1 @chapter Filtering Introduction
2 @c man begin FILTERING INTRODUCTION
4 Filtering in FFmpeg is enabled through the libavfilter library.
6 In libavfilter, a filter can have multiple inputs and multiple
8 To illustrate the sorts of things that are possible, we consider the
13 input --> split ---------------------> overlay --> output
16 +-----> crop --> vflip -------+
19 This filtergraph splits the input stream in two streams, then sends one
20 stream through the crop filter and the vflip filter, before merging it
21 back with the other stream by overlaying it on top. You can use the
22 following command to achieve this:
25 ffmpeg -i INPUT -vf "split [main][tmp]; [tmp] crop=iw:ih/2:0:0, vflip [flip]; [main][flip] overlay=0:H/2" OUTPUT
28 The result will be that the top half of the video is mirrored
29 onto the bottom half of the output video.
31 Filters in the same linear chain are separated by commas, and distinct
32 linear chains of filters are separated by semicolons. In our example,
33 @var{crop,vflip} are in one linear chain, @var{split} and
34 @var{overlay} are separately in another. The points where the linear
35 chains join are labelled by names enclosed in square brackets. In the
36 example, the split filter generates two outputs that are associated to
37 the labels @var{[main]} and @var{[tmp]}.
39 The stream sent to the second output of @var{split}, labelled as
40 @var{[tmp]}, is processed through the @var{crop} filter, which crops
41 away the lower half part of the video, and then vertically flipped. The
42 @var{overlay} filter takes in input the first unchanged output of the
43 split filter (which was labelled as @var{[main]}), and overlay on its
44 lower half the output generated by the @var{crop,vflip} filterchain.
46 Some filters take in input a list of parameters: they are specified
47 after the filter name and an equal sign, and are separated from each other
50 There exist so-called @var{source filters} that do not have an
51 audio/video input, and @var{sink filters} that will not have audio/video
54 @c man end FILTERING INTRODUCTION
57 @c man begin GRAPH2DOT
59 The @file{graph2dot} program included in the FFmpeg @file{tools}
60 directory can be used to parse a filtergraph description and issue a
61 corresponding textual representation in the dot language.
68 to see how to use @file{graph2dot}.
70 You can then pass the dot description to the @file{dot} program (from
71 the graphviz suite of programs) and obtain a graphical representation
74 For example the sequence of commands:
76 echo @var{GRAPH_DESCRIPTION} | \
77 tools/graph2dot -o graph.tmp && \
78 dot -Tpng graph.tmp -o graph.png && \
82 can be used to create and display an image representing the graph
83 described by the @var{GRAPH_DESCRIPTION} string. Note that this string must be
84 a complete self-contained graph, with its inputs and outputs explicitly defined.
85 For example if your command line is of the form:
87 ffmpeg -i infile -vf scale=640:360 outfile
89 your @var{GRAPH_DESCRIPTION} string will need to be of the form:
91 nullsrc,scale=640:360,nullsink
93 you may also need to set the @var{nullsrc} parameters and add a @var{format}
94 filter in order to simulate a specific input file.
98 @chapter Filtergraph description
99 @c man begin FILTERGRAPH DESCRIPTION
101 A filtergraph is a directed graph of connected filters. It can contain
102 cycles, and there can be multiple links between a pair of
103 filters. Each link has one input pad on one side connecting it to one
104 filter from which it takes its input, and one output pad on the other
105 side connecting it to one filter accepting its output.
107 Each filter in a filtergraph is an instance of a filter class
108 registered in the application, which defines the features and the
109 number of input and output pads of the filter.
111 A filter with no input pads is called a "source", and a filter with no
112 output pads is called a "sink".
114 @anchor{Filtergraph syntax}
115 @section Filtergraph syntax
117 A filtergraph has a textual representation, which is recognized by the
118 @option{-filter}/@option{-vf}/@option{-af} and
119 @option{-filter_complex} options in @command{ffmpeg} and
120 @option{-vf}/@option{-af} in @command{ffplay}, and by the
121 @code{avfilter_graph_parse_ptr()} function defined in
122 @file{libavfilter/avfilter.h}.
124 A filterchain consists of a sequence of connected filters, each one
125 connected to the previous one in the sequence. A filterchain is
126 represented by a list of ","-separated filter descriptions.
