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 Normalize biquad coefficients, by default is disabled.
1553 Enabling it will normalize magnitude response at DC to 0dB.
1556 @subsection Commands
1558 This filter supports the following commands:
1561 Change allpass frequency.
1562 Syntax for the command is : "@var{frequency}"
1565 Change allpass width_type.
1566 Syntax for the command is : "@var{width_type}"
1569 Change allpass width.
1570 Syntax for the command is : "@var{width}"
1574 Syntax for the command is : "@var{mix}"
1581 The filter accepts the following options:
1585 Set the number of loops. Setting this value to -1 will result in infinite loops.
1589 Set maximal number of samples. Default is 0.
1592 Set first sample of loop. Default is 0.
1598 Merge two or more audio streams into a single multi-channel stream.
1600 The filter accepts the following options:
1605 Set the number of inputs. Default is 2.
1609 If the channel layouts of the inputs are disjoint, and therefore compatible,
1610 the channel layout of the output will be set accordingly and the channels
1611 will be reordered as necessary. If the channel layouts of the inputs are not
1612 disjoint, the output will have all the channels of the first input then all
1613 the channels of the second input, in that order, and the channel layout of
1614 the output will be the default value corresponding to the total number of
1617 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1618 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1619 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1620 first input, b1 is the first channel of the second input).
1622 On the other hand, if both input are in stereo, the output channels will be
1623 in the default order: a1, a2, b1, b2, and the channel layout will be
1624 arbitrarily set to 4.0, which may or may not be the expected value.
1626 All inputs must have the same sample rate, and format.
1628 If inputs do not have the same duration, the output will stop with the
1631 @subsection Examples
1635 Merge two mono files into a stereo stream:
1637 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1641 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1643 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
1649 Mixes multiple audio inputs into a single output.
1651 Note that this filter only supports float samples (the @var{amerge}
1652 and @var{pan} audio filters support many formats). If the @var{amix}
1653 input has integer samples then @ref{aresample} will be automatically
1654 inserted to perform the conversion to float samples.
1658 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1660 will mix 3 input audio streams to a single output with the same duration as the
1661 first input and a dropout transition time of 3 seconds.
1663 It accepts the following parameters:
1667 The number of inputs. If unspecified, it defaults to 2.
1670 How to determine the end-of-stream.
1674 The duration of the longest input. (default)
1677 The duration of the shortest input.
1680 The duration of the first input.
1684 @item dropout_transition
1685 The transition time, in seconds, for volume renormalization when an input
1686 stream ends. The default value is 2 seconds.
1689 Specify weight of each input audio stream as sequence.
1690 Each weight is separated by space. By default all inputs have same weight.
1695 Multiply first audio stream with second audio stream and store result
1696 in output audio stream. Multiplication is done by multiplying each
1697 sample from first stream with sample at same position from second stream.
1699 With this element-wise multiplication one can create amplitude fades and
1700 amplitude modulations.
1702 @section anequalizer
1704 High-order parametric multiband equalizer for each channel.
1706 It accepts the following parameters:
1710 This option string is in format:
1711 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1712 Each equalizer band is separated by '|'.
1716 Set channel number to which equalization will be applied.
1717 If input doesn't have that channel the entry is ignored.
1720 Set central frequency for band.
1721 If input doesn't have that frequency the entry is ignored.
1724 Set band width in hertz.
1727 Set band gain in dB.
1730 Set filter type for band, optional, can be:
1734 Butterworth, this is default.
1745 With this option activated frequency response of anequalizer is displayed
1749 Set video stream size. Only useful if curves option is activated.
1752 Set max gain that will be displayed. Only useful if curves option is activated.
1753 Setting this to a reasonable value makes it possible to display gain which is derived from
1754 neighbour bands which are too close to each other and thus produce higher gain
1755 when both are activated.
1758 Set frequency scale used to draw frequency response in video output.
1759 Can be linear or logarithmic. Default is logarithmic.
1762 Set color for each channel curve which is going to be displayed in video stream.
1763 This is list of color names separated by space or by '|'.
1764 Unrecognised or missing colors will be replaced by white color.
1767 @subsection Examples
1771 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1772 for first 2 channels using Chebyshev type 1 filter:
1774 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1778 @subsection Commands
1780 This filter supports the following commands:
1783 Alter existing filter parameters.
1784 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1786 @var{fN} is existing filter number, starting from 0, if no such filter is available
1788 @var{freq} set new frequency parameter.
1789 @var{width} set new width parameter in herz.
1790 @var{gain} set new gain parameter in dB.
1792 Full filter invocation with asendcmd may look like this:
1793 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1798 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1800 Each sample is adjusted by looking for other samples with similar contexts. This
1801 context similarity is defined by comparing their surrounding patches of size
1802 @option{p}. Patches are searched in an area of @option{r} around the sample.
1804 The filter accepts the following options:
1808 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
1811 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
1812 Default value is 2 milliseconds.
1815 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
1816 Default value is 6 milliseconds.
1819 Set the output mode.
1821 It accepts the following values:
1824 Pass input unchanged.
1827 Pass noise filtered out.
1832 Default value is @var{o}.
1836 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
1839 @subsection Commands
1841 This filter supports the following commands:
1844 Change denoise strength. Argument is single float number.
1845 Syntax for the command is : "@var{s}"
1849 Syntax for the command is : "i", "o" or "n" string.
1853 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
1855 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
1856 relate to producing the least mean square of the error signal (difference between the desired,
1857 2nd input audio stream and the actual signal, the 1st input audio stream).
1859 A description of the accepted options follows.
1872 Set the filter leakage.
1875 It accepts the following values:
1884 Pass filtered samples.
1887 Pass difference between desired and filtered samples.
1889 Default value is @var{o}.
1893 @subsection Examples
1897 One of many usages of this filter is noise reduction, input audio is filtered
1898 with same samples that are delayed by fixed amount, one such example for stereo audio is:
1900 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
1904 @subsection Commands
1906 This filter supports the same commands as options, excluding option @code{order}.
1910 Pass the audio source unchanged to the output.
1914 Pad the end of an audio stream with silence.
1916 This can be used together with @command{ffmpeg} @option{-shortest} to
1917 extend audio streams to the same length as the video stream.
1919 A description of the accepted options follows.
1923 Set silence packet size. Default value is 4096.
1926 Set the number of samples of silence to add to the end. After the
1927 value is reached, the stream is terminated. This option is mutually
1928 exclusive with @option{whole_len}.
1931 Set the minimum total number of samples in the output audio stream. If
1932 the value is longer than the input audio length, silence is added to
1933 the end, until the value is reached. This option is mutually exclusive
1934 with @option{pad_len}.
1937 Specify the duration of samples of silence to add. See
1938 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1939 for the accepted syntax. Used only if set to non-zero value.
1942 Specify the minimum total duration in the output audio stream. See
1943 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1944 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
1945 the input audio length, silence is added to the end, until the value is reached.
1946 This option is mutually exclusive with @option{pad_dur}
1949 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
1950 nor @option{whole_dur} option is set, the filter will add silence to the end of
1951 the input stream indefinitely.
1953 @subsection Examples
1957 Add 1024 samples of silence to the end of the input:
1963 Make sure the audio output will contain at least 10000 samples, pad
1964 the input with silence if required:
1966 apad=whole_len=10000
1970 Use @command{ffmpeg} to pad the audio input with silence, so that the
1971 video stream will always result the shortest and will be converted
1972 until the end in the output file when using the @option{shortest}
1975 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
1980 Add a phasing effect to the input audio.
1982 A phaser filter creates series of peaks and troughs in the frequency spectrum.
1983 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
1985 A description of the accepted parameters follows.
1989 Set input gain. Default is 0.4.
1992 Set output gain. Default is 0.74
1995 Set delay in milliseconds. Default is 3.0.
1998 Set decay. Default is 0.4.
2001 Set modulation speed in Hz. Default is 0.5.
2004 Set modulation type. Default is triangular.
2006 It accepts the following values:
2015 Audio pulsator is something between an autopanner and a tremolo.
2016 But it can produce funny stereo effects as well. Pulsator changes the volume
2017 of the left and right channel based on a LFO (low frequency oscillator) with
2018 different waveforms and shifted phases.
2019 This filter have the ability to define an offset between left and right
2020 channel. An offset of 0 means that both LFO shapes match each other.
2021 The left and right channel are altered equally - a conventional tremolo.
2022 An offset of 50% means that the shape of the right channel is exactly shifted
2023 in phase (or moved backwards about half of the frequency) - pulsator acts as
2024 an autopanner. At 1 both curves match again. Every setting in between moves the
2025 phase shift gapless between all stages and produces some "bypassing" sounds with
2026 sine and triangle waveforms. The more you set the offset near 1 (starting from
2027 the 0.5) the faster the signal passes from the left to the right speaker.
2029 The filter accepts the following options:
2033 Set input gain. By default it is 1. Range is [0.015625 - 64].
2036 Set output gain. By default it is 1. Range is [0.015625 - 64].
2039 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2040 sawup or sawdown. Default is sine.
2043 Set modulation. Define how much of original signal is affected by the LFO.
2046 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2049 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2052 Set pulse width. Default is 1. Allowed range is [0 - 2].
2055 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2058 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2062 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2066 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2067 if timing is set to hz.
2073 Resample the input audio to the specified parameters, using the
2074 libswresample library. If none are specified then the filter will
2075 automatically convert between its input and output.
2077 This filter is also able to stretch/squeeze the audio data to make it match
2078 the timestamps or to inject silence / cut out audio to make it match the
2079 timestamps, do a combination of both or do neither.
2081 The filter accepts the syntax
2082 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2083 expresses a sample rate and @var{resampler_options} is a list of
2084 @var{key}=@var{value} pairs, separated by ":". See the
2085 @ref{Resampler Options,,"Resampler Options" section in the
2086 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2087 for the complete list of supported options.
2089 @subsection Examples
2093 Resample the input audio to 44100Hz:
2099 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2100 samples per second compensation:
2102 aresample=async=1000
2108 Reverse an audio clip.
2110 Warning: This filter requires memory to buffer the entire clip, so trimming
2113 @subsection Examples
2117 Take the first 5 seconds of a clip, and reverse it.
2119 atrim=end=5,areverse
2125 Reduce noise from speech using Recurrent Neural Networks.
2127 This filter accepts the following options:
2131 Set train model file to load. This option is always required.
2134 @section asetnsamples
2136 Set the number of samples per each output audio frame.
2138 The last output packet may contain a different number of samples, as
2139 the filter will flush all the remaining samples when the input audio
2142 The filter accepts the following options:
2146 @item nb_out_samples, n
2147 Set the number of frames per each output audio frame. The number is
2148 intended as the number of samples @emph{per each channel}.
2149 Default value is 1024.
2152 If set to 1, the filter will pad the last audio frame with zeroes, so
2153 that the last frame will contain the same number of samples as the
2154 previous ones. Default value is 1.
2157 For example, to set the number of per-frame samples to 1234 and
2158 disable padding for the last frame, use:
2160 asetnsamples=n=1234:p=0
2165 Set the sample rate without altering the PCM data.
2166 This will result in a change of speed and pitch.
2168 The filter accepts the following options:
2171 @item sample_rate, r
2172 Set the output sample rate. Default is 44100 Hz.
2177 Show a line containing various information for each input audio frame.
2178 The input audio is not modified.
2180 The shown line contains a sequence of key/value pairs of the form
2181 @var{key}:@var{value}.
2183 The following values are shown in the output:
2187 The (sequential) number of the input frame, starting from 0.
2190 The presentation timestamp of the input frame, in time base units; the time base
2191 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2194 The presentation timestamp of the input frame in seconds.
2197 position of the frame in the input stream, -1 if this information in
2198 unavailable and/or meaningless (for example in case of synthetic audio)
2207 The sample rate for the audio frame.
2210 The number of samples (per channel) in the frame.
2213 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2214 audio, the data is treated as if all the planes were concatenated.
2216 @item plane_checksums
2217 A list of Adler-32 checksums for each data plane.
2221 Apply audio soft clipping.
2223 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2224 along a smooth curve, rather than the abrupt shape of hard-clipping.
2226 This filter accepts the following options:
2230 Set type of soft-clipping.
2232 It accepts the following values:
2244 Set additional parameter which controls sigmoid function.
2248 Automatic Speech Recognition
2250 This filter uses PocketSphinx for speech recognition. To enable
2251 compilation of this filter, you need to configure FFmpeg with
2252 @code{--enable-pocketsphinx}.
2254 It accepts the following options:
2258 Set sampling rate of input audio. Defaults is @code{16000}.
2259 This need to match speech models, otherwise one will get poor results.
2262 Set dictionary containing acoustic model files.
2265 Set pronunciation dictionary.
2268 Set language model file.
2271 Set language model set.
2274 Set which language model to use.
2277 Set output for log messages.
2280 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2285 Display time domain statistical information about the audio channels.
2286 Statistics are calculated and displayed for each audio channel and,
2287 where applicable, an overall figure is also given.
2289 It accepts the following option:
2292 Short window length in seconds, used for peak and trough RMS measurement.
2293 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2297 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2298 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2301 Available keys for each channel are:
2343 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2344 this @code{lavfi.astats.Overall.Peak_count}.
2346 For description what each key means read below.
2349 Set number of frame after which stats are going to be recalculated.
2350 Default is disabled.
2352 @item measure_perchannel
2353 Select the entries which need to be measured per channel. The metadata keys can
2354 be used as flags, default is @option{all} which measures everything.
2355 @option{none} disables all per channel measurement.
2357 @item measure_overall
2358 Select the entries which need to be measured overall. The metadata keys can
2359 be used as flags, default is @option{all} which measures everything.
2360 @option{none} disables all overall measurement.
2364 A description of each shown parameter follows:
2368 Mean amplitude displacement from zero.
2371 Minimal sample level.
2374 Maximal sample level.
2376 @item Min difference
2377 Minimal difference between two consecutive samples.
2379 @item Max difference
2380 Maximal difference between two consecutive samples.
2382 @item Mean difference
2383 Mean difference between two consecutive samples.
2384 The average of each difference between two consecutive samples.
2386 @item RMS difference
2387 Root Mean Square difference between two consecutive samples.
2391 Standard peak and RMS level measured in dBFS.
2395 Peak and trough values for RMS level measured over a short window.
2398 Standard ratio of peak to RMS level (note: not in dB).
2401 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2402 (i.e. either @var{Min level} or @var{Max level}).
2405 Number of occasions (not the number of samples) that the signal attained either
2406 @var{Min level} or @var{Max level}.
2409 Overall bit depth of audio. Number of bits used for each sample.
2412 Measured dynamic range of audio in dB.
2414 @item Zero crossings
2415 Number of points where the waveform crosses the zero level axis.
2417 @item Zero crossings rate
2418 Rate of Zero crossings and number of audio samples.
2425 The filter accepts exactly one parameter, the audio tempo. If not
2426 specified then the filter will assume nominal 1.0 tempo. Tempo must
2427 be in the [0.5, 100.0] range.
2429 Note that tempo greater than 2 will skip some samples rather than
2430 blend them in. If for any reason this is a concern it is always
2431 possible to daisy-chain several instances of atempo to achieve the
2432 desired product tempo.
2434 @subsection Examples
2438 Slow down audio to 80% tempo:
2444 To speed up audio to 300% tempo:
2450 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2452 atempo=sqrt(3),atempo=sqrt(3)
2456 @subsection Commands
2458 This filter supports the following commands:
2461 Change filter tempo scale factor.
2462 Syntax for the command is : "@var{tempo}"
2467 Trim the input so that the output contains one continuous subpart of the input.
2469 It accepts the following parameters:
2472 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2473 sample with the timestamp @var{start} will be the first sample in the output.
2476 Specify time of the first audio sample that will be dropped, i.e. the
2477 audio sample immediately preceding the one with the timestamp @var{end} will be
2478 the last sample in the output.
2481 Same as @var{start}, except this option sets the start timestamp in samples
2485 Same as @var{end}, except this option sets the end timestamp in samples instead
2489 The maximum duration of the output in seconds.
2492 The number of the first sample that should be output.
2495 The number of the first sample that should be dropped.
2498 @option{start}, @option{end}, and @option{duration} are expressed as time
2499 duration specifications; see
2500 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2502 Note that the first two sets of the start/end options and the @option{duration}
2503 option look at the frame timestamp, while the _sample options simply count the
2504 samples that pass through the filter. So start/end_pts and start/end_sample will
2505 give different results when the timestamps are wrong, inexact or do not start at
2506 zero. Also note that this filter does not modify the timestamps. If you wish
2507 to have the output timestamps start at zero, insert the asetpts filter after the
2510 If multiple start or end options are set, this filter tries to be greedy and
2511 keep all samples that match at least one of the specified constraints. To keep
2512 only the part that matches all the constraints at once, chain multiple atrim
2515 The defaults are such that all the input is kept. So it is possible to set e.g.
2516 just the end values to keep everything before the specified time.
2521 Drop everything except the second minute of input:
2523 ffmpeg -i INPUT -af atrim=60:120
2527 Keep only the first 1000 samples:
2529 ffmpeg -i INPUT -af atrim=end_sample=1000
2534 @section axcorrelate
2535 Calculate normalized cross-correlation between two input audio streams.
2537 Resulted samples are always between -1 and 1 inclusive.
2538 If result is 1 it means two input samples are highly correlated in that selected segment.
2539 Result 0 means they are not correlated at all.
2540 If result is -1 it means two input samples are out of phase, which means they cancel each
2543 The filter accepts the following options:
2547 Set size of segment over which cross-correlation is calculated.
2548 Default is 256. Allowed range is from 2 to 131072.
2551 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2552 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2553 are always zero and thus need much less calculations to make.
2554 This is generally not true, but is valid for typical audio streams.
2557 @subsection Examples
2561 Calculate correlation between channels in stereo audio stream:
2563 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2569 Apply a two-pole Butterworth band-pass filter with central
2570 frequency @var{frequency}, and (3dB-point) band-width width.
2571 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2572 instead of the default: constant 0dB peak gain.
2573 The filter roll off at 6dB per octave (20dB per decade).
2575 The filter accepts the following options:
2579 Set the filter's central frequency. Default is @code{3000}.
2582 Constant skirt gain if set to 1. Defaults to 0.
2585 Set method to specify band-width of filter.
2600 Specify the band-width of a filter in width_type units.
2603 How much to use filtered signal in output. Default is 1.
2604 Range is between 0 and 1.
2607 Specify which channels to filter, by default all available are filtered.
2610 Normalize biquad coefficients, by default is disabled.
2611 Enabling it will normalize magnitude response at DC to 0dB.
2614 @subsection Commands
2616 This filter supports the following commands:
2619 Change bandpass frequency.
2620 Syntax for the command is : "@var{frequency}"
2623 Change bandpass width_type.
2624 Syntax for the command is : "@var{width_type}"
2627 Change bandpass width.
2628 Syntax for the command is : "@var{width}"
2631 Change bandpass mix.
2632 Syntax for the command is : "@var{mix}"
2637 Apply a two-pole Butterworth band-reject filter with central
2638 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
2639 The filter roll off at 6dB per octave (20dB per decade).
2641 The filter accepts the following options:
2645 Set the filter's central frequency. Default is @code{3000}.
2648 Set method to specify band-width of filter.
2663 Specify the band-width of a filter in width_type units.
2666 How much to use filtered signal in output. Default is 1.
2667 Range is between 0 and 1.
2670 Specify which channels to filter, by default all available are filtered.
2673 Normalize biquad coefficients, by default is disabled.
2674 Enabling it will normalize magnitude response at DC to 0dB.
2677 @subsection Commands
2679 This filter supports the following commands:
2682 Change bandreject frequency.
2683 Syntax for the command is : "@var{frequency}"
2686 Change bandreject width_type.
2687 Syntax for the command is : "@var{width_type}"
2690 Change bandreject width.
2691 Syntax for the command is : "@var{width}"
2694 Change bandreject mix.
2695 Syntax for the command is : "@var{mix}"
2698 @section bass, lowshelf
2700 Boost or cut the bass (lower) frequencies of the audio using a two-pole
2701 shelving filter with a response similar to that of a standard
2702 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
2704 The filter accepts the following options:
2708 Give the gain at 0 Hz. Its useful range is about -20
2709 (for a large cut) to +20 (for a large boost).
2710 Beware of clipping when using a positive gain.
2713 Set the filter's central frequency and so can be used
2714 to extend or reduce the frequency range to be boosted or cut.
2715 The default value is @code{100} Hz.
2718 Set method to specify band-width of filter.
2733 Determine how steep is the filter's shelf transition.
2736 How much to use filtered signal in output. Default is 1.
2737 Range is between 0 and 1.
2740 Specify which channels to filter, by default all available are filtered.
2743 Normalize biquad coefficients, by default is disabled.
2744 Enabling it will normalize magnitude response at DC to 0dB.
2747 @subsection Commands
2749 This filter supports the following commands:
2752 Change bass frequency.
2753 Syntax for the command is : "@var{frequency}"
2756 Change bass width_type.
2757 Syntax for the command is : "@var{width_type}"
2761 Syntax for the command is : "@var{width}"
2765 Syntax for the command is : "@var{gain}"
2769 Syntax for the command is : "@var{mix}"
2774 Apply a biquad IIR filter with the given coefficients.
2775 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
2776 are the numerator and denominator coefficients respectively.
2777 and @var{channels}, @var{c} specify which channels to filter, by default all
2778 available are filtered.
2780 @subsection Commands
2782 This filter supports the following commands:
2790 Change biquad parameter.
2791 Syntax for the command is : "@var{value}"
2794 How much to use filtered signal in output. Default is 1.
2795 Range is between 0 and 1.
2798 Specify which channels to filter, by default all available are filtered.
2801 Normalize biquad coefficients, by default is disabled.
2802 Enabling it will normalize magnitude response at DC to 0dB.
2806 Bauer stereo to binaural transformation, which improves headphone listening of
2807 stereo audio records.
2809 To enable compilation of this filter you need to configure FFmpeg with
2810 @code{--enable-libbs2b}.
2812 It accepts the following parameters:
2816 Pre-defined crossfeed level.
2820 Default level (fcut=700, feed=50).
2823 Chu Moy circuit (fcut=700, feed=60).
2826 Jan Meier circuit (fcut=650, feed=95).
2831 Cut frequency (in Hz).
2840 Remap input channels to new locations.
2842 It accepts the following parameters:
2845 Map channels from input to output. The argument is a '|'-separated list of
2846 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
2847 @var{in_channel} form. @var{in_channel} can be either the name of the input
2848 channel (e.g. FL for front left) or its index in the input channel layout.
2849 @var{out_channel} is the name of the output channel or its index in the output
2850 channel layout. If @var{out_channel} is not given then it is implicitly an
2851 index, starting with zero and increasing by one for each mapping.
2853 @item channel_layout
2854 The channel layout of the output stream.
2857 If no mapping is present, the filter will implicitly map input channels to
2858 output channels, preserving indices.
2860 @subsection Examples
2864 For example, assuming a 5.1+downmix input MOV file,
2866 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
2868 will create an output WAV file tagged as stereo from the downmix channels of
2872 To fix a 5.1 WAV improperly encoded in AAC's native channel order
2874 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
2878 @section channelsplit
2880 Split each channel from an input audio stream into a separate output stream.
2882 It accepts the following parameters:
2884 @item channel_layout
2885 The channel layout of the input stream. The default is "stereo".
2887 A channel layout describing the channels to be extracted as separate output streams
2888 or "all" to extract each input channel as a separate stream. The default is "all".
2890 Choosing channels not present in channel layout in the input will result in an error.
2893 @subsection Examples
2897 For example, assuming a stereo input MP3 file,
2899 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
2901 will create an output Matroska file with two audio streams, one containing only
2902 the left channel and the other the right channel.
2905 Split a 5.1 WAV file into per-channel files:
2907 ffmpeg -i in.wav -filter_complex
2908 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
2909 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
2910 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
2915 Extract only LFE from a 5.1 WAV file:
2917 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
2918 -map '[LFE]' lfe.wav
2923 Add a chorus effect to the audio.
2925 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
2927 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
2928 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
2929 The modulation depth defines the range the modulated delay is played before or after
2930 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
2931 sound tuned around the original one, like in a chorus where some vocals are slightly
2934 It accepts the following parameters:
2937 Set input gain. Default is 0.4.
2940 Set output gain. Default is 0.4.
2943 Set delays. A typical delay is around 40ms to 60ms.
2955 @subsection Examples
2961 chorus=0.7:0.9:55:0.4:0.25:2
2967 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
2971 Fuller sounding chorus with three delays:
2973 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
2978 Compress or expand the audio's dynamic range.
2980 It accepts the following parameters:
2986 A list of times in seconds for each channel over which the instantaneous level
2987 of the input signal is averaged to determine its volume. @var{attacks} refers to
2988 increase of volume and @var{decays} refers to decrease of volume. For most
2989 situations, the attack time (response to the audio getting louder) should be
2990 shorter than the decay time, because the human ear is more sensitive to sudden
2991 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
2992 a typical value for decay is 0.8 seconds.
2993 If specified number of attacks & decays is lower than number of channels, the last
2994 set attack/decay will be used for all remaining channels.
2997 A list of points for the transfer function, specified in dB relative to the
2998 maximum possible signal amplitude. Each key points list must be defined using
2999 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3000 @code{x0/y0 x1/y1 x2/y2 ....}
3002 The input values must be in strictly increasing order but the transfer function
3003 does not have to be monotonically rising. The point @code{0/0} is assumed but
3004 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3005 function are @code{-70/-70|-60/-20|1/0}.
3008 Set the curve radius in dB for all joints. It defaults to 0.01.
3011 Set the additional gain in dB to be applied at all points on the transfer
3012 function. This allows for easy adjustment of the overall gain.
3016 Set an initial volume, in dB, to be assumed for each channel when filtering
3017 starts. This permits the user to supply a nominal level initially, so that, for
3018 example, a very large gain is not applied to initial signal levels before the
3019 companding has begun to operate. A typical value for audio which is initially
3020 quiet is -90 dB. It defaults to 0.
3023 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3024 delayed before being fed to the volume adjuster. Specifying a delay
3025 approximately equal to the attack/decay times allows the filter to effectively
3026 operate in predictive rather than reactive mode. It defaults to 0.
3030 @subsection Examples
3034 Make music with both quiet and loud passages suitable for listening to in a
3037 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3040 Another example for audio with whisper and explosion parts:
3042 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3046 A noise gate for when the noise is at a lower level than the signal:
3048 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3052 Here is another noise gate, this time for when the noise is at a higher level
3053 than the signal (making it, in some ways, similar to squelch):
3055 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3059 2:1 compression starting at -6dB:
3061 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3065 2:1 compression starting at -9dB:
3067 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3071 2:1 compression starting at -12dB:
3073 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3077 2:1 compression starting at -18dB:
3079 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3083 3:1 compression starting at -15dB:
3085 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3091 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3097 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
3101 Hard limiter at -6dB:
3103 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3107 Hard limiter at -12dB:
3109 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3113 Hard noise gate at -35 dB:
3115 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3121 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3125 @section compensationdelay
3127 Compensation Delay Line is a metric based delay to compensate differing
3128 positions of microphones or speakers.
3130 For example, you have recorded guitar with two microphones placed in
3131 different locations. Because the front of sound wave has fixed speed in
3132 normal conditions, the phasing of microphones can vary and depends on
3133 their location and interposition. The best sound mix can be achieved when
3134 these microphones are in phase (synchronized). Note that a distance of
3135 ~30 cm between microphones makes one microphone capture the signal in
3136 antiphase to the other microphone. That makes the final mix sound moody.
3137 This filter helps to solve phasing problems by adding different delays
3138 to each microphone track and make them synchronized.
3140 The best result can be reached when you take one track as base and
3141 synchronize other tracks one by one with it.
3142 Remember that synchronization/delay tolerance depends on sample rate, too.
3143 Higher sample rates will give more tolerance.
3145 The filter accepts the following parameters:
3149 Set millimeters distance. This is compensation distance for fine tuning.
3153 Set cm distance. This is compensation distance for tightening distance setup.
3157 Set meters distance. This is compensation distance for hard distance setup.
3161 Set dry amount. Amount of unprocessed (dry) signal.
3165 Set wet amount. Amount of processed (wet) signal.
3169 Set temperature in degrees Celsius. This is the temperature of the environment.
3174 Apply headphone crossfeed filter.
3176 Crossfeed is the process of blending the left and right channels of stereo
3178 It is mainly used to reduce extreme stereo separation of low frequencies.
3180 The intent is to produce more speaker like sound to the listener.
3182 The filter accepts the following options:
3186 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3187 This sets gain of low shelf filter for side part of stereo image.
3188 Default is -6dB. Max allowed is -30db when strength is set to 1.
3191 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3192 This sets cut off frequency of low shelf filter. Default is cut off near
3193 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3196 Set input gain. Default is 0.9.
3199 Set output gain. Default is 1.
3202 @section crystalizer
3203 Simple algorithm to expand audio dynamic range.
3205 The filter accepts the following options:
3209 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3210 (unchanged sound) to 10.0 (maximum effect).
3213 Enable clipping. By default is enabled.
3217 Apply a DC shift to the audio.
3219 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3220 in the recording chain) from the audio. The effect of a DC offset is reduced
3221 headroom and hence volume. The @ref{astats} filter can be used to determine if
3222 a signal has a DC offset.
3226 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3230 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3231 used to prevent clipping.
3236 Apply de-essing to the audio samples.
3240 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3244 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3248 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3252 Set the output mode.
3254 It accepts the following values:
3257 Pass input unchanged.
3260 Pass ess filtered out.
3265 Default value is @var{o}.
3271 Measure audio dynamic range.
3273 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3274 is found in transition material. And anything less that 8 have very poor dynamics
3275 and is very compressed.
3277 The filter accepts the following options:
3281 Set window length in seconds used to split audio into segments of equal length.
3282 Default is 3 seconds.
3286 Dynamic Audio Normalizer.
3288 This filter applies a certain amount of gain to the input audio in order
3289 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3290 contrast to more "simple" normalization algorithms, the Dynamic Audio
3291 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3292 This allows for applying extra gain to the "quiet" sections of the audio
3293 while avoiding distortions or clipping the "loud" sections. In other words:
3294 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3295 sections, in the sense that the volume of each section is brought to the
3296 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3297 this goal *without* applying "dynamic range compressing". It will retain 100%
3298 of the dynamic range *within* each section of the audio file.
3302 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3303 Default is 500 milliseconds.
3304 The Dynamic Audio Normalizer processes the input audio in small chunks,
3305 referred to as frames. This is required, because a peak magnitude has no
3306 meaning for just a single sample value. Instead, we need to determine the
3307 peak magnitude for a contiguous sequence of sample values. While a "standard"
3308 normalizer would simply use the peak magnitude of the complete file, the
3309 Dynamic Audio Normalizer determines the peak magnitude individually for each
3310 frame. The length of a frame is specified in milliseconds. By default, the
3311 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3312 been found to give good results with most files.
3313 Note that the exact frame length, in number of samples, will be determined
3314 automatically, based on the sampling rate of the individual input audio file.
3317 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3318 number. Default is 31.
3319 Probably the most important parameter of the Dynamic Audio Normalizer is the
3320 @code{window size} of the Gaussian smoothing filter. The filter's window size
3321 is specified in frames, centered around the current frame. For the sake of
3322 simplicity, this must be an odd number. Consequently, the default value of 31
3323 takes into account the current frame, as well as the 15 preceding frames and
3324 the 15 subsequent frames. Using a larger window results in a stronger
3325 smoothing effect and thus in less gain variation, i.e. slower gain
3326 adaptation. Conversely, using a smaller window results in a weaker smoothing
3327 effect and thus in more gain variation, i.e. faster gain adaptation.
3328 In other words, the more you increase this value, the more the Dynamic Audio
3329 Normalizer will behave like a "traditional" normalization filter. On the
3330 contrary, the more you decrease this value, the more the Dynamic Audio
3331 Normalizer will behave like a dynamic range compressor.
3334 Set the target peak value. This specifies the highest permissible magnitude
3335 level for the normalized audio input. This filter will try to approach the
3336 target peak magnitude as closely as possible, but at the same time it also
3337 makes sure that the normalized signal will never exceed the peak magnitude.
3338 A frame's maximum local gain factor is imposed directly by the target peak
3339 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3340 It is not recommended to go above this value.
3343 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3344 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3345 factor for each input frame, i.e. the maximum gain factor that does not
3346 result in clipping or distortion. The maximum gain factor is determined by
3347 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3348 additionally bounds the frame's maximum gain factor by a predetermined
3349 (global) maximum gain factor. This is done in order to avoid excessive gain
3350 factors in "silent" or almost silent frames. By default, the maximum gain
3351 factor is 10.0, For most inputs the default value should be sufficient and
3352 it usually is not recommended to increase this value. Though, for input
3353 with an extremely low overall volume level, it may be necessary to allow even
3354 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3355 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3356 Instead, a "sigmoid" threshold function will be applied. This way, the
3357 gain factors will smoothly approach the threshold value, but never exceed that
3361 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3362 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3363 This means that the maximum local gain factor for each frame is defined
3364 (only) by the frame's highest magnitude sample. This way, the samples can
3365 be amplified as much as possible without exceeding the maximum signal
3366 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3367 Normalizer can also take into account the frame's root mean square,
3368 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3369 determine the power of a time-varying signal. It is therefore considered
3370 that the RMS is a better approximation of the "perceived loudness" than
3371 just looking at the signal's peak magnitude. Consequently, by adjusting all
3372 frames to a constant RMS value, a uniform "perceived loudness" can be
3373 established. If a target RMS value has been specified, a frame's local gain
3374 factor is defined as the factor that would result in exactly that RMS value.
3375 Note, however, that the maximum local gain factor is still restricted by the
3376 frame's highest magnitude sample, in order to prevent clipping.
3379 Enable channels coupling. By default is enabled.
3380 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3381 amount. This means the same gain factor will be applied to all channels, i.e.
3382 the maximum possible gain factor is determined by the "loudest" channel.
3383 However, in some recordings, it may happen that the volume of the different
3384 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3385 In this case, this option can be used to disable the channel coupling. This way,
3386 the gain factor will be determined independently for each channel, depending
3387 only on the individual channel's highest magnitude sample. This allows for
3388 harmonizing the volume of the different channels.
3391 Enable DC bias correction. By default is disabled.
3392 An audio signal (in the time domain) is a sequence of sample values.
3393 In the Dynamic Audio Normalizer these sample values are represented in the
3394 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3395 audio signal, or "waveform", should be centered around the zero point.
3396 That means if we calculate the mean value of all samples in a file, or in a
3397 single frame, then the result should be 0.0 or at least very close to that
3398 value. If, however, there is a significant deviation of the mean value from
3399 0.0, in either positive or negative direction, this is referred to as a
3400 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3401 Audio Normalizer provides optional DC bias correction.
3402 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3403 the mean value, or "DC correction" offset, of each input frame and subtract
3404 that value from all of the frame's sample values which ensures those samples
3405 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3406 boundaries, the DC correction offset values will be interpolated smoothly
3407 between neighbouring frames.
3409 @item altboundary, b
3410 Enable alternative boundary mode. By default is disabled.
3411 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3412 around each frame. This includes the preceding frames as well as the
3413 subsequent frames. However, for the "boundary" frames, located at the very
3414 beginning and at the very end of the audio file, not all neighbouring
3415 frames are available. In particular, for the first few frames in the audio
3416 file, the preceding frames are not known. And, similarly, for the last few
3417 frames in the audio file, the subsequent frames are not known. Thus, the
3418 question arises which gain factors should be assumed for the missing frames
3419 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3420 to deal with this situation. The default boundary mode assumes a gain factor
3421 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3422 "fade out" at the beginning and at the end of the input, respectively.
3425 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3426 By default, the Dynamic Audio Normalizer does not apply "traditional"
3427 compression. This means that signal peaks will not be pruned and thus the
3428 full dynamic range will be retained within each local neighbourhood. However,
3429 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3430 normalization algorithm with a more "traditional" compression.
3431 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3432 (thresholding) function. If (and only if) the compression feature is enabled,
3433 all input frames will be processed by a soft knee thresholding function prior
3434 to the actual normalization process. Put simply, the thresholding function is
3435 going to prune all samples whose magnitude exceeds a certain threshold value.
3436 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3437 value. Instead, the threshold value will be adjusted for each individual
3439 In general, smaller parameters result in stronger compression, and vice versa.
3440 Values below 3.0 are not recommended, because audible distortion may appear.
3445 Make audio easier to listen to on headphones.
3447 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3448 so that when listened to on headphones the stereo image is moved from
3449 inside your head (standard for headphones) to outside and in front of
3450 the listener (standard for speakers).
3456 Apply a two-pole peaking equalisation (EQ) filter. With this
3457 filter, the signal-level at and around a selected frequency can
3458 be increased or decreased, whilst (unlike bandpass and bandreject
3459 filters) that at all other frequencies is unchanged.
3461 In order to produce complex equalisation curves, this filter can
3462 be given several times, each with a different central frequency.
3464 The filter accepts the following options:
3468 Set the filter's central frequency in Hz.
3471 Set method to specify band-width of filter.
3486 Specify the band-width of a filter in width_type units.
3489 Set the required gain or attenuation in dB.
3490 Beware of clipping when using a positive gain.
3493 How much to use filtered signal in output. Default is 1.
3494 Range is between 0 and 1.
3497 Specify which channels to filter, by default all available are filtered.
3500 Normalize biquad coefficients, by default is disabled.
3501 Enabling it will normalize magnitude response at DC to 0dB.
3504 @subsection Examples
3507 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
3509 equalizer=f=1000:t=h:width=200:g=-10
3513 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
3515 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
3519 @subsection Commands
3521 This filter supports the following commands:
3524 Change equalizer frequency.
3525 Syntax for the command is : "@var{frequency}"
3528 Change equalizer width_type.
3529 Syntax for the command is : "@var{width_type}"
3532 Change equalizer width.
3533 Syntax for the command is : "@var{width}"
3536 Change equalizer gain.
3537 Syntax for the command is : "@var{gain}"
3540 Change equalizer mix.
3541 Syntax for the command is : "@var{mix}"
3544 @section extrastereo
3546 Linearly increases the difference between left and right channels which
3547 adds some sort of "live" effect to playback.
3549 The filter accepts the following options:
3553 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
3554 (average of both channels), with 1.0 sound will be unchanged, with
3555 -1.0 left and right channels will be swapped.
3558 Enable clipping. By default is enabled.
3561 @section firequalizer
3562 Apply FIR Equalization using arbitrary frequency response.
3564 The filter accepts the following option:
3568 Set gain curve equation (in dB). The expression can contain variables:
3571 the evaluated frequency
3575 channel number, set to 0 when multichannels evaluation is disabled
3577 channel id, see libavutil/channel_layout.h, set to the first channel id when
3578 multichannels evaluation is disabled
3582 channel_layout, see libavutil/channel_layout.h
3587 @item gain_interpolate(f)
3588 interpolate gain on frequency f based on gain_entry
3589 @item cubic_interpolate(f)
3590 same as gain_interpolate, but smoother
3592 This option is also available as command. Default is @code{gain_interpolate(f)}.
3595 Set gain entry for gain_interpolate function. The expression can
3599 store gain entry at frequency f with value g
3601 This option is also available as command.
3604 Set filter delay in seconds. Higher value means more accurate.
3605 Default is @code{0.01}.
3608 Set filter accuracy in Hz. Lower value means more accurate.
3609 Default is @code{5}.
3612 Set window function. Acceptable values are:
3615 rectangular window, useful when gain curve is already smooth
3617 hann window (default)
3623 3-terms continuous 1st derivative nuttall window
3625 minimum 3-terms discontinuous nuttall window
3627 4-terms continuous 1st derivative nuttall window
3629 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
3631 blackman-harris window
3637 If enabled, use fixed number of audio samples. This improves speed when
3638 filtering with large delay. Default is disabled.
3641 Enable multichannels evaluation on gain. Default is disabled.
3644 Enable zero phase mode by subtracting timestamp to compensate delay.
3645 Default is disabled.
3648 Set scale used by gain. Acceptable values are:
3651 linear frequency, linear gain
3653 linear frequency, logarithmic (in dB) gain (default)
3655 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
3657 logarithmic frequency, logarithmic gain
3661 Set file for dumping, suitable for gnuplot.
3664 Set scale for dumpfile. Acceptable values are same with scale option.
3668 Enable 2-channel convolution using complex FFT. This improves speed significantly.
3669 Default is disabled.
3672 Enable minimum phase impulse response. Default is disabled.
3675 @subsection Examples
3680 firequalizer=gain='if(lt(f,1000), 0, -INF)'
3683 lowpass at 1000 Hz with gain_entry:
3685 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
3688 custom equalization:
3690 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
3693 higher delay with zero phase to compensate delay:
3695 firequalizer=delay=0.1:fixed=on:zero_phase=on
3698 lowpass on left channel, highpass on right channel:
3700 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
3701 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
3706 Apply a flanging effect to the audio.
3708 The filter accepts the following options:
3712 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
3715 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
3718 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
3722 Set percentage of delayed signal mixed with original. Range from 0 to 100.
3723 Default value is 71.
3726 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
3729 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
3730 Default value is @var{sinusoidal}.
3733 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
3734 Default value is 25.
3737 Set delay-line interpolation, @var{linear} or @var{quadratic}.
3738 Default is @var{linear}.
3742 Apply Haas effect to audio.
3744 Note that this makes most sense to apply on mono signals.
3745 With this filter applied to mono signals it give some directionality and
3746 stretches its stereo image.
3748 The filter accepts the following options:
3752 Set input level. By default is @var{1}, or 0dB
3755 Set output level. By default is @var{1}, or 0dB.
3758 Set gain applied to side part of signal. By default is @var{1}.
3761 Set kind of middle source. Can be one of the following:
3771 Pick middle part signal of stereo image.
3774 Pick side part signal of stereo image.
3778 Change middle phase. By default is disabled.
3781 Set left channel delay. By default is @var{2.05} milliseconds.
3784 Set left channel balance. By default is @var{-1}.
3787 Set left channel gain. By default is @var{1}.
3790 Change left phase. By default is disabled.
3793 Set right channel delay. By defaults is @var{2.12} milliseconds.
3796 Set right channel balance. By default is @var{1}.
3799 Set right channel gain. By default is @var{1}.
3802 Change right phase. By default is enabled.
3807 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
3808 embedded HDCD codes is expanded into a 20-bit PCM stream.
3810 The filter supports the Peak Extend and Low-level Gain Adjustment features
3811 of HDCD, and detects the Transient Filter flag.
3814 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
3817 When using the filter with wav, note the default encoding for wav is 16-bit,
3818 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
3819 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
3821 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
3822 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
3825 The filter accepts the following options:
3828 @item disable_autoconvert
3829 Disable any automatic format conversion or resampling in the filter graph.
3831 @item process_stereo
3832 Process the stereo channels together. If target_gain does not match between
3833 channels, consider it invalid and use the last valid target_gain.
3836 Set the code detect timer period in ms.
3839 Always extend peaks above -3dBFS even if PE isn't signaled.
3842 Replace audio with a solid tone and adjust the amplitude to signal some
3843 specific aspect of the decoding process. The output file can be loaded in
3844 an audio editor alongside the original to aid analysis.
3846 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
3853 Gain adjustment level at each sample
3855 Samples where peak extend occurs
3857 Samples where the code detect timer is active
3859 Samples where the target gain does not match between channels
3865 Apply head-related transfer functions (HRTFs) to create virtual
3866 loudspeakers around the user for binaural listening via headphones.
3867 The HRIRs are provided via additional streams, for each channel
3868 one stereo input stream is needed.
3870 The filter accepts the following options:
3874 Set mapping of input streams for convolution.
3875 The argument is a '|'-separated list of channel names in order as they
3876 are given as additional stream inputs for filter.
3877 This also specify number of input streams. Number of input streams
3878 must be not less than number of channels in first stream plus one.
3881 Set gain applied to audio. Value is in dB. Default is 0.
3884 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
3885 processing audio in time domain which is slow.
3886 @var{freq} is processing audio in frequency domain which is fast.
3887 Default is @var{freq}.
3890 Set custom gain for LFE channels. Value is in dB. Default is 0.
3893 Set size of frame in number of samples which will be processed at once.
3894 Default value is @var{1024}. Allowed range is from 1024 to 96000.
3897 Set format of hrir stream.
3898 Default value is @var{stereo}. Alternative value is @var{multich}.
3899 If value is set to @var{stereo}, number of additional streams should
3900 be greater or equal to number of input channels in first input stream.
3901 Also each additional stream should have stereo number of channels.
3902 If value is set to @var{multich}, number of additional streams should
3903 be exactly one. Also number of input channels of additional stream
3904 should be equal or greater than twice number of channels of first input
3908 @subsection Examples
3912 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3913 each amovie filter use stereo file with IR coefficients as input.
3914 The files give coefficients for each position of virtual loudspeaker:
3917 -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"
3922 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
3923 but now in @var{multich} @var{hrir} format.
3925 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"
3932 Apply a high-pass filter with 3dB point frequency.
3933 The filter can be either single-pole, or double-pole (the default).
3934 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
3936 The filter accepts the following options:
3940 Set frequency in Hz. Default is 3000.
3943 Set number of poles. Default is 2.
3946 Set method to specify band-width of filter.
3961 Specify the band-width of a filter in width_type units.
3962 Applies only to double-pole filter.
3963 The default is 0.707q and gives a Butterworth response.
3966 How much to use filtered signal in output. Default is 1.
3967 Range is between 0 and 1.
3970 Specify which channels to filter, by default all available are filtered.
3973 Normalize biquad coefficients, by default is disabled.
3974 Enabling it will normalize magnitude response at DC to 0dB.
3977 @subsection Commands
3979 This filter supports the following commands:
3982 Change highpass frequency.
3983 Syntax for the command is : "@var{frequency}"
3986 Change highpass width_type.
3987 Syntax for the command is : "@var{width_type}"
3990 Change highpass width.
3991 Syntax for the command is : "@var{width}"
3994 Change highpass mix.
3995 Syntax for the command is : "@var{mix}"
4000 Join multiple input streams into one multi-channel stream.
4002 It accepts the following parameters:
4006 The number of input streams. It defaults to 2.
4008 @item channel_layout
4009 The desired output channel layout. It defaults to stereo.
4012 Map channels from inputs to output. The argument is a '|'-separated list of
4013 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4014 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4015 can be either the name of the input channel (e.g. FL for front left) or its
4016 index in the specified input stream. @var{out_channel} is the name of the output
4020 The filter will attempt to guess the mappings when they are not specified
4021 explicitly. It does so by first trying to find an unused matching input channel
4022 and if that fails it picks the first unused input channel.
4024 Join 3 inputs (with properly set channel layouts):
4026 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4029 Build a 5.1 output from 6 single-channel streams:
4031 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4032 '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'
4038 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4040 To enable compilation of this filter you need to configure FFmpeg with
4041 @code{--enable-ladspa}.
4045 Specifies the name of LADSPA plugin library to load. If the environment
4046 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4047 each one of the directories specified by the colon separated list in
4048 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4049 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4050 @file{/usr/lib/ladspa/}.
4053 Specifies the plugin within the library. Some libraries contain only
4054 one plugin, but others contain many of them. If this is not set filter
4055 will list all available plugins within the specified library.
4058 Set the '|' separated list of controls which are zero or more floating point
4059 values that determine the behavior of the loaded plugin (for example delay,
4061 Controls need to be defined using the following syntax:
4062 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4063 @var{valuei} is the value set on the @var{i}-th control.
4064 Alternatively they can be also defined using the following syntax:
4065 @var{value0}|@var{value1}|@var{value2}|..., where
4066 @var{valuei} is the value set on the @var{i}-th control.
4067 If @option{controls} is set to @code{help}, all available controls and
4068 their valid ranges are printed.
4070 @item sample_rate, s
4071 Specify the sample rate, default to 44100. Only used if plugin have
4075 Set the number of samples per channel per each output frame, default
4076 is 1024. Only used if plugin have zero inputs.
4079 Set the minimum duration of the sourced audio. See
4080 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4081 for the accepted syntax.
4082 Note that the resulting duration may be greater than the specified duration,
4083 as the generated audio is always cut at the end of a complete frame.
4084 If not specified, or the expressed duration is negative, the audio is
4085 supposed to be generated forever.
4086 Only used if plugin have zero inputs.
4090 @subsection Examples
4094 List all available plugins within amp (LADSPA example plugin) library:
4100 List all available controls and their valid ranges for @code{vcf_notch}
4101 plugin from @code{VCF} library:
4103 ladspa=f=vcf:p=vcf_notch:c=help
4107 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4110 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4114 Add reverberation to the audio using TAP-plugins
4115 (Tom's Audio Processing plugins):
4117 ladspa=file=tap_reverb:tap_reverb
4121 Generate white noise, with 0.2 amplitude:
4123 ladspa=file=cmt:noise_source_white:c=c0=.2
4127 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4128 @code{C* Audio Plugin Suite} (CAPS) library:
4130 ladspa=file=caps:Click:c=c1=20'
4134 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4136 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4140 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4141 @code{SWH Plugins} collection:
4143 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4147 Attenuate low frequencies using Multiband EQ from Steve Harris
4148 @code{SWH Plugins} collection:
4150 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4154 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4157 ladspa=caps:Narrower
4161 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4163 ladspa=caps:White:.2
4167 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4169 ladspa=caps:Fractal:c=c1=1
4173 Dynamic volume normalization using @code{VLevel} plugin:
4175 ladspa=vlevel-ladspa:vlevel_mono
4179 @subsection Commands
4181 This filter supports the following commands:
4184 Modify the @var{N}-th control value.
4186 If the specified value is not valid, it is ignored and prior one is kept.
4191 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4192 Support for both single pass (livestreams, files) and double pass (files) modes.
4193 This algorithm can target IL, LRA, and maximum true peak. To accurately detect true peaks,
4194 the audio stream will be upsampled to 192 kHz unless the normalization mode is linear.
4195 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4197 The filter accepts the following options:
4201 Set integrated loudness target.
4202 Range is -70.0 - -5.0. Default value is -24.0.
4205 Set loudness range target.
4206 Range is 1.0 - 20.0. Default value is 7.0.
4209 Set maximum true peak.
4210 Range is -9.0 - +0.0. Default value is -2.0.
4212 @item measured_I, measured_i
4213 Measured IL of input file.
4214 Range is -99.0 - +0.0.
4216 @item measured_LRA, measured_lra
4217 Measured LRA of input file.
4218 Range is 0.0 - 99.0.
4220 @item measured_TP, measured_tp
4221 Measured true peak of input file.
4222 Range is -99.0 - +99.0.
4224 @item measured_thresh
4225 Measured threshold of input file.
4226 Range is -99.0 - +0.0.
4229 Set offset gain. Gain is applied before the true-peak limiter.
4230 Range is -99.0 - +99.0. Default is +0.0.
4233 Normalize linearly if possible.
4234 measured_I, measured_LRA, measured_TP, and measured_thresh must also
4235 to be specified in order to use this mode.
4236 Options are true or false. Default is true.
4239 Treat mono input files as "dual-mono". If a mono file is intended for playback
4240 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4241 If set to @code{true}, this option will compensate for this effect.
4242 Multi-channel input files are not affected by this option.
4243 Options are true or false. Default is false.
4246 Set print format for stats. Options are summary, json, or none.
4247 Default value is none.
4252 Apply a low-pass filter with 3dB point frequency.
4253 The filter can be either single-pole or double-pole (the default).
4254 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4256 The filter accepts the following options:
4260 Set frequency in Hz. Default is 500.
4263 Set number of poles. Default is 2.
4266 Set method to specify band-width of filter.
4281 Specify the band-width of a filter in width_type units.
4282 Applies only to double-pole filter.
4283 The default is 0.707q and gives a Butterworth response.
4286 How much to use filtered signal in output. Default is 1.
4287 Range is between 0 and 1.
4290 Specify which channels to filter, by default all available are filtered.
4293 Normalize biquad coefficients, by default is disabled.
4294 Enabling it will normalize magnitude response at DC to 0dB.
4297 @subsection Examples
4300 Lowpass only LFE channel, it LFE is not present it does nothing:
4306 @subsection Commands
4308 This filter supports the following commands:
4311 Change lowpass frequency.
4312 Syntax for the command is : "@var{frequency}"
4315 Change lowpass width_type.
4316 Syntax for the command is : "@var{width_type}"
4319 Change lowpass width.
4320 Syntax for the command is : "@var{width}"
4324 Syntax for the command is : "@var{mix}"
4329 Load a LV2 (LADSPA Version 2) plugin.
4331 To enable compilation of this filter you need to configure FFmpeg with
4332 @code{--enable-lv2}.
