values:
23304 Reverse logarithmic scale.
23306 Default is @code{lin}.
23309 Set window size. Allowed range is from 16 to 65536.
23311 Default is @code{2048}
23314 Set windowing function.
23316 It accepts the following values:
23339 Default is @code{hanning}.
23342 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
23343 which means optimal overlap for selected window function will be picked.
23346 Set time averaging. Setting this to 0 will display current maximal peaks.
23347 Default is @code{1}, which means time averaging is disabled.
23350 Specify list of colors separated by space or by '|' which will be used to
23351 draw channel frequencies. Unrecognized or missing colors will be replaced
23355 Set channel display mode.
23357 It accepts the following values:
23362 Default is @code{combined}.
23365 Set minimum amplitude used in @code{log} amplitude scaler.
23369 @section showspatial
23371 Convert stereo input audio to a video output, representing the spatial relationship
23372 between two channels.
23374 The filter accepts the following options:
23378 Specify the video size for the output. For the syntax of this option, check the
23379 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23380 Default value is @code{512x512}.
23383 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
23386 Set window function.
23388 It accepts the following values:
23413 Default value is @code{hann}.
23416 Set ratio of overlap window. Default value is @code{0.5}.
23417 When value is @code{1} overlap is set to recommended size for specific
23418 window function currently used.
23421 @anchor{showspectrum}
23422 @section showspectrum
23424 Convert input audio to a video output, representing the audio frequency
23427 The filter accepts the following options:
23431 Specify the video size for the output. For the syntax of this option, check the
23432 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23433 Default value is @code{640x512}.
23436 Specify how the spectrum should slide along the window.
23438 It accepts the following values:
23441 the samples start again on the left when they reach the right
23443 the samples scroll from right to left
23445 frames are only produced when the samples reach the right
23447 the samples scroll from left to right
23450 Default value is @code{replace}.
23453 Specify display mode.
23455 It accepts the following values:
23458 all channels are displayed in the same row
23460 all channels are displayed in separate rows
23463 Default value is @samp{combined}.
23466 Specify display color mode.
23468 It accepts the following values:
23471 each channel is displayed in a separate color
23473 each channel is displayed using the same color scheme
23475 each channel is displayed using the rainbow color scheme
23477 each channel is displayed using the moreland color scheme
23479 each channel is displayed using the nebulae color scheme
23481 each channel is displayed using the fire color scheme
23483 each channel is displayed using the fiery color scheme
23485 each channel is displayed using the fruit color scheme
23487 each channel is displayed using the cool color scheme
23489 each channel is displayed using the magma color scheme
23491 each channel is displayed using the green color scheme
23493 each channel is displayed using the viridis color scheme
23495 each channel is displayed using the plasma color scheme
23497 each channel is displayed using the cividis color scheme
23499 each channel is displayed using the terrain color scheme
23502 Default value is @samp{channel}.
23505 Specify scale used for calculating intensity color values.
23507 It accepts the following values:
23512 square root, default
23523 Default value is @samp{sqrt}.
23526 Specify frequency scale.
23528 It accepts the following values:
23536 Default value is @samp{lin}.
23539 Set saturation modifier for displayed colors. Negative values provide
23540 alternative color scheme. @code{0} is no saturation at all.
23541 Saturation must be in [-10.0, 10.0] range.
23542 Default value is @code{1}.
23545 Set window function.
23547 It accepts the following values:
23572 Default value is @code{hann}.
23575 Set orientation of time vs frequency axis. Can be @code{vertical} or
23576 @code{horizontal}. Default is @code{vertical}.
23579 Set ratio of overlap window. Default value is @code{0}.
23580 When value is @code{1} overlap is set to recommended size for specific
23581 window function currently used.
23584 Set scale gain for calculating intensity color values.
23585 Default value is @code{1}.
23588 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
23591 Set color rotation, must be in [-1.0, 1.0] range.
23592 Default value is @code{0}.
23595 Set start frequency from which to display spectrogram. Default is @code{0}.
23598 Set stop frequency to which to display spectrogram. Default is @code{0}.
23601 Set upper frame rate limit. Default is @code{auto}, unlimited.
23604 Draw time and frequency axes and legends. Default is disabled.
23607 The usage is very similar to the showwaves filter; see the examples in that
23610 @subsection Examples
23614 Large window with logarithmic color scaling:
23616 showspectrum=s=1280x480:scale=log
23620 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
23622 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23623 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
23627 @section showspectrumpic
23629 Convert input audio to a single video frame, representing the audio frequency
23632 The filter accepts the following options:
23636 Specify the video size for the output. For the syntax of this option, check the
23637 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23638 Default value is @code{4096x2048}.