128 A filtergraph consists of a sequence of filterchains. A sequence of
129 filterchains is represented by a list of ";"-separated filterchain
132 A filter is represented by a string of the form:
133 [@var{in_link_1}]...[@var{in_link_N}]@var{filter_name}@@@var{id}=@var{arguments}[@var{out_link_1}]...[@var{out_link_M}]
135 @var{filter_name} is the name of the filter class of which the
136 described filter is an instance of, and has to be the name of one of
137 the filter classes registered in the program optionally followed by "@@@var{id}".
138 The name of the filter class is optionally followed by a string
141 @var{arguments} is a string which contains the parameters used to
142 initialize the filter instance. It may have one of two forms:
146 A ':'-separated list of @var{key=value} pairs.
149 A ':'-separated list of @var{value}. In this case, the keys are assumed to be
150 the option names in the order they are declared. E.g. the @code{fade} filter
151 declares three options in this order -- @option{type}, @option{start_frame} and
152 @option{nb_frames}. Then the parameter list @var{in:0:30} means that the value
153 @var{in} is assigned to the option @option{type}, @var{0} to
154 @option{start_frame} and @var{30} to @option{nb_frames}.
157 A ':'-separated list of mixed direct @var{value} and long @var{key=value}
158 pairs. The direct @var{value} must precede the @var{key=value} pairs, and
159 follow the same constraints order of the previous point. The following
160 @var{key=value} pairs can be set in any preferred order.
164 If the option value itself is a list of items (e.g. the @code{format} filter
165 takes a list of pixel formats), the items in the list are usually separated by
168 The list of arguments can be quoted using the character @samp{'} as initial
169 and ending mark, and the character @samp{\} for escaping the characters
170 within the quoted text; otherwise the argument string is considered
171 terminated when the next special character (belonging to the set
172 @samp{[]=;,}) is encountered.
174 The name and arguments of the filter are optionally preceded and
175 followed by a list of link labels.
176 A link label allows one to name a link and associate it to a filter output
177 or input pad. The preceding labels @var{in_link_1}
178 ... @var{in_link_N}, are associated to the filter input pads,
179 the following labels @var{out_link_1} ... @var{out_link_M}, are
180 associated to the output pads.
182 When two link labels with the same name are found in the
183 filtergraph, a link between the corresponding input and output pad is
186 If an output pad is not labelled, it is linked by default to the first
187 unlabelled input pad of the next filter in the filterchain.
188 For example in the filterchain
190 nullsrc, split[L1], [L2]overlay, nullsink
192 the split filter instance has two output pads, and the overlay filter
193 instance two input pads. The first output pad of split is labelled
194 "L1", the first input pad of overlay is labelled "L2", and the second
195 output pad of split is linked to the second input pad of overlay,
196 which are both unlabelled.
198 In a filter description, if the input label of the first filter is not
199 specified, "in" is assumed; if the output label of the last filter is not
200 specified, "out" is assumed.
202 In a complete filterchain all the unlabelled filter input and output
203 pads must be connected. A filtergraph is considered valid if all the
204 filter input and output pads of all the filterchains are connected.
206 Libavfilter will automatically insert @ref{scale} filters where format
207 conversion is required. It is possible to specify swscale flags
208 for those automatically inserted scalers by prepending
209 @code{sws_flags=@var{flags};}
210 to the filtergraph description.
212 Here is a BNF description of the filtergraph syntax:
214 @var{NAME} ::= sequence of alphanumeric characters and '_'
215 @var{FILTER_NAME} ::= @var{NAME}["@@"@var{NAME}]
216 @var{LINKLABEL} ::= "[" @var{NAME} "]"
217 @var{LINKLABELS} ::= @var{LINKLABEL} [@var{LINKLABELS}]
218 @var{FILTER_ARGUMENTS} ::= sequence of chars (possibly quoted)
219 @var{FILTER} ::= [@var{LINKLABELS}] @var{FILTER_NAME} ["=" @var{FILTER_ARGUMENTS}] [@var{LINKLABELS}]
220 @var{FILTERCHAIN} ::= @var{FILTER} [,@var{FILTERCHAIN}]
221 @var{FILTERGRAPH} ::= [sws_flags=@var{flags};] @var{FILTERCHAIN} [;@var{FILTERGRAPH}]
224 @anchor{filtergraph escaping}
225 @section Notes on filtergraph escaping
227 Filtergraph description composition entails several levels of
228 escaping. See @ref{quoting_and_escaping,,the "Quoting and escaping"
229 section in the ffmpeg-utils(1) manual,ffmpeg-utils} for more
230 information about the employed escaping procedure.