4336 Specifies the plugin URI. You may need to escape ':'.
4339 Set the '|' separated list of controls which are zero or more floating point
4340 values that determine the behavior of the loaded plugin (for example delay,
4342 If @option{controls} is set to @code{help}, all available controls and
4343 their valid ranges are printed.
4345 @item sample_rate, s
4346 Specify the sample rate, default to 44100. Only used if plugin have
4350 Set the number of samples per channel per each output frame, default
4351 is 1024. Only used if plugin have zero inputs.
4354 Set the minimum duration of the sourced audio. See
4355 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4356 for the accepted syntax.
4357 Note that the resulting duration may be greater than the specified duration,
4358 as the generated audio is always cut at the end of a complete frame.
4359 If not specified, or the expressed duration is negative, the audio is
4360 supposed to be generated forever.
4361 Only used if plugin have zero inputs.
4364 @subsection Examples
4368 Apply bass enhancer plugin from Calf:
4370 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4374 Apply vinyl plugin from Calf:
4376 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4380 Apply bit crusher plugin from ArtyFX:
4382 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4387 Multiband Compress or expand the audio's dynamic range.
4389 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4390 This is akin to the crossover of a loudspeaker, and results in flat frequency
4391 response when absent compander action.
4393 It accepts the following parameters:
4397 This option syntax is:
4398 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4399 For explanation of each item refer to compand filter documentation.
4405 Mix channels with specific gain levels. The filter accepts the output
4406 channel layout followed by a set of channels definitions.
4408 This filter is also designed to efficiently remap the channels of an audio
4411 The filter accepts parameters of the form:
4412 "@var{l}|@var{outdef}|@var{outdef}|..."
4416 output channel layout or number of channels
4419 output channel specification, of the form:
4420 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
4423 output channel to define, either a channel name (FL, FR, etc.) or a channel
4424 number (c0, c1, etc.)
4427 multiplicative coefficient for the channel, 1 leaving the volume unchanged
4430 input channel to use, see out_name for details; it is not possible to mix
4431 named and numbered input channels
4434 If the `=' in a channel specification is replaced by `<', then the gains for
4435 that specification will be renormalized so that the total is 1, thus
4436 avoiding clipping noise.
4438 @subsection Mixing examples
4440 For example, if you want to down-mix from stereo to mono, but with a bigger
4441 factor for the left channel:
4443 pan=1c|c0=0.9*c0+0.1*c1
4446 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
4447 7-channels surround:
4449 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
4452 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
4453 that should be preferred (see "-ac" option) unless you have very specific
4456 @subsection Remapping examples
4458 The channel remapping will be effective if, and only if:
4461 @item gain coefficients are zeroes or ones,
4462 @item only one input per channel output,
4465 If all these conditions are satisfied, the filter will notify the user ("Pure
4466 channel mapping detected"), and use an optimized and lossless method to do the
4469 For example, if you have a 5.1 source and want a stereo audio stream by
4470 dropping the extra channels:
4472 pan="stereo| c0=FL | c1=FR"
4475 Given the same source, you can also switch front left and front right channels
4476 and keep the input channel layout:
4478 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
4481 If the input is a stereo audio stream, you can mute the front left channel (and
4482 still keep the stereo channel layout) with:
4487 Still with a stereo audio stream input, you can copy the right channel in both
4488 front left and right:
4490 pan="stereo| c0=FR | c1=FR"
4495 ReplayGain scanner filter. This filter takes an audio stream as an input and
4496 outputs it unchanged.
4497 At end of filtering it displays @code{track_gain} and @code{track_peak}.
4501 Convert the audio sample format, sample rate and channel layout. It is
4502 not meant to be used directly.
4505 Apply time-stretching and pitch-shifting with librubberband.
4507 To enable compilation of this filter, you need to configure FFmpeg with
4508 @code{--enable-librubberband}.
4510 The filter accepts the following options:
4514 Set tempo scale factor.
4517 Set pitch scale factor.
4520 Set transients detector.
4521 Possible values are:
4530 Possible values are:
4539 Possible values are:
4546 Set processing window size.
4547 Possible values are:
4556 Possible values are:
4563 Enable formant preservation when shift pitching.
4564 Possible values are:
4572 Possible values are:
4581 Possible values are:
4588 @subsection Commands
4590 This filter supports the following commands:
4593 Change filter tempo scale factor.
4594 Syntax for the command is : "@var{tempo}"
4597 Change filter pitch scale factor.
4598 Syntax for the command is : "@var{pitch}"
4601 @section sidechaincompress
4603 This filter acts like normal compressor but has the ability to compress
4604 detected signal using second input signal.
4605 It needs two input streams and returns one output stream.
4606 First input stream will be processed depending on second stream signal.
4607 The filtered signal then can be filtered with other filters in later stages of
4608 processing. See @ref{pan} and @ref{amerge} filter.
4610 The filter accepts the following options:
4614 Set input gain. Default is 1. Range is between 0.015625 and 64.
4617 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
4618 Default is @code{downward}.
4621 If a signal of second stream raises above this level it will affect the gain
4622 reduction of first stream.
4623 By default is 0.125. Range is between 0.00097563 and 1.
4626 Set a ratio about which the signal is reduced. 1:2 means that if the level
4627 raised 4dB above the threshold, it will be only 2dB above after the reduction.
4628 Default is 2. Range is between 1 and 20.
4631 Amount of milliseconds the signal has to rise above the threshold before gain
4632 reduction starts. Default is 20. Range is between 0.01 and 2000.
4635 Amount of milliseconds the signal has to fall below the threshold before
4636 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
4639 Set the amount by how much signal will be amplified after processing.
4640 Default is 1. Range is from 1 to 64.
4643 Curve the sharp knee around the threshold to enter gain reduction more softly.
4644 Default is 2.82843. Range is between 1 and 8.
4647 Choose if the @code{average} level between all channels of side-chain stream
4648 or the louder(@code{maximum}) channel of side-chain stream affects the
4649 reduction. Default is @code{average}.
4652 Should the exact signal be taken in case of @code{peak} or an RMS one in case
4653 of @code{rms}. Default is @code{rms} which is mainly smoother.
4656 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
4659 How much to use compressed signal in output. Default is 1.
4660 Range is between 0 and 1.
4663 @subsection Examples
4667 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
4668 depending on the signal of 2nd input and later compressed signal to be
4669 merged with 2nd input:
4671 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
4675 @section sidechaingate
4677 A sidechain gate acts like a normal (wideband) gate but has the ability to
4678 filter the detected signal before sending it to the gain reduction stage.
4679 Normally a gate uses the full range signal to detect a level above the
4681 For example: If you cut all lower frequencies from your sidechain signal
4682 the gate will decrease the volume of your track only if not enough highs
4683 appear. With this technique you are able to reduce the resonation of a
4684 natural drum or remove "rumbling" of muted strokes from a heavily distorted
4686 It needs two input streams and returns one output stream.
4687 First input stream will be processed depending on second stream signal.
4689 The filter accepts the following options:
4693 Set input level before filtering.
4694 Default is 1. Allowed range is from 0.015625 to 64.
4697 Set the mode of operation. Can be @code{upward} or @code{downward}.
4698 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
4699 will be amplified, expanding dynamic range in upward direction.
4700 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
4703 Set the level of gain reduction when the signal is below the threshold.
4704 Default is 0.06125. Allowed range is from 0 to 1.
4705 Setting this to 0 disables reduction and then filter behaves like expander.
4708 If a signal rises above this level the gain reduction is released.
4709 Default is 0.125. Allowed range is from 0 to 1.
4712 Set a ratio about which the signal is reduced.
4713 Default is 2. Allowed range is from 1 to 9000.
4716 Amount of milliseconds the signal has to rise above the threshold before gain
4718 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
4721 Amount of milliseconds the signal has to fall below the threshold before the
4722 reduction is increased again. Default is 250 milliseconds.
4723 Allowed range is from 0.01 to 9000.
4726 Set amount of amplification of signal after processing.
4727 Default is 1. Allowed range is from 1 to 64.
4730 Curve the sharp knee around the threshold to enter gain reduction more softly.
4731 Default is 2.828427125. Allowed range is from 1 to 8.
4734 Choose if exact signal should be taken for detection or an RMS like one.
4735 Default is rms. Can be peak or rms.
4738 Choose if the average level between all channels or the louder channel affects
4740 Default is average. Can be average or maximum.
4743 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
4746 @section silencedetect
4748 Detect silence in an audio stream.
4750 This filter logs a message when it detects that the input audio volume is less
4751 or equal to a noise tolerance value for a duration greater or equal to the
4752 minimum detected noise duration.
4754 The printed times and duration are expressed in seconds. The
4755 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
4756 is set on the first frame whose timestamp equals or exceeds the detection
4757 duration and it contains the timestamp of the first frame of the silence.
4759 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
4760 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
4761 keys are set on the first frame after the silence. If @option{mono} is
4762 enabled, and each channel is evaluated separately, the @code{.X}
4763 suffixed keys are used, and @code{X} corresponds to the channel number.
4765 The filter accepts the following options:
4769 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
4770 specified value) or amplitude ratio. Default is -60dB, or 0.001.
4773 Set silence duration until notification (default is 2 seconds). See
4774 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4775 for the accepted syntax.
4778 Process each channel separately, instead of combined. By default is disabled.
4781 @subsection Examples
4785 Detect 5 seconds of silence with -50dB noise tolerance:
4787 silencedetect=n=-50dB:d=5
4791 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
4792 tolerance in @file{silence.mp3}:
4794 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
4798 @section silenceremove
4800 Remove silence from the beginning, middle or end of the audio.
4802 The filter accepts the following options:
4806 This value is used to indicate if audio should be trimmed at beginning of
4807 the audio. A value of zero indicates no silence should be trimmed from the
4808 beginning. When specifying a non-zero value, it trims audio up until it
4809 finds non-silence. Normally, when trimming silence from beginning of audio
4810 the @var{start_periods} will be @code{1} but it can be increased to higher
4811 values to trim all audio up to specific count of non-silence periods.
4812 Default value is @code{0}.
4814 @item start_duration
4815 Specify the amount of time that non-silence must be detected before it stops
4816 trimming audio. By increasing the duration, bursts of noises can be treated
4817 as silence and trimmed off. Default value is @code{0}.
4819 @item start_threshold
4820 This indicates what sample value should be treated as silence. For digital
4821 audio, a value of @code{0} may be fine but for audio recorded from analog,
4822 you may wish to increase the value to account for background noise.
4823 Can be specified in dB (in case "dB" is appended to the specified value)
4824 or amplitude ratio. Default value is @code{0}.
4827 Specify max duration of silence at beginning that will be kept after
4828 trimming. Default is 0, which is equal to trimming all samples detected
4832 Specify mode of detection of silence end in start of multi-channel audio.
4833 Can be @var{any} or @var{all}. Default is @var{any}.
4834 With @var{any}, any sample that is detected as non-silence will cause
4835 stopped trimming of silence.
4836 With @var{all}, only if all channels are detected as non-silence will cause
4837 stopped trimming of silence.
4840 Set the count for trimming silence from the end of audio.
4841 To remove silence from the middle of a file, specify a @var{stop_periods}
4842 that is negative. This value is then treated as a positive value and is
4843 used to indicate the effect should restart processing as specified by
4844 @var{start_periods}, making it suitable for removing periods of silence
4845 in the middle of the audio.
4846 Default value is @code{0}.
4849 Specify a duration of silence that must exist before audio is not copied any
4850 more. By specifying a higher duration, silence that is wanted can be left in
4852 Default value is @code{0}.
4854 @item stop_threshold
4855 This is the same as @option{start_threshold} but for trimming silence from
4857 Can be specified in dB (in case "dB" is appended to the specified value)
4858 or amplitude ratio. Default value is @code{0}.
4861 Specify max duration of silence at end that will be kept after
4862 trimming. Default is 0, which is equal to trimming all samples detected
4866 Specify mode of detection of silence start in end of multi-channel audio.
4867 Can be @var{any} or @var{all}. Default is @var{any}.
4868 With @var{any}, any sample that is detected as non-silence will cause
4869 stopped trimming of silence.
4870 With @var{all}, only if all channels are detected as non-silence will cause
4871 stopped trimming of silence.
4874 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
4875 and works better with digital silence which is exactly 0.
4876 Default value is @code{rms}.
4879 Set duration in number of seconds used to calculate size of window in number
4880 of samples for detecting silence.
4881 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
4884 @subsection Examples
4888 The following example shows how this filter can be used to start a recording
4889 that does not contain the delay at the start which usually occurs between
4890 pressing the record button and the start of the performance:
4892 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
4896 Trim all silence encountered from beginning to end where there is more than 1
4897 second of silence in audio:
4899 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
4903 Trim all digital silence samples, using peak detection, from beginning to end
4904 where there is more than 0 samples of digital silence in audio and digital
4905 silence is detected in all channels at same positions in stream:
4907 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
4913 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
4914 loudspeakers around the user for binaural listening via headphones (audio
4915 formats up to 9 channels supported).
4916 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
4917 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
4918 Austrian Academy of Sciences.
4920 To enable compilation of this filter you need to configure FFmpeg with
4921 @code{--enable-libmysofa}.
4923 The filter accepts the following options:
4927 Set the SOFA file used for rendering.
4930 Set gain applied to audio. Value is in dB. Default is 0.
4933 Set rotation of virtual loudspeakers in deg. Default is 0.
4936 Set elevation of virtual speakers in deg. Default is 0.
4939 Set distance in meters between loudspeakers and the listener with near-field
4940 HRTFs. Default is 1.
4943 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4944 processing audio in time domain which is slow.
4945 @var{freq} is processing audio in frequency domain which is fast.
4946 Default is @var{freq}.
4949 Set custom positions of virtual loudspeakers. Syntax for this option is:
4950 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
4951 Each virtual loudspeaker is described with short channel name following with
4952 azimuth and elevation in degrees.
4953 Each virtual loudspeaker description is separated by '|'.
4954 For example to override front left and front right channel positions use:
4955 'speakers=FL 45 15|FR 345 15'.
4956 Descriptions with unrecognised channel names are ignored.
4959 Set custom gain for LFE channels. Value is in dB. Default is 0.
4962 Set custom frame size in number of samples. Default is 1024.
4963 Allowed range is from 1024 to 96000. Only used if option @samp{type}
4964 is set to @var{freq}.
4967 Should all IRs be normalized upon importing SOFA file.
4968 By default is enabled.
4971 Should nearest IRs be interpolated with neighbor IRs if exact position
4972 does not match. By default is disabled.
4975 Minphase all IRs upon loading of SOFA file. By default is disabled.
4978 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
4981 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
4984 @subsection Examples
4988 Using ClubFritz6 sofa file:
4990 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
4994 Using ClubFritz12 sofa file and bigger radius with small rotation:
4996 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5000 Similar as above but with custom speaker positions for front left, front right, back left and back right
5001 and also with custom gain:
5003 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5007 @section stereotools
5009 This filter has some handy utilities to manage stereo signals, for converting
5010 M/S stereo recordings to L/R signal while having control over the parameters
5011 or spreading the stereo image of master track.
5013 The filter accepts the following options:
5017 Set input level before filtering for both channels. Defaults is 1.
5018 Allowed range is from 0.015625 to 64.
5021 Set output level after filtering for both channels. Defaults is 1.
5022 Allowed range is from 0.015625 to 64.
5025 Set input balance between both channels. Default is 0.
5026 Allowed range is from -1 to 1.
5029 Set output balance between both channels. Default is 0.
5030 Allowed range is from -1 to 1.
5033 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5034 clipping. Disabled by default.
5037 Mute the left channel. Disabled by default.
5040 Mute the right channel. Disabled by default.
5043 Change the phase of the left channel. Disabled by default.
5046 Change the phase of the right channel. Disabled by default.
5049 Set stereo mode. Available values are:
5053 Left/Right to Left/Right, this is default.
5056 Left/Right to Mid/Side.
5059 Mid/Side to Left/Right.
5062 Left/Right to Left/Left.
5065 Left/Right to Right/Right.
5068 Left/Right to Left + Right.
5071 Left/Right to Right/Left.
5074 Mid/Side to Left/Left.
5077 Mid/Side to Right/Right.
5081 Set level of side signal. Default is 1.
5082 Allowed range is from 0.015625 to 64.
5085 Set balance of side signal. Default is 0.
5086 Allowed range is from -1 to 1.
5089 Set level of the middle signal. Default is 1.
5090 Allowed range is from 0.015625 to 64.
5093 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5096 Set stereo base between mono and inversed channels. Default is 0.
5097 Allowed range is from -1 to 1.
5100 Set delay in milliseconds how much to delay left from right channel and
5101 vice versa. Default is 0. Allowed range is from -20 to 20.
5104 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5107 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5109 @item bmode_in, bmode_out
5110 Set balance mode for balance_in/balance_out option.
5112 Can be one of the following:
5116 Classic balance mode. Attenuate one channel at time.
5117 Gain is raised up to 1.
5120 Similar as classic mode above but gain is raised up to 2.
5123 Equal power distribution, from -6dB to +6dB range.
5127 @subsection Examples
5131 Apply karaoke like effect:
5133 stereotools=mlev=0.015625
5137 Convert M/S signal to L/R:
5139 "stereotools=mode=ms>lr"
5143 @section stereowiden
5145 This filter enhance the stereo effect by suppressing signal common to both
5146 channels and by delaying the signal of left into right and vice versa,
5147 thereby widening the stereo effect.
5149 The filter accepts the following options:
5153 Time in milliseconds of the delay of left signal into right and vice versa.
5154 Default is 20 milliseconds.
5157 Amount of gain in delayed signal into right and vice versa. Gives a delay
5158 effect of left signal in right output and vice versa which gives widening
5159 effect. Default is 0.3.
5162 Cross feed of left into right with inverted phase. This helps in suppressing
5163 the mono. If the value is 1 it will cancel all the signal common to both
5164 channels. Default is 0.3.
5167 Set level of input signal of original channel. Default is 0.8.
5170 @section superequalizer
5171 Apply 18 band equalizer.
5173 The filter accepts the following options:
5180 Set 131Hz band gain.
5182 Set 185Hz band gain.
5184 Set 262Hz band gain.
5186 Set 370Hz band gain.
5188 Set 523Hz band gain.
5190 Set 740Hz band gain.
5192 Set 1047Hz band gain.
5194 Set 1480Hz band gain.
5196 Set 2093Hz band gain.
5198 Set 2960Hz band gain.
5200 Set 4186Hz band gain.
5202 Set 5920Hz band gain.
5204 Set 8372Hz band gain.
5206 Set 11840Hz band gain.
5208 Set 16744Hz band gain.
5210 Set 20000Hz band gain.
5214 Apply audio surround upmix filter.
5216 This filter allows to produce multichannel output from audio stream.
5218 The filter accepts the following options:
5222 Set output channel layout. By default, this is @var{5.1}.
5224 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5225 for the required syntax.
5228 Set input channel layout. By default, this is @var{stereo}.
5230 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5231 for the required syntax.
5234 Set input volume level. By default, this is @var{1}.
5237 Set output volume level. By default, this is @var{1}.
5240 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5243 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5246 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5249 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5250 In @var{add} mode, LFE channel is created from input audio and added to output.
5251 In @var{sub} mode, LFE channel is created from input audio and added to output but
5252 also all non-LFE output channels are subtracted with output LFE channel.
5255 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5256 Default is @var{90}.
5259 Set front center input volume. By default, this is @var{1}.
5262 Set front center output volume. By default, this is @var{1}.
5265 Set front left input volume. By default, this is @var{1}.
5268 Set front left output volume. By default, this is @var{1}.
5271 Set front right input volume. By default, this is @var{1}.
5274 Set front right output volume. By default, this is @var{1}.
5277 Set side left input volume. By default, this is @var{1}.
5280 Set side left output volume. By default, this is @var{1}.
5283 Set side right input volume. By default, this is @var{1}.
5286 Set side right output volume. By default, this is @var{1}.
5289 Set back left input volume. By default, this is @var{1}.
5292 Set back left output volume. By default, this is @var{1}.
5295 Set back right input volume. By default, this is @var{1}.
5298 Set back right output volume. By default, this is @var{1}.
5301 Set back center input volume. By default, this is @var{1}.
5304 Set back center output volume. By default, this is @var{1}.
5307 Set LFE input volume. By default, this is @var{1}.
5310 Set LFE output volume. By default, this is @var{1}.
5313 Set spread usage of stereo image across X axis for all channels.
5316 Set spread usage of stereo image across Y axis for all channels.
5318 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5319 Set spread usage of stereo image across X axis for each channel.
5321 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5322 Set spread usage of stereo image across Y axis for each channel.
5325 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5328 Set window function.
5330 It accepts the following values:
5353 Default is @code{hann}.
5356 Set window overlap. If set to 1, the recommended overlap for selected
5357 window function will be picked. Default is @code{0.5}.
5360 @section treble, highshelf
5362 Boost or cut treble (upper) frequencies of the audio using a two-pole
5363 shelving filter with a response similar to that of a standard
5364 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
5366 The filter accepts the following options:
5370 Give the gain at whichever is the lower of ~22 kHz and the
5371 Nyquist frequency. Its useful range is about -20 (for a large cut)
5372 to +20 (for a large boost). Beware of clipping when using a positive gain.
5375 Set the filter's central frequency and so can be used
5376 to extend or reduce the frequency range to be boosted or cut.
5377 The default value is @code{3000} Hz.
5380 Set method to specify band-width of filter.
5395 Determine how steep is the filter's shelf transition.
5398 How much to use filtered signal in output. Default is 1.
5399 Range is between 0 and 1.
5402 Specify which channels to filter, by default all available are filtered.
5405 Normalize biquad coefficients, by default is disabled.
5406 Enabling it will normalize magnitude response at DC to 0dB.
5409 @subsection Commands
5411 This filter supports the following commands:
5414 Change treble frequency.
5415 Syntax for the command is : "@var{frequency}"
5418 Change treble width_type.
5419 Syntax for the command is : "@var{width_type}"
5422 Change treble width.
5423 Syntax for the command is : "@var{width}"
5427 Syntax for the command is : "@var{gain}"
5431 Syntax for the command is : "@var{mix}"
5436 Sinusoidal amplitude modulation.
5438 The filter accepts the following options:
5442 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
5443 (20 Hz or lower) will result in a tremolo effect.
5444 This filter may also be used as a ring modulator by specifying
5445 a modulation frequency higher than 20 Hz.
5446 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5449 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5450 Default value is 0.5.
5455 Sinusoidal phase modulation.
5457 The filter accepts the following options:
5461 Modulation frequency in Hertz.
5462 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
5465 Depth of modulation as a percentage. Range is 0.0 - 1.0.
5466 Default value is 0.5.
5471 Adjust the input audio volume.
5473 It accepts the following parameters:
5477 Set audio volume expression.
5479 Output values are clipped to the maximum value.
5481 The output audio volume is given by the relation:
5483 @var{output_volume} = @var{volume} * @var{input_volume}
5486 The default value for @var{volume} is "1.0".
5489 This parameter represents the mathematical precision.
5491 It determines which input sample formats will be allowed, which affects the
5492 precision of the volume scaling.
5496 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
5498 32-bit floating-point; this limits input sample format to FLT. (default)
5500 64-bit floating-point; this limits input sample format to DBL.
5504 Choose the behaviour on encountering ReplayGain side data in input frames.
5508 Remove ReplayGain side data, ignoring its contents (the default).
5511 Ignore ReplayGain side data, but leave it in the frame.
5514 Prefer the track gain, if present.
5517 Prefer the album gain, if present.
5520 @item replaygain_preamp
5521 Pre-amplification gain in dB to apply to the selected replaygain gain.
5523 Default value for @var{replaygain_preamp} is 0.0.
5526 Set when the volume expression is evaluated.
5528 It accepts the following values:
5531 only evaluate expression once during the filter initialization, or
5532 when the @samp{volume} command is sent
5535 evaluate expression for each incoming frame
5538 Default value is @samp{once}.
5541 The volume expression can contain the following parameters.
5545 frame number (starting at zero)
5548 @item nb_consumed_samples
5549 number of samples consumed by the filter
5551 number of samples in the current frame
5553 original frame position in the file
5559 PTS at start of stream
5561 time at start of stream
5567 last set volume value
5570 Note that when @option{eval} is set to @samp{once} only the
5571 @var{sample_rate} and @var{tb} variables are available, all other
5572 variables will evaluate to NAN.
5574 @subsection Commands
5576 This filter supports the following commands:
5579 Modify the volume expression.
5580 The command accepts the same syntax of the corresponding option.
5582 If the specified expression is not valid, it is kept at its current
5584 @item replaygain_noclip
5585 Prevent clipping by limiting the gain applied.
5587 Default value for @var{replaygain_noclip} is 1.
5591 @subsection Examples
5595 Halve the input audio volume:
5599 volume=volume=-6.0206dB
5602 In all the above example the named key for @option{volume} can be
5603 omitted, for example like in:
5609 Increase input audio power by 6 decibels using fixed-point precision:
5611 volume=volume=6dB:precision=fixed
5615 Fade volume after time 10 with an annihilation period of 5 seconds:
5617 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
5621 @section volumedetect
5623 Detect the volume of the input video.
5625 The filter has no parameters. The input is not modified. Statistics about
5626 the volume will be printed in the log when the input stream end is reached.
5628 In particular it will show the mean volume (root mean square), maximum
5629 volume (on a per-sample basis), and the beginning of a histogram of the
5630 registered volume values (from the maximum value to a cumulated 1/1000 of
5633 All volumes are in decibels relative to the maximum PCM value.
5635 @subsection Examples
5637 Here is an excerpt of the output:
5639 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
5640 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
5641 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
5642 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
5643 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
5644 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
5645 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
5646 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
5647 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
5653 The mean square energy is approximately -27 dB, or 10^-2.7.
5655 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
5657 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
5660 In other words, raising the volume by +4 dB does not cause any clipping,
5661 raising it by +5 dB causes clipping for 6 samples, etc.
5663 @c man end AUDIO FILTERS
5665 @chapter Audio Sources
5666 @c man begin AUDIO SOURCES
5668 Below is a description of the currently available audio sources.
5672 Buffer audio frames, and make them available to the filter chain.
5674 This source is mainly intended for a programmatic use, in particular
5675 through the interface defined in @file{libavfilter/asrc_abuffer.h}.
5677 It accepts the following parameters:
5681 The timebase which will be used for timestamps of submitted frames. It must be
5682 either a floating-point number or in @var{numerator}/@var{denominator} form.
5685 The sample rate of the incoming audio buffers.
5688 The sample format of the incoming audio buffers.
5689 Either a sample format name or its corresponding integer representation from
5690 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
5692 @item channel_layout
5693 The channel layout of the incoming audio buffers.
5694 Either a channel layout name from channel_layout_map in
5695 @file{libavutil/channel_layout.c} or its corresponding integer representation
5696 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
5699 The number of channels of the incoming audio buffers.
5700 If both @var{channels} and @var{channel_layout} are specified, then they
5705 @subsection Examples
5708 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
5711 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
5712 Since the sample format with name "s16p" corresponds to the number
5713 6 and the "stereo" channel layout corresponds to the value 0x3, this is
5716 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
5721 Generate an audio signal specified by an expression.
5723 This source accepts in input one or more expressions (one for each
5724 channel), which are evaluated and used to generate a corresponding
5727 This source accepts the following options:
5731 Set the '|'-separated expressions list for each separate channel. In case the
5732 @option{channel_layout} option is not specified, the selected channel layout
5733 depends on the number of provided expressions. Otherwise the last
5734 specified expression is applied to the remaining output channels.
5736 @item channel_layout, c
5737 Set the channel layout. The number of channels in the specified layout
5738 must be equal to the number of specified expressions.
5741 Set the minimum duration of the sourced audio. See
5742 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5743 for the accepted syntax.
5744 Note that the resulting duration may be greater than the specified
5745 duration, as the generated audio is always cut at the end of a
5748 If not specified, or the expressed duration is negative, the audio is
5749 supposed to be generated forever.
5752 Set the number of samples per channel per each output frame,
5755 @item sample_rate, s
5756 Specify the sample rate, default to 44100.
5759 Each expression in @var{exprs} can contain the following constants:
5763 number of the evaluated sample, starting from 0
5766 time of the evaluated sample expressed in seconds, starting from 0
5773 @subsection Examples
5783 Generate a sin signal with frequency of 440 Hz, set sample rate to
5786 aevalsrc="sin(440*2*PI*t):s=8000"
5790 Generate a two channels signal, specify the channel layout (Front
5791 Center + Back Center) explicitly:
5793 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
5797 Generate white noise:
5799 aevalsrc="-2+random(0)"
5803 Generate an amplitude modulated signal:
5805 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
5809 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
5811 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
5818 The null audio source, return unprocessed audio frames. It is mainly useful
5819 as a template and to be employed in analysis / debugging tools, or as
5820 the source for filters which ignore the input data (for example the sox
5823 This source accepts the following options:
5827 @item channel_layout, cl
5829 Specifies the channel layout, and can be either an integer or a string
5830 representing a channel layout. The default value of @var{channel_layout}
5833 Check the channel_layout_map definition in
5834 @file{libavutil/channel_layout.c} for the mapping between strings and
5835 channel layout values.
5837 @item sample_rate, r
5838 Specifies the sample rate, and defaults to 44100.
5841 Set the number of samples per requested frames.
5845 @subsection Examples
5849 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
5851 anullsrc=r=48000:cl=4
5855 Do the same operation with a more obvious syntax:
5857 anullsrc=r=48000:cl=mono
5861 All the parameters need to be explicitly defined.
5865 Synthesize a voice utterance using the libflite library.
5867 To enable compilation of this filter you need to configure FFmpeg with
5868 @code{--enable-libflite}.
5870 Note that versions of the flite library prior to 2.0 are not thread-safe.
5872 The filter accepts the following options:
5877 If set to 1, list the names of the available voices and exit
5878 immediately. Default value is 0.
5881 Set the maximum number of samples per frame. Default value is 512.
5884 Set the filename containing the text to speak.
5887 Set the text to speak.
5890 Set the voice to use for the speech synthesis. Default value is
5891 @code{kal}. See also the @var{list_voices} option.
5894 @subsection Examples
5898 Read from file @file{speech.txt}, and synthesize the text using the
5899 standard flite voice:
5901 flite=textfile=speech.txt
5905 Read the specified text selecting the @code{slt} voice:
5907 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
5911 Input text to ffmpeg:
5913 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
5917 Make @file{ffplay} speak the specified text, using @code{flite} and
5918 the @code{lavfi} device:
5920 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
5924 For more information about libflite, check:
5925 @url{http://www.festvox.org/flite/}
5929 Generate a noise audio signal.
5931 The filter accepts the following options:
5934 @item sample_rate, r
5935 Specify the sample rate. Default value is 48000 Hz.
5938 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
5942 Specify the duration of the generated audio stream. Not specifying this option
5943 results in noise with an infinite length.
5945 @item color, colour, c
5946 Specify the color of noise. Available noise colors are white, pink, brown,
5947 blue and violet. Default color is white.
5950 Specify a value used to seed the PRNG.
5953 Set the number of samples per each output frame, default is 1024.
5956 @subsection Examples
5961 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
5963 anoisesrc=d=60:c=pink:r=44100:a=0.5
5969 Generate odd-tap Hilbert transform FIR coefficients.
5971 The resulting stream can be used with @ref{afir} filter for phase-shifting
5972 the signal by 90 degrees.
5974 This is used in many matrix coding schemes and for analytic signal generation.
5975 The process is often written as a multiplication by i (or j), the imaginary unit.
5977 The filter accepts the following options:
5981 @item sample_rate, s
5982 Set sample rate, default is 44100.
5985 Set length of FIR filter, default is 22051.
5988 Set number of samples per each frame.
5991 Set window function to be used when generating FIR coefficients.
5996 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
5998 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6000 The filter accepts the following options:
6003 @item sample_rate, r
6004 Set sample rate, default is 44100.
6007 Set number of samples per each frame. Default is 1024.
6010 Set high-pass frequency. Default is 0.
6013 Set low-pass frequency. Default is 0.
6014 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6015 is higher than 0 then filter will create band-pass filter coefficients,
6016 otherwise band-reject filter coefficients.
6019 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6022 Set Kaiser window beta.
6025 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6028 Enable rounding, by default is disabled.
6031 Set number of taps for high-pass filter.
6034 Set number of taps for low-pass filter.
6039 Generate an audio signal made of a sine wave with amplitude 1/8.
6041 The audio signal is bit-exact.
6043 The filter accepts the following options:
6048 Set the carrier frequency. Default is 440 Hz.
6050 @item beep_factor, b
6051 Enable a periodic beep every second with frequency @var{beep_factor} times
6052 the carrier frequency. Default is 0, meaning the beep is disabled.
6054 @item sample_rate, r
6055 Specify the sample rate, default is 44100.
6058 Specify the duration of the generated audio stream.
6060 @item samples_per_frame
6061 Set the number of samples per output frame.
6063 The expression can contain the following constants:
6067 The (sequential) number of the output audio frame, starting from 0.
6070 The PTS (Presentation TimeStamp) of the output audio frame,
6071 expressed in @var{TB} units.
6074 The PTS of the output audio frame, expressed in seconds.
6077 The timebase of the output audio frames.
6080 Default is @code{1024}.
6083 @subsection Examples
6088 Generate a simple 440 Hz sine wave:
6094 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6098 sine=frequency=220:beep_factor=4:duration=5
6102 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6105 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6109 @c man end AUDIO SOURCES
6111 @chapter Audio Sinks
6112 @c man begin AUDIO SINKS
6114 Below is a description of the currently available audio sinks.
6116 @section abuffersink
6118 Buffer audio frames, and make them available to the end of filter chain.
6120 This sink is mainly intended for programmatic use, in particular
6121 through the interface defined in @file{libavfilter/buffersink.h}
6122 or the options system.
6124 It accepts a pointer to an AVABufferSinkContext structure, which
6125 defines the incoming buffers' formats, to be passed as the opaque
6126 parameter to @code{avfilter_init_filter} for initialization.
6129 Null audio sink; do absolutely nothing with the input audio. It is
6130 mainly useful as a template and for use in analysis / debugging
6133 @c man end AUDIO SINKS
6135 @chapter Video Filters
6136 @c man begin VIDEO FILTERS
6138 When you configure your FFmpeg build, you can disable any of the
6139 existing filters using @code{--disable-filters}.
6140 The configure output will show the video filters included in your
6143 Below is a description of the currently available video filters.
6147 Mark a region of interest in a video frame.
6149 The frame data is passed through unchanged, but metadata is attached
6150 to the frame indicating regions of interest which can affect the
6151 behaviour of later encoding. Multiple regions can be marked by
6152 applying the filter multiple times.
6156 Region distance in pixels from the left edge of the frame.
6158 Region distance in pixels from the top edge of the frame.
6160 Region width in pixels.
6162 Region height in pixels.
6164 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6165 and may contain the following variables:
6168 Width of the input frame.
6170 Height of the input frame.
6174 Quantisation offset to apply within the region.
6176 This must be a real value in the range -1 to +1. A value of zero
6177 indicates no quality change. A negative value asks for better quality
6178 (less quantisation), while a positive value asks for worse quality
6179 (greater quantisation).
6181 The range is calibrated so that the extreme values indicate the
6182 largest possible offset - if the rest of the frame is encoded with the
6183 worst possible quality, an offset of -1 indicates that this region
6184 should be encoded with the best possible quality anyway. Intermediate
6185 values are then interpolated in some codec-dependent way.
6187 For example, in 10-bit H.264 the quantisation parameter varies between
6188 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6189 this region should be encoded with a QP around one-tenth of the full
6190 range better than the rest of the frame. So, if most of the frame
6191 were to be encoded with a QP of around 30, this region would get a QP
6192 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6193 An extreme value of -1 would indicate that this region should be
6194 encoded with the best possible quality regardless of the treatment of
6195 the rest of the frame - that is, should be encoded at a QP of -12.
6197 If set to true, remove any existing regions of interest marked on the
6198 frame before adding the new one.
6201 @subsection Examples
6205 Mark the centre quarter of the frame as interesting.
6207 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6210 Mark the 100-pixel-wide region on the left edge of the frame as very
6211 uninteresting (to be encoded at much lower quality than the rest of
6214 addroi=0:0:100:ih:+1/5
6218 @section alphaextract
6220 Extract the alpha component from the input as a grayscale video. This
6221 is especially useful with the @var{alphamerge} filter.
6225 Add or replace the alpha component of the primary input with the
6226 grayscale value of a second input. This is intended for use with
6227 @var{alphaextract} to allow the transmission or storage of frame
6228 sequences that have alpha in a format that doesn't support an alpha
6231 For example, to reconstruct full frames from a normal YUV-encoded video
6232 and a separate video created with @var{alphaextract}, you might use:
6234 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6237 Since this filter is designed for reconstruction, it operates on frame
6238 sequences without considering timestamps, and terminates when either
6239 input reaches end of stream. This will cause problems if your encoding
6240 pipeline drops frames. If you're trying to apply an image as an
6241 overlay to a video stream, consider the @var{overlay} filter instead.
6245 Amplify differences between current pixel and pixels of adjacent frames in
6246 same pixel location.
6248 This filter accepts the following options:
6252 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6253 For example radius of 3 will instruct filter to calculate average of 7 frames.
6256 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6259 Set threshold for difference amplification. Any difference greater or equal to
6260 this value will not alter source pixel. Default is 10.
6261 Allowed range is from 0 to 65535.
6264 Set tolerance for difference amplification. Any difference lower to
6265 this value will not alter source pixel. Default is 0.
6266 Allowed range is from 0 to 65535.
6269 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6270 This option controls maximum possible value that will decrease source pixel value.
6273 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
6274 This option controls maximum possible value that will increase source pixel value.
6277 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
6280 @subsection Commands
6282 This filter supports the following @ref{commands} that corresponds to option of same name:
6294 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
6295 and libavformat to work. On the other hand, it is limited to ASS (Advanced
6296 Substation Alpha) subtitles files.
6298 This filter accepts the following option in addition to the common options from
6299 the @ref{subtitles} filter:
6303 Set the shaping engine
6305 Available values are:
6308 The default libass shaping engine, which is the best available.
6310 Fast, font-agnostic shaper that can do only substitutions
6312 Slower shaper using OpenType for substitutions and positioning
6315 The default is @code{auto}.
6319 Apply an Adaptive Temporal Averaging Denoiser to the video input.
6321 The filter accepts the following options:
6325 Set threshold A for 1st plane. Default is 0.02.
6326 Valid range is 0 to 0.3.
6329 Set threshold B for 1st plane. Default is 0.04.
6330 Valid range is 0 to 5.
6333 Set threshold A for 2nd plane. Default is 0.02.
6334 Valid range is 0 to 0.3.
6337 Set threshold B for 2nd plane. Default is 0.04.
6338 Valid range is 0 to 5.
6341 Set threshold A for 3rd plane. Default is 0.02.
6342 Valid range is 0 to 0.3.
6345 Set threshold B for 3rd plane. Default is 0.04.
6346 Valid range is 0 to 5.
6348 Threshold A is designed to react on abrupt changes in the input signal and
6349 threshold B is designed to react on continuous changes in the input signal.
6352 Set number of frames filter will use for averaging. Default is 9. Must be odd
6353 number in range [5, 129].
6356 Set what planes of frame filter will use for averaging. Default is all.
6359 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
6360 Alternatively can be set to @code{s} serial.
6362 Parallel can be faster then serial, while other way around is never true.
6363 Parallel will abort early on first change being greater then thresholds, while serial
6364 will continue processing other side of frames if they are equal or bellow thresholds.
6367 @subsection Commands
6368 This filter supports same @ref{commands} as options except option @code{s}.
6369 The command accepts the same syntax of the corresponding option.
6373 Apply average blur filter.
6375 The filter accepts the following options:
6379 Set horizontal radius size.
6382 Set which planes to filter. By default all planes are filtered.
6385 Set vertical radius size, if zero it will be same as @code{sizeX}.
6386 Default is @code{0}.
6389 @subsection Commands
6390 This filter supports same commands as options.
6391 The command accepts the same syntax of the corresponding option.
6393 If the specified expression is not valid, it is kept at its current
6398 Compute the bounding box for the non-black pixels in the input frame
6401 This filter computes the bounding box containing all the pixels with a
6402 luminance value greater than the minimum allowed value.
6403 The parameters describing the bounding box are printed on the filter
6406 The filter accepts the following option:
6410 Set the minimal luminance value. Default is @code{16}.
6414 Apply bilateral filter, spatial smoothing while preserving edges.
6416 The filter accepts the following options:
6419 Set sigma of gaussian function to calculate spatial weight.
6420 Allowed range is 0 to 10. Default is 0.1.
6423 Set sigma of gaussian function to calculate range weight.
6424 Allowed range is 0 to 1. Default is 0.1.
6427 Set planes to filter. Default is first only.
6430 @section bitplanenoise
6432 Show and measure bit plane noise.
6434 The filter accepts the following options:
6438 Set which plane to analyze. Default is @code{1}.
6441 Filter out noisy pixels from @code{bitplane} set above.
6442 Default is disabled.
6445 @section blackdetect
6447 Detect video intervals that are (almost) completely black. Can be
6448 useful to detect chapter transitions, commercials, or invalid
6449 recordings. Output lines contains the time for the start, end and
6450 duration of the detected black interval expressed in seconds.
6452 In order to display the output lines, you need to set the loglevel at
6453 least to the AV_LOG_INFO value.
6455 The filter accepts the following options:
6458 @item black_min_duration, d
6459 Set the minimum detected black duration expressed in seconds. It must
6460 be a non-negative floating point number.
6462 Default value is 2.0.
6464 @item picture_black_ratio_th, pic_th
6465 Set the threshold for considering a picture "black".
6466 Express the minimum value for the ratio:
6468 @var{nb_black_pixels} / @var{nb_pixels}
6471 for which a picture is considered black.
6472 Default value is 0.98.
6474 @item pixel_black_th, pix_th
6475 Set the threshold for considering a pixel "black".
6477 The threshold expresses the maximum pixel luminance value for which a
6478 pixel is considered "black". The provided value is scaled according to
6479 the following equation:
6481 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
6484 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
6485 the input video format, the range is [0-255] for YUV full-range
6486 formats and [16-235] for YUV non full-range formats.
6488 Default value is 0.10.
6491 The following example sets the maximum pixel threshold to the minimum
6492 value, and detects only black intervals of 2 or more seconds:
6494 blackdetect=d=2:pix_th=0.00
6499 Detect frames that are (almost) completely black. Can be useful to
6500 detect chapter transitions or commercials. Output lines consist of
6501 the frame number of the detected frame, the percentage of blackness,
6502 the position in the file if known or -1 and the timestamp in seconds.
6504 In order to display the output lines, you need to set the loglevel at
6505 least to the AV_LOG_INFO value.
6507 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
6508 The value represents the percentage of pixels in the picture that
6509 are below the threshold value.
6511 It accepts the following parameters:
6516 The percentage of the pixels that have to be below the threshold; it defaults to
6519 @item threshold, thresh
6520 The threshold below which a pixel value is considered black; it defaults to
6525 @section blend, tblend
6527 Blend two video frames into each other.
6529 The @code{blend} filter takes two input streams and outputs one
6530 stream, the first input is the "top" layer and second input is
6531 "bottom" layer. By default, the output terminates when the longest input terminates.
6533 The @code{tblend} (time blend) filter takes two consecutive frames
6534 from one single stream, and outputs the result obtained by blending
6535 the new frame on top of the old frame.
6537 A description of the accepted options follows.
6545 Set blend mode for specific pixel component or all pixel components in case
6546 of @var{all_mode}. Default value is @code{normal}.
6548 Available values for component modes are:
6590 Set blend opacity for specific pixel component or all pixel components in case
6591 of @var{all_opacity}. Only used in combination with pixel component blend modes.
6598 Set blend expression for specific pixel component or all pixel components in case
6599 of @var{all_expr}. Note that related mode options will be ignored if those are set.
6601 The expressions can use the following variables:
6605 The sequential number of the filtered frame, starting from @code{0}.
6609 the coordinates of the current sample
6613 the width and height of currently filtered plane
6617 Width and height scale for the plane being filtered. It is the
6618 ratio between the dimensions of the current plane to the luma plane,
6619 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
6620 the luma plane and @code{0.5,0.5} for the chroma planes.
6623 Time of the current frame, expressed in seconds.
6626 Value of pixel component at current location for first video frame (top layer).
6629 Value of pixel component at current location for second video frame (bottom layer).
6633 The @code{blend} filter also supports the @ref{framesync} options.
6635 @subsection Examples
6639 Apply transition from bottom layer to top layer in first 10 seconds:
6641 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
6645 Apply linear horizontal transition from top layer to bottom layer:
6647 blend=all_expr='A*(X/W)+B*(1-X/W)'
6651 Apply 1x1 checkerboard effect:
6653 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
6657 Apply uncover left effect:
6659 blend=all_expr='if(gte(N*SW+X,W),A,B)'
6663 Apply uncover down effect:
6665 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
6669 Apply uncover up-left effect:
6671 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
6675 Split diagonally video and shows top and bottom layer on each side:
6677 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
6681 Display differences between the current and the previous frame:
6683 tblend=all_mode=grainextract
6689 Denoise frames using Block-Matching 3D algorithm.
6691 The filter accepts the following options.
6695 Set denoising strength. Default value is 1.
6696 Allowed range is from 0 to 999.9.
6697 The denoising algorithm is very sensitive to sigma, so adjust it
6698 according to the source.
6701 Set local patch size. This sets dimensions in 2D.
6704 Set sliding step for processing blocks. Default value is 4.
6705 Allowed range is from 1 to 64.
6706 Smaller values allows processing more reference blocks and is slower.
6709 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
6710 When set to 1, no block matching is done. Larger values allows more blocks
6712 Allowed range is from 1 to 256.
6715 Set radius for search block matching. Default is 9.
6716 Allowed range is from 1 to INT32_MAX.
6719 Set step between two search locations for block matching. Default is 1.
6720 Allowed range is from 1 to 64. Smaller is slower.
6723 Set threshold of mean square error for block matching. Valid range is 0 to
6727 Set thresholding parameter for hard thresholding in 3D transformed domain.
6728 Larger values results in stronger hard-thresholding filtering in frequency
6732 Set filtering estimation mode. Can be @code{basic} or @code{final}.
6733 Default is @code{basic}.
6736 If enabled, filter will use 2nd stream for block matching.
6737 Default is disabled for @code{basic} value of @var{estim} option,
6738 and always enabled if value of @var{estim} is @code{final}.
6741 Set planes to filter. Default is all available except alpha.
6744 @subsection Examples
6748 Basic filtering with bm3d:
6750 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
6754 Same as above, but filtering only luma:
6756 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
6760 Same as above, but with both estimation modes:
6762 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
6766 Same as above, but prefilter with @ref{nlmeans} filter instead:
6768 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
6774 Apply a boxblur algorithm to the input video.
6776 It accepts the following parameters:
6780 @item luma_radius, lr
6781 @item luma_power, lp
6782 @item chroma_radius, cr
6783 @item chroma_power, cp
6784 @item alpha_radius, ar
6785 @item alpha_power, ap
6789 A description of the accepted options follows.
6792 @item luma_radius, lr
6793 @item chroma_radius, cr
6794 @item alpha_radius, ar
6795 Set an expression for the box radius in pixels used for blurring the
6796 corresponding input plane.
6798 The radius value must be a non-negative number, and must not be
6799 greater than the value of the expression @code{min(w,h)/2} for the
6800 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
6803 Default value for @option{luma_radius} is "2". If not specified,
6804 @option{chroma_radius} and @option{alpha_radius} default to the
6805 corresponding value set for @option{luma_radius}.
6807 The expressions can contain the following constants:
6811 The input width and height in pixels.
6815 The input chroma image width and height in pixels.
6819 The horizontal and vertical chroma subsample values. For example, for the
6820 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
6823 @item luma_power, lp
6824 @item chroma_power, cp
6825 @item alpha_power, ap
6826 Specify how many times the boxblur filter is applied to the
6827 corresponding plane.
6829 Default value for @option{luma_power} is 2. If not specified,
6830 @option{chroma_power} and @option{alpha_power} default to the
6831 corresponding value set for @option{luma_power}.
6833 A value of 0 will disable the effect.
6836 @subsection Examples
6840 Apply a boxblur filter with the luma, chroma, and alpha radii
6843 boxblur=luma_radius=2:luma_power=1
6848 Set the luma radius to 2, and alpha and chroma radius to 0:
6850 boxblur=2:1:cr=0:ar=0
6854 Set the luma and chroma radii to a fraction of the video dimension:
6856 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
6862 Deinterlace the input video ("bwdif" stands for "Bob Weaver
6863 Deinterlacing Filter").
6865 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
6866 interpolation algorithms.
6867 It accepts the following parameters:
6871 The interlacing mode to adopt. It accepts one of the following values:
6875 Output one frame for each frame.
6877 Output one frame for each field.
6880 The default value is @code{send_field}.
6883 The picture field parity assumed for the input interlaced video. It accepts one
6884 of the following values:
6888 Assume the top field is first.
6890 Assume the bottom field is first.
6892 Enable automatic detection of field parity.
6895 The default value is @code{auto}.
6896 If the interlacing is unknown or the decoder does not export this information,
6897 top field first will be assumed.
6900 Specify which frames to deinterlace. Accepts one of the following
6905 Deinterlace all frames.
6907 Only deinterlace frames marked as interlaced.
6910 The default value is @code{all}.
6914 Remove all color information for all colors except for certain one.
6916 The filter accepts the following options:
6920 The color which will not be replaced with neutral chroma.
6923 Similarity percentage with the above color.
6924 0.01 matches only the exact key color, while 1.0 matches everything.
6928 0.0 makes pixels either fully gray, or not gray at all.
6929 Higher values result in more preserved color.
6932 Signals that the color passed is already in YUV instead of RGB.
6934 Literal colors like "green" or "red" don't make sense with this enabled anymore.
6935 This can be used to pass exact YUV values as hexadecimal numbers.
6938 @subsection Commands
6939 This filter supports same @ref{commands} as options.
6940 The command accepts the same syntax of the corresponding option.
6942 If the specified expression is not valid, it is kept at its current
6946 YUV colorspace color/chroma keying.
6948 The filter accepts the following options:
6952 The color which will be replaced with transparency.
6955 Similarity percentage with the key color.
6957 0.01 matches only the exact key color, while 1.0 matches everything.
6962 0.0 makes pixels either fully transparent, or not transparent at all.
6964 Higher values result in semi-transparent pixels, with a higher transparency
6965 the more similar the pixels color is to the key color.
6968 Signals that the color passed is already in YUV instead of RGB.
6970 Literal colors like "green" or "red" don't make sense with this enabled anymore.
6971 This can be used to pass exact YUV values as hexadecimal numbers.
6974 @subsection Commands
6975 This filter supports same @ref{commands} as options.
6976 The command accepts the same syntax of the corresponding option.
6978 If the specified expression is not valid, it is kept at its current
6981 @subsection Examples
6985 Make every green pixel in the input image transparent:
6987 ffmpeg -i input.png -vf chromakey=green out.png
6991 Overlay a greenscreen-video on top of a static black background.
6993 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
6997 @section chromashift
6998 Shift chroma pixels horizontally and/or vertically.
7000 The filter accepts the following options:
7003 Set amount to shift chroma-blue horizontally.
7005 Set amount to shift chroma-blue vertically.
7007 Set amount to shift chroma-red horizontally.
7009 Set amount to shift chroma-red vertically.
7011 Set edge mode, can be @var{smear}, default, or @var{warp}.
7014 @subsection Commands
7016 This filter supports the all above options as @ref{commands}.
7020 Display CIE color diagram with pixels overlaid onto it.
7022 The filter accepts the following options:
7037 @item uhdtv, rec2020
7051 Set what gamuts to draw.
7053 See @code{system} option for available values.
7056 Set ciescope size, by default set to 512.
7059 Set intensity used to map input pixel values to CIE diagram.
7062 Set contrast used to draw tongue colors that are out of active color system gamut.
7065 Correct gamma displayed on scope, by default enabled.
7068 Show white point on CIE diagram, by default disabled.
7071 Set input gamma. Used only with XYZ input color space.
7076 Visualize information exported by some codecs.
7078 Some codecs can export information through frames using side-data or other
7079 means. For example, some MPEG based codecs export motion vectors through the
7080 @var{export_mvs} flag in the codec @option{flags2} option.