23641 Specify display mode.
23643 It accepts the following values:
23646 all channels are displayed in the same row
23648 all channels are displayed in separate rows
23650 Default value is @samp{combined}.
23653 Specify display color mode.
23655 It accepts the following values:
23658 each channel is displayed in a separate color
23660 each channel is displayed using the same color scheme
23662 each channel is displayed using the rainbow color scheme
23664 each channel is displayed using the moreland color scheme
23666 each channel is displayed using the nebulae color scheme
23668 each channel is displayed using the fire color scheme
23670 each channel is displayed using the fiery color scheme
23672 each channel is displayed using the fruit color scheme
23674 each channel is displayed using the cool color scheme
23676 each channel is displayed using the magma color scheme
23678 each channel is displayed using the green color scheme
23680 each channel is displayed using the viridis color scheme
23682 each channel is displayed using the plasma color scheme
23684 each channel is displayed using the cividis color scheme
23686 each channel is displayed using the terrain color scheme
23688 Default value is @samp{intensity}.
23691 Specify scale used for calculating intensity color values.
23693 It accepts the following values:
23698 square root, default
23708 Default value is @samp{log}.
23711 Specify frequency scale.
23713 It accepts the following values:
23721 Default value is @samp{lin}.
23724 Set saturation modifier for displayed colors. Negative values provide
23725 alternative color scheme. @code{0} is no saturation at all.
23726 Saturation must be in [-10.0, 10.0] range.
23727 Default value is @code{1}.
23730 Set window function.
23732 It accepts the following values:
23756 Default value is @code{hann}.
23759 Set orientation of time vs frequency axis. Can be @code{vertical} or
23760 @code{horizontal}. Default is @code{vertical}.
23763 Set scale gain for calculating intensity color values.
23764 Default value is @code{1}.
23767 Draw time and frequency axes and legends. Default is enabled.
23770 Set color rotation, must be in [-1.0, 1.0] range.
23771 Default value is @code{0}.
23774 Set start frequency from which to display spectrogram. Default is @code{0}.
23777 Set stop frequency to which to display spectrogram. Default is @code{0}.
23780 @subsection Examples
23784 Extract an audio spectrogram of a whole audio track
23785 in a 1024x1024 picture using @command{ffmpeg}:
23787 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
23791 @section showvolume
23793 Convert input audio volume to a video output.
23795 The filter accepts the following options:
23802 Set border width, allowed range is [0, 5]. Default is 1.
23805 Set channel width, allowed range is [80, 8192]. Default is 400.
23808 Set channel height, allowed range is [1, 900]. Default is 20.
23811 Set fade, allowed range is [0, 1]. Default is 0.95.
23814 Set volume color expression.
23816 The expression can use the following variables:
23820 Current max volume of channel in dB.
23826 Current channel number, starting from 0.
23830 If set, displays channel names. Default is enabled.
23833 If set, displays volume values. Default is enabled.
23836 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
23837 default is @code{h}.
23840 Set step size, allowed range is [0, 5]. Default is 0, which means
23844 Set background opacity, allowed range is [0, 1]. Default is 0.
23847 Set metering mode, can be peak: @code{p} or rms: @code{r},
23848 default is @code{p}.
23851 Set display scale, can be linear: @code{lin} or log: @code{log},
23852 default is @code{lin}.
23856 If set to > 0., display a line for the max level
23857 in the previous seconds.
23858 default is disabled: @code{0.}
23861 The color of the max line. Use when @code{dm} option is set to > 0.
23862 default is: @code{orange}
23867 Convert input audio to a video output, representing the samples waves.
23869 The filter accepts the following options:
23873 Specify the video size for the output. For the syntax of this option, check the
23874 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23875 Default value is @code{600x240}.
23880 Available values are:
23883 Draw a point for each sample.
23886 Draw a vertical line for each sample.
23889 Draw a point for each sample and a line between them.
23892 Draw a centered vertical line for each sample.
23895 Default value is @code{point}.
23898 Set the number of samples which are printed on the same column. A
23899 larger value will decrease the frame rate. Must be a positive
23900 integer. This option can be set only if the value for @var{rate}
23901 is not explicitly specified.
23904 Set the (approximate) output frame rate. This is done by setting the
23905 option @var{n}. Default value is "25".
23907 @item split_channels
23908 Set if channels should be drawn separately or overlap. Default value is 0.
23911 Set colors separated by '|' which are going to be used for drawing of each channel.
23914 Set amplitude scale.
23916 Available values are:
23934 Set the draw mode. This is mostly useful to set for high @var{n}.
23936 Available values are:
23939 Scale pixel values for each drawn sample.
23942 Draw every sample directly.
23945 Default value is @code{scale}.
23948 @subsection Examples
23952 Output the input file audio and the corresponding video representation
23955 amovie=a.mp3,asplit[out0],showwaves[out1]
23959 Create a synthetic signal and show it with showwaves, forcing a
23960 frame rate of 30 frames per second:
23962 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
23966 @section showwavespic
23968 Convert input audio to a single video frame, representing the samples waves.
23970 The filter accepts the following options:
23974 Specify the video size for the output. For the syntax of this option, check the
23975 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23976 Default value is @code{600x240}.