232 A first level escaping affects the content of each filter option
233 value, which may contain the special character @code{:} used to
234 separate values, or one of the escaping characters @code{\'}.
236 A second level escaping affects the whole filter description, which
237 may contain the escaping characters @code{\'} or the special
238 characters @code{[],;} used by the filtergraph description.
240 Finally, when you specify a filtergraph on a shell commandline, you
241 need to perform a third level escaping for the shell special
242 characters contained within it.
244 For example, consider the following string to be embedded in
245 the @ref{drawtext} filter description @option{text} value:
247 this is a 'string': may contain one, or more, special characters
250 This string contains the @code{'} special escaping character, and the
251 @code{:} special character, so it needs to be escaped in this way:
253 text=this is a \'string\'\: may contain one, or more, special characters
256 A second level of escaping is required when embedding the filter
257 description in a filtergraph description, in order to escape all the
258 filtergraph special characters. Thus the example above becomes:
260 drawtext=text=this is a \\\'string\\\'\\: may contain one\, or more\, special characters
262 (note that in addition to the @code{\'} escaping special characters,
263 also @code{,} needs to be escaped).
265 Finally an additional level of escaping is needed when writing the
266 filtergraph description in a shell command, which depends on the
267 escaping rules of the adopted shell. For example, assuming that
268 @code{\} is special and needs to be escaped with another @code{\}, the
269 previous string will finally result in:
271 -vf "drawtext=text=this is a \\\\\\'string\\\\\\'\\\\: may contain one\\, or more\\, special characters"
274 @chapter Timeline editing
276 Some filters support a generic @option{enable} option. For the filters
277 supporting timeline editing, this option can be set to an expression which is
278 evaluated before sending a frame to the filter. If the evaluation is non-zero,
279 the filter will be enabled, otherwise the frame will be sent unchanged to the
280 next filter in the filtergraph.
282 The expression accepts the following values:
285 timestamp expressed in seconds, NAN if the input timestamp is unknown
288 sequential number of the input frame, starting from 0
291 the position in the file of the input frame, NAN if unknown
295 width and height of the input frame if video
298 Additionally, these filters support an @option{enable} command that can be used
299 to re-define the expression.
301 Like any other filtering option, the @option{enable} option follows the same
304 For example, to enable a blur filter (@ref{smartblur}) from 10 seconds to 3
305 minutes, and a @ref{curves} filter starting at 3 seconds:
307 smartblur = enable='between(t,10,3*60)',
308 curves = enable='gte(t,3)' : preset=cross_process
311 See @code{ffmpeg -filters} to view which filters have timeline support.
313 @c man end FILTERGRAPH DESCRIPTION
316 @chapter Changing options at runtime with a command
318 Some options can be changed during the operation of the filter using
319 a command. These options are marked 'T' on the output of
320 @command{ffmpeg} @option{-h filter=<name of filter>}.
321 The name of the command is the name of the option and the argument is
325 @chapter Options for filters with several inputs (framesync)
326 @c man begin OPTIONS FOR FILTERS WITH SEVERAL INPUTS
328 Some filters with several inputs support a common set of options.
329 These options can only be set by name, not with the short notation.
333 The action to take when EOF is encountered on the secondary input; it accepts
334 one of the following values:
338 Repeat the last frame (the default).
342 Pass the main input through.
346 If set to 1, force the output to terminate when the shortest input
347 terminates. Default value is 0.
350 If set to 1, force the filter to extend the last frame of secondary streams
351 until the end of the primary stream. A value of 0 disables this behavior.
355 @c man end OPTIONS FOR FILTERS WITH SEVERAL INPUTS
357 @chapter Audio Filters
358 @c man begin AUDIO FILTERS
360 When you configure your FFmpeg build, you can disable any of the
361 existing filters using @code{--disable-filters}.
362 The configure output will show the audio filters included in your
365 Below is a description of the currently available audio filters.
369 A compressor is mainly used to reduce the dynamic range of a signal.
370 Especially modern music is mostly compressed at a high ratio to
371 improve the overall loudness. It's done to get the highest attention
372 of a listener, "fatten" the sound and bring more "power" to the track.
373 If a signal is compressed too much it may sound dull or "dead"
374 afterwards or it may start to "pump" (which could be a powerful effect
375 but can also destroy a track completely).
376 The right compression is the key to reach a professional sound and is
377 the high art of mixing and mastering. Because of its complex settings
378 it may take a long time to get the right feeling for this kind of effect.