7082 The filter accepts the following option:
7086 Set motion vectors to visualize.
7088 Available flags for @var{mv} are:
7092 forward predicted MVs of P-frames
7094 forward predicted MVs of B-frames
7096 backward predicted MVs of B-frames
7100 Display quantization parameters using the chroma planes.
7103 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7105 Available flags for @var{mv_type} are:
7109 forward predicted MVs
7111 backward predicted MVs
7114 @item frame_type, ft
7115 Set frame type to visualize motion vectors of.
7117 Available flags for @var{frame_type} are:
7121 intra-coded frames (I-frames)
7123 predicted frames (P-frames)
7125 bi-directionally predicted frames (B-frames)
7129 @subsection Examples
7133 Visualize forward predicted MVs of all frames using @command{ffplay}:
7135 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7139 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7141 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7145 @section colorbalance
7146 Modify intensity of primary colors (red, green and blue) of input frames.
7148 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7149 regions for the red-cyan, green-magenta or blue-yellow balance.
7151 A positive adjustment value shifts the balance towards the primary color, a negative
7152 value towards the complementary color.
7154 The filter accepts the following options:
7160 Adjust red, green and blue shadows (darkest pixels).
7165 Adjust red, green and blue midtones (medium pixels).
7170 Adjust red, green and blue highlights (brightest pixels).
7172 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7175 Preserve lightness when changing color balance. Default is disabled.
7178 @subsection Examples
7182 Add red color cast to shadows:
7188 @subsection Commands
7190 This filter supports the all above options as @ref{commands}.
7192 @section colorchannelmixer
7194 Adjust video input frames by re-mixing color channels.
7196 This filter modifies a color channel by adding the values associated to
7197 the other channels of the same pixels. For example if the value to
7198 modify is red, the output value will be:
7200 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
7203 The filter accepts the following options:
7210 Adjust contribution of input red, green, blue and alpha channels for output red channel.
7211 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
7217 Adjust contribution of input red, green, blue and alpha channels for output green channel.
7218 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
7224 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
7225 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
7231 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
7232 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
7234 Allowed ranges for options are @code{[-2.0, 2.0]}.
7237 @subsection Examples
7241 Convert source to grayscale:
7243 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
7246 Simulate sepia tones:
7248 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
7252 @subsection Commands
7254 This filter supports the all above options as @ref{commands}.
7257 RGB colorspace color keying.
7259 The filter accepts the following options:
7263 The color which will be replaced with transparency.
7266 Similarity percentage with the key color.
7268 0.01 matches only the exact key color, while 1.0 matches everything.
7273 0.0 makes pixels either fully transparent, or not transparent at all.
7275 Higher values result in semi-transparent pixels, with a higher transparency
7276 the more similar the pixels color is to the key color.
7279 @subsection Examples
7283 Make every green pixel in the input image transparent:
7285 ffmpeg -i input.png -vf colorkey=green out.png
7289 Overlay a greenscreen-video on top of a static background image.
7291 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
7296 Remove all color information for all RGB colors except for certain one.
7298 The filter accepts the following options:
7302 The color which will not be replaced with neutral gray.
7305 Similarity percentage with the above color.
7306 0.01 matches only the exact key color, while 1.0 matches everything.
7309 Blend percentage. 0.0 makes pixels fully gray.
7310 Higher values result in more preserved color.
7313 @section colorlevels
7315 Adjust video input frames using levels.
7317 The filter accepts the following options:
7324 Adjust red, green, blue and alpha input black point.
7325 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7331 Adjust red, green, blue and alpha input white point.
7332 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
7334 Input levels are used to lighten highlights (bright tones), darken shadows
7335 (dark tones), change the balance of bright and dark tones.
7341 Adjust red, green, blue and alpha output black point.
7342 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
7348 Adjust red, green, blue and alpha output white point.
7349 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
7351 Output levels allows manual selection of a constrained output level range.
7354 @subsection Examples
7358 Make video output darker:
7360 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
7366 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
7370 Make video output lighter:
7372 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
7376 Increase brightness:
7378 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
7382 @section colormatrix
7384 Convert color matrix.
7386 The filter accepts the following options:
7391 Specify the source and destination color matrix. Both values must be
7394 The accepted values are:
7422 For example to convert from BT.601 to SMPTE-240M, use the command:
7424 colormatrix=bt601:smpte240m
7429 Convert colorspace, transfer characteristics or color primaries.
7430 Input video needs to have an even size.
7432 The filter accepts the following options:
7437 Specify all color properties at once.
7439 The accepted values are:
7469 Specify output colorspace.
7471 The accepted values are:
7480 BT.470BG or BT.601-6 625
7483 SMPTE-170M or BT.601-6 525
7492 BT.2020 with non-constant luminance
7498 Specify output transfer characteristics.
7500 The accepted values are:
7512 Constant gamma of 2.2
7515 Constant gamma of 2.8
7518 SMPTE-170M, BT.601-6 625 or BT.601-6 525
7536 BT.2020 for 10-bits content
7539 BT.2020 for 12-bits content
7545 Specify output color primaries.
7547 The accepted values are:
7556 BT.470BG or BT.601-6 625
7559 SMPTE-170M or BT.601-6 525
7583 Specify output color range.
7585 The accepted values are:
7588 TV (restricted) range
7591 MPEG (restricted) range
7602 Specify output color format.
7604 The accepted values are:
7607 YUV 4:2:0 planar 8-bits
7610 YUV 4:2:0 planar 10-bits
7613 YUV 4:2:0 planar 12-bits
7616 YUV 4:2:2 planar 8-bits
7619 YUV 4:2:2 planar 10-bits
7622 YUV 4:2:2 planar 12-bits
7625 YUV 4:4:4 planar 8-bits
7628 YUV 4:4:4 planar 10-bits
7631 YUV 4:4:4 planar 12-bits
7636 Do a fast conversion, which skips gamma/primary correction. This will take
7637 significantly less CPU, but will be mathematically incorrect. To get output
7638 compatible with that produced by the colormatrix filter, use fast=1.
7641 Specify dithering mode.
7643 The accepted values are:
7649 Floyd-Steinberg dithering
7653 Whitepoint adaptation mode.
7655 The accepted values are:
7658 Bradford whitepoint adaptation
7661 von Kries whitepoint adaptation
7664 identity whitepoint adaptation (i.e. no whitepoint adaptation)
7668 Override all input properties at once. Same accepted values as @ref{all}.
7671 Override input colorspace. Same accepted values as @ref{space}.
7674 Override input color primaries. Same accepted values as @ref{primaries}.
7677 Override input transfer characteristics. Same accepted values as @ref{trc}.
7680 Override input color range. Same accepted values as @ref{range}.
7684 The filter converts the transfer characteristics, color space and color
7685 primaries to the specified user values. The output value, if not specified,
7686 is set to a default value based on the "all" property. If that property is
7687 also not specified, the filter will log an error. The output color range and
7688 format default to the same value as the input color range and format. The
7689 input transfer characteristics, color space, color primaries and color range
7690 should be set on the input data. If any of these are missing, the filter will
7691 log an error and no conversion will take place.
7693 For example to convert the input to SMPTE-240M, use the command:
7695 colorspace=smpte240m
7698 @section convolution
7700 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
7702 The filter accepts the following options:
7709 Set matrix for each plane.
7710 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
7711 and from 1 to 49 odd number of signed integers in @var{row} mode.
7717 Set multiplier for calculated value for each plane.
7718 If unset or 0, it will be sum of all matrix elements.
7724 Set bias for each plane. This value is added to the result of the multiplication.
7725 Useful for making the overall image brighter or darker. Default is 0.0.
7731 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
7732 Default is @var{square}.
7735 @subsection Examples
7741 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"
7747 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"
7753 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"
7759 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"
7763 Apply laplacian edge detector which includes diagonals:
7765 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"
7771 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"
7777 Apply 2D convolution of video stream in frequency domain using second stream
7780 The filter accepts the following options:
7784 Set which planes to process.
7787 Set which impulse video frames will be processed, can be @var{first}
7788 or @var{all}. Default is @var{all}.
7791 The @code{convolve} filter also supports the @ref{framesync} options.
7795 Copy the input video source unchanged to the output. This is mainly useful for
7800 Video filtering on GPU using Apple's CoreImage API on OSX.
7802 Hardware acceleration is based on an OpenGL context. Usually, this means it is
7803 processed by video hardware. However, software-based OpenGL implementations
7804 exist which means there is no guarantee for hardware processing. It depends on
7807 There are many filters and image generators provided by Apple that come with a
7808 large variety of options. The filter has to be referenced by its name along
7811 The coreimage filter accepts the following options:
7814 List all available filters and generators along with all their respective
7815 options as well as possible minimum and maximum values along with the default
7822 Specify all filters by their respective name and options.
7823 Use @var{list_filters} to determine all valid filter names and options.
7824 Numerical options are specified by a float value and are automatically clamped
7825 to their respective value range. Vector and color options have to be specified
7826 by a list of space separated float values. Character escaping has to be done.
7827 A special option name @code{default} is available to use default options for a
7830 It is required to specify either @code{default} or at least one of the filter options.
7831 All omitted options are used with their default values.
7832 The syntax of the filter string is as follows:
7834 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
7838 Specify a rectangle where the output of the filter chain is copied into the
7839 input image. It is given by a list of space separated float values:
7841 output_rect=x\ y\ width\ height
7843 If not given, the output rectangle equals the dimensions of the input image.
7844 The output rectangle is automatically cropped at the borders of the input
7845 image. Negative values are valid for each component.
7847 output_rect=25\ 25\ 100\ 100
7851 Several filters can be chained for successive processing without GPU-HOST
7852 transfers allowing for fast processing of complex filter chains.
7853 Currently, only filters with zero (generators) or exactly one (filters) input
7854 image and one output image are supported. Also, transition filters are not yet
7857 Some filters generate output images with additional padding depending on the
7858 respective filter kernel. The padding is automatically removed to ensure the
7859 filter output has the same size as the input image.
7861 For image generators, the size of the output image is determined by the
7862 previous output image of the filter chain or the input image of the whole
7863 filterchain, respectively. The generators do not use the pixel information of
7864 this image to generate their output. However, the generated output is
7865 blended onto this image, resulting in partial or complete coverage of the
7868 The @ref{coreimagesrc} video source can be used for generating input images
7869 which are directly fed into the filter chain. By using it, providing input
7870 images by another video source or an input video is not required.
7872 @subsection Examples
7877 List all filters available:
7879 coreimage=list_filters=true
7883 Use the CIBoxBlur filter with default options to blur an image:
7885 coreimage=filter=CIBoxBlur@@default
7889 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
7890 its center at 100x100 and a radius of 50 pixels:
7892 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
7896 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
7897 given as complete and escaped command-line for Apple's standard bash shell:
7899 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
7905 Cover a rectangular object
7907 It accepts the following options:
7911 Filepath of the optional cover image, needs to be in yuv420.
7916 It accepts the following values:
7919 cover it by the supplied image
7921 cover it by interpolating the surrounding pixels
7924 Default value is @var{blur}.
7927 @subsection Examples
7931 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
7933 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
7939 Crop the input video to given dimensions.
7941 It accepts the following parameters:
7945 The width of the output video. It defaults to @code{iw}.
7946 This expression is evaluated only once during the filter
7947 configuration, or when the @samp{w} or @samp{out_w} command is sent.
7950 The height of the output video. It defaults to @code{ih}.
7951 This expression is evaluated only once during the filter
7952 configuration, or when the @samp{h} or @samp{out_h} command is sent.
7955 The horizontal position, in the input video, of the left edge of the output
7956 video. It defaults to @code{(in_w-out_w)/2}.
7957 This expression is evaluated per-frame.
7960 The vertical position, in the input video, of the top edge of the output video.
7961 It defaults to @code{(in_h-out_h)/2}.
7962 This expression is evaluated per-frame.
7965 If set to 1 will force the output display aspect ratio
7966 to be the same of the input, by changing the output sample aspect
7967 ratio. It defaults to 0.
7970 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
7971 width/height/x/y as specified and will not be rounded to nearest smaller value.
7975 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
7976 expressions containing the following constants:
7981 The computed values for @var{x} and @var{y}. They are evaluated for
7986 The input width and height.
7990 These are the same as @var{in_w} and @var{in_h}.
7994 The output (cropped) width and height.
7998 These are the same as @var{out_w} and @var{out_h}.
8001 same as @var{iw} / @var{ih}
8004 input sample aspect ratio
8007 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8011 horizontal and vertical chroma subsample values. For example for the
8012 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8015 The number of the input frame, starting from 0.
8018 the position in the file of the input frame, NAN if unknown
8021 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8025 The expression for @var{out_w} may depend on the value of @var{out_h},
8026 and the expression for @var{out_h} may depend on @var{out_w}, but they
8027 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8028 evaluated after @var{out_w} and @var{out_h}.
8030 The @var{x} and @var{y} parameters specify the expressions for the
8031 position of the top-left corner of the output (non-cropped) area. They
8032 are evaluated for each frame. If the evaluated value is not valid, it
8033 is approximated to the nearest valid value.
8035 The expression for @var{x} may depend on @var{y}, and the expression
8036 for @var{y} may depend on @var{x}.
8038 @subsection Examples
8042 Crop area with size 100x100 at position (12,34).
8047 Using named options, the example above becomes:
8049 crop=w=100:h=100:x=12:y=34
8053 Crop the central input area with size 100x100:
8059 Crop the central input area with size 2/3 of the input video:
8061 crop=2/3*in_w:2/3*in_h
8065 Crop the input video central square:
8072 Delimit the rectangle with the top-left corner placed at position
8073 100:100 and the right-bottom corner corresponding to the right-bottom
8074 corner of the input image.
8076 crop=in_w-100:in_h-100:100:100
8080 Crop 10 pixels from the left and right borders, and 20 pixels from
8081 the top and bottom borders
8083 crop=in_w-2*10:in_h-2*20
8087 Keep only the bottom right quarter of the input image:
8089 crop=in_w/2:in_h/2:in_w/2:in_h/2
8093 Crop height for getting Greek harmony:
8095 crop=in_w:1/PHI*in_w
8099 Apply trembling effect:
8101 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)
8105 Apply erratic camera effect depending on timestamp:
8107 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)"
8111 Set x depending on the value of y:
8113 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8117 @subsection Commands
8119 This filter supports the following commands:
8125 Set width/height of the output video and the horizontal/vertical position
8127 The command accepts the same syntax of the corresponding option.
8129 If the specified expression is not valid, it is kept at its current
8135 Auto-detect the crop size.
8137 It calculates the necessary cropping parameters and prints the
8138 recommended parameters via the logging system. The detected dimensions
8139 correspond to the non-black area of the input video.
8141 It accepts the following parameters:
8146 Set higher black value threshold, which can be optionally specified
8147 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8148 value greater to the set value is considered non-black. It defaults to 24.
8149 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8150 on the bitdepth of the pixel format.
8153 The value which the width/height should be divisible by. It defaults to
8154 16. The offset is automatically adjusted to center the video. Use 2 to
8155 get only even dimensions (needed for 4:2:2 video). 16 is best when
8156 encoding to most video codecs.
8158 @item reset_count, reset
8159 Set the counter that determines after how many frames cropdetect will
8160 reset the previously detected largest video area and start over to
8161 detect the current optimal crop area. Default value is 0.
8163 This can be useful when channel logos distort the video area. 0
8164 indicates 'never reset', and returns the largest area encountered during
8171 Delay video filtering until a given wallclock timestamp. The filter first
8172 passes on @option{preroll} amount of frames, then it buffers at most
8173 @option{buffer} amount of frames and waits for the cue. After reaching the cue
8174 it forwards the buffered frames and also any subsequent frames coming in its
8177 The filter can be used synchronize the output of multiple ffmpeg processes for
8178 realtime output devices like decklink. By putting the delay in the filtering
8179 chain and pre-buffering frames the process can pass on data to output almost
8180 immediately after the target wallclock timestamp is reached.
8182 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
8188 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
8191 The duration of content to pass on as preroll expressed in seconds. Default is 0.
8194 The maximum duration of content to buffer before waiting for the cue expressed
8195 in seconds. Default is 0.
8202 Apply color adjustments using curves.
8204 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
8205 component (red, green and blue) has its values defined by @var{N} key points
8206 tied from each other using a smooth curve. The x-axis represents the pixel
8207 values from the input frame, and the y-axis the new pixel values to be set for
8210 By default, a component curve is defined by the two points @var{(0;0)} and
8211 @var{(1;1)}. This creates a straight line where each original pixel value is
8212 "adjusted" to its own value, which means no change to the image.
8214 The filter allows you to redefine these two points and add some more. A new
8215 curve (using a natural cubic spline interpolation) will be define to pass
8216 smoothly through all these new coordinates. The new defined points needs to be
8217 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
8218 be in the @var{[0;1]} interval. If the computed curves happened to go outside
8219 the vector spaces, the values will be clipped accordingly.
8221 The filter accepts the following options:
8225 Select one of the available color presets. This option can be used in addition
8226 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
8227 options takes priority on the preset values.
8228 Available presets are:
8231 @item color_negative
8234 @item increase_contrast
8236 @item linear_contrast
8237 @item medium_contrast
8239 @item strong_contrast
8242 Default is @code{none}.
8244 Set the master key points. These points will define a second pass mapping. It
8245 is sometimes called a "luminance" or "value" mapping. It can be used with
8246 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
8247 post-processing LUT.
8249 Set the key points for the red component.
8251 Set the key points for the green component.
8253 Set the key points for the blue component.
8255 Set the key points for all components (not including master).
8256 Can be used in addition to the other key points component
8257 options. In this case, the unset component(s) will fallback on this
8258 @option{all} setting.
8260 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
8262 Save Gnuplot script of the curves in specified file.
8265 To avoid some filtergraph syntax conflicts, each key points list need to be
8266 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
8268 @subsection Examples
8272 Increase slightly the middle level of blue:
8274 curves=blue='0/0 0.5/0.58 1/1'
8280 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'
8282 Here we obtain the following coordinates for each components:
8285 @code{(0;0.11) (0.42;0.51) (1;0.95)}
8287 @code{(0;0) (0.50;0.48) (1;1)}
8289 @code{(0;0.22) (0.49;0.44) (1;0.80)}
8293 The previous example can also be achieved with the associated built-in preset:
8295 curves=preset=vintage
8305 Use a Photoshop preset and redefine the points of the green component:
8307 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
8311 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
8312 and @command{gnuplot}:
8314 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
8315 gnuplot -p /tmp/curves.plt
8321 Video data analysis filter.
8323 This filter shows hexadecimal pixel values of part of video.
8325 The filter accepts the following options:
8329 Set output video size.
8332 Set x offset from where to pick pixels.
8335 Set y offset from where to pick pixels.
8338 Set scope mode, can be one of the following:
8341 Draw hexadecimal pixel values with white color on black background.
8344 Draw hexadecimal pixel values with input video pixel color on black
8348 Draw hexadecimal pixel values on color background picked from input video,
8349 the text color is picked in such way so its always visible.
8353 Draw rows and columns numbers on left and top of video.
8356 Set background opacity.
8361 Denoise frames using 2D DCT (frequency domain filtering).
8363 This filter is not designed for real time.
8365 The filter accepts the following options:
8369 Set the noise sigma constant.
8371 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
8372 coefficient (absolute value) below this threshold with be dropped.
8374 If you need a more advanced filtering, see @option{expr}.
8376 Default is @code{0}.
8379 Set number overlapping pixels for each block. Since the filter can be slow, you
8380 may want to reduce this value, at the cost of a less effective filter and the
8381 risk of various artefacts.
8383 If the overlapping value doesn't permit processing the whole input width or
8384 height, a warning will be displayed and according borders won't be denoised.
8386 Default value is @var{blocksize}-1, which is the best possible setting.
8389 Set the coefficient factor expression.
8391 For each coefficient of a DCT block, this expression will be evaluated as a
8392 multiplier value for the coefficient.
8394 If this is option is set, the @option{sigma} option will be ignored.
8396 The absolute value of the coefficient can be accessed through the @var{c}
8400 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
8401 @var{blocksize}, which is the width and height of the processed blocks.
8403 The default value is @var{3} (8x8) and can be raised to @var{4} for a
8404 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
8405 on the speed processing. Also, a larger block size does not necessarily means a
8409 @subsection Examples
8411 Apply a denoise with a @option{sigma} of @code{4.5}:
8416 The same operation can be achieved using the expression system:
8418 dctdnoiz=e='gte(c, 4.5*3)'
8421 Violent denoise using a block size of @code{16x16}:
8428 Remove banding artifacts from input video.
8429 It works by replacing banded pixels with average value of referenced pixels.
8431 The filter accepts the following options:
8438 Set banding detection threshold for each plane. Default is 0.02.
8439 Valid range is 0.00003 to 0.5.
8440 If difference between current pixel and reference pixel is less than threshold,
8441 it will be considered as banded.
8444 Banding detection range in pixels. Default is 16. If positive, random number
8445 in range 0 to set value will be used. If negative, exact absolute value
8447 The range defines square of four pixels around current pixel.
8450 Set direction in radians from which four pixel will be compared. If positive,
8451 random direction from 0 to set direction will be picked. If negative, exact of
8452 absolute value will be picked. For example direction 0, -PI or -2*PI radians
8453 will pick only pixels on same row and -PI/2 will pick only pixels on same
8457 If enabled, current pixel is compared with average value of all four
8458 surrounding pixels. The default is enabled. If disabled current pixel is
8459 compared with all four surrounding pixels. The pixel is considered banded
8460 if only all four differences with surrounding pixels are less than threshold.
8463 If enabled, current pixel is changed if and only if all pixel components are banded,
8464 e.g. banding detection threshold is triggered for all color components.
8465 The default is disabled.
8470 Remove blocking artifacts from input video.
8472 The filter accepts the following options:
8476 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
8477 This controls what kind of deblocking is applied.
8480 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
8486 Set blocking detection thresholds. Allowed range is 0 to 1.
8487 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
8488 Using higher threshold gives more deblocking strength.
8489 Setting @var{alpha} controls threshold detection at exact edge of block.
8490 Remaining options controls threshold detection near the edge. Each one for
8491 below/above or left/right. Setting any of those to @var{0} disables
8495 Set planes to filter. Default is to filter all available planes.
8498 @subsection Examples
8502 Deblock using weak filter and block size of 4 pixels.
8504 deblock=filter=weak:block=4
8508 Deblock using strong filter, block size of 4 pixels and custom thresholds for
8509 deblocking more edges.
8511 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
8515 Similar as above, but filter only first plane.
8517 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
8521 Similar as above, but filter only second and third plane.
8523 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
8530 Drop duplicated frames at regular intervals.
8532 The filter accepts the following options:
8536 Set the number of frames from which one will be dropped. Setting this to
8537 @var{N} means one frame in every batch of @var{N} frames will be dropped.
8538 Default is @code{5}.
8541 Set the threshold for duplicate detection. If the difference metric for a frame
8542 is less than or equal to this value, then it is declared as duplicate. Default
8546 Set scene change threshold. Default is @code{15}.
8550 Set the size of the x and y-axis blocks used during metric calculations.
8551 Larger blocks give better noise suppression, but also give worse detection of
8552 small movements. Must be a power of two. Default is @code{32}.
8555 Mark main input as a pre-processed input and activate clean source input
8556 stream. This allows the input to be pre-processed with various filters to help
8557 the metrics calculation while keeping the frame selection lossless. When set to
8558 @code{1}, the first stream is for the pre-processed input, and the second
8559 stream is the clean source from where the kept frames are chosen. Default is
8563 Set whether or not chroma is considered in the metric calculations. Default is
8569 Apply 2D deconvolution of video stream in frequency domain using second stream
8572 The filter accepts the following options:
8576 Set which planes to process.
8579 Set which impulse video frames will be processed, can be @var{first}
8580 or @var{all}. Default is @var{all}.
8583 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
8584 and height are not same and not power of 2 or if stream prior to convolving
8588 The @code{deconvolve} filter also supports the @ref{framesync} options.
8592 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
8594 It accepts the following options:
8598 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
8599 @var{rainbows} for cross-color reduction.
8602 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
8605 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
8608 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
8611 Set temporal chroma threshold. Lower values increases reduction of cross-color.
8616 Apply deflate effect to the video.
8618 This filter replaces the pixel by the local(3x3) average by taking into account
8619 only values lower than the pixel.
8621 It accepts the following options:
8628 Limit the maximum change for each plane, default is 65535.
8629 If 0, plane will remain unchanged.
8634 Remove temporal frame luminance variations.
8636 It accepts the following options:
8640 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
8643 Set averaging mode to smooth temporal luminance variations.
8645 Available values are:
8670 Do not actually modify frame. Useful when one only wants metadata.
8675 Remove judder produced by partially interlaced telecined content.
8677 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
8678 source was partially telecined content then the output of @code{pullup,dejudder}
8679 will have a variable frame rate. May change the recorded frame rate of the
8680 container. Aside from that change, this filter will not affect constant frame
8683 The option available in this filter is:
8687 Specify the length of the window over which the judder repeats.
8689 Accepts any integer greater than 1. Useful values are:
8693 If the original was telecined from 24 to 30 fps (Film to NTSC).
8696 If the original was telecined from 25 to 30 fps (PAL to NTSC).
8699 If a mixture of the two.
8702 The default is @samp{4}.
8707 Suppress a TV station logo by a simple interpolation of the surrounding
8708 pixels. Just set a rectangle covering the logo and watch it disappear
8709 (and sometimes something even uglier appear - your mileage may vary).
8711 It accepts the following parameters:
8716 Specify the top left corner coordinates of the logo. They must be
8721 Specify the width and height of the logo to clear. They must be
8725 Specify the thickness of the fuzzy edge of the rectangle (added to
8726 @var{w} and @var{h}). The default value is 1. This option is
8727 deprecated, setting higher values should no longer be necessary and
8731 When set to 1, a green rectangle is drawn on the screen to simplify
8732 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
8733 The default value is 0.
8735 The rectangle is drawn on the outermost pixels which will be (partly)
8736 replaced with interpolated values. The values of the next pixels
8737 immediately outside this rectangle in each direction will be used to
8738 compute the interpolated pixel values inside the rectangle.
8742 @subsection Examples
8746 Set a rectangle covering the area with top left corner coordinates 0,0
8747 and size 100x77, and a band of size 10:
8749 delogo=x=0:y=0:w=100:h=77:band=10
8756 Remove the rain in the input image/video by applying the derain methods based on
8757 convolutional neural networks. Supported models:
8761 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
8762 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
8765 Training as well as model generation scripts are provided in
8766 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
8768 Native model files (.model) can be generated from TensorFlow model
8769 files (.pb) by using tools/python/convert.py
8771 The filter accepts the following options:
8775 Specify which filter to use. This option accepts the following values:
8779 Derain filter. To conduct derain filter, you need to use a derain model.
8782 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
8784 Default value is @samp{derain}.
8787 Specify which DNN backend to use for model loading and execution. This option accepts
8788 the following values:
8792 Native implementation of DNN loading and execution.
8795 TensorFlow backend. To enable this backend you
8796 need to install the TensorFlow for C library (see
8797 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
8798 @code{--enable-libtensorflow}
8800 Default value is @samp{native}.
8803 Set path to model file specifying network architecture and its parameters.
8804 Note that different backends use different file formats. TensorFlow and native
8805 backend can load files for only its format.
8810 Attempt to fix small changes in horizontal and/or vertical shift. This
8811 filter helps remove camera shake from hand-holding a camera, bumping a
8812 tripod, moving on a vehicle, etc.
8814 The filter accepts the following options:
8822 Specify a rectangular area where to limit the search for motion
8824 If desired the search for motion vectors can be limited to a
8825 rectangular area of the frame defined by its top left corner, width
8826 and height. These parameters have the same meaning as the drawbox
8827 filter which can be used to visualise the position of the bounding
8830 This is useful when simultaneous movement of subjects within the frame
8831 might be confused for camera motion by the motion vector search.
8833 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
8834 then the full frame is used. This allows later options to be set
8835 without specifying the bounding box for the motion vector search.
8837 Default - search the whole frame.
8841 Specify the maximum extent of movement in x and y directions in the
8842 range 0-64 pixels. Default 16.
8845 Specify how to generate pixels to fill blanks at the edge of the
8846 frame. Available values are:
8849 Fill zeroes at blank locations
8851 Original image at blank locations
8853 Extruded edge value at blank locations
8855 Mirrored edge at blank locations
8857 Default value is @samp{mirror}.
8860 Specify the blocksize to use for motion search. Range 4-128 pixels,
8864 Specify the contrast threshold for blocks. Only blocks with more than
8865 the specified contrast (difference between darkest and lightest
8866 pixels) will be considered. Range 1-255, default 125.
8869 Specify the search strategy. Available values are:
8872 Set exhaustive search
8874 Set less exhaustive search.
8876 Default value is @samp{exhaustive}.
8879 If set then a detailed log of the motion search is written to the
8886 Remove unwanted contamination of foreground colors, caused by reflected color of
8887 greenscreen or bluescreen.
8889 This filter accepts the following options:
8893 Set what type of despill to use.
8896 Set how spillmap will be generated.
8899 Set how much to get rid of still remaining spill.
8902 Controls amount of red in spill area.
8905 Controls amount of green in spill area.
8906 Should be -1 for greenscreen.
8909 Controls amount of blue in spill area.
8910 Should be -1 for bluescreen.
8913 Controls brightness of spill area, preserving colors.
8916 Modify alpha from generated spillmap.
8921 Apply an exact inverse of the telecine operation. It requires a predefined
8922 pattern specified using the pattern option which must be the same as that passed
8923 to the telecine filter.
8925 This filter accepts the following options:
8934 The default value is @code{top}.
8938 A string of numbers representing the pulldown pattern you wish to apply.
8939 The default value is @code{23}.
8942 A number representing position of the first frame with respect to the telecine
8943 pattern. This is to be used if the stream is cut. The default value is @code{0}.
8948 Apply dilation effect to the video.
8950 This filter replaces the pixel by the local(3x3) maximum.
8952 It accepts the following options:
8959 Limit the maximum change for each plane, default is 65535.
8960 If 0, plane will remain unchanged.
8963 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
8966 Flags to local 3x3 coordinates maps like this:
8975 Displace pixels as indicated by second and third input stream.
8977 It takes three input streams and outputs one stream, the first input is the
8978 source, and second and third input are displacement maps.
8980 The second input specifies how much to displace pixels along the
8981 x-axis, while the third input specifies how much to displace pixels
8983 If one of displacement map streams terminates, last frame from that
8984 displacement map will be used.
8986 Note that once generated, displacements maps can be reused over and over again.
8988 A description of the accepted options follows.
8992 Set displace behavior for pixels that are out of range.
8994 Available values are:
8997 Missing pixels are replaced by black pixels.
9000 Adjacent pixels will spread out to replace missing pixels.
9003 Out of range pixels are wrapped so they point to pixels of other side.
9006 Out of range pixels will be replaced with mirrored pixels.
9008 Default is @samp{smear}.
9012 @subsection Examples
9016 Add ripple effect to rgb input of video size hd720:
9018 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
9022 Add wave effect to rgb input of video size hd720:
9024 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
9028 @section dnn_processing
9030 Do image processing with deep neural networks. Currently only AVFrame with RGB24
9031 and BGR24 are supported, more formats will be added later.
9033 The filter accepts the following options:
9037 Specify which DNN backend to use for model loading and execution. This option accepts
9038 the following values:
9042 Native implementation of DNN loading and execution.
9045 TensorFlow backend. To enable this backend you
9046 need to install the TensorFlow for C library (see
9047 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9048 @code{--enable-libtensorflow}
9051 Default value is @samp{native}.
9054 Set path to model file specifying network architecture and its parameters.
9055 Note that different backends use different file formats. TensorFlow and native
9056 backend can load files for only its format.
9058 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9061 Set the input name of the dnn network.
9064 Set the output name of the dnn network.
9067 Set the pixel format for the Frame. Allowed values are @code{AV_PIX_FMT_RGB24}, and @code{AV_PIX_FMT_BGR24}.
9068 Default value is @code{AV_PIX_FMT_RGB24}.
9074 Draw a colored box on the input image.
9076 It accepts the following parameters:
9081 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
9085 The expressions which specify the width and height of the box; if 0 they are interpreted as
9086 the input width and height. It defaults to 0.
9089 Specify the color of the box to write. For the general syntax of this option,
9090 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9091 value @code{invert} is used, the box edge color is the same as the
9092 video with inverted luma.
9095 The expression which sets the thickness of the box edge.
9096 A value of @code{fill} will create a filled box. Default value is @code{3}.
9098 See below for the list of accepted constants.
9101 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
9102 will overwrite the video's color and alpha pixels.
9103 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
9106 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9107 following constants:
9111 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9115 horizontal and vertical chroma subsample values. For example for the
9116 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9120 The input width and height.
9123 The input sample aspect ratio.
9127 The x and y offset coordinates where the box is drawn.
9131 The width and height of the drawn box.
9134 The thickness of the drawn box.
9136 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9137 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9141 @subsection Examples
9145 Draw a black box around the edge of the input image:
9151 Draw a box with color red and an opacity of 50%:
9153 drawbox=10:20:200:60:red@@0.5
9156 The previous example can be specified as:
9158 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
9162 Fill the box with pink color:
9164 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
9168 Draw a 2-pixel red 2.40:1 mask:
9170 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
9174 @subsection Commands
9175 This filter supports same commands as options.
9176 The command accepts the same syntax of the corresponding option.
9178 If the specified expression is not valid, it is kept at its current
9183 Draw a graph using input video metadata.
9185 It accepts the following parameters:
9189 Set 1st frame metadata key from which metadata values will be used to draw a graph.
9192 Set 1st foreground color expression.
9195 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
9198 Set 2nd foreground color expression.
9201 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
9204 Set 3rd foreground color expression.
9207 Set 4th frame metadata key from which metadata values will be used to draw a graph.
9210 Set 4th foreground color expression.
9213 Set minimal value of metadata value.
9216 Set maximal value of metadata value.
9219 Set graph background color. Default is white.
9224 Available values for mode is:
9231 Default is @code{line}.
9236 Available values for slide is:
9239 Draw new frame when right border is reached.
9242 Replace old columns with new ones.
9245 Scroll from right to left.
9248 Scroll from left to right.
9251 Draw single picture.
9254 Default is @code{frame}.
9257 Set size of graph video. For the syntax of this option, check the
9258 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
9259 The default value is @code{900x256}.
9261 The foreground color expressions can use the following variables:
9264 Minimal value of metadata value.
9267 Maximal value of metadata value.
9270 Current metadata key value.
9273 The color is defined as 0xAABBGGRR.
9276 Example using metadata from @ref{signalstats} filter:
9278 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
9281 Example using metadata from @ref{ebur128} filter:
9283 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
9288 Draw a grid on the input image.
9290 It accepts the following parameters:
9295 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
9299 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
9300 input width and height, respectively, minus @code{thickness}, so image gets
9301 framed. Default to 0.
9304 Specify the color of the grid. For the general syntax of this option,
9305 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
9306 value @code{invert} is used, the grid color is the same as the
9307 video with inverted luma.
9310 The expression which sets the thickness of the grid line. Default value is @code{1}.
9312 See below for the list of accepted constants.
9315 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
9316 will overwrite the video's color and alpha pixels.
9317 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
9320 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
9321 following constants:
9325 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
9329 horizontal and vertical chroma subsample values. For example for the
9330 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9334 The input grid cell width and height.
9337 The input sample aspect ratio.
9341 The x and y coordinates of some point of grid intersection (meant to configure offset).
9345 The width and height of the drawn cell.
9348 The thickness of the drawn cell.
9350 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
9351 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
9355 @subsection Examples
9359 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
9361 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
9365 Draw a white 3x3 grid with an opacity of 50%:
9367 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
9371 @subsection Commands
9372 This filter supports same commands as options.
9373 The command accepts the same syntax of the corresponding option.
9375 If the specified expression is not valid, it is kept at its current
9381 Draw a text string or text from a specified file on top of a video, using the
9382 libfreetype library.
9384 To enable compilation of this filter, you need to configure FFmpeg with
9385 @code{--enable-libfreetype}.
9386 To enable default font fallback and the @var{font} option you need to
9387 configure FFmpeg with @code{--enable-libfontconfig}.
9388 To enable the @var{text_shaping} option, you need to configure FFmpeg with
9389 @code{--enable-libfribidi}.
9393 It accepts the following parameters:
9398 Used to draw a box around text using the background color.
9399 The value must be either 1 (enable) or 0 (disable).
9400 The default value of @var{box} is 0.
9403 Set the width of the border to be drawn around the box using @var{boxcolor}.
9404 The default value of @var{boxborderw} is 0.
9407 The color to be used for drawing box around text. For the syntax of this
9408 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9410 The default value of @var{boxcolor} is "white".
9413 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
9414 The default value of @var{line_spacing} is 0.
9417 Set the width of the border to be drawn around the text using @var{bordercolor}.
9418 The default value of @var{borderw} is 0.
9421 Set the color to be used for drawing border around text. For the syntax of this
9422 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9424 The default value of @var{bordercolor} is "black".
9427 Select how the @var{text} is expanded. Can be either @code{none},
9428 @code{strftime} (deprecated) or
9429 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
9433 Set a start time for the count. Value is in microseconds. Only applied
9434 in the deprecated strftime expansion mode. To emulate in normal expansion
9435 mode use the @code{pts} function, supplying the start time (in seconds)
9436 as the second argument.
9439 If true, check and fix text coords to avoid clipping.
9442 The color to be used for drawing fonts. For the syntax of this option, check
9443 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
9445 The default value of @var{fontcolor} is "black".
9447 @item fontcolor_expr
9448 String which is expanded the same way as @var{text} to obtain dynamic
9449 @var{fontcolor} value. By default this option has empty value and is not
9450 processed. When this option is set, it overrides @var{fontcolor} option.
9453 The font family to be used for drawing text. By default Sans.
9456 The font file to be used for drawing text. The path must be included.
9457 This parameter is mandatory if the fontconfig support is disabled.
9460 Draw the text applying alpha blending. The value can
9461 be a number between 0.0 and 1.0.
9462 The expression accepts the same variables @var{x, y} as well.
9463 The default value is 1.
9464 Please see @var{fontcolor_expr}.
9467 The font size to be used for drawing text.
9468 The default value of @var{fontsize} is 16.
9471 If set to 1, attempt to shape the text (for example, reverse the order of
9472 right-to-left text and join Arabic characters) before drawing it.
9473 Otherwise, just draw the text exactly as given.
9474 By default 1 (if supported).
9477 The flags to be used for loading the fonts.
9479 The flags map the corresponding flags supported by libfreetype, and are
9480 a combination of the following values:
9487 @item vertical_layout
9488 @item force_autohint
9491 @item ignore_global_advance_width
9493 @item ignore_transform
9499 Default value is "default".
9501 For more information consult the documentation for the FT_LOAD_*
9505 The color to be used for drawing a shadow behind the drawn text. For the
9506 syntax of this option, check the @ref{color syntax,,"Color" section in the
9507 ffmpeg-utils manual,ffmpeg-utils}.
9509 The default value of @var{shadowcolor} is "black".
9513 The x and y offsets for the text shadow position with respect to the
9514 position of the text. They can be either positive or negative
9515 values. The default value for both is "0".
9518 The starting frame number for the n/frame_num variable. The default value
9522 The size in number of spaces to use for rendering the tab.
9526 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
9527 format. It can be used with or without text parameter. @var{timecode_rate}
9528 option must be specified.
9530 @item timecode_rate, rate, r
9531 Set the timecode frame rate (timecode only). Value will be rounded to nearest
9532 integer. Minimum value is "1".
9533 Drop-frame timecode is supported for frame rates 30 & 60.
9536 If set to 1, the output of the timecode option will wrap around at 24 hours.
9537 Default is 0 (disabled).
9540 The text string to be drawn. The text must be a sequence of UTF-8
9542 This parameter is mandatory if no file is specified with the parameter
9546 A text file containing text to be drawn. The text must be a sequence
9547 of UTF-8 encoded characters.
9549 This parameter is mandatory if no text string is specified with the
9550 parameter @var{text}.
9552 If both @var{text} and @var{textfile} are specified, an error is thrown.
9555 If set to 1, the @var{textfile} will be reloaded before each frame.
9556 Be sure to update it atomically, or it may be read partially, or even fail.
9560 The expressions which specify the offsets where text will be drawn
9561 within the video frame. They are relative to the top/left border of the
9564 The default value of @var{x} and @var{y} is "0".
9566 See below for the list of accepted constants and functions.
9569 The parameters for @var{x} and @var{y} are expressions containing the
9570 following constants and functions:
9574 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
9578 horizontal and vertical chroma subsample values. For example for the
9579 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
9582 the height of each text line
9590 @item max_glyph_a, ascent
9591 the maximum distance from the baseline to the highest/upper grid
9592 coordinate used to place a glyph outline point, for all the rendered
9594 It is a positive value, due to the grid's orientation with the Y axis
9597 @item max_glyph_d, descent
9598 the maximum distance from the baseline to the lowest grid coordinate
9599 used to place a glyph outline point, for all the rendered glyphs.
9600 This is a negative value, due to the grid's orientation, with the Y axis
9604 maximum glyph height, that is the maximum height for all the glyphs
9605 contained in the rendered text, it is equivalent to @var{ascent} -
9609 maximum glyph width, that is the maximum width for all the glyphs
9610 contained in the rendered text
9613 the number of input frame, starting from 0
9615 @item rand(min, max)
9616 return a random number included between @var{min} and @var{max}
9619 The input sample aspect ratio.
9622 timestamp expressed in seconds, NAN if the input timestamp is unknown
9625 the height of the rendered text
9628 the width of the rendered text
9632 the x and y offset coordinates where the text is drawn.
9634 These parameters allow the @var{x} and @var{y} expressions to refer
9635 to each other, so you can for example specify @code{y=x/dar}.
9638 A one character description of the current frame's picture type.
9641 The current packet's position in the input file or stream
9642 (in bytes, from the start of the input). A value of -1 indicates
9643 this info is not available.
9646 The current packet's duration, in seconds.
9649 The current packet's size (in bytes).
9652 @anchor{drawtext_expansion}
9653 @subsection Text expansion
9655 If @option{expansion} is set to @code{strftime},
9656 the filter recognizes strftime() sequences in the provided text and
9657 expands them accordingly. Check the documentation of strftime(). This
9658 feature is deprecated.
9660 If @option{expansion} is set to @code{none}, the text is printed verbatim.
9662 If @option{expansion} is set to @code{normal} (which is the default),
9663 the following expansion mechanism is used.
9665 The backslash character @samp{\}, followed by any character, always expands to
9666 the second character.
9668 Sequences of the form @code{%@{...@}} are expanded. The text between the
9669 braces is a function name, possibly followed by arguments separated by ':'.
9670 If the arguments contain special characters or delimiters (':' or '@}'),
9671 they should be escaped.
9673 Note that they probably must also be escaped as the value for the
9674 @option{text} option in the filter argument string and as the filter
9675 argument in the filtergraph description, and possibly also for the shell,
9676 that makes up to four levels of escaping; using a text file avoids these
9679 The following functions are available:
9684 The expression evaluation result.
9686 It must take one argument specifying the expression to be evaluated,
9687 which accepts the same constants and functions as the @var{x} and
9688 @var{y} values. Note that not all constants should be used, for
9689 example the text size is not known when evaluating the expression, so
9690 the constants @var{text_w} and @var{text_h} will have an undefined
9693 @item expr_int_format, eif
9694 Evaluate the expression's value and output as formatted integer.
9696 The first argument is the expression to be evaluated, just as for the @var{expr} function.
9697 The second argument specifies the output format. Allowed values are @samp{x},
9698 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
9699 @code{printf} function.
9700 The third parameter is optional and sets the number of positions taken by the output.
9701 It can be used to add padding with zeros from the left.
9704 The time at which the filter is running, expressed in UTC.
9705 It can accept an argument: a strftime() format string.
9708 The time at which the filter is running, expressed in the local time zone.
9709 It can accept an argument: a strftime() format string.
9712 Frame metadata. Takes one or two arguments.
9714 The first argument is mandatory and specifies the metadata key.
9716 The second argument is optional and specifies a default value, used when the
9717 metadata key is not found or empty.
9719 Available metadata can be identified by inspecting entries
9720 starting with TAG included within each frame section
9721 printed by running @code{ffprobe -show_frames}.
9723 String metadata generated in filters leading to
9724 the drawtext filter are also available.
9727 The frame number, starting from 0.
9730 A one character description of the current picture type.
9733 The timestamp of the current frame.
9734 It can take up to three arguments.
9736 The first argument is the format of the timestamp; it defaults to @code{flt}
9737 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
9738 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
9739 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
9740 @code{localtime} stands for the timestamp of the frame formatted as
9741 local time zone time.
9743 The second argument is an offset added to the timestamp.
9745 If the format is set to @code{hms}, a third argument @code{24HH} may be
9746 supplied to present the hour part of the formatted timestamp in 24h format
9749 If the format is set to @code{localtime} or @code{gmtime},
9750 a third argument may be supplied: a strftime() format string.
9751 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
9754 @subsection Commands
9756 This filter supports altering parameters via commands:
9759 Alter existing filter parameters.
9761 Syntax for the argument is the same as for filter invocation, e.g.
9764 fontsize=56:fontcolor=green:text='Hello World'
9767 Full filter invocation with sendcmd would look like this:
9770 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
9774 If the entire argument can't be parsed or applied as valid values then the filter will
9775 continue with its existing parameters.
9777 @subsection Examples
9781 Draw "Test Text" with font FreeSerif, using the default values for the
9782 optional parameters.
9785 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
9789 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
9790 and y=50 (counting from the top-left corner of the screen), text is
9791 yellow with a red box around it. Both the text and the box have an
9795 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
9796 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
9799 Note that the double quotes are not necessary if spaces are not used
9800 within the parameter list.
9803 Show the text at the center of the video frame:
9805 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
9809 Show the text at a random position, switching to a new position every 30 seconds:
9811 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)"
9815 Show a text line sliding from right to left in the last row of the video
9816 frame. The file @file{LONG_LINE} is assumed to contain a single line
9819 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
9823 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
9825 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
9829 Draw a single green letter "g", at the center of the input video.
9830 The glyph baseline is placed at half screen height.
9832 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
9836 Show text for 1 second every 3 seconds:
9838 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
9842 Use fontconfig to set the font. Note that the colons need to be escaped.
9844 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
9848 Print the date of a real-time encoding (see strftime(3)):
9850 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
9854 Show text fading in and out (appearing/disappearing):
9857 DS=1.0 # display start
9858 DE=10.0 # display end
9859 FID=1.5 # fade in duration
9860 FOD=5 # fade out duration
9861 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 @}"
9865 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
9866 and the @option{fontsize} value are included in the @option{y} offset.
9868 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
9869 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
9874 For more information about libfreetype, check:
9875 @url{http://www.freetype.org/}.
9877 For more information about fontconfig, check:
9878 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
9880 For more information about libfribidi, check:
9881 @url{http://fribidi.org/}.
9885 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
9887 The filter accepts the following options:
9892 Set low and high threshold values used by the Canny thresholding
9895 The high threshold selects the "strong" edge pixels, which are then
9896 connected through 8-connectivity with the "weak" edge pixels selected
9897 by the low threshold.
9899 @var{low} and @var{high} threshold values must be chosen in the range
9900 [0,1], and @var{low} should be lesser or equal to @var{high}.
9902 Default value for @var{low} is @code{20/255}, and default value for @var{high}
9906 Define the drawing mode.
9910 Draw white/gray wires on black background.
9913 Mix the colors to create a paint/cartoon effect.
9916 Apply Canny edge detector on all selected planes.
9918 Default value is @var{wires}.
9921 Select planes for filtering. By default all available planes are filtered.
9924 @subsection Examples
9928 Standard edge detection with custom values for the hysteresis thresholding:
9930 edgedetect=low=0.1:high=0.4
9934 Painting effect without thresholding:
9936 edgedetect=mode=colormix:high=0
9942 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
9944 For each input image, the filter will compute the optimal mapping from
9945 the input to the output given the codebook length, that is the number
9946 of distinct output colors.
9948 This filter accepts the following options.
9951 @item codebook_length, l
9952 Set codebook length. The value must be a positive integer, and
9953 represents the number of distinct output colors. Default value is 256.
9956 Set the maximum number of iterations to apply for computing the optimal
9957 mapping. The higher the value the better the result and the higher the
9958 computation time. Default value is 1.
9961 Set a random seed, must be an integer included between 0 and
9962 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
9963 will try to use a good random seed on a best effort basis.
9966 Set pal8 output pixel format. This option does not work with codebook
9967 length greater than 256.
9972 Measure graylevel entropy in histogram of color channels of video frames.
9974 It accepts the following parameters:
9978 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
9980 @var{diff} mode measures entropy of histogram delta values, absolute differences
9981 between neighbour histogram values.
9985 Set brightness, contrast, saturation and approximate gamma adjustment.
9987 The filter accepts the following options:
9991 Set the contrast expression. The value must be a float value in range
9992 @code{-1000.0} to @code{1000.0}. The default value is "1".
9995 Set the brightness expression. The value must be a float value in
9996 range @code{-1.0} to @code{1.0}. The default value is "0".
9999 Set the saturation expression. The value must be a float in
10000 range @code{0.0} to @code{3.0}. The default value is "1".
10003 Set the gamma expression. The value must be a float in range
10004 @code{0.1} to @code{10.0}. The default value is "1".
10007 Set the gamma expression for red. The value must be a float in
10008 range @code{0.1} to @code{10.0}. The default value is "1".
10011 Set the gamma expression for green. The value must be a float in range
10012 @code{0.1} to @code{10.0}. The default value is "1".
10015 Set the gamma expression for blue. The value must be a float in range
10016 @code{0.1} to @code{10.0}. The default value is "1".
10019 Set the gamma weight expression. It can be used to reduce the effect
10020 of a high gamma value on bright image areas, e.g. keep them from
10021 getting overamplified and just plain white. The value must be a float
10022 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10023 gamma correction all the way down while @code{1.0} leaves it at its
10024 full strength. Default is "1".
10027 Set when the expressions for brightness, contrast, saturation and
10028 gamma expressions are evaluated.
10030 It accepts the following values:
10033 only evaluate expressions once during the filter initialization or
10034 when a command is processed
10037 evaluate expressions for each incoming frame
10040 Default value is @samp{init}.
10043 The expressions accept the following parameters:
10046 frame count of the input frame starting from 0
10049 byte position of the corresponding packet in the input file, NAN if
10053 frame rate of the input video, NAN if the input frame rate is unknown
10056 timestamp expressed in seconds, NAN if the input timestamp is unknown
10059 @subsection Commands
10060 The filter supports the following commands:
10064 Set the contrast expression.
10067 Set the brightness expression.
10070 Set the saturation expression.
10073 Set the gamma expression.
10076 Set the gamma_r expression.
10079 Set gamma_g expression.
10082 Set gamma_b expression.
10085 Set gamma_weight expression.
10087 The command accepts the same syntax of the corresponding option.
10089 If the specified expression is not valid, it is kept at its current
10096 Apply erosion effect to the video.
10098 This filter replaces the pixel by the local(3x3) minimum.
10100 It accepts the following options:
10107 Limit the maximum change for each plane, default is 65535.
10108 If 0, plane will remain unchanged.
10111 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
10114 Flags to local 3x3 coordinates maps like this:
10121 @section extractplanes
10123 Extract color channel components from input video stream into
10124 separate grayscale video streams.
10126 The filter accepts the following option:
10130 Set plane(s) to extract.
10132 Available values for planes are:
10143 Choosing planes not available in the input will result in an error.
10144 That means you cannot select @code{r}, @code{g}, @code{b} planes
10145 with @code{y}, @code{u}, @code{v} planes at same time.
10148 @subsection Examples
10152 Extract luma, u and v color channel component from input video frame
10153 into 3 grayscale outputs:
10155 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
10161 Apply a fade-in/out effect to the input video.
10163 It accepts the following parameters:
10167 The effect type can be either "in" for a fade-in, or "out" for a fade-out
10169 Default is @code{in}.
10171 @item start_frame, s
10172 Specify the number of the frame to start applying the fade
10173 effect at. Default is 0.
10176 The number of frames that the fade effect lasts. At the end of the
10177 fade-in effect, the output video will have the same intensity as the input video.
10178 At the end of the fade-out transition, the output video will be filled with the
10179 selected @option{color}.
10183 If set to 1, fade only alpha channel, if one exists on the input.
10184 Default value is 0.
10186 @item start_time, st
10187 Specify the timestamp (in seconds) of the frame to start to apply the fade
10188 effect. If both start_frame and start_time are specified, the fade will start at
10189 whichever comes last. Default is 0.
10192 The number of seconds for which the fade effect has to last. At the end of the
10193 fade-in effect the output video will have the same intensity as the input video,
10194 at the end of the fade-out transition the output video will be filled with the
10195 selected @option{color}.
10196 If both duration and nb_frames are specified, duration is used. Default is 0
10197 (nb_frames is used by default).