23978 @item split_channels
23979 Set if channels should be drawn separately or overlap. Default value is 0.
23982 Set colors separated by '|' which are going to be used for drawing of each channel.
23985 Set amplitude scale.
23987 Available values are:
24007 Available values are:
24010 Scale pixel values for each drawn sample.
24013 Draw every sample directly.
24016 Default value is @code{scale}.
24019 @subsection Examples
24023 Extract a channel split representation of the wave form of a whole audio track
24024 in a 1024x800 picture using @command{ffmpeg}:
24026 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
24030 @section sidedata, asidedata
24032 Delete frame side data, or select frames based on it.
24034 This filter accepts the following options:
24038 Set mode of operation of the filter.
24040 Can be one of the following:
24044 Select every frame with side data of @code{type}.
24047 Delete side data of @code{type}. If @code{type} is not set, delete all side
24053 Set side data type used with all modes. Must be set for @code{select} mode. For
24054 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
24055 in @file{libavutil/frame.h}. For example, to choose
24056 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
24060 @section spectrumsynth
24062 Synthesize audio from 2 input video spectrums, first input stream represents
24063 magnitude across time and second represents phase across time.
24064 The filter will transform from frequency domain as displayed in videos back
24065 to time domain as presented in audio output.
24067 This filter is primarily created for reversing processed @ref{showspectrum}
24068 filter outputs, but can synthesize sound from other spectrograms too.
24069 But in such case results are going to be poor if the phase data is not
24070 available, because in such cases phase data need to be recreated, usually
24071 it's just recreated from random noise.
24072 For best results use gray only output (@code{channel} color mode in
24073 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
24074 @code{lin} scale for phase video. To produce phase, for 2nd video, use
24075 @code{data} option. Inputs videos should generally use @code{fullframe}
24076 slide mode as that saves resources needed for decoding video.
24078 The filter accepts the following options:
24082 Specify sample rate of output audio, the sample rate of audio from which
24083 spectrum was generated may differ.
24086 Set number of channels represented in input video spectrums.
24089 Set scale which was used when generating magnitude input spectrum.
24090 Can be @code{lin} or @code{log}. Default is @code{log}.
24093 Set slide which was used when generating inputs spectrums.
24094 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
24095 Default is @code{fullframe}.
24098 Set window function used for resynthesis.
24101 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
24102 which means optimal overlap for selected window function will be picked.
24105 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
24106 Default is @code{vertical}.
24109 @subsection Examples
24113 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
24114 then resynthesize videos back to audio with spectrumsynth:
24116 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
24117 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
24118 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
24122 @section split, asplit
24124 Split input into several identical outputs.
24126 @code{asplit} works with audio input, @code{split} with video.
24128 The filter accepts a single parameter which specifies the number of outputs. If
24129 unspecified, it defaults to 2.
24131 @subsection Examples
24135 Create two separate outputs from the same input:
24137 [in] split [out0][out1]
24141 To create 3 or more outputs, you need to specify the number of
24144 [in] asplit=3 [out0][out1][out2]
24148 Create two separate outputs from the same input, one cropped and
24151 [in] split [splitout1][splitout2];
24152 [splitout1] crop=100:100:0:0 [cropout];
24153 [splitout2] pad=200:200:100:100 [padout];
24157 Create 5 copies of the input audio with @command{ffmpeg}:
24159 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
24165 Receive commands sent through a libzmq client, and forward them to
24166 filters in the filtergraph.
24168 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
24169 must be inserted between two video filters, @code{azmq} between two
24170 audio filters. Both are capable to send messages to any filter type.
24172 To enable these filters you need to install the libzmq library and
24173 headers and configure FFmpeg with @code{--enable-libzmq}.
24175 For more information about libzmq see:
24176 @url{http://www.zeromq.org/}
24178 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
24179 receives messages sent through a network interface defined by the
24180 @option{bind_address} (or the abbreviation "@option{b}") option.
24181 Default value of this option is @file{tcp://localhost:5555}. You may
24182 want to alter this value to your needs, but do not forget to escape any
24183 ':' signs (see @ref{filtergraph escaping}).
24185 The received message must be in the form:
24187 @var{TARGET} @var{COMMAND} [@var{ARG}]
24190 @var{TARGET} specifies the target of the command, usually the name of
24191 the filter class or a specific filter instance name. The default
24192 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
24193 but you can override this by using the @samp{filter_name@@id} syntax
24194 (see @ref{Filtergraph syntax}).