380 Compression is done by detecting the volume above a chosen level
381 @code{threshold} and dividing it by the factor set with @code{ratio}.
382 So if you set the threshold to -12dB and your signal reaches -6dB a ratio
383 of 2:1 will result in a signal at -9dB. Because an exact manipulation of
384 the signal would cause distortion of the waveform the reduction can be
385 levelled over the time. This is done by setting "Attack" and "Release".
386 @code{attack} determines how long the signal has to rise above the threshold
387 before any reduction will occur and @code{release} sets the time the signal
388 has to fall below the threshold to reduce the reduction again. Shorter signals
389 than the chosen attack time will be left untouched.
390 The overall reduction of the signal can be made up afterwards with the
391 @code{makeup} setting. So compressing the peaks of a signal about 6dB and
392 raising the makeup to this level results in a signal twice as loud than the
393 source. To gain a softer entry in the compression the @code{knee} flattens the
394 hard edge at the threshold in the range of the chosen decibels.
396 The filter accepts the following options:
400 Set input gain. Default is 1. Range is between 0.015625 and 64.
403 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
404 Default is @code{downward}.
407 If a signal of stream rises above this level it will affect the gain
409 By default it is 0.125. Range is between 0.00097563 and 1.
412 Set a ratio by which the signal is reduced. 1:2 means that if the level
413 rose 4dB above the threshold, it will be only 2dB above after the reduction.
414 Default is 2. Range is between 1 and 20.
417 Amount of milliseconds the signal has to rise above the threshold before gain
418 reduction starts. Default is 20. Range is between 0.01 and 2000.
421 Amount of milliseconds the signal has to fall below the threshold before
422 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
425 Set the amount by how much signal will be amplified after processing.
426 Default is 1. Range is from 1 to 64.
429 Curve the sharp knee around the threshold to enter gain reduction more softly.
430 Default is 2.82843. Range is between 1 and 8.
433 Choose if the @code{average} level between all channels of input stream
434 or the louder(@code{maximum}) channel of input stream affects the
435 reduction. Default is @code{average}.
438 Should the exact signal be taken in case of @code{peak} or an RMS one in case
439 of @code{rms}. Default is @code{rms} which is mostly smoother.
442 How much to use compressed signal in output. Default is 1.
443 Range is between 0 and 1.
448 This filter supports the all above options as @ref{commands}.
451 Simple audio dynamic range compression/expansion filter.
453 The filter accepts the following options:
457 Set contrast. Default is 33. Allowed range is between 0 and 100.
462 Copy the input audio source unchanged to the output. This is mainly useful for
467 Apply cross fade from one input audio stream to another input audio stream.
468 The cross fade is applied for specified duration near the end of first stream.
470 The filter accepts the following options:
474 Specify the number of samples for which the cross fade effect has to last.
475 At the end of the cross fade effect the first input audio will be completely
476 silent. Default is 44100.
479 Specify the duration of the cross fade effect. See
480 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
481 for the accepted syntax.
482 By default the duration is determined by @var{nb_samples}.
483 If set this option is used instead of @var{nb_samples}.
486 Should first stream end overlap with second stream start. Default is enabled.
489 Set curve for cross fade transition for first stream.
492 Set curve for cross fade transition for second stream.
494 For description of available curve types see @ref{afade} filter description.
501 Cross fade from one input to another:
503 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:c1=exp:c2=exp output.flac
507 Cross fade from one input to another but without overlapping:
509 ffmpeg -i first.flac -i second.flac -filter_complex acrossfade=d=10:o=0:c1=exp:c2=exp output.flac
514 Split audio stream into several bands.
516 This filter splits audio stream into two or more frequency ranges.
517 Summing all streams back will give flat output.
519 The filter accepts the following options:
523 Set split frequencies. Those must be positive and increasing.
526 Set filter order for each band split. This controls filter roll-off or steepness
527 of filter transfer function.