10200 Specify the color of the fade. Default is "black".
10203 @subsection Examples
10207 Fade in the first 30 frames of video:
10212 The command above is equivalent to:
10218 Fade out the last 45 frames of a 200-frame video:
10221 fade=type=out:start_frame=155:nb_frames=45
10225 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
10227 fade=in:0:25, fade=out:975:25
10231 Make the first 5 frames yellow, then fade in from frame 5-24:
10233 fade=in:5:20:color=yellow
10237 Fade in alpha over first 25 frames of video:
10239 fade=in:0:25:alpha=1
10243 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
10245 fade=t=in:st=5.5:d=0.5
10251 Denoise frames using 3D FFT (frequency domain filtering).
10253 The filter accepts the following options:
10257 Set the noise sigma constant. This sets denoising strength.
10258 Default value is 1. Allowed range is from 0 to 30.
10259 Using very high sigma with low overlap may give blocking artifacts.
10262 Set amount of denoising. By default all detected noise is reduced.
10263 Default value is 1. Allowed range is from 0 to 1.
10266 Set size of block, Default is 4, can be 3, 4, 5 or 6.
10267 Actual size of block in pixels is 2 to power of @var{block}, so by default
10268 block size in pixels is 2^4 which is 16.
10271 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
10274 Set number of previous frames to use for denoising. By default is set to 0.
10277 Set number of next frames to to use for denoising. By default is set to 0.
10280 Set planes which will be filtered, by default are all available filtered
10285 Apply arbitrary expressions to samples in frequency domain
10289 Adjust the dc value (gain) of the luma plane of the image. The filter
10290 accepts an integer value in range @code{0} to @code{1000}. The default
10291 value is set to @code{0}.
10294 Adjust the dc value (gain) of the 1st chroma plane of the image. The
10295 filter accepts an integer value in range @code{0} to @code{1000}. The
10296 default value is set to @code{0}.
10299 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
10300 filter accepts an integer value in range @code{0} to @code{1000}. The
10301 default value is set to @code{0}.
10304 Set the frequency domain weight expression for the luma plane.
10307 Set the frequency domain weight expression for the 1st chroma plane.
10310 Set the frequency domain weight expression for the 2nd chroma plane.
10313 Set when the expressions are evaluated.
10315 It accepts the following values:
10318 Only evaluate expressions once during the filter initialization.
10321 Evaluate expressions for each incoming frame.
10324 Default value is @samp{init}.
10326 The filter accepts the following variables:
10329 The coordinates of the current sample.
10333 The width and height of the image.
10336 The number of input frame, starting from 0.
10339 @subsection Examples
10345 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
10351 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
10357 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
10363 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
10370 Extract a single field from an interlaced image using stride
10371 arithmetic to avoid wasting CPU time. The output frames are marked as
10374 The filter accepts the following options:
10378 Specify whether to extract the top (if the value is @code{0} or
10379 @code{top}) or the bottom field (if the value is @code{1} or
10385 Create new frames by copying the top and bottom fields from surrounding frames
10386 supplied as numbers by the hint file.
10390 Set file containing hints: absolute/relative frame numbers.
10392 There must be one line for each frame in a clip. Each line must contain two
10393 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
10394 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
10395 is current frame number for @code{absolute} mode or out of [-1, 1] range
10396 for @code{relative} mode. First number tells from which frame to pick up top
10397 field and second number tells from which frame to pick up bottom field.
10399 If optionally followed by @code{+} output frame will be marked as interlaced,
10400 else if followed by @code{-} output frame will be marked as progressive, else
10401 it will be marked same as input frame.
10402 If optionally followed by @code{t} output frame will use only top field, or in
10403 case of @code{b} it will use only bottom field.
10404 If line starts with @code{#} or @code{;} that line is skipped.
10407 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
10410 Example of first several lines of @code{hint} file for @code{relative} mode:
10412 0,0 - # first frame
10413 1,0 - # second frame, use third's frame top field and second's frame bottom field
10414 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
10429 @section fieldmatch
10431 Field matching filter for inverse telecine. It is meant to reconstruct the
10432 progressive frames from a telecined stream. The filter does not drop duplicated
10433 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
10434 followed by a decimation filter such as @ref{decimate} in the filtergraph.
10436 The separation of the field matching and the decimation is notably motivated by
10437 the possibility of inserting a de-interlacing filter fallback between the two.
10438 If the source has mixed telecined and real interlaced content,
10439 @code{fieldmatch} will not be able to match fields for the interlaced parts.
10440 But these remaining combed frames will be marked as interlaced, and thus can be
10441 de-interlaced by a later filter such as @ref{yadif} before decimation.
10443 In addition to the various configuration options, @code{fieldmatch} can take an
10444 optional second stream, activated through the @option{ppsrc} option. If
10445 enabled, the frames reconstruction will be based on the fields and frames from
10446 this second stream. This allows the first input to be pre-processed in order to
10447 help the various algorithms of the filter, while keeping the output lossless
10448 (assuming the fields are matched properly). Typically, a field-aware denoiser,
10449 or brightness/contrast adjustments can help.
10451 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
10452 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
10453 which @code{fieldmatch} is based on. While the semantic and usage are very
10454 close, some behaviour and options names can differ.
10456 The @ref{decimate} filter currently only works for constant frame rate input.
10457 If your input has mixed telecined (30fps) and progressive content with a lower
10458 framerate like 24fps use the following filterchain to produce the necessary cfr
10459 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
10461 The filter accepts the following options:
10465 Specify the assumed field order of the input stream. Available values are:
10469 Auto detect parity (use FFmpeg's internal parity value).
10471 Assume bottom field first.
10473 Assume top field first.
10476 Note that it is sometimes recommended not to trust the parity announced by the
10479 Default value is @var{auto}.
10482 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
10483 sense that it won't risk creating jerkiness due to duplicate frames when
10484 possible, but if there are bad edits or blended fields it will end up
10485 outputting combed frames when a good match might actually exist. On the other
10486 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
10487 but will almost always find a good frame if there is one. The other values are
10488 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
10489 jerkiness and creating duplicate frames versus finding good matches in sections
10490 with bad edits, orphaned fields, blended fields, etc.
10492 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
10494 Available values are:
10498 2-way matching (p/c)
10500 2-way matching, and trying 3rd match if still combed (p/c + n)
10502 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
10504 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
10505 still combed (p/c + n + u/b)
10507 3-way matching (p/c/n)
10509 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
10510 detected as combed (p/c/n + u/b)
10513 The parenthesis at the end indicate the matches that would be used for that
10514 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
10517 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
10520 Default value is @var{pc_n}.
10523 Mark the main input stream as a pre-processed input, and enable the secondary
10524 input stream as the clean source to pick the fields from. See the filter
10525 introduction for more details. It is similar to the @option{clip2} feature from
10528 Default value is @code{0} (disabled).
10531 Set the field to match from. It is recommended to set this to the same value as
10532 @option{order} unless you experience matching failures with that setting. In
10533 certain circumstances changing the field that is used to match from can have a
10534 large impact on matching performance. Available values are:
10538 Automatic (same value as @option{order}).
10540 Match from the bottom field.
10542 Match from the top field.
10545 Default value is @var{auto}.
10548 Set whether or not chroma is included during the match comparisons. In most
10549 cases it is recommended to leave this enabled. You should set this to @code{0}
10550 only if your clip has bad chroma problems such as heavy rainbowing or other
10551 artifacts. Setting this to @code{0} could also be used to speed things up at
10552 the cost of some accuracy.
10554 Default value is @code{1}.
10558 These define an exclusion band which excludes the lines between @option{y0} and
10559 @option{y1} from being included in the field matching decision. An exclusion
10560 band can be used to ignore subtitles, a logo, or other things that may
10561 interfere with the matching. @option{y0} sets the starting scan line and
10562 @option{y1} sets the ending line; all lines in between @option{y0} and
10563 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
10564 @option{y0} and @option{y1} to the same value will disable the feature.
10565 @option{y0} and @option{y1} defaults to @code{0}.
10568 Set the scene change detection threshold as a percentage of maximum change on
10569 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
10570 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
10571 @option{scthresh} is @code{[0.0, 100.0]}.
10573 Default value is @code{12.0}.
10576 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
10577 account the combed scores of matches when deciding what match to use as the
10578 final match. Available values are:
10582 No final matching based on combed scores.
10584 Combed scores are only used when a scene change is detected.
10586 Use combed scores all the time.
10589 Default is @var{sc}.
10592 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
10593 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
10594 Available values are:
10598 No forced calculation.
10600 Force p/c/n calculations.
10602 Force p/c/n/u/b calculations.
10605 Default value is @var{none}.
10608 This is the area combing threshold used for combed frame detection. This
10609 essentially controls how "strong" or "visible" combing must be to be detected.
10610 Larger values mean combing must be more visible and smaller values mean combing
10611 can be less visible or strong and still be detected. Valid settings are from
10612 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
10613 be detected as combed). This is basically a pixel difference value. A good
10614 range is @code{[8, 12]}.
10616 Default value is @code{9}.
10619 Sets whether or not chroma is considered in the combed frame decision. Only
10620 disable this if your source has chroma problems (rainbowing, etc.) that are
10621 causing problems for the combed frame detection with chroma enabled. Actually,
10622 using @option{chroma}=@var{0} is usually more reliable, except for the case
10623 where there is chroma only combing in the source.
10625 Default value is @code{0}.
10629 Respectively set the x-axis and y-axis size of the window used during combed
10630 frame detection. This has to do with the size of the area in which
10631 @option{combpel} pixels are required to be detected as combed for a frame to be
10632 declared combed. See the @option{combpel} parameter description for more info.
10633 Possible values are any number that is a power of 2 starting at 4 and going up
10636 Default value is @code{16}.
10639 The number of combed pixels inside any of the @option{blocky} by
10640 @option{blockx} size blocks on the frame for the frame to be detected as
10641 combed. While @option{cthresh} controls how "visible" the combing must be, this
10642 setting controls "how much" combing there must be in any localized area (a
10643 window defined by the @option{blockx} and @option{blocky} settings) on the
10644 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
10645 which point no frames will ever be detected as combed). This setting is known
10646 as @option{MI} in TFM/VFM vocabulary.
10648 Default value is @code{80}.
10651 @anchor{p/c/n/u/b meaning}
10652 @subsection p/c/n/u/b meaning
10654 @subsubsection p/c/n
10656 We assume the following telecined stream:
10659 Top fields: 1 2 2 3 4
10660 Bottom fields: 1 2 3 4 4
10663 The numbers correspond to the progressive frame the fields relate to. Here, the
10664 first two frames are progressive, the 3rd and 4th are combed, and so on.
10666 When @code{fieldmatch} is configured to run a matching from bottom
10667 (@option{field}=@var{bottom}) this is how this input stream get transformed:
10672 B 1 2 3 4 4 <-- matching reference
10681 As a result of the field matching, we can see that some frames get duplicated.
10682 To perform a complete inverse telecine, you need to rely on a decimation filter
10683 after this operation. See for instance the @ref{decimate} filter.
10685 The same operation now matching from top fields (@option{field}=@var{top})
10690 T 1 2 2 3 4 <-- matching reference
10700 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
10701 basically, they refer to the frame and field of the opposite parity:
10704 @item @var{p} matches the field of the opposite parity in the previous frame
10705 @item @var{c} matches the field of the opposite parity in the current frame
10706 @item @var{n} matches the field of the opposite parity in the next frame
10711 The @var{u} and @var{b} matching are a bit special in the sense that they match
10712 from the opposite parity flag. In the following examples, we assume that we are
10713 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
10714 'x' is placed above and below each matched fields.
10716 With bottom matching (@option{field}=@var{bottom}):
10721 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10722 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10730 With top matching (@option{field}=@var{top}):
10735 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
10736 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
10744 @subsection Examples
10746 Simple IVTC of a top field first telecined stream:
10748 fieldmatch=order=tff:combmatch=none, decimate
10751 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
10753 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
10756 @section fieldorder
10758 Transform the field order of the input video.
10760 It accepts the following parameters:
10765 The output field order. Valid values are @var{tff} for top field first or @var{bff}
10766 for bottom field first.
10769 The default value is @samp{tff}.
10771 The transformation is done by shifting the picture content up or down
10772 by one line, and filling the remaining line with appropriate picture content.
10773 This method is consistent with most broadcast field order converters.
10775 If the input video is not flagged as being interlaced, or it is already
10776 flagged as being of the required output field order, then this filter does
10777 not alter the incoming video.
10779 It is very useful when converting to or from PAL DV material,
10780 which is bottom field first.
10784 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
10787 @section fifo, afifo
10789 Buffer input images and send them when they are requested.
10791 It is mainly useful when auto-inserted by the libavfilter
10794 It does not take parameters.
10796 @section fillborders
10798 Fill borders of the input video, without changing video stream dimensions.
10799 Sometimes video can have garbage at the four edges and you may not want to
10800 crop video input to keep size multiple of some number.
10802 This filter accepts the following options:
10806 Number of pixels to fill from left border.
10809 Number of pixels to fill from right border.
10812 Number of pixels to fill from top border.
10815 Number of pixels to fill from bottom border.
10820 It accepts the following values:
10823 fill pixels using outermost pixels
10826 fill pixels using mirroring
10829 fill pixels with constant value
10832 Default is @var{smear}.
10835 Set color for pixels in fixed mode. Default is @var{black}.
10838 @subsection Commands
10839 This filter supports same @ref{commands} as options.
10840 The command accepts the same syntax of the corresponding option.
10842 If the specified expression is not valid, it is kept at its current
10847 Find a rectangular object
10849 It accepts the following options:
10853 Filepath of the object image, needs to be in gray8.
10856 Detection threshold, default is 0.5.
10859 Number of mipmaps, default is 3.
10861 @item xmin, ymin, xmax, ymax
10862 Specifies the rectangle in which to search.
10865 @subsection Examples
10869 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
10871 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
10877 Flood area with values of same pixel components with another values.
10879 It accepts the following options:
10882 Set pixel x coordinate.
10885 Set pixel y coordinate.
10888 Set source #0 component value.
10891 Set source #1 component value.
10894 Set source #2 component value.
10897 Set source #3 component value.
10900 Set destination #0 component value.
10903 Set destination #1 component value.
10906 Set destination #2 component value.
10909 Set destination #3 component value.
10915 Convert the input video to one of the specified pixel formats.
10916 Libavfilter will try to pick one that is suitable as input to
10919 It accepts the following parameters:
10923 A '|'-separated list of pixel format names, such as
10924 "pix_fmts=yuv420p|monow|rgb24".
10928 @subsection Examples
10932 Convert the input video to the @var{yuv420p} format
10934 format=pix_fmts=yuv420p
10937 Convert the input video to any of the formats in the list
10939 format=pix_fmts=yuv420p|yuv444p|yuv410p
10946 Convert the video to specified constant frame rate by duplicating or dropping
10947 frames as necessary.
10949 It accepts the following parameters:
10953 The desired output frame rate. The default is @code{25}.
10956 Assume the first PTS should be the given value, in seconds. This allows for
10957 padding/trimming at the start of stream. By default, no assumption is made
10958 about the first frame's expected PTS, so no padding or trimming is done.
10959 For example, this could be set to 0 to pad the beginning with duplicates of
10960 the first frame if a video stream starts after the audio stream or to trim any
10961 frames with a negative PTS.
10964 Timestamp (PTS) rounding method.
10966 Possible values are:
10973 round towards -infinity
10975 round towards +infinity
10979 The default is @code{near}.
10982 Action performed when reading the last frame.
10984 Possible values are:
10987 Use same timestamp rounding method as used for other frames.
10989 Pass through last frame if input duration has not been reached yet.
10991 The default is @code{round}.
10995 Alternatively, the options can be specified as a flat string:
10996 @var{fps}[:@var{start_time}[:@var{round}]].
10998 See also the @ref{setpts} filter.
11000 @subsection Examples
11004 A typical usage in order to set the fps to 25:
11010 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11012 fps=fps=film:round=near
11018 Pack two different video streams into a stereoscopic video, setting proper
11019 metadata on supported codecs. The two views should have the same size and
11020 framerate and processing will stop when the shorter video ends. Please note
11021 that you may conveniently adjust view properties with the @ref{scale} and
11024 It accepts the following parameters:
11028 The desired packing format. Supported values are:
11033 The views are next to each other (default).
11036 The views are on top of each other.
11039 The views are packed by line.
11042 The views are packed by column.
11045 The views are temporally interleaved.
11054 # Convert left and right views into a frame-sequential video
11055 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
11057 # Convert views into a side-by-side video with the same output resolution as the input
11058 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
11063 Change the frame rate by interpolating new video output frames from the source
11066 This filter is not designed to function correctly with interlaced media. If
11067 you wish to change the frame rate of interlaced media then you are required
11068 to deinterlace before this filter and re-interlace after this filter.
11070 A description of the accepted options follows.
11074 Specify the output frames per second. This option can also be specified
11075 as a value alone. The default is @code{50}.
11078 Specify the start of a range where the output frame will be created as a
11079 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11080 the default is @code{15}.
11083 Specify the end of a range where the output frame will be created as a
11084 linear interpolation of two frames. The range is [@code{0}-@code{255}],
11085 the default is @code{240}.
11088 Specify the level at which a scene change is detected as a value between
11089 0 and 100 to indicate a new scene; a low value reflects a low
11090 probability for the current frame to introduce a new scene, while a higher
11091 value means the current frame is more likely to be one.
11092 The default is @code{8.2}.
11095 Specify flags influencing the filter process.
11097 Available value for @var{flags} is:
11100 @item scene_change_detect, scd
11101 Enable scene change detection using the value of the option @var{scene}.
11102 This flag is enabled by default.
11108 Select one frame every N-th frame.
11110 This filter accepts the following option:
11113 Select frame after every @code{step} frames.
11114 Allowed values are positive integers higher than 0. Default value is @code{1}.
11117 @section freezedetect
11119 Detect frozen video.
11121 This filter logs a message and sets frame metadata when it detects that the
11122 input video has no significant change in content during a specified duration.
11123 Video freeze detection calculates the mean average absolute difference of all
11124 the components of video frames and compares it to a noise floor.
11126 The printed times and duration are expressed in seconds. The
11127 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
11128 whose timestamp equals or exceeds the detection duration and it contains the
11129 timestamp of the first frame of the freeze. The
11130 @code{lavfi.freezedetect.freeze_duration} and
11131 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
11134 The filter accepts the following options:
11138 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
11139 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
11143 Set freeze duration until notification (default is 2 seconds).
11149 Apply a frei0r effect to the input video.
11151 To enable the compilation of this filter, you need to install the frei0r
11152 header and configure FFmpeg with @code{--enable-frei0r}.
11154 It accepts the following parameters:
11159 The name of the frei0r effect to load. If the environment variable
11160 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
11161 directories specified by the colon-separated list in @env{FREI0R_PATH}.
11162 Otherwise, the standard frei0r paths are searched, in this order:
11163 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
11164 @file{/usr/lib/frei0r-1/}.
11166 @item filter_params
11167 A '|'-separated list of parameters to pass to the frei0r effect.
11171 A frei0r effect parameter can be a boolean (its value is either
11172 "y" or "n"), a double, a color (specified as
11173 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
11174 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
11175 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
11176 a position (specified as @var{X}/@var{Y}, where
11177 @var{X} and @var{Y} are floating point numbers) and/or a string.
11179 The number and types of parameters depend on the loaded effect. If an
11180 effect parameter is not specified, the default value is set.
11182 @subsection Examples
11186 Apply the distort0r effect, setting the first two double parameters:
11188 frei0r=filter_name=distort0r:filter_params=0.5|0.01
11192 Apply the colordistance effect, taking a color as the first parameter:
11194 frei0r=colordistance:0.2/0.3/0.4
11195 frei0r=colordistance:violet
11196 frei0r=colordistance:0x112233
11200 Apply the perspective effect, specifying the top left and top right image
11203 frei0r=perspective:0.2/0.2|0.8/0.2
11207 For more information, see
11208 @url{http://frei0r.dyne.org}
11212 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
11214 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
11215 processing filter, one of them is performed once per block, not per pixel.
11216 This allows for much higher speed.
11218 The filter accepts the following options:
11222 Set quality. This option defines the number of levels for averaging. It accepts
11223 an integer in the range 4-5. Default value is @code{4}.
11226 Force a constant quantization parameter. It accepts an integer in range 0-63.
11227 If not set, the filter will use the QP from the video stream (if available).
11230 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
11231 more details but also more artifacts, while higher values make the image smoother
11232 but also blurrier. Default value is @code{0} − PSNR optimal.
11234 @item use_bframe_qp
11235 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
11236 option may cause flicker since the B-Frames have often larger QP. Default is
11237 @code{0} (not enabled).
11243 Apply Gaussian blur filter.
11245 The filter accepts the following options:
11249 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
11252 Set number of steps for Gaussian approximation. Default is @code{1}.
11255 Set which planes to filter. By default all planes are filtered.
11258 Set vertical sigma, if negative it will be same as @code{sigma}.
11259 Default is @code{-1}.
11262 @subsection Commands
11263 This filter supports same commands as options.
11264 The command accepts the same syntax of the corresponding option.
11266 If the specified expression is not valid, it is kept at its current
11271 Apply generic equation to each pixel.
11273 The filter accepts the following options:
11276 @item lum_expr, lum
11277 Set the luminance expression.
11279 Set the chrominance blue expression.
11281 Set the chrominance red expression.
11282 @item alpha_expr, a
11283 Set the alpha expression.
11285 Set the red expression.
11286 @item green_expr, g
11287 Set the green expression.
11289 Set the blue expression.
11292 The colorspace is selected according to the specified options. If one
11293 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
11294 options is specified, the filter will automatically select a YCbCr
11295 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
11296 @option{blue_expr} options is specified, it will select an RGB
11299 If one of the chrominance expression is not defined, it falls back on the other
11300 one. If no alpha expression is specified it will evaluate to opaque value.
11301 If none of chrominance expressions are specified, they will evaluate
11302 to the luminance expression.
11304 The expressions can use the following variables and functions:
11308 The sequential number of the filtered frame, starting from @code{0}.
11312 The coordinates of the current sample.
11316 The width and height of the image.
11320 Width and height scale depending on the currently filtered plane. It is the
11321 ratio between the corresponding luma plane number of pixels and the current
11322 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
11323 @code{0.5,0.5} for chroma planes.
11326 Time of the current frame, expressed in seconds.
11329 Return the value of the pixel at location (@var{x},@var{y}) of the current
11333 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
11337 Return the value of the pixel at location (@var{x},@var{y}) of the
11338 blue-difference chroma plane. Return 0 if there is no such plane.
11341 Return the value of the pixel at location (@var{x},@var{y}) of the
11342 red-difference chroma plane. Return 0 if there is no such plane.
11347 Return the value of the pixel at location (@var{x},@var{y}) of the
11348 red/green/blue component. Return 0 if there is no such component.
11351 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
11352 plane. Return 0 if there is no such plane.
11354 @item interpolation
11355 Set one of interpolation methods:
11360 Default is bilinear.
11363 For functions, if @var{x} and @var{y} are outside the area, the value will be
11364 automatically clipped to the closer edge.
11366 @subsection Examples
11370 Flip the image horizontally:
11376 Generate a bidimensional sine wave, with angle @code{PI/3} and a
11377 wavelength of 100 pixels:
11379 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
11383 Generate a fancy enigmatic moving light:
11385 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
11389 Generate a quick emboss effect:
11391 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
11395 Modify RGB components depending on pixel position:
11397 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
11401 Create a radial gradient that is the same size as the input (also see
11402 the @ref{vignette} filter):
11404 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
11410 Fix the banding artifacts that are sometimes introduced into nearly flat
11411 regions by truncation to 8-bit color depth.
11412 Interpolate the gradients that should go where the bands are, and
11415 It is designed for playback only. Do not use it prior to
11416 lossy compression, because compression tends to lose the dither and
11417 bring back the bands.
11419 It accepts the following parameters:
11424 The maximum amount by which the filter will change any one pixel. This is also
11425 the threshold for detecting nearly flat regions. Acceptable values range from
11426 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
11430 The neighborhood to fit the gradient to. A larger radius makes for smoother
11431 gradients, but also prevents the filter from modifying the pixels near detailed
11432 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
11433 values will be clipped to the valid range.
11437 Alternatively, the options can be specified as a flat string:
11438 @var{strength}[:@var{radius}]
11440 @subsection Examples
11444 Apply the filter with a @code{3.5} strength and radius of @code{8}:
11450 Specify radius, omitting the strength (which will fall-back to the default
11458 @anchor{graphmonitor}
11459 @section graphmonitor
11460 Show various filtergraph stats.
11462 With this filter one can debug complete filtergraph.
11463 Especially issues with links filling with queued frames.
11465 The filter accepts the following options:
11469 Set video output size. Default is @var{hd720}.
11472 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
11475 Set output mode, can be @var{fulll} or @var{compact}.
11476 In @var{compact} mode only filters with some queued frames have displayed stats.
11479 Set flags which enable which stats are shown in video.
11481 Available values for flags are:
11484 Display number of queued frames in each link.
11486 @item frame_count_in
11487 Display number of frames taken from filter.
11489 @item frame_count_out
11490 Display number of frames given out from filter.
11493 Display current filtered frame pts.
11496 Display current filtered frame time.
11499 Display time base for filter link.
11502 Display used format for filter link.
11505 Display video size or number of audio channels in case of audio used by filter link.
11508 Display video frame rate or sample rate in case of audio used by filter link.
11512 Set upper limit for video rate of output stream, Default value is @var{25}.
11513 This guarantee that output video frame rate will not be higher than this value.
11517 A color constancy variation filter which estimates scene illumination via grey edge algorithm
11518 and corrects the scene colors accordingly.
11520 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
11522 The filter accepts the following options:
11526 The order of differentiation to be applied on the scene. Must be chosen in the range
11527 [0,2] and default value is 1.
11530 The Minkowski parameter to be used for calculating the Minkowski distance. Must
11531 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
11532 max value instead of calculating Minkowski distance.
11535 The standard deviation of Gaussian blur to be applied on the scene. Must be
11536 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
11537 can't be equal to 0 if @var{difford} is greater than 0.
11540 @subsection Examples
11546 greyedge=difford=1:minknorm=5:sigma=2
11552 greyedge=difford=1:minknorm=0:sigma=2
11560 Apply a Hald CLUT to a video stream.
11562 First input is the video stream to process, and second one is the Hald CLUT.
11563 The Hald CLUT input can be a simple picture or a complete video stream.
11565 The filter accepts the following options:
11569 Force termination when the shortest input terminates. Default is @code{0}.
11571 Continue applying the last CLUT after the end of the stream. A value of
11572 @code{0} disable the filter after the last frame of the CLUT is reached.
11573 Default is @code{1}.
11576 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
11577 filters share the same internals).
11579 This filter also supports the @ref{framesync} options.
11581 More information about the Hald CLUT can be found on Eskil Steenberg's website
11582 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
11584 @subsection Workflow examples
11586 @subsubsection Hald CLUT video stream
11588 Generate an identity Hald CLUT stream altered with various effects:
11590 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
11593 Note: make sure you use a lossless codec.
11595 Then use it with @code{haldclut} to apply it on some random stream:
11597 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
11600 The Hald CLUT will be applied to the 10 first seconds (duration of
11601 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
11602 to the remaining frames of the @code{mandelbrot} stream.
11604 @subsubsection Hald CLUT with preview
11606 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
11607 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
11608 biggest possible square starting at the top left of the picture. The remaining
11609 padding pixels (bottom or right) will be ignored. This area can be used to add
11610 a preview of the Hald CLUT.
11612 Typically, the following generated Hald CLUT will be supported by the
11613 @code{haldclut} filter:
11616 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
11617 pad=iw+320 [padded_clut];
11618 smptebars=s=320x256, split [a][b];
11619 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
11620 [main][b] overlay=W-320" -frames:v 1 clut.png
11623 It contains the original and a preview of the effect of the CLUT: SMPTE color
11624 bars are displayed on the right-top, and below the same color bars processed by
11627 Then, the effect of this Hald CLUT can be visualized with:
11629 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
11634 Flip the input video horizontally.
11636 For example, to horizontally flip the input video with @command{ffmpeg}:
11638 ffmpeg -i in.avi -vf "hflip" out.avi
11642 This filter applies a global color histogram equalization on a
11645 It can be used to correct video that has a compressed range of pixel
11646 intensities. The filter redistributes the pixel intensities to
11647 equalize their distribution across the intensity range. It may be
11648 viewed as an "automatically adjusting contrast filter". This filter is
11649 useful only for correcting degraded or poorly captured source
11652 The filter accepts the following options:
11656 Determine the amount of equalization to be applied. As the strength
11657 is reduced, the distribution of pixel intensities more-and-more
11658 approaches that of the input frame. The value must be a float number
11659 in the range [0,1] and defaults to 0.200.
11662 Set the maximum intensity that can generated and scale the output
11663 values appropriately. The strength should be set as desired and then
11664 the intensity can be limited if needed to avoid washing-out. The value
11665 must be a float number in the range [0,1] and defaults to 0.210.
11668 Set the antibanding level. If enabled the filter will randomly vary
11669 the luminance of output pixels by a small amount to avoid banding of
11670 the histogram. Possible values are @code{none}, @code{weak} or
11671 @code{strong}. It defaults to @code{none}.
11676 Compute and draw a color distribution histogram for the input video.
11678 The computed histogram is a representation of the color component
11679 distribution in an image.
11681 Standard histogram displays the color components distribution in an image.
11682 Displays color graph for each color component. Shows distribution of
11683 the Y, U, V, A or R, G, B components, depending on input format, in the
11684 current frame. Below each graph a color component scale meter is shown.
11686 The filter accepts the following options:
11690 Set height of level. Default value is @code{200}.
11691 Allowed range is [50, 2048].
11694 Set height of color scale. Default value is @code{12}.
11695 Allowed range is [0, 40].
11699 It accepts the following values:
11702 Per color component graphs are placed below each other.
11705 Per color component graphs are placed side by side.
11708 Presents information identical to that in the @code{parade}, except
11709 that the graphs representing color components are superimposed directly
11712 Default is @code{stack}.
11715 Set mode. Can be either @code{linear}, or @code{logarithmic}.
11716 Default is @code{linear}.
11719 Set what color components to display.
11720 Default is @code{7}.
11723 Set foreground opacity. Default is @code{0.7}.
11726 Set background opacity. Default is @code{0.5}.
11729 @subsection Examples
11734 Calculate and draw histogram:
11736 ffplay -i input -vf histogram
11744 This is a high precision/quality 3d denoise filter. It aims to reduce
11745 image noise, producing smooth images and making still images really
11746 still. It should enhance compressibility.
11748 It accepts the following optional parameters:
11752 A non-negative floating point number which specifies spatial luma strength.
11753 It defaults to 4.0.
11755 @item chroma_spatial
11756 A non-negative floating point number which specifies spatial chroma strength.
11757 It defaults to 3.0*@var{luma_spatial}/4.0.
11760 A floating point number which specifies luma temporal strength. It defaults to
11761 6.0*@var{luma_spatial}/4.0.
11764 A floating point number which specifies chroma temporal strength. It defaults to
11765 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
11768 @anchor{hwdownload}
11769 @section hwdownload
11771 Download hardware frames to system memory.
11773 The input must be in hardware frames, and the output a non-hardware format.
11774 Not all formats will be supported on the output - it may be necessary to insert
11775 an additional @option{format} filter immediately following in the graph to get
11776 the output in a supported format.
11780 Map hardware frames to system memory or to another device.
11782 This filter has several different modes of operation; which one is used depends
11783 on the input and output formats:
11786 Hardware frame input, normal frame output
11788 Map the input frames to system memory and pass them to the output. If the
11789 original hardware frame is later required (for example, after overlaying
11790 something else on part of it), the @option{hwmap} filter can be used again
11791 in the next mode to retrieve it.
11793 Normal frame input, hardware frame output
11795 If the input is actually a software-mapped hardware frame, then unmap it -
11796 that is, return the original hardware frame.
11798 Otherwise, a device must be provided. Create new hardware surfaces on that
11799 device for the output, then map them back to the software format at the input
11800 and give those frames to the preceding filter. This will then act like the
11801 @option{hwupload} filter, but may be able to avoid an additional copy when
11802 the input is already in a compatible format.
11804 Hardware frame input and output
11806 A device must be supplied for the output, either directly or with the
11807 @option{derive_device} option. The input and output devices must be of
11808 different types and compatible - the exact meaning of this is
11809 system-dependent, but typically it means that they must refer to the same
11810 underlying hardware context (for example, refer to the same graphics card).
11812 If the input frames were originally created on the output device, then unmap
11813 to retrieve the original frames.
11815 Otherwise, map the frames to the output device - create new hardware frames
11816 on the output corresponding to the frames on the input.
11819 The following additional parameters are accepted:
11823 Set the frame mapping mode. Some combination of:
11826 The mapped frame should be readable.
11828 The mapped frame should be writeable.
11830 The mapping will always overwrite the entire frame.
11832 This may improve performance in some cases, as the original contents of the
11833 frame need not be loaded.
11835 The mapping must not involve any copying.
11837 Indirect mappings to copies of frames are created in some cases where either
11838 direct mapping is not possible or it would have unexpected properties.
11839 Setting this flag ensures that the mapping is direct and will fail if that is
11842 Defaults to @var{read+write} if not specified.
11844 @item derive_device @var{type}
11845 Rather than using the device supplied at initialisation, instead derive a new
11846 device of type @var{type} from the device the input frames exist on.
11849 In a hardware to hardware mapping, map in reverse - create frames in the sink
11850 and map them back to the source. This may be necessary in some cases where
11851 a mapping in one direction is required but only the opposite direction is
11852 supported by the devices being used.
11854 This option is dangerous - it may break the preceding filter in undefined
11855 ways if there are any additional constraints on that filter's output.
11856 Do not use it without fully understanding the implications of its use.
11862 Upload system memory frames to hardware surfaces.
11864 The device to upload to must be supplied when the filter is initialised. If
11865 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
11868 @anchor{hwupload_cuda}
11869 @section hwupload_cuda
11871 Upload system memory frames to a CUDA device.
11873 It accepts the following optional parameters:
11877 The number of the CUDA device to use
11882 Apply a high-quality magnification filter designed for pixel art. This filter
11883 was originally created by Maxim Stepin.
11885 It accepts the following option:
11889 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
11890 @code{hq3x} and @code{4} for @code{hq4x}.
11891 Default is @code{3}.
11895 Stack input videos horizontally.
11897 All streams must be of same pixel format and of same height.
11899 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
11900 to create same output.
11902 The filter accepts the following option:
11906 Set number of input streams. Default is 2.
11909 If set to 1, force the output to terminate when the shortest input
11910 terminates. Default value is 0.
11915 Modify the hue and/or the saturation of the input.
11917 It accepts the following parameters:
11921 Specify the hue angle as a number of degrees. It accepts an expression,
11922 and defaults to "0".
11925 Specify the saturation in the [-10,10] range. It accepts an expression and
11929 Specify the hue angle as a number of radians. It accepts an
11930 expression, and defaults to "0".
11933 Specify the brightness in the [-10,10] range. It accepts an expression and
11937 @option{h} and @option{H} are mutually exclusive, and can't be
11938 specified at the same time.
11940 The @option{b}, @option{h}, @option{H} and @option{s} option values are
11941 expressions containing the following constants:
11945 frame count of the input frame starting from 0
11948 presentation timestamp of the input frame expressed in time base units
11951 frame rate of the input video, NAN if the input frame rate is unknown
11954 timestamp expressed in seconds, NAN if the input timestamp is unknown
11957 time base of the input video
11960 @subsection Examples
11964 Set the hue to 90 degrees and the saturation to 1.0:
11970 Same command but expressing the hue in radians:
11976 Rotate hue and make the saturation swing between 0
11977 and 2 over a period of 1 second:
11979 hue="H=2*PI*t: s=sin(2*PI*t)+1"
11983 Apply a 3 seconds saturation fade-in effect starting at 0:
11985 hue="s=min(t/3\,1)"
11988 The general fade-in expression can be written as:
11990 hue="s=min(0\, max((t-START)/DURATION\, 1))"
11994 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
11996 hue="s=max(0\, min(1\, (8-t)/3))"
11999 The general fade-out expression can be written as:
12001 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
12006 @subsection Commands
12008 This filter supports the following commands:
12014 Modify the hue and/or the saturation and/or brightness of the input video.
12015 The command accepts the same syntax of the corresponding option.
12017 If the specified expression is not valid, it is kept at its current
12021 @section hysteresis
12023 Grow first stream into second stream by connecting components.
12024 This makes it possible to build more robust edge masks.
12026 This filter accepts the following options:
12030 Set which planes will be processed as bitmap, unprocessed planes will be
12031 copied from first stream.
12032 By default value 0xf, all planes will be processed.
12035 Set threshold which is used in filtering. If pixel component value is higher than
12036 this value filter algorithm for connecting components is activated.
12037 By default value is 0.
12042 Detect video interlacing type.
12044 This filter tries to detect if the input frames are interlaced, progressive,
12045 top or bottom field first. It will also try to detect fields that are
12046 repeated between adjacent frames (a sign of telecine).
12048 Single frame detection considers only immediately adjacent frames when classifying each frame.
12049 Multiple frame detection incorporates the classification history of previous frames.
12051 The filter will log these metadata values:
12054 @item single.current_frame
12055 Detected type of current frame using single-frame detection. One of:
12056 ``tff'' (top field first), ``bff'' (bottom field first),
12057 ``progressive'', or ``undetermined''
12060 Cumulative number of frames detected as top field first using single-frame detection.
12063 Cumulative number of frames detected as top field first using multiple-frame detection.
12066 Cumulative number of frames detected as bottom field first using single-frame detection.
12068 @item multiple.current_frame
12069 Detected type of current frame using multiple-frame detection. One of:
12070 ``tff'' (top field first), ``bff'' (bottom field first),
12071 ``progressive'', or ``undetermined''
12074 Cumulative number of frames detected as bottom field first using multiple-frame detection.
12076 @item single.progressive
12077 Cumulative number of frames detected as progressive using single-frame detection.
12079 @item multiple.progressive
12080 Cumulative number of frames detected as progressive using multiple-frame detection.
12082 @item single.undetermined
12083 Cumulative number of frames that could not be classified using single-frame detection.
12085 @item multiple.undetermined
12086 Cumulative number of frames that could not be classified using multiple-frame detection.
12088 @item repeated.current_frame
12089 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
12091 @item repeated.neither
12092 Cumulative number of frames with no repeated field.
12095 Cumulative number of frames with the top field repeated from the previous frame's top field.
12097 @item repeated.bottom
12098 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
12101 The filter accepts the following options:
12105 Set interlacing threshold.
12107 Set progressive threshold.
12109 Threshold for repeated field detection.
12111 Number of frames after which a given frame's contribution to the
12112 statistics is halved (i.e., it contributes only 0.5 to its
12113 classification). The default of 0 means that all frames seen are given
12114 full weight of 1.0 forever.
12115 @item analyze_interlaced_flag
12116 When this is not 0 then idet will use the specified number of frames to determine
12117 if the interlaced flag is accurate, it will not count undetermined frames.
12118 If the flag is found to be accurate it will be used without any further
12119 computations, if it is found to be inaccurate it will be cleared without any
12120 further computations. This allows inserting the idet filter as a low computational
12121 method to clean up the interlaced flag
12126 Deinterleave or interleave fields.
12128 This filter allows one to process interlaced images fields without
12129 deinterlacing them. Deinterleaving splits the input frame into 2
12130 fields (so called half pictures). Odd lines are moved to the top
12131 half of the output image, even lines to the bottom half.
12132 You can process (filter) them independently and then re-interleave them.
12134 The filter accepts the following options:
12138 @item chroma_mode, c
12139 @item alpha_mode, a
12140 Available values for @var{luma_mode}, @var{chroma_mode} and
12141 @var{alpha_mode} are:
12147 @item deinterleave, d
12148 Deinterleave fields, placing one above the other.
12150 @item interleave, i
12151 Interleave fields. Reverse the effect of deinterleaving.
12153 Default value is @code{none}.
12155 @item luma_swap, ls
12156 @item chroma_swap, cs
12157 @item alpha_swap, as
12158 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
12163 Apply inflate effect to the video.
12165 This filter replaces the pixel by the local(3x3) average by taking into account
12166 only values higher than the pixel.
12168 It accepts the following options:
12175 Limit the maximum change for each plane, default is 65535.
12176 If 0, plane will remain unchanged.
12181 Simple interlacing filter from progressive contents. This interleaves upper (or
12182 lower) lines from odd frames with lower (or upper) lines from even frames,
12183 halving the frame rate and preserving image height.
12186 Original Original New Frame
12187 Frame 'j' Frame 'j+1' (tff)
12188 ========== =========== ==================
12189 Line 0 --------------------> Frame 'j' Line 0
12190 Line 1 Line 1 ----> Frame 'j+1' Line 1
12191 Line 2 ---------------------> Frame 'j' Line 2
12192 Line 3 Line 3 ----> Frame 'j+1' Line 3
12194 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
12197 It accepts the following optional parameters:
12201 This determines whether the interlaced frame is taken from the even
12202 (tff - default) or odd (bff) lines of the progressive frame.
12205 Vertical lowpass filter to avoid twitter interlacing and
12206 reduce moire patterns.
12210 Disable vertical lowpass filter
12213 Enable linear filter (default)
12216 Enable complex filter. This will slightly less reduce twitter and moire
12217 but better retain detail and subjective sharpness impression.
12224 Deinterlace input video by applying Donald Graft's adaptive kernel
12225 deinterling. Work on interlaced parts of a video to produce
12226 progressive frames.
12228 The description of the accepted parameters follows.
12232 Set the threshold which affects the filter's tolerance when
12233 determining if a pixel line must be processed. It must be an integer
12234 in the range [0,255] and defaults to 10. A value of 0 will result in
12235 applying the process on every pixels.
12238 Paint pixels exceeding the threshold value to white if set to 1.
12242 Set the fields order. Swap fields if set to 1, leave fields alone if
12246 Enable additional sharpening if set to 1. Default is 0.
12249 Enable twoway sharpening if set to 1. Default is 0.
12252 @subsection Examples
12256 Apply default values:
12258 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
12262 Enable additional sharpening:
12268 Paint processed pixels in white:
12276 Slowly update darker pixels.
12278 This filter makes short flashes of light appear longer.
12279 This filter accepts the following options:
12283 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
12286 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
12289 @section lenscorrection
12291 Correct radial lens distortion
12293 This filter can be used to correct for radial distortion as can result from the use
12294 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
12295 one can use tools available for example as part of opencv or simply trial-and-error.
12296 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
12297 and extract the k1 and k2 coefficients from the resulting matrix.
12299 Note that effectively the same filter is available in the open-source tools Krita and
12300 Digikam from the KDE project.
12302 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
12303 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
12304 brightness distribution, so you may want to use both filters together in certain
12305 cases, though you will have to take care of ordering, i.e. whether vignetting should
12306 be applied before or after lens correction.
12308 @subsection Options
12310 The filter accepts the following options:
12314 Relative x-coordinate of the focal point of the image, and thereby the center of the
12315 distortion. This value has a range [0,1] and is expressed as fractions of the image
12316 width. Default is 0.5.
12318 Relative y-coordinate of the focal point of the image, and thereby the center of the
12319 distortion. This value has a range [0,1] and is expressed as fractions of the image
12320 height. Default is 0.5.
12322 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
12323 no correction. Default is 0.
12325 Coefficient of the double quadratic correction term. This value has a range [-1,1].
12326 0 means no correction. Default is 0.
12329 The formula that generates the correction is:
12331 @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)
12333 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
12334 distances from the focal point in the source and target images, respectively.
12338 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
12340 The @code{lensfun} filter requires the camera make, camera model, and lens model
12341 to apply the lens correction. The filter will load the lensfun database and
12342 query it to find the corresponding camera and lens entries in the database. As
12343 long as these entries can be found with the given options, the filter can
12344 perform corrections on frames. Note that incomplete strings will result in the
12345 filter choosing the best match with the given options, and the filter will
12346 output the chosen camera and lens models (logged with level "info"). You must
12347 provide the make, camera model, and lens model as they are required.
12349 The filter accepts the following options:
12353 The make of the camera (for example, "Canon"). This option is required.
12356 The model of the camera (for example, "Canon EOS 100D"). This option is
12360 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
12361 option is required.
12364 The type of correction to apply. The following values are valid options:
12368 Enables fixing lens vignetting.
12371 Enables fixing lens geometry. This is the default.
12374 Enables fixing chromatic aberrations.
12377 Enables fixing lens vignetting and lens geometry.
12380 Enables fixing lens vignetting and chromatic aberrations.
12383 Enables fixing both lens geometry and chromatic aberrations.
12386 Enables all possible corrections.
12390 The focal length of the image/video (zoom; expected constant for video). For
12391 example, a 18--55mm lens has focal length range of [18--55], so a value in that
12392 range should be chosen when using that lens. Default 18.
12395 The aperture of the image/video (expected constant for video). Note that
12396 aperture is only used for vignetting correction. Default 3.5.
12398 @item focus_distance
12399 The focus distance of the image/video (expected constant for video). Note that
12400 focus distance is only used for vignetting and only slightly affects the
12401 vignetting correction process. If unknown, leave it at the default value (which
12405 The scale factor which is applied after transformation. After correction the
12406 video is no longer necessarily rectangular. This parameter controls how much of
12407 the resulting image is visible. The value 0 means that a value will be chosen
12408 automatically such that there is little or no unmapped area in the output
12409 image. 1.0 means that no additional scaling is done. Lower values may result
12410 in more of the corrected image being visible, while higher values may avoid
12411 unmapped areas in the output.
12413 @item target_geometry
12414 The target geometry of the output image/video. The following values are valid
12418 @item rectilinear (default)
12421 @item equirectangular
12422 @item fisheye_orthographic
12423 @item fisheye_stereographic
12424 @item fisheye_equisolid
12425 @item fisheye_thoby
12428 Apply the reverse of image correction (instead of correcting distortion, apply
12431 @item interpolation
12432 The type of interpolation used when correcting distortion. The following values
12437 @item linear (default)
12442 @subsection Examples
12446 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
12447 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
12451 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
12455 Apply the same as before, but only for the first 5 seconds of video.
12458 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
12465 Obtain the VMAF (Video Multi-Method Assessment Fusion)
12466 score between two input videos.
12468 The obtained VMAF score is printed through the logging system.
12470 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
12471 After installing the library it can be enabled using:
12472 @code{./configure --enable-libvmaf --enable-version3}.
12473 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
12475 The filter has following options:
12479 Set the model path which is to be used for SVM.
12480 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
12483 Set the file path to be used to store logs.
12486 Set the format of the log file (xml or json).
12488 @item enable_transform
12489 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
12490 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
12491 Default value: @code{false}
12494 Invokes the phone model which will generate VMAF scores higher than in the
12495 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
12496 Default value: @code{false}
12499 Enables computing psnr along with vmaf.
12500 Default value: @code{false}
12503 Enables computing ssim along with vmaf.
12504 Default value: @code{false}
12507 Enables computing ms_ssim along with vmaf.
12508 Default value: @code{false}
12511 Set the pool method to be used for computing vmaf.
12512 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
12515 Set number of threads to be used when computing vmaf.
12516 Default value: @code{0}, which makes use of all available logical processors.
12519 Set interval for frame subsampling used when computing vmaf.
12520 Default value: @code{1}
12522 @item enable_conf_interval
12523 Enables confidence interval.
12524 Default value: @code{false}
12527 This filter also supports the @ref{framesync} options.
12529 @subsection Examples
12532 On the below examples the input file @file{main.mpg} being processed is
12533 compared with the reference file @file{ref.mpg}.
12536 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
12540 Example with options:
12542 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
12546 Example with options and different containers:
12548 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]libvmaf=psnr=1:log_fmt=json" -f null -
12554 Limits the pixel components values to the specified range [min, max].
12556 The filter accepts the following options:
12560 Lower bound. Defaults to the lowest allowed value for the input.
12563 Upper bound. Defaults to the highest allowed value for the input.
12566 Specify which planes will be processed. Defaults to all available.
12573 The filter accepts the following options:
12577 Set the number of loops. Setting this value to -1 will result in infinite loops.
12581 Set maximal size in number of frames. Default is 0.
12584 Set first frame of loop. Default is 0.
12587 @subsection Examples
12591 Loop single first frame infinitely:
12593 loop=loop=-1:size=1:start=0
12597 Loop single first frame 10 times:
12599 loop=loop=10:size=1:start=0
12603 Loop 10 first frames 5 times:
12605 loop=loop=5:size=10:start=0
12611 Apply a 1D LUT to an input video.
12613 The filter accepts the following options:
12617 Set the 1D LUT file name.
12619 Currently supported formats:
12628 Select interpolation mode.
12630 Available values are:
12634 Use values from the nearest defined point.
12636 Interpolate values using the linear interpolation.
12638 Interpolate values using the cosine interpolation.
12640 Interpolate values using the cubic interpolation.
12642 Interpolate values using the spline interpolation.
12649 Apply a 3D LUT to an input video.
12651 The filter accepts the following options:
12655 Set the 3D LUT file name.
12657 Currently supported formats:
12671 Select interpolation mode.
12673 Available values are:
12677 Use values from the nearest defined point.
12679 Interpolate values using the 8 points defining a cube.
12681 Interpolate values using a tetrahedron.
12687 Turn certain luma values into transparency.
12689 The filter accepts the following options:
12693 Set the luma which will be used as base for transparency.
12694 Default value is @code{0}.
12697 Set the range of luma values to be keyed out.
12698 Default value is @code{0.01}.
12701 Set the range of softness. Default value is @code{0}.
12702 Use this to control gradual transition from zero to full transparency.
12705 @subsection Commands
12706 This filter supports same @ref{commands} as options.
12707 The command accepts the same syntax of the corresponding option.
12709 If the specified expression is not valid, it is kept at its current
12712 @section lut, lutrgb, lutyuv
12714 Compute a look-up table for binding each pixel component input value
12715 to an output value, and apply it to the input video.
12717 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
12718 to an RGB input video.
12720 These filters accept the following parameters:
12723 set first pixel component expression
12725 set second pixel component expression
12727 set third pixel component expression
12729 set fourth pixel component expression, corresponds to the alpha component
12732 set red component expression
12734 set green component expression
12736 set blue component expression
12738 alpha component expression
12741 set Y/luminance component expression
12743 set U/Cb component expression
12745 set V/Cr component expression
12748 Each of them specifies the expression to use for computing the lookup table for
12749 the corresponding pixel component values.
12751 The exact component associated to each of the @var{c*} options depends on the
12754 The @var{lut} filter requires either YUV or RGB pixel formats in input,
12755 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
12757 The expressions can contain the following constants and functions:
12762 The input width and height.
12765 The input value for the pixel component.
12768 The input value, clipped to the @var{minval}-@var{maxval} range.
12771 The maximum value for the pixel component.
12774 The minimum value for the pixel component.
12777 The negated value for the pixel component value, clipped to the
12778 @var{minval}-@var{maxval} range; it corresponds to the expression
12779 "maxval-clipval+minval".
12782 The computed value in @var{val}, clipped to the
12783 @var{minval}-@var{maxval} range.
12785 @item gammaval(gamma)
12786 The computed gamma correction value of the pixel component value,
12787 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
12789 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
12793 All expressions default to "val".
12795 @subsection Examples
12799 Negate input video:
12801 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
12802 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
12805 The above is the same as:
12807 lutrgb="r=negval:g=negval:b=negval"
12808 lutyuv="y=negval:u=negval:v=negval"
12818 Remove chroma components, turning the video into a graytone image:
12820 lutyuv="u=128:v=128"
12824 Apply a luma burning effect:
12830 Remove green and blue components:
12836 Set a constant alpha channel value on input:
12838 format=rgba,lutrgb=a="maxval-minval/2"
12842 Correct luminance gamma by a factor of 0.5:
12844 lutyuv=y=gammaval(0.5)
12848 Discard least significant bits of luma:
12850 lutyuv=y='bitand(val, 128+64+32)'
12854 Technicolor like effect:
12856 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
12860 @section lut2, tlut2
12862 The @code{lut2} filter takes two input streams and outputs one
12865 The @code{tlut2} (time lut2) filter takes two consecutive frames
12866 from one single stream.
12868 This filter accepts the following parameters:
12871 set first pixel component expression
12873 set second pixel component expression
12875 set third pixel component expression
12877 set fourth pixel component expression, corresponds to the alpha component
12880 set output bit depth, only available for @code{lut2} filter. By default is 0,
12881 which means bit depth is automatically picked from first input format.
12884 Each of them specifies the expression to use for computing the lookup table for
12885 the corresponding pixel component values.