24196 @var{COMMAND} specifies the name of the command for the target filter.
24198 @var{ARG} is optional and specifies the optional argument list for the
24199 given @var{COMMAND}.
24201 Upon reception, the message is processed and the corresponding command
24202 is injected into the filtergraph. Depending on the result, the filter
24203 will send a reply to the client, adopting the format:
24205 @var{ERROR_CODE} @var{ERROR_REASON}
24209 @var{MESSAGE} is optional.
24211 @subsection Examples
24213 Look at @file{tools/zmqsend} for an example of a zmq client which can
24214 be used to send commands processed by these filters.
24216 Consider the following filtergraph generated by @command{ffplay}.
24217 In this example the last overlay filter has an instance name. All other
24218 filters will have default instance names.
24221 ffplay -dumpgraph 1 -f lavfi "
24222 color=s=100x100:c=red [l];
24223 color=s=100x100:c=blue [r];
24224 nullsrc=s=200x100, zmq [bg];
24225 [bg][l] overlay [bg+l];
24226 [bg+l][r] overlay@@my=x=100 "
24229 To change the color of the left side of the video, the following
24230 command can be used:
24232 echo Parsed_color_0 c yellow | tools/zmqsend
24235 To change the right side:
24237 echo Parsed_color_1 c pink | tools/zmqsend
24240 To change the position of the right side:
24242 echo overlay@@my x 150 | tools/zmqsend
24246 @c man end MULTIMEDIA FILTERS
24248 @chapter Multimedia Sources
24249 @c man begin MULTIMEDIA SOURCES
24251 Below is a description of the currently available multimedia sources.
24255 This is the same as @ref{movie} source, except it selects an audio
24261 Read audio and/or video stream(s) from a movie container.
24263 It accepts the following parameters:
24267 The name of the resource to read (not necessarily a file; it can also be a
24268 device or a stream accessed through some protocol).
24270 @item format_name, f
24271 Specifies the format assumed for the movie to read, and can be either
24272 the name of a container or an input device. If not specified, the
24273 format is guessed from @var{movie_name} or by probing.
24275 @item seek_point, sp
24276 Specifies the seek point in seconds. The frames will be output
24277 starting from this seek point. The parameter is evaluated with
24278 @code{av_strtod}, so the numerical value may be suffixed by an IS
24279 postfix. The default value is "0".
24282 Specifies the streams to read. Several streams can be specified,
24283 separated by "+". The source will then have as many outputs, in the
24284 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
24285 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
24286 respectively the default (best suited) video and audio stream. Default
24287 is "dv", or "da" if the filter is called as "amovie".
24289 @item stream_index, si
24290 Specifies the index of the video stream to read. If the value is -1,
24291 the most suitable video stream will be automatically selected. The default
24292 value is "-1". Deprecated. If the filter is called "amovie", it will select
24293 audio instead of video.
24296 Specifies how many times to read the stream in sequence.
24297 If the value is 0, the stream will be looped infinitely.
24298 Default value is "1".
24300 Note that when the movie is looped the source timestamps are not
24301 changed, so it will generate non monotonically increasing timestamps.
24303 @item discontinuity
24304 Specifies the time difference between frames above which the point is
24305 considered a timestamp discontinuity which is removed by adjusting the later
24309 It allows overlaying a second video on top of the main input of
24310 a filtergraph, as shown in this graph:
24312 input -----------> deltapts0 --> overlay --> output
24315 movie --> scale--> deltapts1 -------+
24317 @subsection Examples
24321 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
24322 on top of the input labelled "in":
24324 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
24325 [in] setpts=PTS-STARTPTS [main];
24326 [main][over] overlay=16:16 [out]
24330 Read from a video4linux2 device, and overlay it on top of the input
24333 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
24334 [in] setpts=PTS-STARTPTS [main];
24335 [main][over] overlay=16:16 [out]
24339 Read the first video stream and the audio stream with id 0x81 from
24340 dvd.vob; the video is connected to the pad named "video" and the audio is
24341 connected to the pad named "audio":
24343 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
24347 @subsection Commands
24349 Both movie and amovie support the following commands:
24352 Perform seek using "av_seek_frame".
24353 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
24356 @var{stream_index}: If stream_index is -1, a default
24357 stream is selected, and @var{timestamp} is automatically converted
24358 from AV_TIME_BASE units to the stream specific time_base.
24360 @var{timestamp}: Timestamp in AVStream.time_base units
24361 or, if no stream is specified, in AV_TIME_BASE units.
24363 @var{flags}: Flags which select direction and seeking mode.
24367 Get movie duration in AV_TIME_BASE units.
24371 @c man end MULTIMEDIA SOURCES