12887 The exact component associated to each of the @var{c*} options depends on the
12890 The expressions can contain the following constants:
12895 The input width and height.
12898 The first input value for the pixel component.
12901 The second input value for the pixel component.
12904 The first input video bit depth.
12907 The second input video bit depth.
12910 All expressions default to "x".
12912 @subsection Examples
12916 Highlight differences between two RGB video streams:
12918 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)'
12922 Highlight differences between two YUV video streams:
12924 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)'
12928 Show max difference between two video streams:
12930 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)))'
12934 @section maskedclamp
12936 Clamp the first input stream with the second input and third input stream.
12938 Returns the value of first stream to be between second input
12939 stream - @code{undershoot} and third input stream + @code{overshoot}.
12941 This filter accepts the following options:
12944 Default value is @code{0}.
12947 Default value is @code{0}.
12950 Set which planes will be processed as bitmap, unprocessed planes will be
12951 copied from first stream.
12952 By default value 0xf, all planes will be processed.
12957 Merge the second and third input stream into output stream using absolute differences
12958 between second input stream and first input stream and absolute difference between
12959 third input stream and first input stream. The picked value will be from second input
12960 stream if second absolute difference is greater than first one or from third input stream
12963 This filter accepts the following options:
12966 Set which planes will be processed as bitmap, unprocessed planes will be
12967 copied from first stream.
12968 By default value 0xf, all planes will be processed.
12971 @section maskedmerge
12973 Merge the first input stream with the second input stream using per pixel
12974 weights in the third input stream.
12976 A value of 0 in the third stream pixel component means that pixel component
12977 from first stream is returned unchanged, while maximum value (eg. 255 for
12978 8-bit videos) means that pixel component from second stream is returned
12979 unchanged. Intermediate values define the amount of merging between both
12980 input stream's pixel components.
12982 This filter accepts the following options:
12985 Set which planes will be processed as bitmap, unprocessed planes will be
12986 copied from first stream.
12987 By default value 0xf, all planes will be processed.
12992 Merge the second and third input stream into output stream using absolute differences
12993 between second input stream and first input stream and absolute difference between
12994 third input stream and first input stream. The picked value will be from second input
12995 stream if second absolute difference is less than first one or from third input stream
12998 This filter accepts the following options:
13001 Set which planes will be processed as bitmap, unprocessed planes will be
13002 copied from first stream.
13003 By default value 0xf, all planes will be processed.
13007 Create mask from input video.
13009 For example it is useful to create motion masks after @code{tblend} filter.
13011 This filter accepts the following options:
13015 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
13018 Set high threshold. Any pixel component higher than this value will be set to max value
13019 allowed for current pixel format.
13022 Set planes to filter, by default all available planes are filtered.
13025 Fill all frame pixels with this value.
13028 Set max average pixel value for frame. If sum of all pixel components is higher that this
13029 average, output frame will be completely filled with value set by @var{fill} option.
13030 Typically useful for scene changes when used in combination with @code{tblend} filter.
13035 Apply motion-compensation deinterlacing.
13037 It needs one field per frame as input and must thus be used together
13038 with yadif=1/3 or equivalent.
13040 This filter accepts the following options:
13043 Set the deinterlacing mode.
13045 It accepts one of the following values:
13050 use iterative motion estimation
13052 like @samp{slow}, but use multiple reference frames.
13054 Default value is @samp{fast}.
13057 Set the picture field parity assumed for the input video. It must be
13058 one of the following values:
13062 assume top field first
13064 assume bottom field first
13067 Default value is @samp{bff}.
13070 Set per-block quantization parameter (QP) used by the internal
13073 Higher values should result in a smoother motion vector field but less
13074 optimal individual vectors. Default value is 1.
13079 Pick median pixel from certain rectangle defined by radius.
13081 This filter accepts the following options:
13085 Set horizontal radius size. Default value is @code{1}.
13086 Allowed range is integer from 1 to 127.
13089 Set which planes to process. Default is @code{15}, which is all available planes.
13092 Set vertical radius size. Default value is @code{0}.
13093 Allowed range is integer from 0 to 127.
13094 If it is 0, value will be picked from horizontal @code{radius} option.
13097 @subsection Commands
13098 This filter supports same @ref{commands} as options.
13099 The command accepts the same syntax of the corresponding option.
13101 If the specified expression is not valid, it is kept at its current
13104 @section mergeplanes
13106 Merge color channel components from several video streams.
13108 The filter accepts up to 4 input streams, and merge selected input
13109 planes to the output video.
13111 This filter accepts the following options:
13114 Set input to output plane mapping. Default is @code{0}.
13116 The mappings is specified as a bitmap. It should be specified as a
13117 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
13118 mapping for the first plane of the output stream. 'A' sets the number of
13119 the input stream to use (from 0 to 3), and 'a' the plane number of the
13120 corresponding input to use (from 0 to 3). The rest of the mappings is
13121 similar, 'Bb' describes the mapping for the output stream second
13122 plane, 'Cc' describes the mapping for the output stream third plane and
13123 'Dd' describes the mapping for the output stream fourth plane.
13126 Set output pixel format. Default is @code{yuva444p}.
13129 @subsection Examples
13133 Merge three gray video streams of same width and height into single video stream:
13135 [a0][a1][a2]mergeplanes=0x001020:yuv444p
13139 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
13141 [a0][a1]mergeplanes=0x00010210:yuva444p
13145 Swap Y and A plane in yuva444p stream:
13147 format=yuva444p,mergeplanes=0x03010200:yuva444p
13151 Swap U and V plane in yuv420p stream:
13153 format=yuv420p,mergeplanes=0x000201:yuv420p
13157 Cast a rgb24 clip to yuv444p:
13159 format=rgb24,mergeplanes=0x000102:yuv444p
13165 Estimate and export motion vectors using block matching algorithms.
13166 Motion vectors are stored in frame side data to be used by other filters.
13168 This filter accepts the following options:
13171 Specify the motion estimation method. Accepts one of the following values:
13175 Exhaustive search algorithm.
13177 Three step search algorithm.
13179 Two dimensional logarithmic search algorithm.
13181 New three step search algorithm.
13183 Four step search algorithm.
13185 Diamond search algorithm.
13187 Hexagon-based search algorithm.
13189 Enhanced predictive zonal search algorithm.
13191 Uneven multi-hexagon search algorithm.
13193 Default value is @samp{esa}.
13196 Macroblock size. Default @code{16}.
13199 Search parameter. Default @code{7}.
13202 @section midequalizer
13204 Apply Midway Image Equalization effect using two video streams.
13206 Midway Image Equalization adjusts a pair of images to have the same
13207 histogram, while maintaining their dynamics as much as possible. It's
13208 useful for e.g. matching exposures from a pair of stereo cameras.
13210 This filter has two inputs and one output, which must be of same pixel format, but
13211 may be of different sizes. The output of filter is first input adjusted with
13212 midway histogram of both inputs.
13214 This filter accepts the following option:
13218 Set which planes to process. Default is @code{15}, which is all available planes.
13221 @section minterpolate
13223 Convert the video to specified frame rate using motion interpolation.
13225 This filter accepts the following options:
13228 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}.
13231 Motion interpolation mode. Following values are accepted:
13234 Duplicate previous or next frame for interpolating new ones.
13236 Blend source frames. Interpolated frame is mean of previous and next frames.
13238 Motion compensated interpolation. Following options are effective when this mode is selected:
13242 Motion compensation mode. Following values are accepted:
13245 Overlapped block motion compensation.
13247 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
13249 Default mode is @samp{obmc}.
13252 Motion estimation mode. Following values are accepted:
13255 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
13257 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
13259 Default mode is @samp{bilat}.
13262 The algorithm to be used for motion estimation. Following values are accepted:
13265 Exhaustive search algorithm.
13267 Three step search algorithm.
13269 Two dimensional logarithmic search algorithm.
13271 New three step search algorithm.
13273 Four step search algorithm.
13275 Diamond search algorithm.
13277 Hexagon-based search algorithm.
13279 Enhanced predictive zonal search algorithm.
13281 Uneven multi-hexagon search algorithm.
13283 Default algorithm is @samp{epzs}.
13286 Macroblock size. Default @code{16}.
13289 Motion estimation search parameter. Default @code{32}.
13292 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).
13297 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:
13300 Disable scene change detection.
13302 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
13304 Default method is @samp{fdiff}.
13306 @item scd_threshold
13307 Scene change detection threshold. Default is @code{5.0}.
13312 Mix several video input streams into one video stream.
13314 A description of the accepted options follows.
13318 The number of inputs. If unspecified, it defaults to 2.
13321 Specify weight of each input video stream as sequence.
13322 Each weight is separated by space. If number of weights
13323 is smaller than number of @var{frames} last specified
13324 weight will be used for all remaining unset weights.
13327 Specify scale, if it is set it will be multiplied with sum
13328 of each weight multiplied with pixel values to give final destination
13329 pixel value. By default @var{scale} is auto scaled to sum of weights.
13332 Specify how end of stream is determined.
13335 The duration of the longest input. (default)
13338 The duration of the shortest input.
13341 The duration of the first input.
13345 @section mpdecimate
13347 Drop frames that do not differ greatly from the previous frame in
13348 order to reduce frame rate.
13350 The main use of this filter is for very-low-bitrate encoding
13351 (e.g. streaming over dialup modem), but it could in theory be used for
13352 fixing movies that were inverse-telecined incorrectly.
13354 A description of the accepted options follows.
13358 Set the maximum number of consecutive frames which can be dropped (if
13359 positive), or the minimum interval between dropped frames (if
13360 negative). If the value is 0, the frame is dropped disregarding the
13361 number of previous sequentially dropped frames.
13363 Default value is 0.
13368 Set the dropping threshold values.
13370 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
13371 represent actual pixel value differences, so a threshold of 64
13372 corresponds to 1 unit of difference for each pixel, or the same spread
13373 out differently over the block.
13375 A frame is a candidate for dropping if no 8x8 blocks differ by more
13376 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
13377 meaning the whole image) differ by more than a threshold of @option{lo}.
13379 Default value for @option{hi} is 64*12, default value for @option{lo} is
13380 64*5, and default value for @option{frac} is 0.33.
13386 Negate (invert) the input video.
13388 It accepts the following option:
13393 With value 1, it negates the alpha component, if present. Default value is 0.
13399 Denoise frames using Non-Local Means algorithm.
13401 Each pixel is adjusted by looking for other pixels with similar contexts. This
13402 context similarity is defined by comparing their surrounding patches of size
13403 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
13406 Note that the research area defines centers for patches, which means some
13407 patches will be made of pixels outside that research area.
13409 The filter accepts the following options.
13413 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
13416 Set patch size. Default is 7. Must be odd number in range [0, 99].
13419 Same as @option{p} but for chroma planes.
13421 The default value is @var{0} and means automatic.
13424 Set research size. Default is 15. Must be odd number in range [0, 99].
13427 Same as @option{r} but for chroma planes.
13429 The default value is @var{0} and means automatic.
13434 Deinterlace video using neural network edge directed interpolation.
13436 This filter accepts the following options:
13440 Mandatory option, without binary file filter can not work.
13441 Currently file can be found here:
13442 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
13445 Set which frames to deinterlace, by default it is @code{all}.
13446 Can be @code{all} or @code{interlaced}.
13449 Set mode of operation.
13451 Can be one of the following:
13455 Use frame flags, both fields.
13457 Use frame flags, single field.
13459 Use top field only.
13461 Use bottom field only.
13463 Use both fields, top first.
13465 Use both fields, bottom first.
13469 Set which planes to process, by default filter process all frames.
13472 Set size of local neighborhood around each pixel, used by the predictor neural
13475 Can be one of the following:
13488 Set the number of neurons in predictor neural network.
13489 Can be one of the following:
13500 Controls the number of different neural network predictions that are blended
13501 together to compute the final output value. Can be @code{fast}, default or
13505 Set which set of weights to use in the predictor.
13506 Can be one of the following:
13510 weights trained to minimize absolute error
13512 weights trained to minimize squared error
13516 Controls whether or not the prescreener neural network is used to decide
13517 which pixels should be processed by the predictor neural network and which
13518 can be handled by simple cubic interpolation.
13519 The prescreener is trained to know whether cubic interpolation will be
13520 sufficient for a pixel or whether it should be predicted by the predictor nn.
13521 The computational complexity of the prescreener nn is much less than that of
13522 the predictor nn. Since most pixels can be handled by cubic interpolation,
13523 using the prescreener generally results in much faster processing.
13524 The prescreener is pretty accurate, so the difference between using it and not
13525 using it is almost always unnoticeable.
13527 Can be one of the following:
13535 Default is @code{new}.
13538 Set various debugging flags.
13543 Force libavfilter not to use any of the specified pixel formats for the
13544 input to the next filter.
13546 It accepts the following parameters:
13550 A '|'-separated list of pixel format names, such as
13551 pix_fmts=yuv420p|monow|rgb24".
13555 @subsection Examples
13559 Force libavfilter to use a format different from @var{yuv420p} for the
13560 input to the vflip filter:
13562 noformat=pix_fmts=yuv420p,vflip
13566 Convert the input video to any of the formats not contained in the list:
13568 noformat=yuv420p|yuv444p|yuv410p
13574 Add noise on video input frame.
13576 The filter accepts the following options:
13584 Set noise seed for specific pixel component or all pixel components in case
13585 of @var{all_seed}. Default value is @code{123457}.
13587 @item all_strength, alls
13588 @item c0_strength, c0s
13589 @item c1_strength, c1s
13590 @item c2_strength, c2s
13591 @item c3_strength, c3s
13592 Set noise strength for specific pixel component or all pixel components in case
13593 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
13595 @item all_flags, allf
13596 @item c0_flags, c0f
13597 @item c1_flags, c1f
13598 @item c2_flags, c2f
13599 @item c3_flags, c3f
13600 Set pixel component flags or set flags for all components if @var{all_flags}.
13601 Available values for component flags are:
13604 averaged temporal noise (smoother)
13606 mix random noise with a (semi)regular pattern
13608 temporal noise (noise pattern changes between frames)
13610 uniform noise (gaussian otherwise)
13614 @subsection Examples
13616 Add temporal and uniform noise to input video:
13618 noise=alls=20:allf=t+u
13623 Normalize RGB video (aka histogram stretching, contrast stretching).
13624 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
13626 For each channel of each frame, the filter computes the input range and maps
13627 it linearly to the user-specified output range. The output range defaults
13628 to the full dynamic range from pure black to pure white.
13630 Temporal smoothing can be used on the input range to reduce flickering (rapid
13631 changes in brightness) caused when small dark or bright objects enter or leave
13632 the scene. This is similar to the auto-exposure (automatic gain control) on a
13633 video camera, and, like a video camera, it may cause a period of over- or
13634 under-exposure of the video.
13636 The R,G,B channels can be normalized independently, which may cause some
13637 color shifting, or linked together as a single channel, which prevents
13638 color shifting. Linked normalization preserves hue. Independent normalization
13639 does not, so it can be used to remove some color casts. Independent and linked
13640 normalization can be combined in any ratio.
13642 The normalize filter accepts the following options:
13647 Colors which define the output range. The minimum input value is mapped to
13648 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
13649 The defaults are black and white respectively. Specifying white for
13650 @var{blackpt} and black for @var{whitept} will give color-inverted,
13651 normalized video. Shades of grey can be used to reduce the dynamic range
13652 (contrast). Specifying saturated colors here can create some interesting
13656 The number of previous frames to use for temporal smoothing. The input range
13657 of each channel is smoothed using a rolling average over the current frame
13658 and the @var{smoothing} previous frames. The default is 0 (no temporal
13662 Controls the ratio of independent (color shifting) channel normalization to
13663 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
13664 independent. Defaults to 1.0 (fully independent).
13667 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
13668 expensive no-op. Defaults to 1.0 (full strength).
13672 @subsection Commands
13673 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
13674 The command accepts the same syntax of the corresponding option.
13676 If the specified expression is not valid, it is kept at its current
13679 @subsection Examples
13681 Stretch video contrast to use the full dynamic range, with no temporal
13682 smoothing; may flicker depending on the source content:
13684 normalize=blackpt=black:whitept=white:smoothing=0
13687 As above, but with 50 frames of temporal smoothing; flicker should be
13688 reduced, depending on the source content:
13690 normalize=blackpt=black:whitept=white:smoothing=50
13693 As above, but with hue-preserving linked channel normalization:
13695 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
13698 As above, but with half strength:
13700 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
13703 Map the darkest input color to red, the brightest input color to cyan:
13705 normalize=blackpt=red:whitept=cyan
13710 Pass the video source unchanged to the output.
13713 Optical Character Recognition
13715 This filter uses Tesseract for optical character recognition. To enable
13716 compilation of this filter, you need to configure FFmpeg with
13717 @code{--enable-libtesseract}.
13719 It accepts the following options:
13723 Set datapath to tesseract data. Default is to use whatever was
13724 set at installation.
13727 Set language, default is "eng".
13730 Set character whitelist.
13733 Set character blacklist.
13736 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
13737 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
13741 Apply a video transform using libopencv.
13743 To enable this filter, install the libopencv library and headers and
13744 configure FFmpeg with @code{--enable-libopencv}.
13746 It accepts the following parameters:
13751 The name of the libopencv filter to apply.
13753 @item filter_params
13754 The parameters to pass to the libopencv filter. If not specified, the default
13755 values are assumed.
13759 Refer to the official libopencv documentation for more precise
13761 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
13763 Several libopencv filters are supported; see the following subsections.
13768 Dilate an image by using a specific structuring element.
13769 It corresponds to the libopencv function @code{cvDilate}.
13771 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
13773 @var{struct_el} represents a structuring element, and has the syntax:
13774 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
13776 @var{cols} and @var{rows} represent the number of columns and rows of
13777 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
13778 point, and @var{shape} the shape for the structuring element. @var{shape}
13779 must be "rect", "cross", "ellipse", or "custom".
13781 If the value for @var{shape} is "custom", it must be followed by a
13782 string of the form "=@var{filename}". The file with name
13783 @var{filename} is assumed to represent a binary image, with each
13784 printable character corresponding to a bright pixel. When a custom
13785 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
13786 or columns and rows of the read file are assumed instead.
13788 The default value for @var{struct_el} is "3x3+0x0/rect".
13790 @var{nb_iterations} specifies the number of times the transform is
13791 applied to the image, and defaults to 1.
13795 # Use the default values
13798 # Dilate using a structuring element with a 5x5 cross, iterating two times
13799 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
13801 # Read the shape from the file diamond.shape, iterating two times.
13802 # The file diamond.shape may contain a pattern of characters like this
13808 # The specified columns and rows are ignored
13809 # but the anchor point coordinates are not
13810 ocv=dilate:0x0+2x2/custom=diamond.shape|2
13815 Erode an image by using a specific structuring element.
13816 It corresponds to the libopencv function @code{cvErode}.
13818 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
13819 with the same syntax and semantics as the @ref{dilate} filter.
13823 Smooth the input video.
13825 The filter takes the following parameters:
13826 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
13828 @var{type} is the type of smooth filter to apply, and must be one of
13829 the following values: "blur", "blur_no_scale", "median", "gaussian",
13830 or "bilateral". The default value is "gaussian".
13832 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
13833 depends on the smooth type. @var{param1} and
13834 @var{param2} accept integer positive values or 0. @var{param3} and
13835 @var{param4} accept floating point values.
13837 The default value for @var{param1} is 3. The default value for the
13838 other parameters is 0.
13840 These parameters correspond to the parameters assigned to the
13841 libopencv function @code{cvSmooth}.
13843 @section oscilloscope
13845 2D Video Oscilloscope.
13847 Useful to measure spatial impulse, step responses, chroma delays, etc.
13849 It accepts the following parameters:
13853 Set scope center x position.
13856 Set scope center y position.
13859 Set scope size, relative to frame diagonal.
13862 Set scope tilt/rotation.
13868 Set trace center x position.
13871 Set trace center y position.
13874 Set trace width, relative to width of frame.
13877 Set trace height, relative to height of frame.
13880 Set which components to trace. By default it traces first three components.
13883 Draw trace grid. By default is enabled.
13886 Draw some statistics. By default is enabled.
13889 Draw scope. By default is enabled.
13892 @subsection Commands
13893 This filter supports same @ref{commands} as options.
13894 The command accepts the same syntax of the corresponding option.
13896 If the specified expression is not valid, it is kept at its current
13899 @subsection Examples
13903 Inspect full first row of video frame.
13905 oscilloscope=x=0.5:y=0:s=1
13909 Inspect full last row of video frame.
13911 oscilloscope=x=0.5:y=1:s=1
13915 Inspect full 5th line of video frame of height 1080.
13917 oscilloscope=x=0.5:y=5/1080:s=1
13921 Inspect full last column of video frame.
13923 oscilloscope=x=1:y=0.5:s=1:t=1
13931 Overlay one video on top of another.
13933 It takes two inputs and has one output. The first input is the "main"
13934 video on which the second input is overlaid.
13936 It accepts the following parameters:
13938 A description of the accepted options follows.
13943 Set the expression for the x and y coordinates of the overlaid video
13944 on the main video. Default value is "0" for both expressions. In case
13945 the expression is invalid, it is set to a huge value (meaning that the
13946 overlay will not be displayed within the output visible area).
13949 See @ref{framesync}.
13952 Set when the expressions for @option{x}, and @option{y} are evaluated.
13954 It accepts the following values:
13957 only evaluate expressions once during the filter initialization or
13958 when a command is processed
13961 evaluate expressions for each incoming frame
13964 Default value is @samp{frame}.
13967 See @ref{framesync}.
13970 Set the format for the output video.
13972 It accepts the following values:
13975 force YUV420 output
13978 force YUV422 output
13981 force YUV444 output
13984 force packed RGB output
13987 force planar RGB output
13990 automatically pick format
13993 Default value is @samp{yuv420}.
13996 See @ref{framesync}.
13999 Set format of alpha of the overlaid video, it can be @var{straight} or
14000 @var{premultiplied}. Default is @var{straight}.
14003 The @option{x}, and @option{y} expressions can contain the following
14009 The main input width and height.
14013 The overlay input width and height.
14017 The computed values for @var{x} and @var{y}. They are evaluated for
14022 horizontal and vertical chroma subsample values of the output
14023 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
14027 the number of input frame, starting from 0
14030 the position in the file of the input frame, NAN if unknown
14033 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
14037 This filter also supports the @ref{framesync} options.
14039 Note that the @var{n}, @var{pos}, @var{t} variables are available only
14040 when evaluation is done @emph{per frame}, and will evaluate to NAN
14041 when @option{eval} is set to @samp{init}.
14043 Be aware that frames are taken from each input video in timestamp
14044 order, hence, if their initial timestamps differ, it is a good idea
14045 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
14046 have them begin in the same zero timestamp, as the example for
14047 the @var{movie} filter does.
14049 You can chain together more overlays but you should test the
14050 efficiency of such approach.
14052 @subsection Commands
14054 This filter supports the following commands:
14058 Modify the x and y of the overlay input.
14059 The command accepts the same syntax of the corresponding option.
14061 If the specified expression is not valid, it is kept at its current
14065 @subsection Examples
14069 Draw the overlay at 10 pixels from the bottom right corner of the main
14072 overlay=main_w-overlay_w-10:main_h-overlay_h-10
14075 Using named options the example above becomes:
14077 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
14081 Insert a transparent PNG logo in the bottom left corner of the input,
14082 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
14084 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
14088 Insert 2 different transparent PNG logos (second logo on bottom
14089 right corner) using the @command{ffmpeg} tool:
14091 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
14095 Add a transparent color layer on top of the main video; @code{WxH}
14096 must specify the size of the main input to the overlay filter:
14098 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
14102 Play an original video and a filtered version (here with the deshake
14103 filter) side by side using the @command{ffplay} tool:
14105 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
14108 The above command is the same as:
14110 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
14114 Make a sliding overlay appearing from the left to the right top part of the
14115 screen starting since time 2:
14117 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
14121 Compose output by putting two input videos side to side:
14123 ffmpeg -i left.avi -i right.avi -filter_complex "
14124 nullsrc=size=200x100 [background];
14125 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
14126 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
14127 [background][left] overlay=shortest=1 [background+left];
14128 [background+left][right] overlay=shortest=1:x=100 [left+right]
14133 Mask 10-20 seconds of a video by applying the delogo filter to a section
14135 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
14136 -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]'
14141 Chain several overlays in cascade:
14143 nullsrc=s=200x200 [bg];
14144 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
14145 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
14146 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
14147 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
14148 [in3] null, [mid2] overlay=100:100 [out0]
14155 Apply Overcomplete Wavelet denoiser.
14157 The filter accepts the following options:
14163 Larger depth values will denoise lower frequency components more, but
14164 slow down filtering.
14166 Must be an int in the range 8-16, default is @code{8}.
14168 @item luma_strength, ls
14171 Must be a double value in the range 0-1000, default is @code{1.0}.
14173 @item chroma_strength, cs
14174 Set chroma strength.
14176 Must be a double value in the range 0-1000, default is @code{1.0}.
14182 Add paddings to the input image, and place the original input at the
14183 provided @var{x}, @var{y} coordinates.
14185 It accepts the following parameters:
14190 Specify an expression for the size of the output image with the
14191 paddings added. If the value for @var{width} or @var{height} is 0, the
14192 corresponding input size is used for the output.
14194 The @var{width} expression can reference the value set by the
14195 @var{height} expression, and vice versa.
14197 The default value of @var{width} and @var{height} is 0.
14201 Specify the offsets to place the input image at within the padded area,
14202 with respect to the top/left border of the output image.
14204 The @var{x} expression can reference the value set by the @var{y}
14205 expression, and vice versa.
14207 The default value of @var{x} and @var{y} is 0.
14209 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
14210 so the input image is centered on the padded area.
14213 Specify the color of the padded area. For the syntax of this option,
14214 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
14215 manual,ffmpeg-utils}.
14217 The default value of @var{color} is "black".
14220 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
14222 It accepts the following values:
14226 Only evaluate expressions once during the filter initialization or when
14227 a command is processed.
14230 Evaluate expressions for each incoming frame.
14234 Default value is @samp{init}.
14237 Pad to aspect instead to a resolution.
14241 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
14242 options are expressions containing the following constants:
14247 The input video width and height.
14251 These are the same as @var{in_w} and @var{in_h}.
14255 The output width and height (the size of the padded area), as
14256 specified by the @var{width} and @var{height} expressions.
14260 These are the same as @var{out_w} and @var{out_h}.
14264 The x and y offsets as specified by the @var{x} and @var{y}
14265 expressions, or NAN if not yet specified.
14268 same as @var{iw} / @var{ih}
14271 input sample aspect ratio
14274 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
14278 The horizontal and vertical chroma subsample values. For example for the
14279 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
14282 @subsection Examples
14286 Add paddings with the color "violet" to the input video. The output video
14287 size is 640x480, and the top-left corner of the input video is placed at
14290 pad=640:480:0:40:violet
14293 The example above is equivalent to the following command:
14295 pad=width=640:height=480:x=0:y=40:color=violet
14299 Pad the input to get an output with dimensions increased by 3/2,
14300 and put the input video at the center of the padded area:
14302 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
14306 Pad the input to get a squared output with size equal to the maximum
14307 value between the input width and height, and put the input video at
14308 the center of the padded area:
14310 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
14314 Pad the input to get a final w/h ratio of 16:9:
14316 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
14320 In case of anamorphic video, in order to set the output display aspect
14321 correctly, it is necessary to use @var{sar} in the expression,
14322 according to the relation:
14324 (ih * X / ih) * sar = output_dar
14325 X = output_dar / sar
14328 Thus the previous example needs to be modified to:
14330 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
14334 Double the output size and put the input video in the bottom-right
14335 corner of the output padded area:
14337 pad="2*iw:2*ih:ow-iw:oh-ih"
14341 @anchor{palettegen}
14342 @section palettegen
14344 Generate one palette for a whole video stream.
14346 It accepts the following options:
14350 Set the maximum number of colors to quantize in the palette.
14351 Note: the palette will still contain 256 colors; the unused palette entries
14354 @item reserve_transparent
14355 Create a palette of 255 colors maximum and reserve the last one for
14356 transparency. Reserving the transparency color is useful for GIF optimization.
14357 If not set, the maximum of colors in the palette will be 256. You probably want
14358 to disable this option for a standalone image.
14361 @item transparency_color
14362 Set the color that will be used as background for transparency.
14365 Set statistics mode.
14367 It accepts the following values:
14370 Compute full frame histograms.
14372 Compute histograms only for the part that differs from previous frame. This
14373 might be relevant to give more importance to the moving part of your input if
14374 the background is static.
14376 Compute new histogram for each frame.
14379 Default value is @var{full}.
14382 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
14383 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
14384 color quantization of the palette. This information is also visible at
14385 @var{info} logging level.
14387 @subsection Examples
14391 Generate a representative palette of a given video using @command{ffmpeg}:
14393 ffmpeg -i input.mkv -vf palettegen palette.png
14397 @section paletteuse
14399 Use a palette to downsample an input video stream.
14401 The filter takes two inputs: one video stream and a palette. The palette must
14402 be a 256 pixels image.
14404 It accepts the following options:
14408 Select dithering mode. Available algorithms are:
14411 Ordered 8x8 bayer dithering (deterministic)
14413 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
14414 Note: this dithering is sometimes considered "wrong" and is included as a
14416 @item floyd_steinberg
14417 Floyd and Steingberg dithering (error diffusion)
14419 Frankie Sierra dithering v2 (error diffusion)
14421 Frankie Sierra dithering v2 "Lite" (error diffusion)
14424 Default is @var{sierra2_4a}.
14427 When @var{bayer} dithering is selected, this option defines the scale of the
14428 pattern (how much the crosshatch pattern is visible). A low value means more
14429 visible pattern for less banding, and higher value means less visible pattern
14430 at the cost of more banding.
14432 The option must be an integer value in the range [0,5]. Default is @var{2}.
14435 If set, define the zone to process
14439 Only the changing rectangle will be reprocessed. This is similar to GIF
14440 cropping/offsetting compression mechanism. This option can be useful for speed
14441 if only a part of the image is changing, and has use cases such as limiting the
14442 scope of the error diffusal @option{dither} to the rectangle that bounds the
14443 moving scene (it leads to more deterministic output if the scene doesn't change
14444 much, and as a result less moving noise and better GIF compression).
14447 Default is @var{none}.
14450 Take new palette for each output frame.
14452 @item alpha_threshold
14453 Sets the alpha threshold for transparency. Alpha values above this threshold
14454 will be treated as completely opaque, and values below this threshold will be
14455 treated as completely transparent.
14457 The option must be an integer value in the range [0,255]. Default is @var{128}.
14460 @subsection Examples
14464 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
14465 using @command{ffmpeg}:
14467 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
14471 @section perspective
14473 Correct perspective of video not recorded perpendicular to the screen.
14475 A description of the accepted parameters follows.
14486 Set coordinates expression for top left, top right, bottom left and bottom right corners.
14487 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
14488 If the @code{sense} option is set to @code{source}, then the specified points will be sent
14489 to the corners of the destination. If the @code{sense} option is set to @code{destination},
14490 then the corners of the source will be sent to the specified coordinates.
14492 The expressions can use the following variables:
14497 the width and height of video frame.
14501 Output frame count.
14504 @item interpolation
14505 Set interpolation for perspective correction.
14507 It accepts the following values:
14513 Default value is @samp{linear}.
14516 Set interpretation of coordinate options.
14518 It accepts the following values:
14522 Send point in the source specified by the given coordinates to
14523 the corners of the destination.
14525 @item 1, destination
14527 Send the corners of the source to the point in the destination specified
14528 by the given coordinates.
14530 Default value is @samp{source}.
14534 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
14536 It accepts the following values:
14539 only evaluate expressions once during the filter initialization or
14540 when a command is processed
14543 evaluate expressions for each incoming frame
14546 Default value is @samp{init}.
14551 Delay interlaced video by one field time so that the field order changes.
14553 The intended use is to fix PAL movies that have been captured with the
14554 opposite field order to the film-to-video transfer.
14556 A description of the accepted parameters follows.
14562 It accepts the following values:
14565 Capture field order top-first, transfer bottom-first.
14566 Filter will delay the bottom field.
14569 Capture field order bottom-first, transfer top-first.
14570 Filter will delay the top field.
14573 Capture and transfer with the same field order. This mode only exists
14574 for the documentation of the other options to refer to, but if you
14575 actually select it, the filter will faithfully do nothing.
14578 Capture field order determined automatically by field flags, transfer
14580 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
14581 basis using field flags. If no field information is available,
14582 then this works just like @samp{u}.
14585 Capture unknown or varying, transfer opposite.
14586 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
14587 analyzing the images and selecting the alternative that produces best
14588 match between the fields.
14591 Capture top-first, transfer unknown or varying.
14592 Filter selects among @samp{t} and @samp{p} using image analysis.
14595 Capture bottom-first, transfer unknown or varying.
14596 Filter selects among @samp{b} and @samp{p} using image analysis.
14599 Capture determined by field flags, transfer unknown or varying.
14600 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
14601 image analysis. If no field information is available, then this works just
14602 like @samp{U}. This is the default mode.
14605 Both capture and transfer unknown or varying.
14606 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
14610 @section photosensitivity
14611 Reduce various flashes in video, so to help users with epilepsy.
14613 It accepts the following options:
14616 Set how many frames to use when filtering. Default is 30.
14619 Set detection threshold factor. Default is 1.
14623 Set how many pixels to skip when sampling frames. Default is 1.
14624 Allowed range is from 1 to 1024.
14627 Leave frames unchanged. Default is disabled.
14630 @section pixdesctest
14632 Pixel format descriptor test filter, mainly useful for internal
14633 testing. The output video should be equal to the input video.
14637 format=monow, pixdesctest
14640 can be used to test the monowhite pixel format descriptor definition.
14644 Display sample values of color channels. Mainly useful for checking color
14645 and levels. Minimum supported resolution is 640x480.
14647 The filters accept the following options:
14651 Set scope X position, relative offset on X axis.
14654 Set scope Y position, relative offset on Y axis.
14663 Set window opacity. This window also holds statistics about pixel area.
14666 Set window X position, relative offset on X axis.
14669 Set window Y position, relative offset on Y axis.
14674 Enable the specified chain of postprocessing subfilters using libpostproc. This
14675 library should be automatically selected with a GPL build (@code{--enable-gpl}).
14676 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
14677 Each subfilter and some options have a short and a long name that can be used
14678 interchangeably, i.e. dr/dering are the same.
14680 The filters accept the following options:
14684 Set postprocessing subfilters string.
14687 All subfilters share common options to determine their scope:
14691 Honor the quality commands for this subfilter.
14694 Do chrominance filtering, too (default).
14697 Do luminance filtering only (no chrominance).
14700 Do chrominance filtering only (no luminance).
14703 These options can be appended after the subfilter name, separated by a '|'.
14705 Available subfilters are:
14708 @item hb/hdeblock[|difference[|flatness]]
14709 Horizontal deblocking filter
14712 Difference factor where higher values mean more deblocking (default: @code{32}).
14714 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14717 @item vb/vdeblock[|difference[|flatness]]
14718 Vertical deblocking filter
14721 Difference factor where higher values mean more deblocking (default: @code{32}).
14723 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14726 @item ha/hadeblock[|difference[|flatness]]
14727 Accurate horizontal deblocking filter
14730 Difference factor where higher values mean more deblocking (default: @code{32}).
14732 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14735 @item va/vadeblock[|difference[|flatness]]
14736 Accurate vertical deblocking filter
14739 Difference factor where higher values mean more deblocking (default: @code{32}).
14741 Flatness threshold where lower values mean more deblocking (default: @code{39}).
14745 The horizontal and vertical deblocking filters share the difference and
14746 flatness values so you cannot set different horizontal and vertical
14750 @item h1/x1hdeblock
14751 Experimental horizontal deblocking filter
14753 @item v1/x1vdeblock
14754 Experimental vertical deblocking filter
14759 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
14762 larger -> stronger filtering
14764 larger -> stronger filtering
14766 larger -> stronger filtering
14769 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
14772 Stretch luminance to @code{0-255}.
14775 @item lb/linblenddeint
14776 Linear blend deinterlacing filter that deinterlaces the given block by
14777 filtering all lines with a @code{(1 2 1)} filter.
14779 @item li/linipoldeint
14780 Linear interpolating deinterlacing filter that deinterlaces the given block by
14781 linearly interpolating every second line.
14783 @item ci/cubicipoldeint
14784 Cubic interpolating deinterlacing filter deinterlaces the given block by
14785 cubically interpolating every second line.
14787 @item md/mediandeint
14788 Median deinterlacing filter that deinterlaces the given block by applying a
14789 median filter to every second line.
14791 @item fd/ffmpegdeint
14792 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
14793 second line with a @code{(-1 4 2 4 -1)} filter.
14796 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
14797 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
14799 @item fq/forceQuant[|quantizer]
14800 Overrides the quantizer table from the input with the constant quantizer you
14808 Default pp filter combination (@code{hb|a,vb|a,dr|a})
14811 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
14814 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
14817 @subsection Examples
14821 Apply horizontal and vertical deblocking, deringing and automatic
14822 brightness/contrast:
14828 Apply default filters without brightness/contrast correction:
14834 Apply default filters and temporal denoiser:
14836 pp=default/tmpnoise|1|2|3
14840 Apply deblocking on luminance only, and switch vertical deblocking on or off
14841 automatically depending on available CPU time:
14848 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
14849 similar to spp = 6 with 7 point DCT, where only the center sample is
14852 The filter accepts the following options:
14856 Force a constant quantization parameter. It accepts an integer in range
14857 0 to 63. If not set, the filter will use the QP from the video stream
14861 Set thresholding mode. Available modes are:
14865 Set hard thresholding.
14867 Set soft thresholding (better de-ringing effect, but likely blurrier).
14869 Set medium thresholding (good results, default).
14873 @section premultiply
14874 Apply alpha premultiply effect to input video stream using first plane
14875 of second stream as alpha.
14877 Both streams must have same dimensions and same pixel format.
14879 The filter accepts the following option:
14883 Set which planes will be processed, unprocessed planes will be copied.
14884 By default value 0xf, all planes will be processed.
14887 Do not require 2nd input for processing, instead use alpha plane from input stream.
14891 Apply prewitt operator to input video stream.
14893 The filter accepts the following option:
14897 Set which planes will be processed, unprocessed planes will be copied.
14898 By default value 0xf, all planes will be processed.
14901 Set value which will be multiplied with filtered result.
14904 Set value which will be added to filtered result.
14907 @anchor{program_opencl}
14908 @section program_opencl
14910 Filter video using an OpenCL program.
14915 OpenCL program source file.
14918 Kernel name in program.
14921 Number of inputs to the filter. Defaults to 1.
14924 Size of output frames. Defaults to the same as the first input.
14928 The program source file must contain a kernel function with the given name,
14929 which will be run once for each plane of the output. Each run on a plane
14930 gets enqueued as a separate 2D global NDRange with one work-item for each
14931 pixel to be generated. The global ID offset for each work-item is therefore
14932 the coordinates of a pixel in the destination image.
14934 The kernel function needs to take the following arguments:
14937 Destination image, @var{__write_only image2d_t}.
14939 This image will become the output; the kernel should write all of it.
14941 Frame index, @var{unsigned int}.
14943 This is a counter starting from zero and increasing by one for each frame.
14945 Source images, @var{__read_only image2d_t}.
14947 These are the most recent images on each input. The kernel may read from
14948 them to generate the output, but they can't be written to.
14955 Copy the input to the output (output must be the same size as the input).
14957 __kernel void copy(__write_only image2d_t destination,
14958 unsigned int index,
14959 __read_only image2d_t source)
14961 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
14963 int2 location = (int2)(get_global_id(0), get_global_id(1));
14965 float4 value = read_imagef(source, sampler, location);
14967 write_imagef(destination, location, value);
14972 Apply a simple transformation, rotating the input by an amount increasing
14973 with the index counter. Pixel values are linearly interpolated by the
14974 sampler, and the output need not have the same dimensions as the input.
14976 __kernel void rotate_image(__write_only image2d_t dst,
14977 unsigned int index,
14978 __read_only image2d_t src)
14980 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
14981 CLK_FILTER_LINEAR);
14983 float angle = (float)index / 100.0f;
14985 float2 dst_dim = convert_float2(get_image_dim(dst));
14986 float2 src_dim = convert_float2(get_image_dim(src));
14988 float2 dst_cen = dst_dim / 2.0f;
14989 float2 src_cen = src_dim / 2.0f;
14991 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
14993 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
14995 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
14996 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
14998 src_pos = src_pos * src_dim / dst_dim;
15000 float2 src_loc = src_pos + src_cen;
15002 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
15003 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
15004 write_imagef(dst, dst_loc, 0.5f);
15006 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
15011 Blend two inputs together, with the amount of each input used varying
15012 with the index counter.
15014 __kernel void blend_images(__write_only image2d_t dst,
15015 unsigned int index,
15016 __read_only image2d_t src1,
15017 __read_only image2d_t src2)
15019 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
15020 CLK_FILTER_LINEAR);
15022 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
15024 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
15025 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
15026 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
15028 float4 val1 = read_imagef(src1, sampler, src1_loc);
15029 float4 val2 = read_imagef(src2, sampler, src2_loc);
15031 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
15037 @section pseudocolor
15039 Alter frame colors in video with pseudocolors.
15041 This filter accepts the following options:
15045 set pixel first component expression
15048 set pixel second component expression
15051 set pixel third component expression
15054 set pixel fourth component expression, corresponds to the alpha component
15057 set component to use as base for altering colors
15060 Each of them specifies the expression to use for computing the lookup table for
15061 the corresponding pixel component values.
15063 The expressions can contain the following constants and functions:
15068 The input width and height.
15071 The input value for the pixel component.
15073 @item ymin, umin, vmin, amin
15074 The minimum allowed component value.
15076 @item ymax, umax, vmax, amax
15077 The maximum allowed component value.
15080 All expressions default to "val".
15082 @subsection Examples
15086 Change too high luma values to gradient:
15088 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'"
15094 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
15095 Ratio) between two input videos.
15097 This filter takes in input two input videos, the first input is
15098 considered the "main" source and is passed unchanged to the
15099 output. The second input is used as a "reference" video for computing
15102 Both video inputs must have the same resolution and pixel format for
15103 this filter to work correctly. Also it assumes that both inputs
15104 have the same number of frames, which are compared one by one.
15106 The obtained average PSNR is printed through the logging system.
15108 The filter stores the accumulated MSE (mean squared error) of each
15109 frame, and at the end of the processing it is averaged across all frames
15110 equally, and the following formula is applied to obtain the PSNR:
15113 PSNR = 10*log10(MAX^2/MSE)
15116 Where MAX is the average of the maximum values of each component of the
15119 The description of the accepted parameters follows.
15122 @item stats_file, f
15123 If specified the filter will use the named file to save the PSNR of
15124 each individual frame. When filename equals "-" the data is sent to
15127 @item stats_version
15128 Specifies which version of the stats file format to use. Details of
15129 each format are written below.
15130 Default value is 1.
15132 @item stats_add_max
15133 Determines whether the max value is output to the stats log.
15134 Default value is 0.
15135 Requires stats_version >= 2. If this is set and stats_version < 2,
15136 the filter will return an error.
15139 This filter also supports the @ref{framesync} options.
15141 The file printed if @var{stats_file} is selected, contains a sequence of
15142 key/value pairs of the form @var{key}:@var{value} for each compared
15145 If a @var{stats_version} greater than 1 is specified, a header line precedes
15146 the list of per-frame-pair stats, with key value pairs following the frame
15147 format with the following parameters:
15150 @item psnr_log_version
15151 The version of the log file format. Will match @var{stats_version}.
15154 A comma separated list of the per-frame-pair parameters included in
15158 A description of each shown per-frame-pair parameter follows:
15162 sequential number of the input frame, starting from 1
15165 Mean Square Error pixel-by-pixel average difference of the compared
15166 frames, averaged over all the image components.
15168 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
15169 Mean Square Error pixel-by-pixel average difference of the compared
15170 frames for the component specified by the suffix.
15172 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
15173 Peak Signal to Noise ratio of the compared frames for the component
15174 specified by the suffix.
15176 @item max_avg, max_y, max_u, max_v
15177 Maximum allowed value for each channel, and average over all
15181 @subsection Examples
15186 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
15187 [main][ref] psnr="stats_file=stats.log" [out]
15190 On this example the input file being processed is compared with the
15191 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
15192 is stored in @file{stats.log}.
15195 Another example with different containers:
15197 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 -
15204 Pulldown reversal (inverse telecine) filter, capable of handling mixed
15205 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
15208 The pullup filter is designed to take advantage of future context in making
15209 its decisions. This filter is stateless in the sense that it does not lock
15210 onto a pattern to follow, but it instead looks forward to the following
15211 fields in order to identify matches and rebuild progressive frames.
15213 To produce content with an even framerate, insert the fps filter after
15214 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
15215 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
15217 The filter accepts the following options:
15224 These options set the amount of "junk" to ignore at the left, right, top, and
15225 bottom of the image, respectively. Left and right are in units of 8 pixels,
15226 while top and bottom are in units of 2 lines.
15227 The default is 8 pixels on each side.
15230 Set the strict breaks. Setting this option to 1 will reduce the chances of
15231 filter generating an occasional mismatched frame, but it may also cause an
15232 excessive number of frames to be dropped during high motion sequences.
15233 Conversely, setting it to -1 will make filter match fields more easily.
15234 This may help processing of video where there is slight blurring between
15235 the fields, but may also cause there to be interlaced frames in the output.
15236 Default value is @code{0}.
15239 Set the metric plane to use. It accepts the following values:
15245 Use chroma blue plane.
15248 Use chroma red plane.
15251 This option may be set to use chroma plane instead of the default luma plane
15252 for doing filter's computations. This may improve accuracy on very clean
15253 source material, but more likely will decrease accuracy, especially if there
15254 is chroma noise (rainbow effect) or any grayscale video.
15255 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
15256 load and make pullup usable in realtime on slow machines.
15259 For best results (without duplicated frames in the output file) it is
15260 necessary to change the output frame rate. For example, to inverse
15261 telecine NTSC input:
15263 ffmpeg -i input -vf pullup -r 24000/1001 ...
15268 Change video quantization parameters (QP).
15270 The filter accepts the following option:
15274 Set expression for quantization parameter.
15277 The expression is evaluated through the eval API and can contain, among others,
15278 the following constants:
15282 1 if index is not 129, 0 otherwise.
15285 Sequential index starting from -129 to 128.
15288 @subsection Examples
15292 Some equation like:
15300 Flush video frames from internal cache of frames into a random order.
15301 No frame is discarded.
15302 Inspired by @ref{frei0r} nervous filter.
15306 Set size in number of frames of internal cache, in range from @code{2} to
15307 @code{512}. Default is @code{30}.
15310 Set seed for random number generator, must be an integer included between
15311 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
15312 less than @code{0}, the filter will try to use a good random seed on a
15316 @section readeia608
15318 Read closed captioning (EIA-608) information from the top lines of a video frame.
15320 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
15321 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
15322 with EIA-608 data (starting from 0). A description of each metadata value follows:
15325 @item lavfi.readeia608.X.cc
15326 The two bytes stored as EIA-608 data (printed in hexadecimal).
15328 @item lavfi.readeia608.X.line
15329 The number of the line on which the EIA-608 data was identified and read.
15332 This filter accepts the following options:
15336 Set the line to start scanning for EIA-608 data. Default is @code{0}.
15339 Set the line to end scanning for EIA-608 data. Default is @code{29}.
15342 Set minimal acceptable amplitude change for sync codes detection.
15343 Default is @code{0.2}. Allowed range is @code{[0.001 - 1]}.
15346 Set the ratio of width reserved for sync code detection.
15347 Default is @code{0.27}. Allowed range is @code{[0.01 - 0.7]}.
15350 Set the max peaks height difference for sync code detection.
15351 Default is @code{0.1}. Allowed range is @code{[0.0 - 0.5]}.
15354 Set max peaks period difference for sync code detection.
15355 Default is @code{0.1}. Allowed range is @code{[0.0 - 0.5]}.
15358 Set the first two max start code bits differences.
15359 Default is @code{0.02}. Allowed range is @code{[0.0 - 0.5]}.
15362 Set the minimum ratio of bits height compared to 3rd start code bit.
15363 Default is @code{0.75}. Allowed range is @code{[0.01 - 1]}.
15366 Set the white color threshold. Default is @code{0.35}. Allowed range is @code{[0.1 - 1]}.
15369 Set the black color threshold. Default is @code{0.15}. Allowed range is @code{[0.0 - 0.5]}.
15372 Enable checking the parity bit. In the event of a parity error, the filter will output
15373 @code{0x00} for that character. Default is false.
15376 Lowpass lines prior to further processing. Default is disabled.
15379 @subsection Examples
15383 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
15385 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
15391 Read vertical interval timecode (VITC) information from the top lines of a
15394 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
15395 timecode value, if a valid timecode has been detected. Further metadata key
15396 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
15397 timecode data has been found or not.
15399 This filter accepts the following options:
15403 Set the maximum number of lines to scan for VITC data. If the value is set to
15404 @code{-1} the full video frame is scanned. Default is @code{45}.
15407 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
15408 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
15411 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
15412 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
15415 @subsection Examples
15419 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
15420 draw @code{--:--:--:--} as a placeholder:
15422 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
15428 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
15430 Destination pixel at position (X, Y) will be picked from source (x, y) position
15431 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
15432 value for pixel will be used for destination pixel.
15434 Xmap and Ymap input video streams must be of same dimensions. Output video stream
15435 will have Xmap/Ymap video stream dimensions.
15436 Xmap and Ymap input video streams are 16bit depth, single channel.
15440 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
15441 Default is @code{color}.
15444 @section removegrain
15446 The removegrain filter is a spatial denoiser for progressive video.
15450 Set mode for the first plane.
15453 Set mode for the second plane.
15456 Set mode for the third plane.
15459 Set mode for the fourth plane.
15462 Range of mode is from 0 to 24. Description of each mode follows:
15466 Leave input plane unchanged. Default.
15469 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
15472 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
15475 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
15478 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
15479 This is equivalent to a median filter.
15482 Line-sensitive clipping giving the minimal change.
15485 Line-sensitive clipping, intermediate.
15488 Line-sensitive clipping, intermediate.
15491 Line-sensitive clipping, intermediate.
15494 Line-sensitive clipping on a line where the neighbours pixels are the closest.
15497 Replaces the target pixel with the closest neighbour.
15500 [1 2 1] horizontal and vertical kernel blur.
15506 Bob mode, interpolates top field from the line where the neighbours
15507 pixels are the closest.
15510 Bob mode, interpolates bottom field from the line where the neighbours
15511 pixels are the closest.
15514 Bob mode, interpolates top field. Same as 13 but with a more complicated
15515 interpolation formula.
15518 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
15519 interpolation formula.
15522 Clips the pixel with the minimum and maximum of respectively the maximum and
15523 minimum of each pair of opposite neighbour pixels.
15526 Line-sensitive clipping using opposite neighbours whose greatest distance from
15527 the current pixel is minimal.
15530 Replaces the pixel with the average of its 8 neighbours.
15533 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
15536 Clips pixels using the averages of opposite neighbour.
15539 Same as mode 21 but simpler and faster.
15542 Small edge and halo removal, but reputed useless.
15548 @section removelogo
15550 Suppress a TV station logo, using an image file to determine which
15551 pixels comprise the logo. It works by filling in the pixels that
15552 comprise the logo with neighboring pixels.
15554 The filter accepts the following options:
15558 Set the filter bitmap file, which can be any image format supported by
15559 libavformat. The width and height of the image file must match those of the
15560 video stream being processed.
15563 Pixels in the provided bitmap image with a value of zero are not
15564 considered part of the logo, non-zero pixels are considered part of
15565 the logo. If you use white (255) for the logo and black (0) for the
15566 rest, you will be safe. For making the filter bitmap, it is
15567 recommended to take a screen capture of a black frame with the logo
15568 visible, and then using a threshold filter followed by the erode
15569 filter once or twice.
15571 If needed, little splotches can be fixed manually. Remember that if
15572 logo pixels are not covered, the filter quality will be much
15573 reduced. Marking too many pixels as part of the logo does not hurt as
15574 much, but it will increase the amount of blurring needed to cover over
15575 the image and will destroy more information than necessary, and extra
15576 pixels will slow things down on a large logo.
15578 @section repeatfields
15580 This filter uses the repeat_field flag from the Video ES headers and hard repeats
15581 fields based on its value.
15585 Reverse a video clip.
15587 Warning: This filter requires memory to buffer the entire clip, so trimming
15590 @subsection Examples
15594 Take the first 5 seconds of a clip, and reverse it.
15601 Shift R/G/B/A pixels horizontally and/or vertically.
15603 The filter accepts the following options:
15606 Set amount to shift red horizontally.
15608 Set amount to shift red vertically.
15610 Set amount to shift green horizontally.
15612 Set amount to shift green vertically.
15614 Set amount to shift blue horizontally.
15616 Set amount to shift blue vertically.
15618 Set amount to shift alpha horizontally.
15620 Set amount to shift alpha vertically.
15622 Set edge mode, can be @var{smear}, default, or @var{warp}.
15625 @subsection Commands
15627 This filter supports the all above options as @ref{commands}.
15630 Apply roberts cross operator to input video stream.
15632 The filter accepts the following option:
15636 Set which planes will be processed, unprocessed planes will be copied.
15637 By default value 0xf, all planes will be processed.
15640 Set value which will be multiplied with filtered result.
15643 Set value which will be added to filtered result.
15648 Rotate video by an arbitrary angle expressed in radians.
15650 The filter accepts the following options:
15652 A description of the optional parameters follows.
15655 Set an expression for the angle by which to rotate the input video
15656 clockwise, expressed as a number of radians. A negative value will
15657 result in a counter-clockwise rotation. By default it is set to "0".
15659 This expression is evaluated for each frame.
15662 Set the output width expression, default value is "iw".
15663 This expression is evaluated just once during configuration.
15666 Set the output height expression, default value is "ih".
15667 This expression is evaluated just once during configuration.
15670 Enable bilinear interpolation if set to 1, a value of 0 disables
15671 it. Default value is 1.
15674 Set the color used to fill the output area not covered by the rotated
15675 image. For the general syntax of this option, check the
15676 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
15677 If the special value "none" is selected then no
15678 background is printed (useful for example if the background is never shown).
15680 Default value is "black".
15683 The expressions for the angle and the output size can contain the
15684 following constants and functions:
15688 sequential number of the input frame, starting from 0. It is always NAN
15689 before the first frame is filtered.
15692 time in seconds of the input frame, it is set to 0 when the filter is
15693 configured. It is always NAN before the first frame is filtered.
15697 horizontal and vertical chroma subsample values. For example for the
15698 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15702 the input video width and height
15706 the output width and height, that is the size of the padded area as
15707 specified by the @var{width} and @var{height} expressions
15711 the minimal width/height required for completely containing the input
15712 video rotated by @var{a} radians.
15714 These are only available when computing the @option{out_w} and
15715 @option{out_h} expressions.
15718 @subsection Examples
15722 Rotate the input by PI/6 radians clockwise:
15728 Rotate the input by PI/6 radians counter-clockwise:
15734 Rotate the input by 45 degrees clockwise:
15740 Apply a constant rotation with period T, starting from an angle of PI/3:
15742 rotate=PI/3+2*PI*t/T
15746 Make the input video rotation oscillating with a period of T
15747 seconds and an amplitude of A radians:
15749 rotate=A*sin(2*PI/T*t)
15753 Rotate the video, output size is chosen so that the whole rotating
15754 input video is always completely contained in the output:
15756 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
15760 Rotate the video, reduce the output size so that no background is ever
15763 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
15767 @subsection Commands
15769 The filter supports the following commands:
15773 Set the angle expression.
15774 The command accepts the same syntax of the corresponding option.
15776 If the specified expression is not valid, it is kept at its current
15782 Apply Shape Adaptive Blur.
15784 The filter accepts the following options:
15787 @item luma_radius, lr
15788 Set luma blur filter strength, must be a value in range 0.1-4.0, default
15789 value is 1.0. A greater value will result in a more blurred image, and
15790 in slower processing.
15792 @item luma_pre_filter_radius, lpfr
15793 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
15796 @item luma_strength, ls
15797 Set luma maximum difference between pixels to still be considered, must
15798 be a value in the 0.1-100.0 range, default value is 1.0.
15800 @item chroma_radius, cr
15801 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
15802 greater value will result in a more blurred image, and in slower
15805 @item chroma_pre_filter_radius, cpfr
15806 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
15808 @item chroma_strength, cs
15809 Set chroma maximum difference between pixels to still be considered,
15810 must be a value in the -0.9-100.0 range.
15813 Each chroma option value, if not explicitly specified, is set to the
15814 corresponding luma option value.
15819 Scale (resize) the input video, using the libswscale library.
15821 The scale filter forces the output display aspect ratio to be the same
15822 of the input, by changing the output sample aspect ratio.
15824 If the input image format is different from the format requested by
15825 the next filter, the scale filter will convert the input to the
15828 @subsection Options
15829 The filter accepts the following options, or any of the options
15830 supported by the libswscale scaler.
15832 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
15833 the complete list of scaler options.
15838 Set the output video dimension expression. Default value is the input
15841 If the @var{width} or @var{w} value is 0, the input width is used for
15842 the output. If the @var{height} or @var{h} value is 0, the input height
15843 is used for the output.
15845 If one and only one of the values is -n with n >= 1, the scale filter
15846 will use a value that maintains the aspect ratio of the input image,
15847 calculated from the other specified dimension. After that it will,
15848 however, make sure that the calculated dimension is divisible by n and
15849 adjust the value if necessary.
15851 If both values are -n with n >= 1, the behavior will be identical to
15852 both values being set to 0 as previously detailed.
15854 See below for the list of accepted constants for use in the dimension
15858 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
15862 Only evaluate expressions once during the filter initialization or when a command is processed.
15865 Evaluate expressions for each incoming frame.
15869 Default value is @samp{init}.
15873 Set the interlacing mode. It accepts the following values:
15877 Force interlaced aware scaling.
15880 Do not apply interlaced scaling.
15883 Select interlaced aware scaling depending on whether the source frames
15884 are flagged as interlaced or not.
15887 Default value is @samp{0}.
15890 Set libswscale scaling flags. See
15891 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15892 complete list of values. If not explicitly specified the filter applies
15896 @item param0, param1
15897 Set libswscale input parameters for scaling algorithms that need them. See
15898 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
15899 complete documentation. If not explicitly specified the filter applies
15905 Set the video size. For the syntax of this option, check the
15906 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
15908 @item in_color_matrix
15909 @item out_color_matrix
15910 Set in/output YCbCr color space type.
15912 This allows the autodetected value to be overridden as well as allows forcing
15913 a specific value used for the output and encoder.
15915 If not specified, the color space type depends on the pixel format.
15921 Choose automatically.
15924 Format conforming to International Telecommunication Union (ITU)
15925 Recommendation BT.709.
15928 Set color space conforming to the United States Federal Communications
15929 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
15934 Set color space conforming to:
15938 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
15941 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
15944 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
15949 Set color space conforming to SMPTE ST 240:1999.
15952 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
15957 Set in/output YCbCr sample range.
15959 This allows the autodetected value to be overridden as well as allows forcing
15960 a specific value used for the output and encoder. If not specified, the
15961 range depends on the pixel format. Possible values:
15965 Choose automatically.
15968 Set full range (0-255 in case of 8-bit luma).
15970 @item mpeg/limited/tv
15971 Set "MPEG" range (16-235 in case of 8-bit luma).
15974 @item force_original_aspect_ratio
15975 Enable decreasing or increasing output video width or height if necessary to
15976 keep the original aspect ratio. Possible values:
15980 Scale the video as specified and disable this feature.
15983 The output video dimensions will automatically be decreased if needed.
15986 The output video dimensions will automatically be increased if needed.
15990 One useful instance of this option is that when you know a specific device's
15991 maximum allowed resolution, you can use this to limit the output video to
15992 that, while retaining the aspect ratio. For example, device A allows
15993 1280x720 playback, and your video is 1920x800. Using this option (set it to
15994 decrease) and specifying 1280x720 to the command line makes the output
15997 Please note that this is a different thing than specifying -1 for @option{w}
15998 or @option{h}, you still need to specify the output resolution for this option
16001 @item force_divisible_by
16002 Ensures that both the output dimensions, width and height, are divisible by the
16003 given integer when used together with @option{force_original_aspect_ratio}. This
16004 works similar to using @code{-n} in the @option{w} and @option{h} options.
16006 This option respects the value set for @option{force_original_aspect_ratio},
16007 increasing or decreasing the resolution accordingly. The video's aspect ratio
16008 may be slightly modified.
16010 This option can be handy if you need to have a video fit within or exceed
16011 a defined resolution using @option{force_original_aspect_ratio} but also have
16012 encoder restrictions on width or height divisibility.
16016 The values of the @option{w} and @option{h} options are expressions
16017 containing the following constants:
16022 The input width and height
16026 These are the same as @var{in_w} and @var{in_h}.
16030 The output (scaled) width and height
16034 These are the same as @var{out_w} and @var{out_h}
16037 The same as @var{iw} / @var{ih}
16040 input sample aspect ratio
16043 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16047 horizontal and vertical input chroma subsample values. For example for the
16048 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16052 horizontal and vertical output chroma subsample values. For example for the
16053 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16056 @subsection Examples
16060 Scale the input video to a size of 200x100
16065 This is equivalent to:
16076 Specify a size abbreviation for the output size:
16081 which can also be written as:
16087 Scale the input to 2x:
16089 scale=w=2*iw:h=2*ih
16093 The above is the same as:
16095 scale=2*in_w:2*in_h
16099 Scale the input to 2x with forced interlaced scaling:
16101 scale=2*iw:2*ih:interl=1
16105 Scale the input to half size:
16107 scale=w=iw/2:h=ih/2
16111 Increase the width, and set the height to the same size:
16117 Seek Greek harmony:
16124 Increase the height, and set the width to 3/2 of the height:
16126 scale=w=3/2*oh:h=3/5*ih
16130 Increase the size, making the size a multiple of the chroma
16133 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
16137 Increase the width to a maximum of 500 pixels,
16138 keeping the same aspect ratio as the input:
16140 scale=w='min(500\, iw*3/2):h=-1'
16144 Make pixels square by combining scale and setsar:
16146 scale='trunc(ih*dar):ih',setsar=1/1
16150 Make pixels square by combining scale and setsar,
16151 making sure the resulting resolution is even (required by some codecs):
16153 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
16157 @subsection Commands
16159 This filter supports the following commands:
16163 Set the output video dimension expression.
16164 The command accepts the same syntax of the corresponding option.
16166 If the specified expression is not valid, it is kept at its current
16172 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
16173 format conversion on CUDA video frames. Setting the output width and height
16174 works in the same way as for the @var{scale} filter.
16176 The following additional options are accepted:
16179 The pixel format of the output CUDA frames. If set to the string "same" (the
16180 default), the input format will be kept. Note that automatic format negotiation
16181 and conversion is not yet supported for hardware frames
16184 The interpolation algorithm used for resizing. One of the following:
16191 @item cubic2p_bspline
16192 2-parameter cubic (B=1, C=0)
16194 @item cubic2p_catmullrom
16195 2-parameter cubic (B=0, C=1/2)
16197 @item cubic2p_b05c03
16198 2-parameter cubic (B=1/2, C=3/10)
16210 Scale (resize) the input video, based on a reference video.
16212 See the scale filter for available options, scale2ref supports the same but
16213 uses the reference video instead of the main input as basis. scale2ref also
16214 supports the following additional constants for the @option{w} and
16215 @option{h} options:
16220 The main input video's width and height
16223 The same as @var{main_w} / @var{main_h}
16226 The main input video's sample aspect ratio
16228 @item main_dar, mdar
16229 The main input video's display aspect ratio. Calculated from
16230 @code{(main_w / main_h) * main_sar}.
16234 The main input video's horizontal and vertical chroma subsample values.
16235 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
16239 @subsection Examples
16243 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
16245 'scale2ref[b][a];[a][b]overlay'
16249 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
16251 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
16256 Scroll input video horizontally and/or vertically by constant speed.
16258 The filter accepts the following options:
16260 @item horizontal, h
16261 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
16262 Negative values changes scrolling direction.
16265 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
16266 Negative values changes scrolling direction.
16269 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
16272 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
16275 @subsection Commands
16277 This filter supports the following @ref{commands}:
16279 @item horizontal, h
16280 Set the horizontal scrolling speed.
16282 Set the vertical scrolling speed.
16285 @anchor{selectivecolor}
16286 @section selectivecolor
16288 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
16289 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
16290 by the "purity" of the color (that is, how saturated it already is).
16292 This filter is similar to the Adobe Photoshop Selective Color tool.
16294 The filter accepts the following options:
16297 @item correction_method
16298 Select color correction method.
16300 Available values are:
16303 Specified adjustments are applied "as-is" (added/subtracted to original pixel
16306 Specified adjustments are relative to the original component value.
16308 Default is @code{absolute}.
16310 Adjustments for red pixels (pixels where the red component is the maximum)
16312 Adjustments for yellow pixels (pixels where the blue component is the minimum)
16314 Adjustments for green pixels (pixels where the green component is the maximum)
16316 Adjustments for cyan pixels (pixels where the red component is the minimum)
16318 Adjustments for blue pixels (pixels where the blue component is the maximum)
16320 Adjustments for magenta pixels (pixels where the green component is the minimum)
16322 Adjustments for white pixels (pixels where all components are greater than 128)
16324 Adjustments for all pixels except pure black and pure white
16326 Adjustments for black pixels (pixels where all components are lesser than 128)
16328 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
16331 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
16332 4 space separated floating point adjustment values in the [-1,1] range,
16333 respectively to adjust the amount of cyan, magenta, yellow and black for the
16334 pixels of its range.
16336 @subsection Examples
16340 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
16341 increase magenta by 27% in blue areas:
16343 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
16347 Use a Photoshop selective color preset:
16349 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
16353 @anchor{separatefields}
16354 @section separatefields
16356 The @code{separatefields} takes a frame-based video input and splits
16357 each frame into its components fields, producing a new half height clip
16358 with twice the frame rate and twice the frame count.
16360 This filter use field-dominance information in frame to decide which
16361 of each pair of fields to place first in the output.
16362 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
16364 @section setdar, setsar
16366 The @code{setdar} filter sets the Display Aspect Ratio for the filter
16369 This is done by changing the specified Sample (aka Pixel) Aspect
16370 Ratio, according to the following equation:
16372 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
16375 Keep in mind that the @code{setdar} filter does not modify the pixel
16376 dimensions of the video frame. Also, the display aspect ratio set by
16377 this filter may be changed by later filters in the filterchain,
16378 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
16381 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
16382 the filter output video.
16384 Note that as a consequence of the application of this filter, the
16385 output display aspect ratio will change according to the equation
16388 Keep in mind that the sample aspect ratio set by the @code{setsar}
16389 filter may be changed by later filters in the filterchain, e.g. if
16390 another "setsar" or a "setdar" filter is applied.
16392 It accepts the following parameters:
16395 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
16396 Set the aspect ratio used by the filter.
16398 The parameter can be a floating point number string, an expression, or
16399 a string of the form @var{num}:@var{den}, where @var{num} and
16400 @var{den} are the numerator and denominator of the aspect ratio. If
16401 the parameter is not specified, it is assumed the value "0".
16402 In case the form "@var{num}:@var{den}" is used, the @code{:} character
16406 Set the maximum integer value to use for expressing numerator and
16407 denominator when reducing the expressed aspect ratio to a rational.
16408 Default value is @code{100}.
16412 The parameter @var{sar} is an expression containing
16413 the following constants:
16417 These are approximated values for the mathematical constants e
16418 (Euler's number), pi (Greek pi), and phi (the golden ratio).
16421 The input width and height.
16424 These are the same as @var{w} / @var{h}.
16427 The input sample aspect ratio.
16430 The input display aspect ratio. It is the same as
16431 (@var{w} / @var{h}) * @var{sar}.
16434 Horizontal and vertical chroma subsample values. For example, for the
16435 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16438 @subsection Examples
16443 To change the display aspect ratio to 16:9, specify one of the following:
16450 To change the sample aspect ratio to 10:11, specify:
16456 To set a display aspect ratio of 16:9, and specify a maximum integer value of
16457 1000 in the aspect ratio reduction, use the command:
16459 setdar=ratio=16/9:max=1000
16467 Force field for the output video frame.
16469 The @code{setfield} filter marks the interlace type field for the
16470 output frames. It does not change the input frame, but only sets the
16471 corresponding property, which affects how the frame is treated by
16472 following filters (e.g. @code{fieldorder} or @code{yadif}).
16474 The filter accepts the following options:
16479 Available values are:
16483 Keep the same field property.
16486 Mark the frame as bottom-field-first.
16489 Mark the frame as top-field-first.
16492 Mark the frame as progressive.
16499 Force frame parameter for the output video frame.
16501 The @code{setparams} filter marks interlace and color range for the
16502 output frames. It does not change the input frame, but only sets the
16503 corresponding property, which affects how the frame is treated by
16508 Available values are:
16512 Keep the same field property (default).
16515 Mark the frame as bottom-field-first.
16518 Mark the frame as top-field-first.
16521 Mark the frame as progressive.
16525 Available values are:
16529 Keep the same color range property (default).
16531 @item unspecified, unknown
16532 Mark the frame as unspecified color range.
16534 @item limited, tv, mpeg
16535 Mark the frame as limited range.
16537 @item full, pc, jpeg
16538 Mark the frame as full range.
16541 @item color_primaries
16542 Set the color primaries.
16543 Available values are:
16547 Keep the same color primaries property (default).
16564 Set the color transfer.
16565 Available values are:
16569 Keep the same color trc property (default).
16591 Set the colorspace.
16592 Available values are:
16596 Keep the same colorspace property (default).
16609 @item chroma-derived-nc
16610 @item chroma-derived-c
16617 Show a line containing various information for each input video frame.
16618 The input video is not modified.
16620 This filter supports the following options:
16624 Calculate checksums of each plane. By default enabled.
16627 The shown line contains a sequence of key/value pairs of the form
16628 @var{key}:@var{value}.
16630 The following values are shown in the output:
16634 The (sequential) number of the input frame, starting from 0.
16637 The Presentation TimeStamp of the input frame, expressed as a number of
16638 time base units. The time base unit depends on the filter input pad.
16641 The Presentation TimeStamp of the input frame, expressed as a number of
16645 The position of the frame in the input stream, or -1 if this information is
16646 unavailable and/or meaningless (for example in case of synthetic video).
16649 The pixel format name.
16652 The sample aspect ratio of the input frame, expressed in the form
16653 @var{num}/@var{den}.
16656 The size of the input frame. For the syntax of this option, check the
16657 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16660 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
16661 for bottom field first).
16664 This is 1 if the frame is a key frame, 0 otherwise.
16667 The picture type of the input frame ("I" for an I-frame, "P" for a
16668 P-frame, "B" for a B-frame, or "?" for an unknown type).
16669 Also refer to the documentation of the @code{AVPictureType} enum and of
16670 the @code{av_get_picture_type_char} function defined in
16671 @file{libavutil/avutil.h}.
16674 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
16676 @item plane_checksum
16677 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
16678 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
16681 @section showpalette
16683 Displays the 256 colors palette of each frame. This filter is only relevant for
16684 @var{pal8} pixel format frames.
16686 It accepts the following option:
16690 Set the size of the box used to represent one palette color entry. Default is
16691 @code{30} (for a @code{30x30} pixel box).
16694 @section shuffleframes
16696 Reorder and/or duplicate and/or drop video frames.
16698 It accepts the following parameters:
16702 Set the destination indexes of input frames.
16703 This is space or '|' separated list of indexes that maps input frames to output
16704 frames. Number of indexes also sets maximal value that each index may have.
16705 '-1' index have special meaning and that is to drop frame.
16708 The first frame has the index 0. The default is to keep the input unchanged.
16710 @subsection Examples
16714 Swap second and third frame of every three frames of the input:
16716 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
16720 Swap 10th and 1st frame of every ten frames of the input:
16722 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
16726 @section shuffleplanes
16728 Reorder and/or duplicate video planes.
16730 It accepts the following parameters:
16735 The index of the input plane to be used as the first output plane.
16738 The index of the input plane to be used as the second output plane.
16741 The index of the input plane to be used as the third output plane.
16744 The index of the input plane to be used as the fourth output plane.
16748 The first plane has the index 0. The default is to keep the input unchanged.
16750 @subsection Examples
16754 Swap the second and third planes of the input:
16756 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
16760 @anchor{signalstats}
16761 @section signalstats
16762 Evaluate various visual metrics that assist in determining issues associated
16763 with the digitization of analog video media.
16765 By default the filter will log these metadata values:
16769 Display the minimal Y value contained within the input frame. Expressed in
16773 Display the Y value at the 10% percentile within the input frame. Expressed in
16777 Display the average Y value within the input frame. Expressed in range of
16781 Display the Y value at the 90% percentile within the input frame. Expressed in
16785 Display the maximum Y value contained within the input frame. Expressed in
16789 Display the minimal U value contained within the input frame. Expressed in
16793 Display the U value at the 10% percentile within the input frame. Expressed in
16797 Display the average U value within the input frame. Expressed in range of
16801 Display the U value at the 90% percentile within the input frame. Expressed in
16805 Display the maximum U value contained within the input frame. Expressed in
16809 Display the minimal V value contained within the input frame. Expressed in
16813 Display the V value at the 10% percentile within the input frame. Expressed in
16817 Display the average V value within the input frame. Expressed in range of
16821 Display the V value at the 90% percentile within the input frame. Expressed in
16825 Display the maximum V value contained within the input frame. Expressed in
16829 Display the minimal saturation value contained within the input frame.
16830 Expressed in range of [0-~181.02].
16833 Display the saturation value at the 10% percentile within the input frame.
16834 Expressed in range of [0-~181.02].
16837 Display the average saturation value within the input frame. Expressed in range
16841 Display the saturation value at the 90% percentile within the input frame.
16842 Expressed in range of [0-~181.02].
16845 Display the maximum saturation value contained within the input frame.
16846 Expressed in range of [0-~181.02].
16849 Display the median value for hue within the input frame. Expressed in range of
16853 Display the average value for hue within the input frame. Expressed in range of
16857 Display the average of sample value difference between all values of the Y
16858 plane in the current frame and corresponding values of the previous input frame.
16859 Expressed in range of [0-255].
16862 Display the average of sample value difference between all values of the U
16863 plane in the current frame and corresponding values of the previous input frame.
16864 Expressed in range of [0-255].
16867 Display the average of sample value difference between all values of the V
16868 plane in the current frame and corresponding values of the previous input frame.
16869 Expressed in range of [0-255].
16872 Display bit depth of Y plane in current frame.
16873 Expressed in range of [0-16].
16876 Display bit depth of U plane in current frame.
16877 Expressed in range of [0-16].
16880 Display bit depth of V plane in current frame.
16881 Expressed in range of [0-16].
16884 The filter accepts the following options:
16890 @option{stat} specify an additional form of image analysis.
16891 @option{out} output video with the specified type of pixel highlighted.
16893 Both options accept the following values:
16897 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
16898 unlike the neighboring pixels of the same field. Examples of temporal outliers
16899 include the results of video dropouts, head clogs, or tape tracking issues.
16902 Identify @var{vertical line repetition}. Vertical line repetition includes
16903 similar rows of pixels within a frame. In born-digital video vertical line
16904 repetition is common, but this pattern is uncommon in video digitized from an
16905 analog source. When it occurs in video that results from the digitization of an
16906 analog source it can indicate concealment from a dropout compensator.
16909 Identify pixels that fall outside of legal broadcast range.
16913 Set the highlight color for the @option{out} option. The default color is
16917 @subsection Examples
16921 Output data of various video metrics:
16923 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
16927 Output specific data about the minimum and maximum values of the Y plane per frame:
16929 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
16933 Playback video while highlighting pixels that are outside of broadcast range in red.
16935 ffplay example.mov -vf signalstats="out=brng:color=red"
16939 Playback video with signalstats metadata drawn over the frame.
16941 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
16944 The contents of signalstat_drawtext.txt used in the command are:
16947 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
16948 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
16949 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
16950 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
16958 Calculates the MPEG-7 Video Signature. The filter can handle more than one
16959 input. In this case the matching between the inputs can be calculated additionally.
16960 The filter always passes through the first input. The signature of each stream can
16961 be written into a file.
16963 It accepts the following options:
16967 Enable or disable the matching process.
16969 Available values are:
16973 Disable the calculation of a matching (default).
16975 Calculate the matching for the whole video and output whether the whole video
16976 matches or only parts.
16978 Calculate only until a matching is found or the video ends. Should be faster in
16983 Set the number of inputs. The option value must be a non negative integer.
16984 Default value is 1.
16987 Set the path to which the output is written. If there is more than one input,
16988 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
16989 integer), that will be replaced with the input number. If no filename is
16990 specified, no output will be written. This is the default.
16993 Choose the output format.
16995 Available values are:
16999 Use the specified binary representation (default).
17001 Use the specified xml representation.
17005 Set threshold to detect one word as similar. The option value must be an integer
17006 greater than zero. The default value is 9000.
17009 Set threshold to detect all words as similar. The option value must be an integer
17010 greater than zero. The default value is 60000.
17013 Set threshold to detect frames as similar. The option value must be an integer
17014 greater than zero. The default value is 116.
17017 Set the minimum length of a sequence in frames to recognize it as matching
17018 sequence. The option value must be a non negative integer value.
17019 The default value is 0.
17022 Set the minimum relation, that matching frames to all frames must have.
17023 The option value must be a double value between 0 and 1. The default value is 0.5.
17026 @subsection Examples
17030 To calculate the signature of an input video and store it in signature.bin:
17032 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
17036 To detect whether two videos match and store the signatures in XML format in
17037 signature0.xml and signature1.xml:
17039 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 -
17047 Blur the input video without impacting the outlines.
17049 It accepts the following options:
17052 @item luma_radius, lr
17053 Set the luma radius. The option value must be a float number in
17054 the range [0.1,5.0] that specifies the variance of the gaussian filter
17055 used to blur the image (slower if larger). Default value is 1.0.
17057 @item luma_strength, ls
17058 Set the luma strength. The option value must be a float number
17059 in the range [-1.0,1.0] that configures the blurring. A value included
17060 in [0.0,1.0] will blur the image whereas a value included in
17061 [-1.0,0.0] will sharpen the image. Default value is 1.0.
17063 @item luma_threshold, lt
17064 Set the luma threshold used as a coefficient to determine
17065 whether a pixel should be blurred or not. The option value must be an
17066 integer in the range [-30,30]. A value of 0 will filter all the image,
17067 a value included in [0,30] will filter flat areas and a value included
17068 in [-30,0] will filter edges. Default value is 0.
17070 @item chroma_radius, cr
17071 Set the chroma radius. The option value must be a float number in
17072 the range [0.1,5.0] that specifies the variance of the gaussian filter
17073 used to blur the image (slower if larger). Default value is @option{luma_radius}.
17075 @item chroma_strength, cs
17076 Set the chroma strength. The option value must be a float number
17077 in the range [-1.0,1.0] that configures the blurring. A value included
17078 in [0.0,1.0] will blur the image whereas a value included in
17079 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
17081 @item chroma_threshold, ct
17082 Set the chroma threshold used as a coefficient to determine
17083 whether a pixel should be blurred or not. The option value must be an
17084 integer in the range [-30,30]. A value of 0 will filter all the image,
17085 a value included in [0,30] will filter flat areas and a value included
17086 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
17089 If a chroma option is not explicitly set, the corresponding luma value
17093 Apply sobel operator to input video stream.
17095 The filter accepts the following option:
17099 Set which planes will be processed, unprocessed planes will be copied.
17100 By default value 0xf, all planes will be processed.
17103 Set value which will be multiplied with filtered result.
17106 Set value which will be added to filtered result.
17112 Apply a simple postprocessing filter that compresses and decompresses the image
17113 at several (or - in the case of @option{quality} level @code{6} - all) shifts
17114 and average the results.
17116 The filter accepts the following options:
17120 Set quality. This option defines the number of levels for averaging. It accepts
17121 an integer in the range 0-6. If set to @code{0}, the filter will have no
17122 effect. A value of @code{6} means the higher quality. For each increment of
17123 that value the speed drops by a factor of approximately 2. Default value is
17127 Force a constant quantization parameter. If not set, the filter will use the QP
17128 from the video stream (if available).
17131 Set thresholding mode. Available modes are:
17135 Set hard thresholding (default).
17137 Set soft thresholding (better de-ringing effect, but likely blurrier).
17140 @item use_bframe_qp
17141 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
17142 option may cause flicker since the B-Frames have often larger QP. Default is
17143 @code{0} (not enabled).
17148 Scale the input by applying one of the super-resolution methods based on
17149 convolutional neural networks. Supported models:
17153 Super-Resolution Convolutional Neural Network model (SRCNN).
17154 See @url{https://arxiv.org/abs/1501.00092}.
17157 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
17158 See @url{https://arxiv.org/abs/1609.05158}.
17161 Training scripts as well as scripts for model file (.pb) saving can be found at
17162 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
17163 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
17165 Native model files (.model) can be generated from TensorFlow model
17166 files (.pb) by using tools/python/convert.py
17168 The filter accepts the following options:
17172 Specify which DNN backend to use for model loading and execution. This option accepts
17173 the following values:
17177 Native implementation of DNN loading and execution.
17180 TensorFlow backend. To enable this backend you
17181 need to install the TensorFlow for C library (see
17182 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
17183 @code{--enable-libtensorflow}
17186 Default value is @samp{native}.
17189 Set path to model file specifying network architecture and its parameters.
17190 Note that different backends use different file formats. TensorFlow backend
17191 can load files for both formats, while native backend can load files for only
17195 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
17196 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
17197 input upscaled using bicubic upscaling with proper scale factor.
17202 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
17204 This filter takes in input two input videos, the first input is
17205 considered the "main" source and is passed unchanged to the
17206 output. The second input is used as a "reference" video for computing
17209 Both video inputs must have the same resolution and pixel format for
17210 this filter to work correctly. Also it assumes that both inputs
17211 have the same number of frames, which are compared one by one.
17213 The filter stores the calculated SSIM of each frame.
17215 The description of the accepted parameters follows.
17218 @item stats_file, f
17219 If specified the filter will use the named file to save the SSIM of
17220 each individual frame. When filename equals "-" the data is sent to
17224 The file printed if @var{stats_file} is selected, contains a sequence of
17225 key/value pairs of the form @var{key}:@var{value} for each compared
17228 A description of each shown parameter follows:
17232 sequential number of the input frame, starting from 1
17234 @item Y, U, V, R, G, B
17235 SSIM of the compared frames for the component specified by the suffix.
17238 SSIM of the compared frames for the whole frame.
17241 Same as above but in dB representation.
17244 This filter also supports the @ref{framesync} options.
17246 @subsection Examples
17251 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
17252 [main][ref] ssim="stats_file=stats.log" [out]
17255 On this example the input file being processed is compared with the
17256 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
17257 is stored in @file{stats.log}.
17260 Another example with both psnr and ssim at same time:
17262 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
17266 Another example with different containers:
17268 ffmpeg -i main.mpg -i ref.mkv -lavfi "[0:v]settb=AVTB,setpts=PTS-STARTPTS[main];[1:v]settb=AVTB,setpts=PTS-STARTPTS[ref];[main][ref]ssim" -f null -
17274 Convert between different stereoscopic image formats.
17276 The filters accept the following options:
17280 Set stereoscopic image format of input.
17282 Available values for input image formats are:
17285 side by side parallel (left eye left, right eye right)
17288 side by side crosseye (right eye left, left eye right)
17291 side by side parallel with half width resolution
17292 (left eye left, right eye right)
17295 side by side crosseye with half width resolution
17296 (right eye left, left eye right)
17300 above-below (left eye above, right eye below)
17304 above-below (right eye above, left eye below)
17308 above-below with half height resolution
17309 (left eye above, right eye below)
17313 above-below with half height resolution
17314 (right eye above, left eye below)
17317 alternating frames (left eye first, right eye second)
17320 alternating frames (right eye first, left eye second)
17323 interleaved rows (left eye has top row, right eye starts on next row)
17326 interleaved rows (right eye has top row, left eye starts on next row)
17329 interleaved columns, left eye first
17332 interleaved columns, right eye first
17334 Default value is @samp{sbsl}.
17338 Set stereoscopic image format of output.
17342 side by side parallel (left eye left, right eye right)
17345 side by side crosseye (right eye left, left eye right)
17348 side by side parallel with half width resolution
17349 (left eye left, right eye right)
17352 side by side crosseye with half width resolution
17353 (right eye left, left eye right)
17357 above-below (left eye above, right eye below)
17361 above-below (right eye above, left eye below)
17365 above-below with half height resolution
17366 (left eye above, right eye below)
17370 above-below with half height resolution
17371 (right eye above, left eye below)
17374 alternating frames (left eye first, right eye second)
17377 alternating frames (right eye first, left eye second)
17380 interleaved rows (left eye has top row, right eye starts on next row)
17383 interleaved rows (right eye has top row, left eye starts on next row)
17386 anaglyph red/blue gray
17387 (red filter on left eye, blue filter on right eye)
17390 anaglyph red/green gray
17391 (red filter on left eye, green filter on right eye)
17394 anaglyph red/cyan gray
17395 (red filter on left eye, cyan filter on right eye)
17398 anaglyph red/cyan half colored
17399 (red filter on left eye, cyan filter on right eye)
17402 anaglyph red/cyan color
17403 (red filter on left eye, cyan filter on right eye)
17406 anaglyph red/cyan color optimized with the least squares projection of dubois
17407 (red filter on left eye, cyan filter on right eye)
17410 anaglyph green/magenta gray
17411 (green filter on left eye, magenta filter on right eye)
17414 anaglyph green/magenta half colored
17415 (green filter on left eye, magenta filter on right eye)
17418 anaglyph green/magenta colored
17419 (green filter on left eye, magenta filter on right eye)
17422 anaglyph green/magenta color optimized with the least squares projection of dubois
17423 (green filter on left eye, magenta filter on right eye)
17426 anaglyph yellow/blue gray
17427 (yellow filter on left eye, blue filter on right eye)
17430 anaglyph yellow/blue half colored
17431 (yellow filter on left eye, blue filter on right eye)
17434 anaglyph yellow/blue colored
17435 (yellow filter on left eye, blue filter on right eye)
17438 anaglyph yellow/blue color optimized with the least squares projection of dubois
17439 (yellow filter on left eye, blue filter on right eye)
17442 mono output (left eye only)
17445 mono output (right eye only)
17448 checkerboard, left eye first
17451 checkerboard, right eye first
17454 interleaved columns, left eye first
17457 interleaved columns, right eye first
17463 Default value is @samp{arcd}.
17466 @subsection Examples
17470 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
17476 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
17482 @section streamselect, astreamselect
17483 Select video or audio streams.
17485 The filter accepts the following options:
17489 Set number of inputs. Default is 2.
17492 Set input indexes to remap to outputs.
17495 @subsection Commands
17497 The @code{streamselect} and @code{astreamselect} filter supports the following
17502 Set input indexes to remap to outputs.
17505 @subsection Examples
17509 Select first 5 seconds 1st stream and rest of time 2nd stream:
17511 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
17515 Same as above, but for audio:
17517 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
17524 Draw subtitles on top of input video using the libass library.
17526 To enable compilation of this filter you need to configure FFmpeg with
17527 @code{--enable-libass}. This filter also requires a build with libavcodec and
17528 libavformat to convert the passed subtitles file to ASS (Advanced Substation
17529 Alpha) subtitles format.
17531 The filter accepts the following options:
17535 Set the filename of the subtitle file to read. It must be specified.
17537 @item original_size
17538 Specify the size of the original video, the video for which the ASS file
17539 was composed. For the syntax of this option, check the
17540 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17541 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
17542 correctly scale the fonts if the aspect ratio has been changed.
17545 Set a directory path containing fonts that can be used by the filter.
17546 These fonts will be used in addition to whatever the font provider uses.
17549 Process alpha channel, by default alpha channel is untouched.
17552 Set subtitles input character encoding. @code{subtitles} filter only. Only
17553 useful if not UTF-8.
17555 @item stream_index, si
17556 Set subtitles stream index. @code{subtitles} filter only.
17559 Override default style or script info parameters of the subtitles. It accepts a
17560 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
17563 If the first key is not specified, it is assumed that the first value
17564 specifies the @option{filename}.
17566 For example, to render the file @file{sub.srt} on top of the input
17567 video, use the command:
17572 which is equivalent to:
17574 subtitles=filename=sub.srt
17577 To render the default subtitles stream from file @file{video.mkv}, use:
17579 subtitles=video.mkv
17582 To render the second subtitles stream from that file, use:
17584 subtitles=video.mkv:si=1
17587 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
17588 @code{DejaVu Serif}, use:
17590 subtitles=sub.srt:force_style='FontName=DejaVu Serif,PrimaryColour=&HCCFF0000'
17593 @section super2xsai
17595 Scale the input by 2x and smooth using the Super2xSaI (Scale and
17596 Interpolate) pixel art scaling algorithm.
17598 Useful for enlarging pixel art images without reducing sharpness.
17602 Swap two rectangular objects in video.
17604 This filter accepts the following options:
17614 Set 1st rect x coordinate.
17617 Set 1st rect y coordinate.
17620 Set 2nd rect x coordinate.
17623 Set 2nd rect y coordinate.
17625 All expressions are evaluated once for each frame.
17628 The all options are expressions containing the following constants:
17633 The input width and height.
17636 same as @var{w} / @var{h}
17639 input sample aspect ratio
17642 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
17645 The number of the input frame, starting from 0.
17648 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
17651 the position in the file of the input frame, NAN if unknown
17659 Apply telecine process to the video.
17661 This filter accepts the following options:
17670 The default value is @code{top}.
17674 A string of numbers representing the pulldown pattern you wish to apply.
17675 The default value is @code{23}.
17679 Some typical patterns:
17684 24p: 2332 (preferred)
17691 24p: 222222222223 ("Euro pulldown")
17698 Apply threshold effect to video stream.
17700 This filter needs four video streams to perform thresholding.
17701 First stream is stream we are filtering.
17702 Second stream is holding threshold values, third stream is holding min values,
17703 and last, fourth stream is holding max values.
17705 The filter accepts the following option:
17709 Set which planes will be processed, unprocessed planes will be copied.
17710 By default value 0xf, all planes will be processed.
17713 For example if first stream pixel's component value is less then threshold value
17714 of pixel component from 2nd threshold stream, third stream value will picked,
17715 otherwise fourth stream pixel component value will be picked.
17717 Using color source filter one can perform various types of thresholding:
17719 @subsection Examples
17723 Binary threshold, using gray color as threshold:
17725 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
17729 Inverted binary threshold, using gray color as threshold:
17731 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
17735 Truncate binary threshold, using gray color as threshold:
17737 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
17741 Threshold to zero, using gray color as threshold:
17743 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
17747 Inverted threshold to zero, using gray color as threshold:
17749 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
17754 Select the most representative frame in a given sequence of consecutive frames.
17756 The filter accepts the following options:
17760 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
17761 will pick one of them, and then handle the next batch of @var{n} frames until
17762 the end. Default is @code{100}.
17765 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
17766 value will result in a higher memory usage, so a high value is not recommended.
17768 @subsection Examples
17772 Extract one picture each 50 frames:
17778 Complete example of a thumbnail creation with @command{ffmpeg}:
17780 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
17786 Tile several successive frames together.
17788 The filter accepts the following options:
17793 Set the grid size (i.e. the number of lines and columns). For the syntax of
17794 this option, check the
17795 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17798 Set the maximum number of frames to render in the given area. It must be less
17799 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
17800 the area will be used.
17803 Set the outer border margin in pixels.
17806 Set the inner border thickness (i.e. the number of pixels between frames). For
17807 more advanced padding options (such as having different values for the edges),
17808 refer to the pad video filter.
17811 Specify the color of the unused area. For the syntax of this option, check the
17812 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
17813 The default value of @var{color} is "black".
17816 Set the number of frames to overlap when tiling several successive frames together.
17817 The value must be between @code{0} and @var{nb_frames - 1}.
17820 Set the number of frames to initially be empty before displaying first output frame.
17821 This controls how soon will one get first output frame.
17822 The value must be between @code{0} and @var{nb_frames - 1}.
17825 @subsection Examples
17829 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
17831 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
17833 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
17834 duplicating each output frame to accommodate the originally detected frame
17838 Display @code{5} pictures in an area of @code{3x2} frames,
17839 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
17840 mixed flat and named options:
17842 tile=3x2:nb_frames=5:padding=7:margin=2
17846 @section tinterlace
17848 Perform various types of temporal field interlacing.
17850 Frames are counted starting from 1, so the first input frame is
17853 The filter accepts the following options:
17858 Specify the mode of the interlacing. This option can also be specified
17859 as a value alone. See below for a list of values for this option.
17861 Available values are:
17865 Move odd frames into the upper field, even into the lower field,
17866 generating a double height frame at half frame rate.
17870 Frame 1 Frame 2 Frame 3 Frame 4
17872 11111 22222 33333 44444
17873 11111 22222 33333 44444
17874 11111 22222 33333 44444
17875 11111 22222 33333 44444
17889 Only output odd frames, even frames are dropped, generating a frame with
17890 unchanged height at half frame rate.
17895 Frame 1 Frame 2 Frame 3 Frame 4
17897 11111 22222 33333 44444
17898 11111 22222 33333 44444
17899 11111 22222 33333 44444
17900 11111 22222 33333 44444
17910 Only output even frames, odd frames are dropped, generating a frame with
17911 unchanged height at half frame rate.
17916 Frame 1 Frame 2 Frame 3 Frame 4
17918 11111 22222 33333 44444
17919 11111 22222 33333 44444
17920 11111 22222 33333 44444
17921 11111 22222 33333 44444
17931 Expand each frame to full height, but pad alternate lines with black,
17932 generating a frame with double height at the same input frame rate.
17937 Frame 1 Frame 2 Frame 3 Frame 4
17939 11111 22222 33333 44444
17940 11111 22222 33333 44444
17941 11111 22222 33333 44444
17942 11111 22222 33333 44444
17945 11111 ..... 33333 .....
17946 ..... 22222 ..... 44444
17947 11111 ..... 33333 .....
17948 ..... 22222 ..... 44444
17949 11111 ..... 33333 .....
17950 ..... 22222 ..... 44444
17951 11111 ..... 33333 .....
17952 ..... 22222 ..... 44444
17956 @item interleave_top, 4
17957 Interleave the upper field from odd frames with the lower field from
17958 even frames, generating a frame with unchanged height at half frame rate.
17963 Frame 1 Frame 2 Frame 3 Frame 4
17965 11111<- 22222 33333<- 44444
17966 11111 22222<- 33333 44444<-
17967 11111<- 22222 33333<- 44444
17968 11111 22222<- 33333 44444<-
17978 @item interleave_bottom, 5
17979 Interleave the lower field from odd frames with the upper field from
17980 even frames, generating a frame with unchanged height at half frame rate.
17985 Frame 1 Frame 2 Frame 3 Frame 4
17987 11111 22222<- 33333 44444<-
17988 11111<- 22222 33333<- 44444
17989 11111 22222<- 33333 44444<-
17990 11111<- 22222 33333<- 44444
18000 @item interlacex2, 6
18001 Double frame rate with unchanged height. Frames are inserted each
18002 containing the second temporal field from the previous input frame and
18003 the first temporal field from the next input frame. This mode relies on
18004 the top_field_first flag. Useful for interlaced video displays with no
18005 field synchronisation.
18010 Frame 1 Frame 2 Frame 3 Frame 4
18012 11111 22222 33333 44444
18013 11111 22222 33333 44444
18014 11111 22222 33333 44444
18015 11111 22222 33333 44444
18018 11111 22222 22222 33333 33333 44444 44444
18019 11111 11111 22222 22222 33333 33333 44444
18020 11111 22222 22222 33333 33333 44444 44444
18021 11111 11111 22222 22222 33333 33333 44444
18026 Move odd frames into the upper field, even into the lower field,
18027 generating a double height frame at same frame rate.
18032 Frame 1 Frame 2 Frame 3 Frame 4
18034 11111 22222 33333 44444
18035 11111 22222 33333 44444
18036 11111 22222 33333 44444
18037 11111 22222 33333 44444
18040 11111 33333 33333 55555
18041 22222 22222 44444 44444
18042 11111 33333 33333 55555
18043 22222 22222 44444 44444
18044 11111 33333 33333 55555
18045 22222 22222 44444 44444
18046 11111 33333 33333 55555
18047 22222 22222 44444 44444
18052 Numeric values are deprecated but are accepted for backward
18053 compatibility reasons.
18055 Default mode is @code{merge}.
18058 Specify flags influencing the filter process.
18060 Available value for @var{flags} is:
18063 @item low_pass_filter, vlpf
18064 Enable linear vertical low-pass filtering in the filter.
18065 Vertical low-pass filtering is required when creating an interlaced
18066 destination from a progressive source which contains high-frequency
18067 vertical detail. Filtering will reduce interlace 'twitter' and Moire
18070 @item complex_filter, cvlpf
18071 Enable complex vertical low-pass filtering.
18072 This will slightly less reduce interlace 'twitter' and Moire
18073 patterning but better retain detail and subjective sharpness impression.
18077 Vertical low-pass filtering can only be enabled for @option{mode}
18078 @var{interleave_top} and @var{interleave_bottom}.
18084 Mix successive video frames.
18086 A description of the accepted options follows.
18090 The number of successive frames to mix. If unspecified, it defaults to 3.
18093 Specify weight of each input video frame.
18094 Each weight is separated by space. If number of weights is smaller than
18095 number of @var{frames} last specified weight will be used for all remaining
18099 Specify scale, if it is set it will be multiplied with sum
18100 of each weight multiplied with pixel values to give final destination
18101 pixel value. By default @var{scale} is auto scaled to sum of weights.
18104 @subsection Examples
18108 Average 7 successive frames:
18110 tmix=frames=7:weights="1 1 1 1 1 1 1"
18114 Apply simple temporal convolution:
18116 tmix=frames=3:weights="-1 3 -1"
18120 Similar as above but only showing temporal differences:
18122 tmix=frames=3:weights="-1 2 -1":scale=1
18128 Tone map colors from different dynamic ranges.
18130 This filter expects data in single precision floating point, as it needs to
18131 operate on (and can output) out-of-range values. Another filter, such as
18132 @ref{zscale}, is needed to convert the resulting frame to a usable format.
18134 The tonemapping algorithms implemented only work on linear light, so input
18135 data should be linearized beforehand (and possibly correctly tagged).
18138 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
18141 @subsection Options
18142 The filter accepts the following options.
18146 Set the tone map algorithm to use.
18148 Possible values are:
18151 Do not apply any tone map, only desaturate overbright pixels.
18154 Hard-clip any out-of-range values. Use it for perfect color accuracy for
18155 in-range values, while distorting out-of-range values.
18158 Stretch the entire reference gamut to a linear multiple of the display.
18161 Fit a logarithmic transfer between the tone curves.
18164 Preserve overall image brightness with a simple curve, using nonlinear
18165 contrast, which results in flattening details and degrading color accuracy.
18168 Preserve both dark and bright details better than @var{reinhard}, at the cost
18169 of slightly darkening everything. Use it when detail preservation is more
18170 important than color and brightness accuracy.
18173 Smoothly map out-of-range values, while retaining contrast and colors for
18174 in-range material as much as possible. Use it when color accuracy is more
18175 important than detail preservation.
18181 Tune the tone mapping algorithm.
18183 This affects the following algorithms:
18189 Specifies the scale factor to use while stretching.
18193 Specifies the exponent of the function.
18197 Specify an extra linear coefficient to multiply into the signal before clipping.
18201 Specify the local contrast coefficient at the display peak.
18202 Default to 0.5, which means that in-gamut values will be about half as bright
18209 Specify the transition point from linear to mobius transform. Every value
18210 below this point is guaranteed to be mapped 1:1. The higher the value, the
18211 more accurate the result will be, at the cost of losing bright details.
18212 Default to 0.3, which due to the steep initial slope still preserves in-range
18213 colors fairly accurately.
18217 Apply desaturation for highlights that exceed this level of brightness. The
18218 higher the parameter, the more color information will be preserved. This
18219 setting helps prevent unnaturally blown-out colors for super-highlights, by
18220 (smoothly) turning into white instead. This makes images feel more natural,
18221 at the cost of reducing information about out-of-range colors.
18223 The default of 2.0 is somewhat conservative and will mostly just apply to
18224 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
18226 This option works only if the input frame has a supported color tag.
18229 Override signal/nominal/reference peak with this value. Useful when the
18230 embedded peak information in display metadata is not reliable or when tone
18231 mapping from a lower range to a higher range.
18236 Temporarily pad video frames.
18238 The filter accepts the following options:
18242 Specify number of delay frames before input video stream.
18245 Specify number of padding frames after input video stream.
18246 Set to -1 to pad indefinitely.
18249 Set kind of frames added to beginning of stream.
18250 Can be either @var{add} or @var{clone}.
18251 With @var{add} frames of solid-color are added.
18252 With @var{clone} frames are clones of first frame.
18255 Set kind of frames added to end of stream.
18256 Can be either @var{add} or @var{clone}.
18257 With @var{add} frames of solid-color are added.
18258 With @var{clone} frames are clones of last frame.
18260 @item start_duration, stop_duration
18261 Specify the duration of the start/stop delay. See
18262 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18263 for the accepted syntax.
18264 These options override @var{start} and @var{stop}.
18267 Specify the color of the padded area. For the syntax of this option,
18268 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
18269 manual,ffmpeg-utils}.
18271 The default value of @var{color} is "black".
18277 Transpose rows with columns in the input video and optionally flip it.
18279 It accepts the following parameters:
18284 Specify the transposition direction.
18286 Can assume the following values:
18288 @item 0, 4, cclock_flip
18289 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
18297 Rotate by 90 degrees clockwise, that is:
18305 Rotate by 90 degrees counterclockwise, that is:
18312 @item 3, 7, clock_flip
18313 Rotate by 90 degrees clockwise and vertically flip, that is:
18321 For values between 4-7, the transposition is only done if the input
18322 video geometry is portrait and not landscape. These values are
18323 deprecated, the @code{passthrough} option should be used instead.
18325 Numerical values are deprecated, and should be dropped in favor of
18326 symbolic constants.
18329 Do not apply the transposition if the input geometry matches the one
18330 specified by the specified value. It accepts the following values:
18333 Always apply transposition.
18335 Preserve portrait geometry (when @var{height} >= @var{width}).
18337 Preserve landscape geometry (when @var{width} >= @var{height}).
18340 Default value is @code{none}.
18343 For example to rotate by 90 degrees clockwise and preserve portrait
18346 transpose=dir=1:passthrough=portrait
18349 The command above can also be specified as:
18351 transpose=1:portrait
18354 @section transpose_npp
18356 Transpose rows with columns in the input video and optionally flip it.
18357 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
18359 It accepts the following parameters:
18364 Specify the transposition direction.
18366 Can assume the following values:
18369 Rotate by 90 degrees counterclockwise and vertically flip. (default)
18372 Rotate by 90 degrees clockwise.
18375 Rotate by 90 degrees counterclockwise.
18378 Rotate by 90 degrees clockwise and vertically flip.
18382 Do not apply the transposition if the input geometry matches the one
18383 specified by the specified value. It accepts the following values:
18386 Always apply transposition. (default)
18388 Preserve portrait geometry (when @var{height} >= @var{width}).
18390 Preserve landscape geometry (when @var{width} >= @var{height}).
18396 Trim the input so that the output contains one continuous subpart of the input.
18398 It accepts the following parameters:
18401 Specify the time of the start of the kept section, i.e. the frame with the
18402 timestamp @var{start} will be the first frame in the output.
18405 Specify the time of the first frame that will be dropped, i.e. the frame
18406 immediately preceding the one with the timestamp @var{end} will be the last
18407 frame in the output.
18410 This is the same as @var{start}, except this option sets the start timestamp
18411 in timebase units instead of seconds.
18414 This is the same as @var{end}, except this option sets the end timestamp
18415 in timebase units instead of seconds.
18418 The maximum duration of the output in seconds.
18421 The number of the first frame that should be passed to the output.
18424 The number of the first frame that should be dropped.
18427 @option{start}, @option{end}, and @option{duration} are expressed as time
18428 duration specifications; see
18429 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
18430 for the accepted syntax.
18432 Note that the first two sets of the start/end options and the @option{duration}
18433 option look at the frame timestamp, while the _frame variants simply count the
18434 frames that pass through the filter. Also note that this filter does not modify
18435 the timestamps. If you wish for the output timestamps to start at zero, insert a
18436 setpts filter after the trim filter.
18438 If multiple start or end options are set, this filter tries to be greedy and
18439 keep all the frames that match at least one of the specified constraints. To keep
18440 only the part that matches all the constraints at once, chain multiple trim
18443 The defaults are such that all the input is kept. So it is possible to set e.g.
18444 just the end values to keep everything before the specified time.
18449 Drop everything except the second minute of input:
18451 ffmpeg -i INPUT -vf trim=60:120
18455 Keep only the first second:
18457 ffmpeg -i INPUT -vf trim=duration=1
18462 @section unpremultiply
18463 Apply alpha unpremultiply effect to input video stream using first plane
18464 of second stream as alpha.
18466 Both streams must have same dimensions and same pixel format.
18468 The filter accepts the following option:
18472 Set which planes will be processed, unprocessed planes will be copied.
18473 By default value 0xf, all planes will be processed.
18475 If the format has 1 or 2 components, then luma is bit 0.
18476 If the format has 3 or 4 components:
18477 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
18478 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
18479 If present, the alpha channel is always the last bit.
18482 Do not require 2nd input for processing, instead use alpha plane from input stream.
18488 Sharpen or blur the input video.
18490 It accepts the following parameters:
18493 @item luma_msize_x, lx
18494 Set the luma matrix horizontal size. It must be an odd integer between
18495 3 and 23. The default value is 5.
18497 @item luma_msize_y, ly
18498 Set the luma matrix vertical size. It must be an odd integer between 3
18499 and 23. The default value is 5.
18501 @item luma_amount, la
18502 Set the luma effect strength. It must be a floating point number, reasonable
18503 values lay between -1.5 and 1.5.
18505 Negative values will blur the input video, while positive values will
18506 sharpen it, a value of zero will disable the effect.
18508 Default value is 1.0.
18510 @item chroma_msize_x, cx
18511 Set the chroma matrix horizontal size. It must be an odd integer
18512 between 3 and 23. The default value is 5.
18514 @item chroma_msize_y, cy
18515 Set the chroma matrix vertical size. It must be an odd integer
18516 between 3 and 23. The default value is 5.
18518 @item chroma_amount, ca
18519 Set the chroma effect strength. It must be a floating point number, reasonable
18520 values lay between -1.5 and 1.5.
18522 Negative values will blur the input video, while positive values will
18523 sharpen it, a value of zero will disable the effect.
18525 Default value is 0.0.
18529 All parameters are optional and default to the equivalent of the
18530 string '5:5:1.0:5:5:0.0'.
18532 @subsection Examples
18536 Apply strong luma sharpen effect:
18538 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
18542 Apply a strong blur of both luma and chroma parameters:
18544 unsharp=7:7:-2:7:7:-2
18550 Apply ultra slow/simple postprocessing filter that compresses and decompresses
18551 the image at several (or - in the case of @option{quality} level @code{8} - all)
18552 shifts and average the results.
18554 The way this differs from the behavior of spp is that uspp actually encodes &
18555 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
18556 DCT similar to MJPEG.
18558 The filter accepts the following options:
18562 Set quality. This option defines the number of levels for averaging. It accepts
18563 an integer in the range 0-8. If set to @code{0}, the filter will have no
18564 effect. A value of @code{8} means the higher quality. For each increment of
18565 that value the speed drops by a factor of approximately 2. Default value is
18569 Force a constant quantization parameter. If not set, the filter will use the QP
18570 from the video stream (if available).
18575 Convert 360 videos between various formats.
18577 The filter accepts the following options:
18583 Set format of the input/output video.
18591 Equirectangular projection.
18596 Cubemap with 3x2/6x1/1x6 layout.
18598 Format specific options:
18603 Set padding proportion for the input/output cubemap. Values in decimals.
18610 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)
18613 Default value is @b{@samp{0}}.
18617 Set fixed padding for the input/output cubemap. Values in pixels.
18619 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
18623 Set order of faces for the input/output cubemap. Choose one direction for each position.
18625 Designation of directions:
18641 Default value is @b{@samp{rludfb}}.
18645 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
18647 Designation of angles:
18650 0 degrees clockwise
18652 90 degrees clockwise
18654 180 degrees clockwise
18656 270 degrees clockwise
18659 Default value is @b{@samp{000000}}.
18663 Equi-Angular Cubemap.
18668 Regular video. @i{(output only)}
18670 Format specific options:
18675 Set horizontal/vertical/diagonal field of view. Values in degrees.
18677 If diagonal field of view is set it overrides horizontal and vertical field of view.
18683 Format specific options:
18687 Set padding proportion. Values in decimals.
18697 Default value is @b{@samp{0}}.
18702 Facebook's 360 format.
18705 Stereographic format.
18707 Format specific options:
18712 Set horizontal/vertical/diagonal field of view. Values in degrees.
18714 If diagonal field of view is set it overrides horizontal and vertical field of view.
18721 Ball format, gives significant distortion toward the back.
18724 Hammer-Aitoff map projection format.
18727 Sinusoidal map projection format.
18732 Set interpolation method.@*
18733 @i{Note: more complex interpolation methods require much more memory to run.}
18743 Bilinear interpolation.
18746 Bicubic interpolation.
18749 Lanczos interpolation.
18752 Default value is @b{@samp{line}}.
18756 Set the output video resolution.
18758 Default resolution depends on formats.
18762 Set the input/output stereo format.
18773 Default value is @b{@samp{2d}} for input and output format.
18778 Set rotation for the output video. Values in degrees.
18781 Set rotation order for the output video. Choose one item for each position.
18792 Default value is @b{@samp{ypr}}.
18797 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
18801 Set if input video is flipped horizontally/vertically. Boolean values.
18804 Set if input video is transposed. Boolean value, by default disabled.
18807 Set if output video needs to be transposed. Boolean value, by default disabled.
18811 @subsection Examples
18815 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
18817 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
18820 Extract back view of Equi-Angular Cubemap:
18822 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
18825 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
18827 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
18831 @section vaguedenoiser
18833 Apply a wavelet based denoiser.
18835 It transforms each frame from the video input into the wavelet domain,
18836 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
18837 the obtained coefficients. It does an inverse wavelet transform after.
18838 Due to wavelet properties, it should give a nice smoothed result, and
18839 reduced noise, without blurring picture features.
18841 This filter accepts the following options:
18845 The filtering strength. The higher, the more filtered the video will be.
18846 Hard thresholding can use a higher threshold than soft thresholding
18847 before the video looks overfiltered. Default value is 2.
18850 The filtering method the filter will use.
18852 It accepts the following values:
18855 All values under the threshold will be zeroed.
18858 All values under the threshold will be zeroed. All values above will be
18859 reduced by the threshold.
18862 Scales or nullifies coefficients - intermediary between (more) soft and
18863 (less) hard thresholding.
18866 Default is garrote.
18869 Number of times, the wavelet will decompose the picture. Picture can't
18870 be decomposed beyond a particular point (typically, 8 for a 640x480
18871 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
18874 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
18877 A list of the planes to process. By default all planes are processed.
18880 @section vectorscope
18882 Display 2 color component values in the two dimensional graph (which is called
18885 This filter accepts the following options:
18889 Set vectorscope mode.
18891 It accepts the following values:
18894 Gray values are displayed on graph, higher brightness means more pixels have
18895 same component color value on location in graph. This is the default mode.
18898 Gray values are displayed on graph. Surrounding pixels values which are not
18899 present in video frame are drawn in gradient of 2 color components which are
18900 set by option @code{x} and @code{y}. The 3rd color component is static.
18903 Actual color components values present in video frame are displayed on graph.
18906 Similar as color2 but higher frequency of same values @code{x} and @code{y}
18907 on graph increases value of another color component, which is luminance by
18908 default values of @code{x} and @code{y}.
18911 Actual colors present in video frame are displayed on graph. If two different
18912 colors map to same position on graph then color with higher value of component
18913 not present in graph is picked.
18916 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
18917 component picked from radial gradient.
18921 Set which color component will be represented on X-axis. Default is @code{1}.
18924 Set which color component will be represented on Y-axis. Default is @code{2}.
18927 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
18928 of color component which represents frequency of (X, Y) location in graph.
18933 No envelope, this is default.
18936 Instant envelope, even darkest single pixel will be clearly highlighted.
18939 Hold maximum and minimum values presented in graph over time. This way you
18940 can still spot out of range values without constantly looking at vectorscope.
18943 Peak and instant envelope combined together.
18947 Set what kind of graticule to draw.
18955 Set graticule opacity.
18958 Set graticule flags.
18962 Draw graticule for white point.
18965 Draw graticule for black point.
18968 Draw color points short names.
18972 Set background opacity.
18974 @item lthreshold, l
18975 Set low threshold for color component not represented on X or Y axis.
18976 Values lower than this value will be ignored. Default is 0.
18977 Note this value is multiplied with actual max possible value one pixel component
18978 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
18981 @item hthreshold, h
18982 Set high threshold for color component not represented on X or Y axis.
18983 Values higher than this value will be ignored. Default is 1.
18984 Note this value is multiplied with actual max possible value one pixel component
18985 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
18986 is 0.9 * 255 = 230.
18988 @item colorspace, c
18989 Set what kind of colorspace to use when drawing graticule.
18998 @anchor{vidstabdetect}
18999 @section vidstabdetect
19001 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
19002 @ref{vidstabtransform} for pass 2.
19004 This filter generates a file with relative translation and rotation
19005 transform information about subsequent frames, which is then used by
19006 the @ref{vidstabtransform} filter.
19008 To enable compilation of this filter you need to configure FFmpeg with
19009 @code{--enable-libvidstab}.
19011 This filter accepts the following options:
19015 Set the path to the file used to write the transforms information.
19016 Default value is @file{transforms.trf}.
19019 Set how shaky the video is and how quick the camera is. It accepts an
19020 integer in the range 1-10, a value of 1 means little shakiness, a
19021 value of 10 means strong shakiness. Default value is 5.
19024 Set the accuracy of the detection process. It must be a value in the
19025 range 1-15. A value of 1 means low accuracy, a value of 15 means high
19026 accuracy. Default value is 15.
19029 Set stepsize of the search process. The region around minimum is
19030 scanned with 1 pixel resolution. Default value is 6.
19033 Set minimum contrast. Below this value a local measurement field is
19034 discarded. Must be a floating point value in the range 0-1. Default
19038 Set reference frame number for tripod mode.
19040 If enabled, the motion of the frames is compared to a reference frame
19041 in the filtered stream, identified by the specified number. The idea
19042 is to compensate all movements in a more-or-less static scene and keep
19043 the camera view absolutely still.
19045 If set to 0, it is disabled. The frames are counted starting from 1.
19048 Show fields and transforms in the resulting frames. It accepts an
19049 integer in the range 0-2. Default value is 0, which disables any
19053 @subsection Examples
19057 Use default values:
19063 Analyze strongly shaky movie and put the results in file
19064 @file{mytransforms.trf}:
19066 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
19070 Visualize the result of internal transformations in the resulting
19073 vidstabdetect=show=1
19077 Analyze a video with medium shakiness using @command{ffmpeg}:
19079 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
19083 @anchor{vidstabtransform}
19084 @section vidstabtransform
19086 Video stabilization/deshaking: pass 2 of 2,
19087 see @ref{vidstabdetect} for pass 1.
19089 Read a file with transform information for each frame and
19090 apply/compensate them. Together with the @ref{vidstabdetect}
19091 filter this can be used to deshake videos. See also
19092 @url{http://public.hronopik.de/vid.stab}. It is important to also use
19093 the @ref{unsharp} filter, see below.
19095 To enable compilation of this filter you need to configure FFmpeg with
19096 @code{--enable-libvidstab}.
19098 @subsection Options
19102 Set path to the file used to read the transforms. Default value is
19103 @file{transforms.trf}.
19106 Set the number of frames (value*2 + 1) used for lowpass filtering the
19107 camera movements. Default value is 10.
19109 For example a number of 10 means that 21 frames are used (10 in the
19110 past and 10 in the future) to smoothen the motion in the video. A
19111 larger value leads to a smoother video, but limits the acceleration of
19112 the camera (pan/tilt movements). 0 is a special case where a static
19113 camera is simulated.
19116 Set the camera path optimization algorithm.
19118 Accepted values are:
19121 gaussian kernel low-pass filter on camera motion (default)
19123 averaging on transformations
19127 Set maximal number of pixels to translate frames. Default value is -1,
19131 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
19132 value is -1, meaning no limit.
19135 Specify how to deal with borders that may be visible due to movement
19138 Available values are:
19141 keep image information from previous frame (default)
19143 fill the border black
19147 Invert transforms if set to 1. Default value is 0.
19150 Consider transforms as relative to previous frame if set to 1,
19151 absolute if set to 0. Default value is 0.
19154 Set percentage to zoom. A positive value will result in a zoom-in
19155 effect, a negative value in a zoom-out effect. Default value is 0 (no
19159 Set optimal zooming to avoid borders.
19161 Accepted values are:
19166 optimal static zoom value is determined (only very strong movements
19167 will lead to visible borders) (default)
19169 optimal adaptive zoom value is determined (no borders will be
19170 visible), see @option{zoomspeed}
19173 Note that the value given at zoom is added to the one calculated here.
19176 Set percent to zoom maximally each frame (enabled when
19177 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
19181 Specify type of interpolation.
19183 Available values are:
19188 linear only horizontal
19190 linear in both directions (default)
19192 cubic in both directions (slow)
19196 Enable virtual tripod mode if set to 1, which is equivalent to
19197 @code{relative=0:smoothing=0}. Default value is 0.
19199 Use also @code{tripod} option of @ref{vidstabdetect}.
19202 Increase log verbosity if set to 1. Also the detected global motions
19203 are written to the temporary file @file{global_motions.trf}. Default
19207 @subsection Examples
19211 Use @command{ffmpeg} for a typical stabilization with default values:
19213 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
19216 Note the use of the @ref{unsharp} filter which is always recommended.
19219 Zoom in a bit more and load transform data from a given file:
19221 vidstabtransform=zoom=5:input="mytransforms.trf"
19225 Smoothen the video even more:
19227 vidstabtransform=smoothing=30
19233 Flip the input video vertically.
19235 For example, to vertically flip a video with @command{ffmpeg}:
19237 ffmpeg -i in.avi -vf "vflip" out.avi
19242 Detect variable frame rate video.
19244 This filter tries to detect if the input is variable or constant frame rate.
19246 At end it will output number of frames detected as having variable delta pts,
19247 and ones with constant delta pts.
19248 If there was frames with variable delta, than it will also show min, max and
19249 average delta encountered.
19253 Boost or alter saturation.
19255 The filter accepts the following options:
19258 Set strength of boost if positive value or strength of alter if negative value.
19259 Default is 0. Allowed range is from -2 to 2.
19262 Set the red balance. Default is 1. Allowed range is from -10 to 10.
19265 Set the green balance. Default is 1. Allowed range is from -10 to 10.
19268 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
19271 Set the red luma coefficient.
19274 Set the green luma coefficient.
19277 Set the blue luma coefficient.
19280 If @code{intensity} is negative and this is set to 1, colors will change,
19281 otherwise colors will be less saturated, more towards gray.
19287 Make or reverse a natural vignetting effect.
19289 The filter accepts the following options:
19293 Set lens angle expression as a number of radians.
19295 The value is clipped in the @code{[0,PI/2]} range.
19297 Default value: @code{"PI/5"}
19301 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
19305 Set forward/backward mode.
19307 Available modes are:
19310 The larger the distance from the central point, the darker the image becomes.
19313 The larger the distance from the central point, the brighter the image becomes.
19314 This can be used to reverse a vignette effect, though there is no automatic
19315 detection to extract the lens @option{angle} and other settings (yet). It can
19316 also be used to create a burning effect.
19319 Default value is @samp{forward}.
19322 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
19324 It accepts the following values:
19327 Evaluate expressions only once during the filter initialization.
19330 Evaluate expressions for each incoming frame. This is way slower than the
19331 @samp{init} mode since it requires all the scalers to be re-computed, but it
19332 allows advanced dynamic expressions.
19335 Default value is @samp{init}.
19338 Set dithering to reduce the circular banding effects. Default is @code{1}
19342 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
19343 Setting this value to the SAR of the input will make a rectangular vignetting
19344 following the dimensions of the video.
19346 Default is @code{1/1}.
19349 @subsection Expressions
19351 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
19352 following parameters.
19357 input width and height
19360 the number of input frame, starting from 0
19363 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
19364 @var{TB} units, NAN if undefined
19367 frame rate of the input video, NAN if the input frame rate is unknown
19370 the PTS (Presentation TimeStamp) of the filtered video frame,
19371 expressed in seconds, NAN if undefined
19374 time base of the input video
19378 @subsection Examples
19382 Apply simple strong vignetting effect:
19388 Make a flickering vignetting:
19390 vignette='PI/4+random(1)*PI/50':eval=frame
19395 @section vmafmotion
19397 Obtain the average VMAF motion score of a video.
19398 It is one of the component metrics of VMAF.
19400 The obtained average motion score is printed through the logging system.
19402 The filter accepts the following options:
19406 If specified, the filter will use the named file to save the motion score of
19407 each frame with respect to the previous frame.
19408 When filename equals "-" the data is sent to standard output.
19413 ffmpeg -i ref.mpg -vf vmafmotion -f null -
19417 Stack input videos vertically.
19419 All streams must be of same pixel format and of same width.
19421 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
19422 to create same output.
19424 The filter accepts the following options:
19428 Set number of input streams. Default is 2.
19431 If set to 1, force the output to terminate when the shortest input
19432 terminates. Default value is 0.
19437 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
19438 Deinterlacing Filter").
19440 Based on the process described by Martin Weston for BBC R&D, and
19441 implemented based on the de-interlace algorithm written by Jim
19442 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
19443 uses filter coefficients calculated by BBC R&D.
19445 This filter uses field-dominance information in frame to decide which
19446 of each pair of fields to place first in the output.
19447 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
19449 There are two sets of filter coefficients, so called "simple"
19450 and "complex". Which set of filter coefficients is used can
19451 be set by passing an optional parameter:
19455 Set the interlacing filter coefficients. Accepts one of the following values:
19459 Simple filter coefficient set.
19461 More-complex filter coefficient set.
19463 Default value is @samp{complex}.
19466 Specify which frames to deinterlace. Accepts one of the following values:
19470 Deinterlace all frames,
19472 Only deinterlace frames marked as interlaced.
19475 Default value is @samp{all}.
19479 Video waveform monitor.
19481 The waveform monitor plots color component intensity. By default luminance
19482 only. Each column of the waveform corresponds to a column of pixels in the
19485 It accepts the following options:
19489 Can be either @code{row}, or @code{column}. Default is @code{column}.
19490 In row mode, the graph on the left side represents color component value 0 and
19491 the right side represents value = 255. In column mode, the top side represents
19492 color component value = 0 and bottom side represents value = 255.
19495 Set intensity. Smaller values are useful to find out how many values of the same
19496 luminance are distributed across input rows/columns.
19497 Default value is @code{0.04}. Allowed range is [0, 1].
19500 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
19501 In mirrored mode, higher values will be represented on the left
19502 side for @code{row} mode and at the top for @code{column} mode. Default is
19503 @code{1} (mirrored).
19507 It accepts the following values:
19510 Presents information identical to that in the @code{parade}, except
19511 that the graphs representing color components are superimposed directly
19514 This display mode makes it easier to spot relative differences or similarities
19515 in overlapping areas of the color components that are supposed to be identical,
19516 such as neutral whites, grays, or blacks.
19519 Display separate graph for the color components side by side in
19520 @code{row} mode or one below the other in @code{column} mode.
19523 Display separate graph for the color components side by side in
19524 @code{column} mode or one below the other in @code{row} mode.
19526 Using this display mode makes it easy to spot color casts in the highlights
19527 and shadows of an image, by comparing the contours of the top and the bottom
19528 graphs of each waveform. Since whites, grays, and blacks are characterized
19529 by exactly equal amounts of red, green, and blue, neutral areas of the picture
19530 should display three waveforms of roughly equal width/height. If not, the
19531 correction is easy to perform by making level adjustments the three waveforms.
19533 Default is @code{stack}.
19535 @item components, c
19536 Set which color components to display. Default is 1, which means only luminance
19537 or red color component if input is in RGB colorspace. If is set for example to
19538 7 it will display all 3 (if) available color components.
19543 No envelope, this is default.
19546 Instant envelope, minimum and maximum values presented in graph will be easily
19547 visible even with small @code{step} value.
19550 Hold minimum and maximum values presented in graph across time. This way you
19551 can still spot out of range values without constantly looking at waveforms.
19554 Peak and instant envelope combined together.
19560 No filtering, this is default.
19563 Luma and chroma combined together.
19566 Similar as above, but shows difference between blue and red chroma.
19569 Similar as above, but use different colors.
19572 Similar as above, but again with different colors.
19575 Displays only chroma.
19578 Displays actual color value on waveform.
19581 Similar as above, but with luma showing frequency of chroma values.
19585 Set which graticule to display.
19589 Do not display graticule.
19592 Display green graticule showing legal broadcast ranges.
19595 Display orange graticule showing legal broadcast ranges.
19598 Display invert graticule showing legal broadcast ranges.
19602 Set graticule opacity.
19605 Set graticule flags.
19609 Draw numbers above lines. By default enabled.
19612 Draw dots instead of lines.
19616 Set scale used for displaying graticule.
19623 Default is digital.
19626 Set background opacity.
19629 @section weave, doubleweave
19631 The @code{weave} takes a field-based video input and join
19632 each two sequential fields into single frame, producing a new double
19633 height clip with half the frame rate and half the frame count.
19635 The @code{doubleweave} works same as @code{weave} but without
19636 halving frame rate and frame count.
19638 It accepts the following option:
19642 Set first field. Available values are:
19646 Set the frame as top-field-first.
19649 Set the frame as bottom-field-first.
19653 @subsection Examples
19657 Interlace video using @ref{select} and @ref{separatefields} filter:
19659 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
19664 Apply the xBR high-quality magnification filter which is designed for pixel
19665 art. It follows a set of edge-detection rules, see
19666 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
19668 It accepts the following option:
19672 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
19673 @code{3xBR} and @code{4} for @code{4xBR}.
19674 Default is @code{3}.
19678 Pick median pixels from several input videos.
19680 The filter accepts the following options:
19684 Set number of inputs.
19685 Default is 3. Allowed range is from 3 to 255.
19686 If number of inputs is even number, than result will be mean value between two median values.
19689 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19693 Stack video inputs into custom layout.
19695 All streams must be of same pixel format.
19697 The filter accepts the following options:
19701 Set number of input streams. Default is 2.
19704 Specify layout of inputs.
19705 This option requires the desired layout configuration to be explicitly set by the user.
19706 This sets position of each video input in output. Each input
19707 is separated by '|'.
19708 The first number represents the column, and the second number represents the row.
19709 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
19710 where X is video input from which to take width or height.
19711 Multiple values can be used when separated by '+'. In such
19712 case values are summed together.
19714 Note that if inputs are of different sizes gaps may appear, as not all of
19715 the output video frame will be filled. Similarly, videos can overlap each
19716 other if their position doesn't leave enough space for the full frame of
19719 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
19720 a layout must be set by the user.
19723 If set to 1, force the output to terminate when the shortest input
19724 terminates. Default value is 0.
19727 @subsection Examples
19731 Display 4 inputs into 2x2 grid.
19735 input1(0, 0) | input3(w0, 0)
19736 input2(0, h0) | input4(w0, h0)
19740 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
19743 Note that if inputs are of different sizes, gaps or overlaps may occur.
19746 Display 4 inputs into 1x4 grid.
19753 input4(0, h0+h1+h2)
19757 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
19760 Note that if inputs are of different widths, unused space will appear.
19763 Display 9 inputs into 3x3 grid.
19767 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
19768 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
19769 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
19773 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
19776 Note that if inputs are of different sizes, gaps or overlaps may occur.
19779 Display 16 inputs into 4x4 grid.
19783 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
19784 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
19785 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
19786 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
19790 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|
19791 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
19794 Note that if inputs are of different sizes, gaps or overlaps may occur.
19801 Deinterlace the input video ("yadif" means "yet another deinterlacing
19804 It accepts the following parameters:
19810 The interlacing mode to adopt. It accepts one of the following values:
19813 @item 0, send_frame
19814 Output one frame for each frame.
19815 @item 1, send_field
19816 Output one frame for each field.
19817 @item 2, send_frame_nospatial
19818 Like @code{send_frame}, but it skips the spatial interlacing check.
19819 @item 3, send_field_nospatial
19820 Like @code{send_field}, but it skips the spatial interlacing check.
19823 The default value is @code{send_frame}.
19826 The picture field parity assumed for the input interlaced video. It accepts one
19827 of the following values:
19831 Assume the top field is first.
19833 Assume the bottom field is first.
19835 Enable automatic detection of field parity.
19838 The default value is @code{auto}.
19839 If the interlacing is unknown or the decoder does not export this information,
19840 top field first will be assumed.
19843 Specify which frames to deinterlace. Accepts one of the following
19848 Deinterlace all frames.
19849 @item 1, interlaced
19850 Only deinterlace frames marked as interlaced.
19853 The default value is @code{all}.
19856 @section yadif_cuda
19858 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
19859 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
19862 It accepts the following parameters:
19868 The interlacing mode to adopt. It accepts one of the following values:
19871 @item 0, send_frame
19872 Output one frame for each frame.
19873 @item 1, send_field
19874 Output one frame for each field.
19875 @item 2, send_frame_nospatial
19876 Like @code{send_frame}, but it skips the spatial interlacing check.
19877 @item 3, send_field_nospatial
19878 Like @code{send_field}, but it skips the spatial interlacing check.
19881 The default value is @code{send_frame}.
19884 The picture field parity assumed for the input interlaced video. It accepts one
19885 of the following values:
19889 Assume the top field is first.
19891 Assume the bottom field is first.
19893 Enable automatic detection of field parity.
19896 The default value is @code{auto}.
19897 If the interlacing is unknown or the decoder does not export this information,
19898 top field first will be assumed.
19901 Specify which frames to deinterlace. Accepts one of the following
19906 Deinterlace all frames.
19907 @item 1, interlaced
19908 Only deinterlace frames marked as interlaced.
19911 The default value is @code{all}.
19916 Apply Zoom & Pan effect.
19918 This filter accepts the following options:
19922 Set the zoom expression. Range is 1-10. Default is 1.
19926 Set the x and y expression. Default is 0.
19929 Set the duration expression in number of frames.
19930 This sets for how many number of frames effect will last for
19931 single input image.
19934 Set the output image size, default is 'hd720'.
19937 Set the output frame rate, default is '25'.
19940 Each expression can contain the following constants:
19959 Output frame count.
19963 Last calculated 'x' and 'y' position from 'x' and 'y' expression
19964 for current input frame.
19968 'x' and 'y' of last output frame of previous input frame or 0 when there was
19969 not yet such frame (first input frame).
19972 Last calculated zoom from 'z' expression for current input frame.
19975 Last calculated zoom of last output frame of previous input frame.
19978 Number of output frames for current input frame. Calculated from 'd' expression
19979 for each input frame.
19982 number of output frames created for previous input frame
19985 Rational number: input width / input height
19988 sample aspect ratio
19991 display aspect ratio
19995 @subsection Examples
19999 Zoom-in up to 1.5 and pan at same time to some spot near center of picture:
20001 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
20005 Zoom-in up to 1.5 and pan always at center of picture:
20007 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20011 Same as above but without pausing:
20013 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
20019 Scale (resize) the input video, using the z.lib library:
20020 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
20021 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
20023 The zscale filter forces the output display aspect ratio to be the same
20024 as the input, by changing the output sample aspect ratio.
20026 If the input image format is different from the format requested by
20027 the next filter, the zscale filter will convert the input to the
20030 @subsection Options
20031 The filter accepts the following options.
20036 Set the output video dimension expression. Default value is the input
20039 If the @var{width} or @var{w} value is 0, the input width is used for
20040 the output. If the @var{height} or @var{h} value is 0, the input height
20041 is used for the output.
20043 If one and only one of the values is -n with n >= 1, the zscale filter
20044 will use a value that maintains the aspect ratio of the input image,
20045 calculated from the other specified dimension. After that it will,
20046 however, make sure that the calculated dimension is divisible by n and
20047 adjust the value if necessary.
20049 If both values are -n with n >= 1, the behavior will be identical to
20050 both values being set to 0 as previously detailed.
20052 See below for the list of accepted constants for use in the dimension
20056 Set the video size. For the syntax of this option, check the
20057 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
20060 Set the dither type.
20062 Possible values are:
20067 @item error_diffusion
20073 Set the resize filter type.
20075 Possible values are:
20085 Default is bilinear.
20088 Set the color range.
20090 Possible values are:
20097 Default is same as input.
20100 Set the color primaries.
20102 Possible values are:
20112 Default is same as input.
20115 Set the transfer characteristics.
20117 Possible values are:
20131 Default is same as input.
20134 Set the colorspace matrix.
20136 Possible value are:
20147 Default is same as input.
20150 Set the input color range.
20152 Possible values are:
20159 Default is same as input.
20161 @item primariesin, pin
20162 Set the input color primaries.
20164 Possible values are:
20174 Default is same as input.
20176 @item transferin, tin
20177 Set the input transfer characteristics.
20179 Possible values are:
20190 Default is same as input.
20192 @item matrixin, min
20193 Set the input colorspace matrix.
20195 Possible value are:
20207 Set the output chroma location.
20209 Possible values are:
20220 @item chromalin, cin
20221 Set the input chroma location.
20223 Possible values are:
20235 Set the nominal peak luminance.
20238 The values of the @option{w} and @option{h} options are expressions
20239 containing the following constants:
20244 The input width and height
20248 These are the same as @var{in_w} and @var{in_h}.
20252 The output (scaled) width and height
20256 These are the same as @var{out_w} and @var{out_h}
20259 The same as @var{iw} / @var{ih}
20262 input sample aspect ratio
20265 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
20269 horizontal and vertical input chroma subsample values. For example for the
20270 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20274 horizontal and vertical output chroma subsample values. For example for the
20275 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
20281 @c man end VIDEO FILTERS
20283 @chapter OpenCL Video Filters
20284 @c man begin OPENCL VIDEO FILTERS
20286 Below is a description of the currently available OpenCL video filters.
20288 To enable compilation of these filters you need to configure FFmpeg with
20289 @code{--enable-opencl}.
20291 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
20294 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
20295 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
20296 given device parameters.
20298 @item -filter_hw_device @var{name}
20299 Pass the hardware device called @var{name} to all filters in any filter graph.
20303 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
20307 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
20309 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
20313 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.
20315 @section avgblur_opencl
20317 Apply average blur filter.
20319 The filter accepts the following options:
20323 Set horizontal radius size.
20324 Range is @code{[1, 1024]} and default value is @code{1}.
20327 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20330 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
20333 @subsection Example
20337 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.
20339 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
20343 @section boxblur_opencl
20345 Apply a boxblur algorithm to the input video.
20347 It accepts the following parameters:
20351 @item luma_radius, lr
20352 @item luma_power, lp
20353 @item chroma_radius, cr
20354 @item chroma_power, cp
20355 @item alpha_radius, ar
20356 @item alpha_power, ap
20360 A description of the accepted options follows.
20363 @item luma_radius, lr
20364 @item chroma_radius, cr
20365 @item alpha_radius, ar
20366 Set an expression for the box radius in pixels used for blurring the
20367 corresponding input plane.
20369 The radius value must be a non-negative number, and must not be
20370 greater than the value of the expression @code{min(w,h)/2} for the
20371 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
20374 Default value for @option{luma_radius} is "2". If not specified,
20375 @option{chroma_radius} and @option{alpha_radius} default to the
20376 corresponding value set for @option{luma_radius}.
20378 The expressions can contain the following constants:
20382 The input width and height in pixels.
20386 The input chroma image width and height in pixels.
20390 The horizontal and vertical chroma subsample values. For example, for the
20391 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
20394 @item luma_power, lp
20395 @item chroma_power, cp
20396 @item alpha_power, ap
20397 Specify how many times the boxblur filter is applied to the
20398 corresponding plane.
20400 Default value for @option{luma_power} is 2. If not specified,
20401 @option{chroma_power} and @option{alpha_power} default to the
20402 corresponding value set for @option{luma_power}.
20404 A value of 0 will disable the effect.
20407 @subsection Examples
20409 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.
20413 Apply a boxblur filter with the luma, chroma, and alpha radius
20414 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.
20416 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
20417 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
20421 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.
20423 For the luma plane, a 2x2 box radius will be run once.
20425 For the chroma plane, a 4x4 box radius will be run 5 times.
20427 For the alpha plane, a 3x3 box radius will be run 7 times.
20429 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
20433 @section convolution_opencl
20435 Apply convolution of 3x3, 5x5, 7x7 matrix.
20437 The filter accepts the following options:
20444 Set matrix for each plane.
20445 Matrix is sequence of 9, 25 or 49 signed numbers.
20446 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
20452 Set multiplier for calculated value for each plane.
20453 If unset or 0, it will be sum of all matrix elements.
20454 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
20460 Set bias for each plane. This value is added to the result of the multiplication.
20461 Useful for making the overall image brighter or darker.
20462 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
20466 @subsection Examples
20472 -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
20478 -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
20482 Apply edge enhance:
20484 -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
20490 -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
20494 Apply laplacian edge detector which includes diagonals:
20496 -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
20502 -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
20506 @section dilation_opencl
20508 Apply dilation effect to the video.
20510 This filter replaces the pixel by the local(3x3) maximum.
20512 It accepts the following options:
20519 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
20520 If @code{0}, plane will remain unchanged.
20523 Flag which specifies the pixel to refer to.
20524 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
20526 Flags to local 3x3 coordinates region centered on @code{x}:
20535 @subsection Example
20539 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.
20541 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
20545 @section erosion_opencl
20547 Apply erosion effect to the video.
20549 This filter replaces the pixel by the local(3x3) minimum.
20551 It accepts the following options:
20558 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
20559 If @code{0}, plane will remain unchanged.
20562 Flag which specifies the pixel to refer to.
20563 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
20565 Flags to local 3x3 coordinates region centered on @code{x}:
20574 @subsection Example
20578 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.
20580 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
20584 @section colorkey_opencl
20585 RGB colorspace color keying.
20587 The filter accepts the following options:
20591 The color which will be replaced with transparency.
20594 Similarity percentage with the key color.
20596 0.01 matches only the exact key color, while 1.0 matches everything.
20601 0.0 makes pixels either fully transparent, or not transparent at all.
20603 Higher values result in semi-transparent pixels, with a higher transparency
20604 the more similar the pixels color is to the key color.
20607 @subsection Examples
20611 Make every semi-green pixel in the input transparent with some slight blending:
20613 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
20617 @section deshake_opencl
20618 Feature-point based video stabilization filter.
20620 The filter accepts the following options:
20624 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
20627 Whether or not additional debug info should be displayed, both in the processed output and in the console.
20629 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
20631 Viewing point matches in the output video is only supported for RGB input.
20633 Defaults to @code{0}.
20635 @item adaptive_crop
20636 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
20638 Defaults to @code{1}.
20640 @item refine_features
20641 Whether or not feature points should be refined at a sub-pixel level.
20643 This can be turned off for a slight performance gain at the cost of precision.
20645 Defaults to @code{1}.
20647 @item smooth_strength
20648 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
20650 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
20652 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
20654 Defaults to @code{0.0}.
20656 @item smooth_window_multiplier
20657 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
20659 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
20661 Acceptable values range from @code{0.1} to @code{10.0}.
20663 Larger values increase the amount of motion data available for determining how to smooth the camera path,
20664 potentially improving smoothness, but also increase latency and memory usage.
20666 Defaults to @code{2.0}.
20670 @subsection Examples
20674 Stabilize a video with a fixed, medium smoothing strength:
20676 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
20680 Stabilize a video with debugging (both in console and in rendered video):
20682 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
20686 @section nlmeans_opencl
20688 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
20690 @section overlay_opencl
20692 Overlay one video on top of another.
20694 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
20695 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
20697 The filter accepts the following options:
20702 Set the x coordinate of the overlaid video on the main video.
20703 Default value is @code{0}.
20706 Set the y coordinate of the overlaid video on the main video.
20707 Default value is @code{0}.
20711 @subsection Examples
20715 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
20717 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
20720 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
20722 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
20727 @section prewitt_opencl
20729 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
20731 The filter accepts the following option:
20735 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20738 Set value which will be multiplied with filtered result.
20739 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20742 Set value which will be added to filtered result.
20743 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20746 @subsection Example
20750 Apply the Prewitt operator with scale set to 2 and delta set to 10.
20752 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
20756 @section roberts_opencl
20757 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
20759 The filter accepts the following option:
20763 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20766 Set value which will be multiplied with filtered result.
20767 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20770 Set value which will be added to filtered result.
20771 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20774 @subsection Example
20778 Apply the Roberts cross operator with scale set to 2 and delta set to 10
20780 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
20784 @section sobel_opencl
20786 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
20788 The filter accepts the following option:
20792 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
20795 Set value which will be multiplied with filtered result.
20796 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
20799 Set value which will be added to filtered result.
20800 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
20803 @subsection Example
20807 Apply sobel operator with scale set to 2 and delta set to 10
20809 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
20813 @section tonemap_opencl
20815 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
20817 It accepts the following parameters:
20821 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
20824 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
20827 Apply desaturation for highlights that exceed this level of brightness. The
20828 higher the parameter, the more color information will be preserved. This
20829 setting helps prevent unnaturally blown-out colors for super-highlights, by
20830 (smoothly) turning into white instead. This makes images feel more natural,
20831 at the cost of reducing information about out-of-range colors.
20833 The default value is 0.5, and the algorithm here is a little different from
20834 the cpu version tonemap currently. A setting of 0.0 disables this option.
20837 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
20838 is used to detect whether the scene has changed or not. If the distance between
20839 the current frame average brightness and the current running average exceeds
20840 a threshold value, we would re-calculate scene average and peak brightness.
20841 The default value is 0.2.
20844 Specify the output pixel format.
20846 Currently supported formats are:
20853 Set the output color range.
20855 Possible values are:
20861 Default is same as input.
20864 Set the output color primaries.
20866 Possible values are:
20872 Default is same as input.
20875 Set the output transfer characteristics.
20877 Possible values are:
20886 Set the output colorspace matrix.
20888 Possible value are:
20894 Default is same as input.
20898 @subsection Example
20902 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
20904 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
20908 @section unsharp_opencl
20910 Sharpen or blur the input video.
20912 It accepts the following parameters:
20915 @item luma_msize_x, lx
20916 Set the luma matrix horizontal size.
20917 Range is @code{[1, 23]} and default value is @code{5}.
20919 @item luma_msize_y, ly
20920 Set the luma matrix vertical size.
20921 Range is @code{[1, 23]} and default value is @code{5}.
20923 @item luma_amount, la
20924 Set the luma effect strength.
20925 Range is @code{[-10, 10]} and default value is @code{1.0}.
20927 Negative values will blur the input video, while positive values will
20928 sharpen it, a value of zero will disable the effect.
20930 @item chroma_msize_x, cx
20931 Set the chroma matrix horizontal size.
20932 Range is @code{[1, 23]} and default value is @code{5}.
20934 @item chroma_msize_y, cy
20935 Set the chroma matrix vertical size.
20936 Range is @code{[1, 23]} and default value is @code{5}.
20938 @item chroma_amount, ca
20939 Set the chroma effect strength.
20940 Range is @code{[-10, 10]} and default value is @code{0.0}.
20942 Negative values will blur the input video, while positive values will
20943 sharpen it, a value of zero will disable the effect.
20947 All parameters are optional and default to the equivalent of the
20948 string '5:5:1.0:5:5:0.0'.
20950 @subsection Examples
20954 Apply strong luma sharpen effect:
20956 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
20960 Apply a strong blur of both luma and chroma parameters:
20962 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
20966 @c man end OPENCL VIDEO FILTERS
20968 @chapter Video Sources
20969 @c man begin VIDEO SOURCES
20971 Below is a description of the currently available video sources.
20975 Buffer video frames, and make them available to the filter chain.
20977 This source is mainly intended for a programmatic use, in particular
20978 through the interface defined in @file{libavfilter/vsrc_buffer.h}.
20980 It accepts the following parameters:
20985 Specify the size (width and height) of the buffered video frames. For the
20986 syntax of this option, check the
20987 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
20990 The input video width.
20993 The input video height.
20996 A string representing the pixel format of the buffered video frames.
20997 It may be a number corresponding to a pixel format, or a pixel format
21001 Specify the timebase assumed by the timestamps of the buffered frames.
21004 Specify the frame rate expected for the video stream.
21006 @item pixel_aspect, sar
21007 The sample (pixel) aspect ratio of the input video.
21010 Specify the optional parameters to be used for the scale filter which
21011 is automatically inserted when an input change is detected in the
21012 input size or format.
21014 @item hw_frames_ctx
21015 When using a hardware pixel format, this should be a reference to an
21016 AVHWFramesContext describing input frames.
21021 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
21024 will instruct the source to accept video frames with size 320x240 and
21025 with format "yuv410p", assuming 1/24 as the timestamps timebase and
21026 square pixels (1:1 sample aspect ratio).
21027 Since the pixel format with name "yuv410p" corresponds to the number 6
21028 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
21029 this example corresponds to:
21031 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
21034 Alternatively, the options can be specified as a flat string, but this
21035 syntax is deprecated:
21037 @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}]
21041 Create a pattern generated by an elementary cellular automaton.
21043 The initial state of the cellular automaton can be defined through the
21044 @option{filename} and @option{pattern} options. If such options are
21045 not specified an initial state is created randomly.
21047 At each new frame a new row in the video is filled with the result of
21048 the cellular automaton next generation. The behavior when the whole
21049 frame is filled is defined by the @option{scroll} option.
21051 This source accepts the following options:
21055 Read the initial cellular automaton state, i.e. the starting row, from
21056 the specified file.
21057 In the file, each non-whitespace character is considered an alive
21058 cell, a newline will terminate the row, and further characters in the
21059 file will be ignored.
21062 Read the initial cellular automaton state, i.e. the starting row, from
21063 the specified string.
21065 Each non-whitespace character in the string is considered an alive
21066 cell, a newline will terminate the row, and further characters in the
21067 string will be ignored.
21070 Set the video rate, that is the number of frames generated per second.
21073 @item random_fill_ratio, ratio
21074 Set the random fill ratio for the initial cellular automaton row. It
21075 is a floating point number value ranging from 0 to 1, defaults to
21078 This option is ignored when a file or a pattern is specified.
21080 @item random_seed, seed
21081 Set the seed for filling randomly the initial row, must be an integer
21082 included between 0 and UINT32_MAX. If not specified, or if explicitly
21083 set to -1, the filter will try to use a good random seed on a best
21087 Set the cellular automaton rule, it is a number ranging from 0 to 255.
21088 Default value is 110.
21091 Set the size of the output video. For the syntax of this option, check the
21092 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21094 If @option{filename} or @option{pattern} is specified, the size is set
21095 by default to the width of the specified initial state row, and the
21096 height is set to @var{width} * PHI.
21098 If @option{size} is set, it must contain the width of the specified
21099 pattern string, and the specified pattern will be centered in the
21102 If a filename or a pattern string is not specified, the size value
21103 defaults to "320x518" (used for a randomly generated initial state).
21106 If set to 1, scroll the output upward when all the rows in the output
21107 have been already filled. If set to 0, the new generated row will be
21108 written over the top row just after the bottom row is filled.
21111 @item start_full, full
21112 If set to 1, completely fill the output with generated rows before
21113 outputting the first frame.
21114 This is the default behavior, for disabling set the value to 0.
21117 If set to 1, stitch the left and right row edges together.
21118 This is the default behavior, for disabling set the value to 0.
21121 @subsection Examples
21125 Read the initial state from @file{pattern}, and specify an output of
21128 cellauto=f=pattern:s=200x400
21132 Generate a random initial row with a width of 200 cells, with a fill
21135 cellauto=ratio=2/3:s=200x200
21139 Create a pattern generated by rule 18 starting by a single alive cell
21140 centered on an initial row with width 100:
21142 cellauto=p=@@:s=100x400:full=0:rule=18
21146 Specify a more elaborated initial pattern:
21148 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
21153 @anchor{coreimagesrc}
21154 @section coreimagesrc
21155 Video source generated on GPU using Apple's CoreImage API on OSX.
21157 This video source is a specialized version of the @ref{coreimage} video filter.
21158 Use a core image generator at the beginning of the applied filterchain to
21159 generate the content.
21161 The coreimagesrc video source accepts the following options:
21163 @item list_generators
21164 List all available generators along with all their respective options as well as
21165 possible minimum and maximum values along with the default values.
21167 list_generators=true
21171 Specify the size of the sourced video. For the syntax of this option, check the
21172 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21173 The default value is @code{320x240}.
21176 Specify the frame rate of the sourced video, as the number of frames
21177 generated per second. It has to be a string in the format
21178 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21179 number or a valid video frame rate abbreviation. The default value is
21183 Set the sample aspect ratio of the sourced video.
21186 Set the duration of the sourced video. See
21187 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21188 for the accepted syntax.
21190 If not specified, or the expressed duration is negative, the video is
21191 supposed to be generated forever.
21194 Additionally, all options of the @ref{coreimage} video filter are accepted.
21195 A complete filterchain can be used for further processing of the
21196 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
21197 and examples for details.
21199 @subsection Examples
21204 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
21205 given as complete and escaped command-line for Apple's standard bash shell:
21207 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
21209 This example is equivalent to the QRCode example of @ref{coreimage} without the
21210 need for a nullsrc video source.
21214 @section mandelbrot
21216 Generate a Mandelbrot set fractal, and progressively zoom towards the
21217 point specified with @var{start_x} and @var{start_y}.
21219 This source accepts the following options:
21224 Set the terminal pts value. Default value is 400.
21227 Set the terminal scale value.
21228 Must be a floating point value. Default value is 0.3.
21231 Set the inner coloring mode, that is the algorithm used to draw the
21232 Mandelbrot fractal internal region.
21234 It shall assume one of the following values:
21239 Show time until convergence.
21241 Set color based on point closest to the origin of the iterations.
21246 Default value is @var{mincol}.
21249 Set the bailout value. Default value is 10.0.
21252 Set the maximum of iterations performed by the rendering
21253 algorithm. Default value is 7189.
21256 Set outer coloring mode.
21257 It shall assume one of following values:
21259 @item iteration_count
21260 Set iteration count mode.
21261 @item normalized_iteration_count
21262 set normalized iteration count mode.
21264 Default value is @var{normalized_iteration_count}.
21267 Set frame rate, expressed as number of frames per second. Default
21271 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
21272 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
21275 Set the initial scale value. Default value is 3.0.
21278 Set the initial x position. Must be a floating point value between
21279 -100 and 100. Default value is -0.743643887037158704752191506114774.
21282 Set the initial y position. Must be a floating point value between
21283 -100 and 100. Default value is -0.131825904205311970493132056385139.
21288 Generate various test patterns, as generated by the MPlayer test filter.
21290 The size of the generated video is fixed, and is 256x256.
21291 This source is useful in particular for testing encoding features.
21293 This source accepts the following options:
21298 Specify the frame rate of the sourced video, as the number of frames
21299 generated per second. It has to be a string in the format
21300 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21301 number or a valid video frame rate abbreviation. The default value is
21305 Set the duration of the sourced video. See
21306 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21307 for the accepted syntax.
21309 If not specified, or the expressed duration is negative, the video is
21310 supposed to be generated forever.
21314 Set the number or the name of the test to perform. Supported tests are:
21328 @item max_frames, m
21329 Set the maximum number of frames generated for each test, default value is 30.
21333 Default value is "all", which will cycle through the list of all tests.
21338 mptestsrc=t=dc_luma
21341 will generate a "dc_luma" test pattern.
21343 @section frei0r_src
21345 Provide a frei0r source.
21347 To enable compilation of this filter you need to install the frei0r
21348 header and configure FFmpeg with @code{--enable-frei0r}.
21350 This source accepts the following parameters:
21355 The size of the video to generate. For the syntax of this option, check the
21356 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21359 The framerate of the generated video. It may be a string of the form
21360 @var{num}/@var{den} or a frame rate abbreviation.
21363 The name to the frei0r source to load. For more information regarding frei0r and
21364 how to set the parameters, read the @ref{frei0r} section in the video filters
21367 @item filter_params
21368 A '|'-separated list of parameters to pass to the frei0r source.
21372 For example, to generate a frei0r partik0l source with size 200x200
21373 and frame rate 10 which is overlaid on the overlay filter main input:
21375 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
21380 Generate a life pattern.
21382 This source is based on a generalization of John Conway's life game.
21384 The sourced input represents a life grid, each pixel represents a cell
21385 which can be in one of two possible states, alive or dead. Every cell
21386 interacts with its eight neighbours, which are the cells that are
21387 horizontally, vertically, or diagonally adjacent.
21389 At each interaction the grid evolves according to the adopted rule,
21390 which specifies the number of neighbor alive cells which will make a
21391 cell stay alive or born. The @option{rule} option allows one to specify
21394 This source accepts the following options:
21398 Set the file from which to read the initial grid state. In the file,
21399 each non-whitespace character is considered an alive cell, and newline
21400 is used to delimit the end of each row.
21402 If this option is not specified, the initial grid is generated
21406 Set the video rate, that is the number of frames generated per second.
21409 @item random_fill_ratio, ratio
21410 Set the random fill ratio for the initial random grid. It is a
21411 floating point number value ranging from 0 to 1, defaults to 1/PHI.
21412 It is ignored when a file is specified.
21414 @item random_seed, seed
21415 Set the seed for filling the initial random grid, must be an integer
21416 included between 0 and UINT32_MAX. If not specified, or if explicitly
21417 set to -1, the filter will try to use a good random seed on a best
21423 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
21424 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
21425 @var{NS} specifies the number of alive neighbor cells which make a
21426 live cell stay alive, and @var{NB} the number of alive neighbor cells
21427 which make a dead cell to become alive (i.e. to "born").
21428 "s" and "b" can be used in place of "S" and "B", respectively.
21430 Alternatively a rule can be specified by an 18-bits integer. The 9
21431 high order bits are used to encode the next cell state if it is alive
21432 for each number of neighbor alive cells, the low order bits specify
21433 the rule for "borning" new cells. Higher order bits encode for an
21434 higher number of neighbor cells.
21435 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
21436 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
21438 Default value is "S23/B3", which is the original Conway's game of life
21439 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
21440 cells, and will born a new cell if there are three alive cells around
21444 Set the size of the output video. For the syntax of this option, check the
21445 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21447 If @option{filename} is specified, the size is set by default to the
21448 same size of the input file. If @option{size} is set, it must contain
21449 the size specified in the input file, and the initial grid defined in
21450 that file is centered in the larger resulting area.
21452 If a filename is not specified, the size value defaults to "320x240"
21453 (used for a randomly generated initial grid).
21456 If set to 1, stitch the left and right grid edges together, and the
21457 top and bottom edges also. Defaults to 1.
21460 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
21461 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
21462 value from 0 to 255.
21465 Set the color of living (or new born) cells.
21468 Set the color of dead cells. If @option{mold} is set, this is the first color
21469 used to represent a dead cell.
21472 Set mold color, for definitely dead and moldy cells.
21474 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
21475 ffmpeg-utils manual,ffmpeg-utils}.
21478 @subsection Examples
21482 Read a grid from @file{pattern}, and center it on a grid of size
21485 life=f=pattern:s=300x300
21489 Generate a random grid of size 200x200, with a fill ratio of 2/3:
21491 life=ratio=2/3:s=200x200
21495 Specify a custom rule for evolving a randomly generated grid:
21501 Full example with slow death effect (mold) using @command{ffplay}:
21503 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
21510 @anchor{haldclutsrc}
21513 @anchor{pal100bars}
21514 @anchor{rgbtestsrc}
21516 @anchor{smptehdbars}
21519 @anchor{yuvtestsrc}
21520 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
21522 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
21524 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
21526 The @code{color} source provides an uniformly colored input.
21528 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
21529 @ref{haldclut} filter.
21531 The @code{nullsrc} source returns unprocessed video frames. It is
21532 mainly useful to be employed in analysis / debugging tools, or as the
21533 source for filters which ignore the input data.
21535 The @code{pal75bars} source generates a color bars pattern, based on
21536 EBU PAL recommendations with 75% color levels.
21538 The @code{pal100bars} source generates a color bars pattern, based on
21539 EBU PAL recommendations with 100% color levels.
21541 The @code{rgbtestsrc} source generates an RGB test pattern useful for
21542 detecting RGB vs BGR issues. You should see a red, green and blue
21543 stripe from top to bottom.
21545 The @code{smptebars} source generates a color bars pattern, based on
21546 the SMPTE Engineering Guideline EG 1-1990.
21548 The @code{smptehdbars} source generates a color bars pattern, based on
21549 the SMPTE RP 219-2002.
21551 The @code{testsrc} source generates a test video pattern, showing a
21552 color pattern, a scrolling gradient and a timestamp. This is mainly
21553 intended for testing purposes.
21555 The @code{testsrc2} source is similar to testsrc, but supports more
21556 pixel formats instead of just @code{rgb24}. This allows using it as an
21557 input for other tests without requiring a format conversion.
21559 The @code{yuvtestsrc} source generates an YUV test pattern. You should
21560 see a y, cb and cr stripe from top to bottom.
21562 The sources accept the following parameters:
21567 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
21568 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
21569 pixels to be used as identity matrix for 3D lookup tables. Each component is
21570 coded on a @code{1/(N*N)} scale.
21573 Specify the color of the source, only available in the @code{color}
21574 source. For the syntax of this option, check the
21575 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
21578 Specify the size of the sourced video. For the syntax of this option, check the
21579 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21580 The default value is @code{320x240}.
21582 This option is not available with the @code{allrgb}, @code{allyuv}, and
21583 @code{haldclutsrc} filters.
21586 Specify the frame rate of the sourced video, as the number of frames
21587 generated per second. It has to be a string in the format
21588 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
21589 number or a valid video frame rate abbreviation. The default value is
21593 Set the duration of the sourced video. See
21594 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
21595 for the accepted syntax.
21597 If not specified, or the expressed duration is negative, the video is
21598 supposed to be generated forever.
21601 Set the sample aspect ratio of the sourced video.
21604 Specify the alpha (opacity) of the background, only available in the
21605 @code{testsrc2} source. The value must be between 0 (fully transparent) and
21606 255 (fully opaque, the default).
21609 Set the number of decimals to show in the timestamp, only available in the
21610 @code{testsrc} source.
21612 The displayed timestamp value will correspond to the original
21613 timestamp value multiplied by the power of 10 of the specified
21614 value. Default value is 0.
21617 @subsection Examples
21621 Generate a video with a duration of 5.3 seconds, with size
21622 176x144 and a frame rate of 10 frames per second:
21624 testsrc=duration=5.3:size=qcif:rate=10
21628 The following graph description will generate a red source
21629 with an opacity of 0.2, with size "qcif" and a frame rate of 10
21632 color=c=red@@0.2:s=qcif:r=10
21636 If the input content is to be ignored, @code{nullsrc} can be used. The
21637 following command generates noise in the luminance plane by employing
21638 the @code{geq} filter:
21640 nullsrc=s=256x256, geq=random(1)*255:128:128
21644 @subsection Commands
21646 The @code{color} source supports the following commands:
21650 Set the color of the created image. Accepts the same syntax of the
21651 corresponding @option{color} option.
21656 Generate video using an OpenCL program.
21661 OpenCL program source file.
21664 Kernel name in program.
21667 Size of frames to generate. This must be set.
21670 Pixel format to use for the generated frames. This must be set.
21673 Number of frames generated every second. Default value is '25'.
21677 For details of how the program loading works, see the @ref{program_opencl}
21684 Generate a colour ramp by setting pixel values from the position of the pixel
21685 in the output image. (Note that this will work with all pixel formats, but
21686 the generated output will not be the same.)
21688 __kernel void ramp(__write_only image2d_t dst,
21689 unsigned int index)
21691 int2 loc = (int2)(get_global_id(0), get_global_id(1));
21694 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
21696 write_imagef(dst, loc, val);
21701 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
21703 __kernel void sierpinski_carpet(__write_only image2d_t dst,
21704 unsigned int index)
21706 int2 loc = (int2)(get_global_id(0), get_global_id(1));
21708 float4 value = 0.0f;
21709 int x = loc.x + index;
21710 int y = loc.y + index;
21711 while (x > 0 || y > 0) {
21712 if (x % 3 == 1 && y % 3 == 1) {
21720 write_imagef(dst, loc, value);
21726 @section sierpinski
21728 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
21730 This source accepts the following options:
21734 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
21735 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
21738 Set frame rate, expressed as number of frames per second. Default
21742 Set seed which is used for random panning.
21745 Set max jump for single pan destination. Allowed range is from 1 to 10000.
21748 Set fractal type, can be default @code{carpet} or @code{triangle}.
21751 @c man end VIDEO SOURCES
21753 @chapter Video Sinks
21754 @c man begin VIDEO SINKS
21756 Below is a description of the currently available video sinks.
21758 @section buffersink
21760 Buffer video frames, and make them available to the end of the filter
21763 This sink is mainly intended for programmatic use, in particular
21764 through the interface defined in @file{libavfilter/buffersink.h}
21765 or the options system.
21767 It accepts a pointer to an AVBufferSinkContext structure, which
21768 defines the incoming buffers' formats, to be passed as the opaque
21769 parameter to @code{avfilter_init_filter} for initialization.
21773 Null video sink: do absolutely nothing with the input video. It is
21774 mainly useful as a template and for use in analysis / debugging
21777 @c man end VIDEO SINKS
21779 @chapter Multimedia Filters
21780 @c man begin MULTIMEDIA FILTERS
21782 Below is a description of the currently available multimedia filters.
21786 Convert input audio to a video output, displaying the audio bit scope.
21788 The filter accepts the following options:
21792 Set frame rate, expressed as number of frames per second. Default
21796 Specify the video size for the output. For the syntax of this option, check the
21797 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21798 Default value is @code{1024x256}.
21801 Specify list of colors separated by space or by '|' which will be used to
21802 draw channels. Unrecognized or missing colors will be replaced
21806 @section adrawgraph
21807 Draw a graph using input audio metadata.
21809 See @ref{drawgraph}
21811 @section agraphmonitor
21813 See @ref{graphmonitor}.
21815 @section ahistogram
21817 Convert input audio to a video output, displaying the volume histogram.
21819 The filter accepts the following options:
21823 Specify how histogram is calculated.
21825 It accepts the following values:
21828 Use single histogram for all channels.
21830 Use separate histogram for each channel.
21832 Default is @code{single}.
21835 Set frame rate, expressed as number of frames per second. Default
21839 Specify the video size for the output. For the syntax of this option, check the
21840 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21841 Default value is @code{hd720}.
21846 It accepts the following values:
21857 reverse logarithmic
21859 Default is @code{log}.
21862 Set amplitude scale.
21864 It accepts the following values:
21871 Default is @code{log}.
21874 Set how much frames to accumulate in histogram.
21875 Default is 1. Setting this to -1 accumulates all frames.
21878 Set histogram ratio of window height.
21881 Set sonogram sliding.
21883 It accepts the following values:
21886 replace old rows with new ones.
21888 scroll from top to bottom.
21890 Default is @code{replace}.
21893 @section aphasemeter
21895 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
21896 representing mean phase of current audio frame. A video output can also be produced and is
21897 enabled by default. The audio is passed through as first output.
21899 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
21900 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
21901 and @code{1} means channels are in phase.
21903 The filter accepts the following options, all related to its video output:
21907 Set the output frame rate. Default value is @code{25}.
21910 Set the video size for the output. For the syntax of this option, check the
21911 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21912 Default value is @code{800x400}.
21917 Specify the red, green, blue contrast. Default values are @code{2},
21918 @code{7} and @code{1}.
21919 Allowed range is @code{[0, 255]}.
21922 Set color which will be used for drawing median phase. If color is
21923 @code{none} which is default, no median phase value will be drawn.
21926 Enable video output. Default is enabled.
21929 @section avectorscope
21931 Convert input audio to a video output, representing the audio vector
21934 The filter is used to measure the difference between channels of stereo
21935 audio stream. A monaural signal, consisting of identical left and right
21936 signal, results in straight vertical line. Any stereo separation is visible
21937 as a deviation from this line, creating a Lissajous figure.
21938 If the straight (or deviation from it) but horizontal line appears this
21939 indicates that the left and right channels are out of phase.
21941 The filter accepts the following options:
21945 Set the vectorscope mode.
21947 Available values are:
21950 Lissajous rotated by 45 degrees.
21953 Same as above but not rotated.
21956 Shape resembling half of circle.
21959 Default value is @samp{lissajous}.
21962 Set the video size for the output. For the syntax of this option, check the
21963 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21964 Default value is @code{400x400}.
21967 Set the output frame rate. Default value is @code{25}.
21973 Specify the red, green, blue and alpha contrast. Default values are @code{40},
21974 @code{160}, @code{80} and @code{255}.
21975 Allowed range is @code{[0, 255]}.
21981 Specify the red, green, blue and alpha fade. Default values are @code{15},
21982 @code{10}, @code{5} and @code{5}.
21983 Allowed range is @code{[0, 255]}.
21986 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
21987 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
21990 Set the vectorscope drawing mode.
21992 Available values are:
21995 Draw dot for each sample.
21998 Draw line between previous and current sample.
22001 Default value is @samp{dot}.
22004 Specify amplitude scale of audio samples.
22006 Available values are:
22022 Swap left channel axis with right channel axis.
22032 Mirror only x axis.
22035 Mirror only y axis.
22043 @subsection Examples
22047 Complete example using @command{ffplay}:
22049 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
22050 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
22054 @section bench, abench
22056 Benchmark part of a filtergraph.
22058 The filter accepts the following options:
22062 Start or stop a timer.
22064 Available values are:
22067 Get the current time, set it as frame metadata (using the key
22068 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
22071 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
22072 the input frame metadata to get the time difference. Time difference, average,
22073 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
22074 @code{min}) are then printed. The timestamps are expressed in seconds.
22078 @subsection Examples
22082 Benchmark @ref{selectivecolor} filter:
22084 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
22090 Concatenate audio and video streams, joining them together one after the
22093 The filter works on segments of synchronized video and audio streams. All
22094 segments must have the same number of streams of each type, and that will
22095 also be the number of streams at output.
22097 The filter accepts the following options:
22102 Set the number of segments. Default is 2.
22105 Set the number of output video streams, that is also the number of video
22106 streams in each segment. Default is 1.
22109 Set the number of output audio streams, that is also the number of audio
22110 streams in each segment. Default is 0.
22113 Activate unsafe mode: do not fail if segments have a different format.
22117 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
22118 @var{a} audio outputs.
22120 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
22121 segment, in the same order as the outputs, then the inputs for the second
22124 Related streams do not always have exactly the same duration, for various
22125 reasons including codec frame size or sloppy authoring. For that reason,
22126 related synchronized streams (e.g. a video and its audio track) should be
22127 concatenated at once. The concat filter will use the duration of the longest
22128 stream in each segment (except the last one), and if necessary pad shorter
22129 audio streams with silence.
22131 For this filter to work correctly, all segments must start at timestamp 0.
22133 All corresponding streams must have the same parameters in all segments; the
22134 filtering system will automatically select a common pixel format for video
22135 streams, and a common sample format, sample rate and channel layout for
22136 audio streams, but other settings, such as resolution, must be converted
22137 explicitly by the user.
22139 Different frame rates are acceptable but will result in variable frame rate
22140 at output; be sure to configure the output file to handle it.
22142 @subsection Examples
22146 Concatenate an opening, an episode and an ending, all in bilingual version
22147 (video in stream 0, audio in streams 1 and 2):
22149 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
22150 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
22151 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
22152 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
22156 Concatenate two parts, handling audio and video separately, using the
22157 (a)movie sources, and adjusting the resolution:
22159 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
22160 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
22161 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
22163 Note that a desync will happen at the stitch if the audio and video streams
22164 do not have exactly the same duration in the first file.
22168 @subsection Commands
22170 This filter supports the following commands:
22173 Close the current segment and step to the next one
22179 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
22180 level. By default, it logs a message at a frequency of 10Hz with the
22181 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
22182 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
22184 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
22185 sample format is double-precision floating point. The input stream will be converted to
22186 this specification, if needed. Users may need to insert aformat and/or aresample filters
22187 after this filter to obtain the original parameters.
22189 The filter also has a video output (see the @var{video} option) with a real
22190 time graph to observe the loudness evolution. The graphic contains the logged
22191 message mentioned above, so it is not printed anymore when this option is set,
22192 unless the verbose logging is set. The main graphing area contains the
22193 short-term loudness (3 seconds of analysis), and the gauge on the right is for
22194 the momentary loudness (400 milliseconds), but can optionally be configured
22195 to instead display short-term loudness (see @var{gauge}).
22197 The green area marks a +/- 1LU target range around the target loudness
22198 (-23LUFS by default, unless modified through @var{target}).
22200 More information about the Loudness Recommendation EBU R128 on
22201 @url{http://tech.ebu.ch/loudness}.
22203 The filter accepts the following options:
22208 Activate the video output. The audio stream is passed unchanged whether this
22209 option is set or no. The video stream will be the first output stream if
22210 activated. Default is @code{0}.
22213 Set the video size. This option is for video only. For the syntax of this
22215 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22216 Default and minimum resolution is @code{640x480}.
22219 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
22220 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
22221 other integer value between this range is allowed.
22224 Set metadata injection. If set to @code{1}, the audio input will be segmented
22225 into 100ms output frames, each of them containing various loudness information
22226 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
22228 Default is @code{0}.
22231 Force the frame logging level.
22233 Available values are:
22236 information logging level
22238 verbose logging level
22241 By default, the logging level is set to @var{info}. If the @option{video} or
22242 the @option{metadata} options are set, it switches to @var{verbose}.
22247 Available modes can be cumulated (the option is a @code{flag} type). Possible
22251 Disable any peak mode (default).
22253 Enable sample-peak mode.
22255 Simple peak mode looking for the higher sample value. It logs a message
22256 for sample-peak (identified by @code{SPK}).
22258 Enable true-peak mode.
22260 If enabled, the peak lookup is done on an over-sampled version of the input
22261 stream for better peak accuracy. It logs a message for true-peak.
22262 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
22263 This mode requires a build with @code{libswresample}.
22267 Treat mono input files as "dual mono". If a mono file is intended for playback
22268 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
22269 If set to @code{true}, this option will compensate for this effect.
22270 Multi-channel input files are not affected by this option.
22273 Set a specific pan law to be used for the measurement of dual mono files.
22274 This parameter is optional, and has a default value of -3.01dB.
22277 Set a specific target level (in LUFS) used as relative zero in the visualization.
22278 This parameter is optional and has a default value of -23LUFS as specified
22279 by EBU R128. However, material published online may prefer a level of -16LUFS
22280 (e.g. for use with podcasts or video platforms).
22283 Set the value displayed by the gauge. Valid values are @code{momentary} and s
22284 @code{shortterm}. By default the momentary value will be used, but in certain
22285 scenarios it may be more useful to observe the short term value instead (e.g.
22289 Sets the display scale for the loudness. Valid parameters are @code{absolute}
22290 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
22291 video output, not the summary or continuous log output.
22294 @subsection Examples
22298 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
22300 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
22304 Run an analysis with @command{ffmpeg}:
22306 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
22310 @section interleave, ainterleave
22312 Temporally interleave frames from several inputs.
22314 @code{interleave} works with video inputs, @code{ainterleave} with audio.
22316 These filters read frames from several inputs and send the oldest
22317 queued frame to the output.
22319 Input streams must have well defined, monotonically increasing frame
22322 In order to submit one frame to output, these filters need to enqueue
22323 at least one frame for each input, so they cannot work in case one
22324 input is not yet terminated and will not receive incoming frames.
22326 For example consider the case when one input is a @code{select} filter
22327 which always drops input frames. The @code{interleave} filter will keep
22328 reading from that input, but it will never be able to send new frames
22329 to output until the input sends an end-of-stream signal.
22331 Also, depending on inputs synchronization, the filters will drop
22332 frames in case one input receives more frames than the other ones, and
22333 the queue is already filled.
22335 These filters accept the following options:
22339 Set the number of different inputs, it is 2 by default.
22342 @subsection Examples
22346 Interleave frames belonging to different streams using @command{ffmpeg}:
22348 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
22352 Add flickering blur effect:
22354 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
22358 @section metadata, ametadata
22360 Manipulate frame metadata.
22362 This filter accepts the following options:
22366 Set mode of operation of the filter.
22368 Can be one of the following:
22372 If both @code{value} and @code{key} is set, select frames
22373 which have such metadata. If only @code{key} is set, select
22374 every frame that has such key in metadata.
22377 Add new metadata @code{key} and @code{value}. If key is already available
22381 Modify value of already present key.
22384 If @code{value} is set, delete only keys that have such value.
22385 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
22389 Print key and its value if metadata was found. If @code{key} is not set print all
22390 metadata values available in frame.
22394 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
22397 Set metadata value which will be used. This option is mandatory for
22398 @code{modify} and @code{add} mode.
22401 Which function to use when comparing metadata value and @code{value}.
22403 Can be one of following:
22407 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
22410 Values are interpreted as strings, returns true if metadata value starts with
22411 the @code{value} option string.
22414 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
22417 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
22420 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
22423 Values are interpreted as floats, returns true if expression from option @code{expr}
22427 Values are interpreted as strings, returns true if metadata value ends with
22428 the @code{value} option string.
22432 Set expression which is used when @code{function} is set to @code{expr}.
22433 The expression is evaluated through the eval API and can contain the following
22438 Float representation of @code{value} from metadata key.
22441 Float representation of @code{value} as supplied by user in @code{value} option.
22445 If specified in @code{print} mode, output is written to the named file. Instead of
22446 plain filename any writable url can be specified. Filename ``-'' is a shorthand
22447 for standard output. If @code{file} option is not set, output is written to the log
22448 with AV_LOG_INFO loglevel.
22452 @subsection Examples
22456 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
22459 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
22462 Print silencedetect output to file @file{metadata.txt}.
22464 silencedetect,ametadata=mode=print:file=metadata.txt
22467 Direct all metadata to a pipe with file descriptor 4.
22469 metadata=mode=print:file='pipe\:4'
22473 @section perms, aperms
22475 Set read/write permissions for the output frames.
22477 These filters are mainly aimed at developers to test direct path in the
22478 following filter in the filtergraph.
22480 The filters accept the following options:
22484 Select the permissions mode.
22486 It accepts the following values:
22489 Do nothing. This is the default.
22491 Set all the output frames read-only.
22493 Set all the output frames directly writable.
22495 Make the frame read-only if writable, and writable if read-only.
22497 Set each output frame read-only or writable randomly.
22501 Set the seed for the @var{random} mode, must be an integer included between
22502 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
22503 @code{-1}, the filter will try to use a good random seed on a best effort
22507 Note: in case of auto-inserted filter between the permission filter and the
22508 following one, the permission might not be received as expected in that
22509 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
22510 perms/aperms filter can avoid this problem.
22512 @section realtime, arealtime
22514 Slow down filtering to match real time approximately.
22516 These filters will pause the filtering for a variable amount of time to
22517 match the output rate with the input timestamps.
22518 They are similar to the @option{re} option to @code{ffmpeg}.
22520 They accept the following options:
22524 Time limit for the pauses. Any pause longer than that will be considered
22525 a timestamp discontinuity and reset the timer. Default is 2 seconds.
22527 Speed factor for processing. The value must be a float larger than zero.
22528 Values larger than 1.0 will result in faster than realtime processing,
22529 smaller will slow processing down. The @var{limit} is automatically adapted
22530 accordingly. Default is 1.0.
22532 A processing speed faster than what is possible without these filters cannot
22537 @section select, aselect
22539 Select frames to pass in output.
22541 This filter accepts the following options:
22546 Set expression, which is evaluated for each input frame.
22548 If the expression is evaluated to zero, the frame is discarded.
22550 If the evaluation result is negative or NaN, the frame is sent to the
22551 first output; otherwise it is sent to the output with index
22552 @code{ceil(val)-1}, assuming that the input index starts from 0.
22554 For example a value of @code{1.2} corresponds to the output with index
22555 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
22558 Set the number of outputs. The output to which to send the selected
22559 frame is based on the result of the evaluation. Default value is 1.
22562 The expression can contain the following constants:
22566 The (sequential) number of the filtered frame, starting from 0.
22569 The (sequential) number of the selected frame, starting from 0.
22571 @item prev_selected_n
22572 The sequential number of the last selected frame. It's NAN if undefined.
22575 The timebase of the input timestamps.
22578 The PTS (Presentation TimeStamp) of the filtered video frame,
22579 expressed in @var{TB} units. It's NAN if undefined.
22582 The PTS of the filtered video frame,
22583 expressed in seconds. It's NAN if undefined.
22586 The PTS of the previously filtered video frame. It's NAN if undefined.
22588 @item prev_selected_pts
22589 The PTS of the last previously filtered video frame. It's NAN if undefined.
22591 @item prev_selected_t
22592 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
22595 The PTS of the first video frame in the video. It's NAN if undefined.
22598 The time of the first video frame in the video. It's NAN if undefined.
22600 @item pict_type @emph{(video only)}
22601 The type of the filtered frame. It can assume one of the following
22613 @item interlace_type @emph{(video only)}
22614 The frame interlace type. It can assume one of the following values:
22617 The frame is progressive (not interlaced).
22619 The frame is top-field-first.
22621 The frame is bottom-field-first.
22624 @item consumed_sample_n @emph{(audio only)}
22625 the number of selected samples before the current frame
22627 @item samples_n @emph{(audio only)}
22628 the number of samples in the current frame
22630 @item sample_rate @emph{(audio only)}
22631 the input sample rate
22634 This is 1 if the filtered frame is a key-frame, 0 otherwise.
22637 the position in the file of the filtered frame, -1 if the information
22638 is not available (e.g. for synthetic video)
22640 @item scene @emph{(video only)}
22641 value between 0 and 1 to indicate a new scene; a low value reflects a low
22642 probability for the current frame to introduce a new scene, while a higher
22643 value means the current frame is more likely to be one (see the example below)
22645 @item concatdec_select
22646 The concat demuxer can select only part of a concat input file by setting an
22647 inpoint and an outpoint, but the output packets may not be entirely contained
22648 in the selected interval. By using this variable, it is possible to skip frames
22649 generated by the concat demuxer which are not exactly contained in the selected
22652 This works by comparing the frame pts against the @var{lavf.concat.start_time}
22653 and the @var{lavf.concat.duration} packet metadata values which are also
22654 present in the decoded frames.
22656 The @var{concatdec_select} variable is -1 if the frame pts is at least
22657 start_time and either the duration metadata is missing or the frame pts is less
22658 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
22661 That basically means that an input frame is selected if its pts is within the
22662 interval set by the concat demuxer.
22666 The default value of the select expression is "1".
22668 @subsection Examples
22672 Select all frames in input:
22677 The example above is the same as:
22689 Select only I-frames:
22691 select='eq(pict_type\,I)'
22695 Select one frame every 100:
22697 select='not(mod(n\,100))'
22701 Select only frames contained in the 10-20 time interval:
22703 select=between(t\,10\,20)
22707 Select only I-frames contained in the 10-20 time interval:
22709 select=between(t\,10\,20)*eq(pict_type\,I)
22713 Select frames with a minimum distance of 10 seconds:
22715 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
22719 Use aselect to select only audio frames with samples number > 100:
22721 aselect='gt(samples_n\,100)'
22725 Create a mosaic of the first scenes:
22727 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
22730 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
22734 Send even and odd frames to separate outputs, and compose them:
22736 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
22740 Select useful frames from an ffconcat file which is using inpoints and
22741 outpoints but where the source files are not intra frame only.
22743 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
22747 @section sendcmd, asendcmd
22749 Send commands to filters in the filtergraph.
22751 These filters read commands to be sent to other filters in the
22754 @code{sendcmd} must be inserted between two video filters,
22755 @code{asendcmd} must be inserted between two audio filters, but apart
22756 from that they act the same way.
22758 The specification of commands can be provided in the filter arguments
22759 with the @var{commands} option, or in a file specified by the
22760 @var{filename} option.
22762 These filters accept the following options:
22765 Set the commands to be read and sent to the other filters.
22767 Set the filename of the commands to be read and sent to the other
22771 @subsection Commands syntax
22773 A commands description consists of a sequence of interval
22774 specifications, comprising a list of commands to be executed when a
22775 particular event related to that interval occurs. The occurring event
22776 is typically the current frame time entering or leaving a given time
22779 An interval is specified by the following syntax:
22781 @var{START}[-@var{END}] @var{COMMANDS};
22784 The time interval is specified by the @var{START} and @var{END} times.
22785 @var{END} is optional and defaults to the maximum time.
22787 The current frame time is considered within the specified interval if
22788 it is included in the interval [@var{START}, @var{END}), that is when
22789 the time is greater or equal to @var{START} and is lesser than
22792 @var{COMMANDS} consists of a sequence of one or more command
22793 specifications, separated by ",", relating to that interval. The
22794 syntax of a command specification is given by:
22796 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
22799 @var{FLAGS} is optional and specifies the type of events relating to
22800 the time interval which enable sending the specified command, and must
22801 be a non-null sequence of identifier flags separated by "+" or "|" and
22802 enclosed between "[" and "]".
22804 The following flags are recognized:
22807 The command is sent when the current frame timestamp enters the
22808 specified interval. In other words, the command is sent when the
22809 previous frame timestamp was not in the given interval, and the
22813 The command is sent when the current frame timestamp leaves the
22814 specified interval. In other words, the command is sent when the
22815 previous frame timestamp was in the given interval, and the
22819 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
22822 @var{TARGET} specifies the target of the command, usually the name of
22823 the filter class or a specific filter instance name.
22825 @var{COMMAND} specifies the name of the command for the target filter.
22827 @var{ARG} is optional and specifies the optional list of argument for
22828 the given @var{COMMAND}.
22830 Between one interval specification and another, whitespaces, or
22831 sequences of characters starting with @code{#} until the end of line,
22832 are ignored and can be used to annotate comments.
22834 A simplified BNF description of the commands specification syntax
22837 @var{COMMAND_FLAG} ::= "enter" | "leave"
22838 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
22839 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
22840 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
22841 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
22842 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
22845 @subsection Examples
22849 Specify audio tempo change at second 4:
22851 asendcmd=c='4.0 atempo tempo 1.5',atempo
22855 Target a specific filter instance:
22857 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
22861 Specify a list of drawtext and hue commands in a file.
22863 # show text in the interval 5-10
22864 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
22865 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
22867 # desaturate the image in the interval 15-20
22868 15.0-20.0 [enter] hue s 0,
22869 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
22871 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
22873 # apply an exponential saturation fade-out effect, starting from time 25
22874 25 [enter] hue s exp(25-t)
22877 A filtergraph allowing to read and process the above command list
22878 stored in a file @file{test.cmd}, can be specified with:
22880 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
22885 @section setpts, asetpts
22887 Change the PTS (presentation timestamp) of the input frames.
22889 @code{setpts} works on video frames, @code{asetpts} on audio frames.
22891 This filter accepts the following options:
22896 The expression which is evaluated for each frame to construct its timestamp.
22900 The expression is evaluated through the eval API and can contain the following
22904 @item FRAME_RATE, FR
22905 frame rate, only defined for constant frame-rate video
22908 The presentation timestamp in input
22911 The count of the input frame for video or the number of consumed samples,
22912 not including the current frame for audio, starting from 0.
22914 @item NB_CONSUMED_SAMPLES
22915 The number of consumed samples, not including the current frame (only
22918 @item NB_SAMPLES, S
22919 The number of samples in the current frame (only audio)
22921 @item SAMPLE_RATE, SR
22922 The audio sample rate.
22925 The PTS of the first frame.
22928 the time in seconds of the first frame
22931 State whether the current frame is interlaced.
22934 the time in seconds of the current frame
22937 original position in the file of the frame, or undefined if undefined
22938 for the current frame
22941 The previous input PTS.
22944 previous input time in seconds
22947 The previous output PTS.
22950 previous output time in seconds
22953 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
22957 The wallclock (RTC) time at the start of the movie in microseconds.
22960 The timebase of the input timestamps.
22964 @subsection Examples
22968 Start counting PTS from zero
22970 setpts=PTS-STARTPTS
22974 Apply fast motion effect:
22980 Apply slow motion effect:
22986 Set fixed rate of 25 frames per second:
22992 Set fixed rate 25 fps with some jitter:
22994 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
22998 Apply an offset of 10 seconds to the input PTS:
23004 Generate timestamps from a "live source" and rebase onto the current timebase:
23006 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
23010 Generate timestamps by counting samples:
23019 Force color range for the output video frame.
23021 The @code{setrange} filter marks the color range property for the
23022 output frames. It does not change the input frame, but only sets the
23023 corresponding property, which affects how the frame is treated by
23026 The filter accepts the following options:
23031 Available values are:
23035 Keep the same color range property.
23037 @item unspecified, unknown
23038 Set the color range as unspecified.
23040 @item limited, tv, mpeg
23041 Set the color range as limited.
23043 @item full, pc, jpeg
23044 Set the color range as full.
23048 @section settb, asettb
23050 Set the timebase to use for the output frames timestamps.
23051 It is mainly useful for testing timebase configuration.
23053 It accepts the following parameters:
23058 The expression which is evaluated into the output timebase.
23062 The value for @option{tb} is an arithmetic expression representing a
23063 rational. The expression can contain the constants "AVTB" (the default
23064 timebase), "intb" (the input timebase) and "sr" (the sample rate,
23065 audio only). Default value is "intb".
23067 @subsection Examples
23071 Set the timebase to 1/25:
23077 Set the timebase to 1/10:
23083 Set the timebase to 1001/1000:
23089 Set the timebase to 2*intb:
23095 Set the default timebase value:
23102 Convert input audio to a video output representing frequency spectrum
23103 logarithmically using Brown-Puckette constant Q transform algorithm with
23104 direct frequency domain coefficient calculation (but the transform itself
23105 is not really constant Q, instead the Q factor is actually variable/clamped),
23106 with musical tone scale, from E0 to D#10.
23108 The filter accepts the following options:
23112 Specify the video size for the output. It must be even. For the syntax of this option,
23113 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23114 Default value is @code{1920x1080}.
23117 Set the output frame rate. Default value is @code{25}.
23120 Set the bargraph height. It must be even. Default value is @code{-1} which
23121 computes the bargraph height automatically.
23124 Set the axis height. It must be even. Default value is @code{-1} which computes
23125 the axis height automatically.
23128 Set the sonogram height. It must be even. Default value is @code{-1} which
23129 computes the sonogram height automatically.
23132 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
23133 instead. Default value is @code{1}.
23135 @item sono_v, volume
23136 Specify the sonogram volume expression. It can contain variables:
23139 the @var{bar_v} evaluated expression
23140 @item frequency, freq, f
23141 the frequency where it is evaluated
23142 @item timeclamp, tc
23143 the value of @var{timeclamp} option
23147 @item a_weighting(f)
23148 A-weighting of equal loudness
23149 @item b_weighting(f)
23150 B-weighting of equal loudness
23151 @item c_weighting(f)
23152 C-weighting of equal loudness.
23154 Default value is @code{16}.
23156 @item bar_v, volume2
23157 Specify the bargraph volume expression. It can contain variables:
23160 the @var{sono_v} evaluated expression
23161 @item frequency, freq, f
23162 the frequency where it is evaluated
23163 @item timeclamp, tc
23164 the value of @var{timeclamp} option
23168 @item a_weighting(f)
23169 A-weighting of equal loudness
23170 @item b_weighting(f)
23171 B-weighting of equal loudness
23172 @item c_weighting(f)
23173 C-weighting of equal loudness.
23175 Default value is @code{sono_v}.
23177 @item sono_g, gamma
23178 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
23179 higher gamma makes the spectrum having more range. Default value is @code{3}.
23180 Acceptable range is @code{[1, 7]}.
23182 @item bar_g, gamma2
23183 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
23187 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
23188 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
23190 @item timeclamp, tc
23191 Specify the transform timeclamp. At low frequency, there is trade-off between
23192 accuracy in time domain and frequency domain. If timeclamp is lower,
23193 event in time domain is represented more accurately (such as fast bass drum),
23194 otherwise event in frequency domain is represented more accurately
23195 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
23198 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
23199 limits future samples by applying asymmetric windowing in time domain, useful
23200 when low latency is required. Accepted range is @code{[0, 1]}.
23203 Specify the transform base frequency. Default value is @code{20.01523126408007475},
23204 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
23207 Specify the transform end frequency. Default value is @code{20495.59681441799654},
23208 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
23211 This option is deprecated and ignored.
23214 Specify the transform length in time domain. Use this option to control accuracy
23215 trade-off between time domain and frequency domain at every frequency sample.
23216 It can contain variables:
23218 @item frequency, freq, f
23219 the frequency where it is evaluated
23220 @item timeclamp, tc
23221 the value of @var{timeclamp} option.
23223 Default value is @code{384*tc/(384+tc*f)}.
23226 Specify the transform count for every video frame. Default value is @code{6}.
23227 Acceptable range is @code{[1, 30]}.
23230 Specify the transform count for every single pixel. Default value is @code{0},
23231 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
23234 Specify font file for use with freetype to draw the axis. If not specified,
23235 use embedded font. Note that drawing with font file or embedded font is not
23236 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
23240 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
23241 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
23245 Specify font color expression. This is arithmetic expression that should return
23246 integer value 0xRRGGBB. It can contain variables:
23248 @item frequency, freq, f
23249 the frequency where it is evaluated
23250 @item timeclamp, tc
23251 the value of @var{timeclamp} option
23256 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
23257 @item r(x), g(x), b(x)
23258 red, green, and blue value of intensity x.
23260 Default value is @code{st(0, (midi(f)-59.5)/12);
23261 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
23262 r(1-ld(1)) + b(ld(1))}.
23265 Specify image file to draw the axis. This option override @var{fontfile} and
23266 @var{fontcolor} option.
23269 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
23270 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
23271 Default value is @code{1}.
23274 Set colorspace. The accepted values are:
23277 Unspecified (default)
23286 BT.470BG or BT.601-6 625
23289 SMPTE-170M or BT.601-6 525
23295 BT.2020 with non-constant luminance
23300 Set spectrogram color scheme. This is list of floating point values with format
23301 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
23302 The default is @code{1|0.5|0|0|0.5|1}.
23306 @subsection Examples
23310 Playing audio while showing the spectrum:
23312 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
23316 Same as above, but with frame rate 30 fps:
23318 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
23322 Playing at 1280x720:
23324 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
23328 Disable sonogram display:
23334 A1 and its harmonics: A1, A2, (near)E3, A3:
23336 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),
23337 asplit[a][out1]; [a] showcqt [out0]'
23341 Same as above, but with more accuracy in frequency domain:
23343 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),
23344 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
23350 bar_v=10:sono_v=bar_v*a_weighting(f)
23354 Custom gamma, now spectrum is linear to the amplitude.
23360 Custom tlength equation:
23362 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)))'
23366 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
23368 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
23372 Custom font using fontconfig:
23374 font='Courier New,Monospace,mono|bold'
23378 Custom frequency range with custom axis using image file:
23380 axisfile=myaxis.png:basefreq=40:endfreq=10000
23386 Convert input audio to video output representing the audio power spectrum.
23387 Audio amplitude is on Y-axis while frequency is on X-axis.
23389 The filter accepts the following options:
23393 Specify size of video. For the syntax of this option, check the
23394 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23395 Default is @code{1024x512}.
23399 This set how each frequency bin will be represented.
23401 It accepts the following values:
23407 Default is @code{bar}.
23410 Set amplitude scale.
23412 It accepts the following values:
23426 Default is @code{log}.
23429 Set frequency scale.
23431 It accepts the following values:
23440 Reverse logarithmic scale.
23442 Default is @code{lin}.
23445 Set window size. Allowed range is from 16 to 65536.
23447 Default is @code{2048}
23450 Set windowing function.
23452 It accepts the following values:
23475 Default is @code{hanning}.
23478 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
23479 which means optimal overlap for selected window function will be picked.
23482 Set time averaging. Setting this to 0 will display current maximal peaks.
23483 Default is @code{1}, which means time averaging is disabled.
23486 Specify list of colors separated by space or by '|' which will be used to
23487 draw channel frequencies. Unrecognized or missing colors will be replaced
23491 Set channel display mode.
23493 It accepts the following values:
23498 Default is @code{combined}.
23501 Set minimum amplitude used in @code{log} amplitude scaler.
23505 @section showspatial
23507 Convert stereo input audio to a video output, representing the spatial relationship
23508 between two channels.
23510 The filter accepts the following options:
23514 Specify the video size for the output. For the syntax of this option, check the
23515 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23516 Default value is @code{512x512}.
23519 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
23522 Set window function.
23524 It accepts the following values:
23549 Default value is @code{hann}.
23552 Set ratio of overlap window. Default value is @code{0.5}.
23553 When value is @code{1} overlap is set to recommended size for specific
23554 window function currently used.
23557 @anchor{showspectrum}
23558 @section showspectrum
23560 Convert input audio to a video output, representing the audio frequency
23563 The filter accepts the following options:
23567 Specify the video size for the output. For the syntax of this option, check the
23568 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23569 Default value is @code{640x512}.
23572 Specify how the spectrum should slide along the window.
23574 It accepts the following values:
23577 the samples start again on the left when they reach the right
23579 the samples scroll from right to left
23581 frames are only produced when the samples reach the right
23583 the samples scroll from left to right
23586 Default value is @code{replace}.
23589 Specify display mode.
23591 It accepts the following values:
23594 all channels are displayed in the same row
23596 all channels are displayed in separate rows
23599 Default value is @samp{combined}.
23602 Specify display color mode.
23604 It accepts the following values:
23607 each channel is displayed in a separate color
23609 each channel is displayed using the same color scheme
23611 each channel is displayed using the rainbow color scheme
23613 each channel is displayed using the moreland color scheme
23615 each channel is displayed using the nebulae color scheme
23617 each channel is displayed using the fire color scheme
23619 each channel is displayed using the fiery color scheme
23621 each channel is displayed using the fruit color scheme
23623 each channel is displayed using the cool color scheme
23625 each channel is displayed using the magma color scheme
23627 each channel is displayed using the green color scheme
23629 each channel is displayed using the viridis color scheme
23631 each channel is displayed using the plasma color scheme
23633 each channel is displayed using the cividis color scheme
23635 each channel is displayed using the terrain color scheme
23638 Default value is @samp{channel}.
23641 Specify scale used for calculating intensity color values.
23643 It accepts the following values:
23648 square root, default
23659 Default value is @samp{sqrt}.
23662 Specify frequency scale.
23664 It accepts the following values:
23672 Default value is @samp{lin}.
23675 Set saturation modifier for displayed colors. Negative values provide
23676 alternative color scheme. @code{0} is no saturation at all.
23677 Saturation must be in [-10.0, 10.0] range.
23678 Default value is @code{1}.
23681 Set window function.
23683 It accepts the following values:
23708 Default value is @code{hann}.
23711 Set orientation of time vs frequency axis. Can be @code{vertical} or
23712 @code{horizontal}. Default is @code{vertical}.
23715 Set ratio of overlap window. Default value is @code{0}.
23716 When value is @code{1} overlap is set to recommended size for specific
23717 window function currently used.
23720 Set scale gain for calculating intensity color values.
23721 Default value is @code{1}.
23724 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
23727 Set color rotation, must be in [-1.0, 1.0] range.
23728 Default value is @code{0}.
23731 Set start frequency from which to display spectrogram. Default is @code{0}.
23734 Set stop frequency to which to display spectrogram. Default is @code{0}.
23737 Set upper frame rate limit. Default is @code{auto}, unlimited.
23740 Draw time and frequency axes and legends. Default is disabled.
23743 The usage is very similar to the showwaves filter; see the examples in that
23746 @subsection Examples
23750 Large window with logarithmic color scaling:
23752 showspectrum=s=1280x480:scale=log
23756 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
23758 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
23759 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
23763 @section showspectrumpic
23765 Convert input audio to a single video frame, representing the audio frequency
23768 The filter accepts the following options:
23772 Specify the video size for the output. For the syntax of this option, check the
23773 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23774 Default value is @code{4096x2048}.
23777 Specify display mode.
23779 It accepts the following values:
23782 all channels are displayed in the same row
23784 all channels are displayed in separate rows
23786 Default value is @samp{combined}.
23789 Specify display color mode.
23791 It accepts the following values:
23794 each channel is displayed in a separate color
23796 each channel is displayed using the same color scheme
23798 each channel is displayed using the rainbow color scheme
23800 each channel is displayed using the moreland color scheme
23802 each channel is displayed using the nebulae color scheme
23804 each channel is displayed using the fire color scheme
23806 each channel is displayed using the fiery color scheme
23808 each channel is displayed using the fruit color scheme
23810 each channel is displayed using the cool color scheme
23812 each channel is displayed using the magma color scheme
23814 each channel is displayed using the green color scheme
23816 each channel is displayed using the viridis color scheme
23818 each channel is displayed using the plasma color scheme
23820 each channel is displayed using the cividis color scheme
23822 each channel is displayed using the terrain color scheme
23824 Default value is @samp{intensity}.
23827 Specify scale used for calculating intensity color values.
23829 It accepts the following values:
23834 square root, default
23844 Default value is @samp{log}.
23847 Specify frequency scale.
23849 It accepts the following values:
23857 Default value is @samp{lin}.
23860 Set saturation modifier for displayed colors. Negative values provide
23861 alternative color scheme. @code{0} is no saturation at all.
23862 Saturation must be in [-10.0, 10.0] range.
23863 Default value is @code{1}.
23866 Set window function.
23868 It accepts the following values:
23892 Default value is @code{hann}.
23895 Set orientation of time vs frequency axis. Can be @code{vertical} or
23896 @code{horizontal}. Default is @code{vertical}.
23899 Set scale gain for calculating intensity color values.
23900 Default value is @code{1}.
23903 Draw time and frequency axes and legends. Default is enabled.
23906 Set color rotation, must be in [-1.0, 1.0] range.
23907 Default value is @code{0}.
23910 Set start frequency from which to display spectrogram. Default is @code{0}.
23913 Set stop frequency to which to display spectrogram. Default is @code{0}.
23916 @subsection Examples
23920 Extract an audio spectrogram of a whole audio track
23921 in a 1024x1024 picture using @command{ffmpeg}:
23923 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
23927 @section showvolume
23929 Convert input audio volume to a video output.
23931 The filter accepts the following options:
23938 Set border width, allowed range is [0, 5]. Default is 1.
23941 Set channel width, allowed range is [80, 8192]. Default is 400.
23944 Set channel height, allowed range is [1, 900]. Default is 20.
23947 Set fade, allowed range is [0, 1]. Default is 0.95.
23950 Set volume color expression.
23952 The expression can use the following variables:
23956 Current max volume of channel in dB.
23962 Current channel number, starting from 0.
23966 If set, displays channel names. Default is enabled.
23969 If set, displays volume values. Default is enabled.
23972 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
23973 default is @code{h}.
23976 Set step size, allowed range is [0, 5]. Default is 0, which means
23980 Set background opacity, allowed range is [0, 1]. Default is 0.
23983 Set metering mode, can be peak: @code{p} or rms: @code{r},
23984 default is @code{p}.
23987 Set display scale, can be linear: @code{lin} or log: @code{log},
23988 default is @code{lin}.
23992 If set to > 0., display a line for the max level
23993 in the previous seconds.
23994 default is disabled: @code{0.}
23997 The color of the max line. Use when @code{dm} option is set to > 0.
23998 default is: @code{orange}
24003 Convert input audio to a video output, representing the samples waves.
24005 The filter accepts the following options:
24009 Specify the video size for the output. For the syntax of this option, check the
24010 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24011 Default value is @code{600x240}.
24016 Available values are:
24019 Draw a point for each sample.
24022 Draw a vertical line for each sample.
24025 Draw a point for each sample and a line between them.
24028 Draw a centered vertical line for each sample.
24031 Default value is @code{point}.
24034 Set the number of samples which are printed on the same column. A
24035 larger value will decrease the frame rate. Must be a positive
24036 integer. This option can be set only if the value for @var{rate}
24037 is not explicitly specified.
24040 Set the (approximate) output frame rate. This is done by setting the
24041 option @var{n}. Default value is "25".
24043 @item split_channels
24044 Set if channels should be drawn separately or overlap. Default value is 0.
24047 Set colors separated by '|' which are going to be used for drawing of each channel.
24050 Set amplitude scale.
24052 Available values are:
24070 Set the draw mode. This is mostly useful to set for high @var{n}.
24072 Available values are:
24075 Scale pixel values for each drawn sample.
24078 Draw every sample directly.
24081 Default value is @code{scale}.
24084 @subsection Examples
24088 Output the input file audio and the corresponding video representation
24091 amovie=a.mp3,asplit[out0],showwaves[out1]
24095 Create a synthetic signal and show it with showwaves, forcing a
24096 frame rate of 30 frames per second:
24098 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
24102 @section showwavespic
24104 Convert input audio to a single video frame, representing the samples waves.
24106 The filter accepts the following options:
24110 Specify the video size for the output. For the syntax of this option, check the
24111 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24112 Default value is @code{600x240}.
24114 @item split_channels
24115 Set if channels should be drawn separately or overlap. Default value is 0.
24118 Set colors separated by '|' which are going to be used for drawing of each channel.
24121 Set amplitude scale.
24123 Available values are:
24143 Available values are:
24146 Scale pixel values for each drawn sample.
24149 Draw every sample directly.
24152 Default value is @code{scale}.
24155 @subsection Examples
24159 Extract a channel split representation of the wave form of a whole audio track
24160 in a 1024x800 picture using @command{ffmpeg}:
24162 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
24166 @section sidedata, asidedata
24168 Delete frame side data, or select frames based on it.
24170 This filter accepts the following options:
24174 Set mode of operation of the filter.
24176 Can be one of the following:
24180 Select every frame with side data of @code{type}.
24183 Delete side data of @code{type}. If @code{type} is not set, delete all side
24189 Set side data type used with all modes. Must be set for @code{select} mode. For
24190 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
24191 in @file{libavutil/frame.h}. For example, to choose
24192 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
24196 @section spectrumsynth
24198 Synthesize audio from 2 input video spectrums, first input stream represents
24199 magnitude across time and second represents phase across time.
24200 The filter will transform from frequency domain as displayed in videos back
24201 to time domain as presented in audio output.
24203 This filter is primarily created for reversing processed @ref{showspectrum}
24204 filter outputs, but can synthesize sound from other spectrograms too.
24205 But in such case results are going to be poor if the phase data is not
24206 available, because in such cases phase data need to be recreated, usually
24207 it's just recreated from random noise.
24208 For best results use gray only output (@code{channel} color mode in
24209 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
24210 @code{lin} scale for phase video. To produce phase, for 2nd video, use
24211 @code{data} option. Inputs videos should generally use @code{fullframe}
24212 slide mode as that saves resources needed for decoding video.
24214 The filter accepts the following options:
24218 Specify sample rate of output audio, the sample rate of audio from which
24219 spectrum was generated may differ.
24222 Set number of channels represented in input video spectrums.
24225 Set scale which was used when generating magnitude input spectrum.
24226 Can be @code{lin} or @code{log}. Default is @code{log}.
24229 Set slide which was used when generating inputs spectrums.
24230 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
24231 Default is @code{fullframe}.
24234 Set window function used for resynthesis.
24237 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
24238 which means optimal overlap for selected window function will be picked.
24241 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
24242 Default is @code{vertical}.
24245 @subsection Examples
24249 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
24250 then resynthesize videos back to audio with spectrumsynth:
24252 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
24253 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
24254 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
24258 @section split, asplit
24260 Split input into several identical outputs.
24262 @code{asplit} works with audio input, @code{split} with video.
24264 The filter accepts a single parameter which specifies the number of outputs. If
24265 unspecified, it defaults to 2.
24267 @subsection Examples
24271 Create two separate outputs from the same input:
24273 [in] split [out0][out1]
24277 To create 3 or more outputs, you need to specify the number of
24280 [in] asplit=3 [out0][out1][out2]
24284 Create two separate outputs from the same input, one cropped and
24287 [in] split [splitout1][splitout2];
24288 [splitout1] crop=100:100:0:0 [cropout];
24289 [splitout2] pad=200:200:100:100 [padout];
24293 Create 5 copies of the input audio with @command{ffmpeg}:
24295 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
24301 Receive commands sent through a libzmq client, and forward them to
24302 filters in the filtergraph.
24304 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
24305 must be inserted between two video filters, @code{azmq} between two
24306 audio filters. Both are capable to send messages to any filter type.
24308 To enable these filters you need to install the libzmq library and
24309 headers and configure FFmpeg with @code{--enable-libzmq}.
24311 For more information about libzmq see:
24312 @url{http://www.zeromq.org/}
24314 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
24315 receives messages sent through a network interface defined by the
24316 @option{bind_address} (or the abbreviation "@option{b}") option.
24317 Default value of this option is @file{tcp://localhost:5555}. You may
24318 want to alter this value to your needs, but do not forget to escape any
24319 ':' signs (see @ref{filtergraph escaping}).
24321 The received message must be in the form:
24323 @var{TARGET} @var{COMMAND} [@var{ARG}]
24326 @var{TARGET} specifies the target of the command, usually the name of
24327 the filter class or a specific filter instance name. The default
24328 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
24329 but you can override this by using the @samp{filter_name@@id} syntax
24330 (see @ref{Filtergraph syntax}).
24332 @var{COMMAND} specifies the name of the command for the target filter.
24334 @var{ARG} is optional and specifies the optional argument list for the
24335 given @var{COMMAND}.
24337 Upon reception, the message is processed and the corresponding command
24338 is injected into the filtergraph. Depending on the result, the filter
24339 will send a reply to the client, adopting the format:
24341 @var{ERROR_CODE} @var{ERROR_REASON}
24345 @var{MESSAGE} is optional.
24347 @subsection Examples
24349 Look at @file{tools/zmqsend} for an example of a zmq client which can
24350 be used to send commands processed by these filters.
24352 Consider the following filtergraph generated by @command{ffplay}.
24353 In this example the last overlay filter has an instance name. All other
24354 filters will have default instance names.
24357 ffplay -dumpgraph 1 -f lavfi "
24358 color=s=100x100:c=red [l];
24359 color=s=100x100:c=blue [r];
24360 nullsrc=s=200x100, zmq [bg];
24361 [bg][l] overlay [bg+l];
24362 [bg+l][r] overlay@@my=x=100 "
24365 To change the color of the left side of the video, the following
24366 command can be used:
24368 echo Parsed_color_0 c yellow | tools/zmqsend
24371 To change the right side:
24373 echo Parsed_color_1 c pink | tools/zmqsend
24376 To change the position of the right side:
24378 echo overlay@@my x 150 | tools/zmqsend
24382 @c man end MULTIMEDIA FILTERS
24384 @chapter Multimedia Sources
24385 @c man begin MULTIMEDIA SOURCES
24387 Below is a description of the currently available multimedia sources.
24391 This is the same as @ref{movie} source, except it selects an audio
24397 Read audio and/or video stream(s) from a movie container.
24399 It accepts the following parameters:
24403 The name of the resource to read (not necessarily a file; it can also be a
24404 device or a stream accessed through some protocol).
24406 @item format_name, f
24407 Specifies the format assumed for the movie to read, and can be either
24408 the name of a container or an input device. If not specified, the
24409 format is guessed from @var{movie_name} or by probing.
24411 @item seek_point, sp
24412 Specifies the seek point in seconds. The frames will be output
24413 starting from this seek point. The parameter is evaluated with
24414 @code{av_strtod}, so the numerical value may be suffixed by an IS
24415 postfix. The default value is "0".
24418 Specifies the streams to read. Several streams can be specified,
24419 separated by "+". The source will then have as many outputs, in the
24420 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
24421 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
24422 respectively the default (best suited) video and audio stream. Default
24423 is "dv", or "da" if the filter is called as "amovie".
24425 @item stream_index, si
24426 Specifies the index of the video stream to read. If the value is -1,
24427 the most suitable video stream will be automatically selected. The default
24428 value is "-1". Deprecated. If the filter is called "amovie", it will select
24429 audio instead of video.
24432 Specifies how many times to read the stream in sequence.
24433 If the value is 0, the stream will be looped infinitely.
24434 Default value is "1".
24436 Note that when the movie is looped the source timestamps are not
24437 changed, so it will generate non monotonically increasing timestamps.
24439 @item discontinuity
24440 Specifies the time difference between frames above which the point is
24441 considered a timestamp discontinuity which is removed by adjusting the later
24445 It allows overlaying a second video on top of the main input of
24446 a filtergraph, as shown in this graph:
24448 input -----------> deltapts0 --> overlay --> output
24451 movie --> scale--> deltapts1 -------+
24453 @subsection Examples
24457 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
24458 on top of the input labelled "in":
24460 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
24461 [in] setpts=PTS-STARTPTS [main];
24462 [main][over] overlay=16:16 [out]
24466 Read from a video4linux2 device, and overlay it on top of the input
24469 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
24470 [in] setpts=PTS-STARTPTS [main];
24471 [main][over] overlay=16:16 [out]
24475 Read the first video stream and the audio stream with id 0x81 from
24476 dvd.vob; the video is connected to the pad named "video" and the audio is
24477 connected to the pad named "audio":
24479 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
24483 @subsection Commands
24485 Both movie and amovie support the following commands:
24488 Perform seek using "av_seek_frame".
24489 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
24492 @var{stream_index}: If stream_index is -1, a default
24493 stream is selected, and @var{timestamp} is automatically converted
24494 from AV_TIME_BASE units to the stream specific time_base.
24496 @var{timestamp}: Timestamp in AVStream.time_base units
24497 or, if no stream is specified, in AV_TIME_BASE units.
24499 @var{flags}: Flags which select direction and seeking mode.
24503 Get movie duration in AV_TIME_BASE units.
24507 @c man end MULTIMEDIA SOURCES