528 Available values are:
553 Default is @var{4th}.
556 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
559 Set output gain for each band. Default value is 1 for all bands.
566 Split input audio stream into two bands (low and high) with split frequency of 1500 Hz,
567 each band will be in separate stream:
569 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
573 Same as above, but with higher filter order:
575 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500:order=8th[LOW][HIGH]' -map '[LOW]' low.wav -map '[HIGH]' high.wav
579 Same as above, but also with additional middle band (frequencies between 1500 and 8000):
581 ffmpeg -i in.flac -filter_complex 'acrossover=split=1500 8000:order=8th[LOW][MID][HIGH]' -map '[LOW]' low.wav -map '[MID]' mid.wav -map '[HIGH]' high.wav
587 Reduce audio bit resolution.
589 This filter is bit crusher with enhanced functionality. A bit crusher
590 is used to audibly reduce number of bits an audio signal is sampled
591 with. This doesn't change the bit depth at all, it just produces the
592 effect. Material reduced in bit depth sounds more harsh and "digital".
593 This filter is able to even round to continuous values instead of discrete
595 Additionally it has a D/C offset which results in different crushing of
596 the lower and the upper half of the signal.
597 An Anti-Aliasing setting is able to produce "softer" crushing sounds.
599 Another feature of this filter is the logarithmic mode.
600 This setting switches from linear distances between bits to logarithmic ones.
601 The result is a much more "natural" sounding crusher which doesn't gate low
602 signals for example. The human ear has a logarithmic perception,
603 so this kind of crushing is much more pleasant.
604 Logarithmic crushing is also able to get anti-aliased.
606 The filter accepts the following options:
622 Can be linear: @code{lin} or logarithmic: @code{log}.
631 Set sample reduction.
634 Enable LFO. By default disabled.
645 Delay audio filtering until a given wallclock timestamp. See the @ref{cue}
649 Remove impulsive noise from input audio.
651 Samples detected as impulsive noise are replaced by interpolated samples using
652 autoregressive modelling.
656 Set window size, in milliseconds. Allowed range is from @code{10} to
657 @code{100}. Default value is @code{55} milliseconds.
658 This sets size of window which will be processed at once.
661 Set window overlap, in percentage of window size. Allowed range is from
662 @code{50} to @code{95}. Default value is @code{75} percent.
663 Setting this to a very high value increases impulsive noise removal but makes
664 whole process much slower.
667 Set autoregression order, in percentage of window size. Allowed range is from
668 @code{0} to @code{25}. Default value is @code{2} percent. This option also
669 controls quality of interpolated samples using neighbour good samples.
672 Set threshold value. Allowed range is from @code{1} to @code{100}.
673 Default value is @code{2}.
674 This controls the strength of impulsive noise which is going to be removed.
675 The lower value, the more samples will be detected as impulsive noise.
678 Set burst fusion, in percentage of window size. Allowed range is @code{0} to
679 @code{10}. Default value is @code{2}.
680 If any two samples detected as noise are spaced less than this value then any
681 sample between those two samples will be also detected as noise.
686 It accepts the following values:
689 Select overlap-add method. Even not interpolated samples are slightly
690 changed with this method.
693 Select overlap-save method. Not interpolated samples remain unchanged.
696 Default value is @code{a}.
700 Remove clipped samples from input audio.
702 Samples detected as clipped are replaced by interpolated samples using
703 autoregressive modelling.
707 Set window size, in milliseconds. Allowed range is from @code{10} to @code{100}.
708 Default value is @code{55} milliseconds.
709 This sets size of window which will be processed at once.
712 Set window overlap, in percentage of window size. Allowed range is from @code{50}
713 to @code{95}. Default value is @code{75} percent.
716 Set autoregression order, in percentage of window size. Allowed range is from
717 @code{0} to @code{25}. Default value is @code{8} percent. This option also controls
718 quality of interpolated samples using neighbour good samples.
721 Set threshold value. Allowed range is from @code{1} to @code{100}.
722 Default value is @code{10}. Higher values make clip detection less aggressive.
725 Set size of histogram used to detect clips. Allowed range is from @code{100} to @code{9999}.
726 Default value is @code{1000}. Higher values make clip detection less aggressive.
731 It accepts the following values:
734 Select overlap-add method. Even not interpolated samples are slightly changed
738 Select overlap-save method. Not interpolated samples remain unchanged.
741 Default value is @code{a}.
746 Delay one or more audio channels.
748 Samples in delayed channel are filled with silence.
750 The filter accepts the following option:
754 Set list of delays in milliseconds for each channel separated by '|'.
755 Unused delays will be silently ignored. If number of given delays is
756 smaller than number of channels all remaining channels will not be delayed.
757 If you want to delay exact number of samples, append 'S' to number.
758 If you want instead to delay in seconds, append 's' to number.
761 Use last set delay for all remaining channels. By default is disabled.
762 This option if enabled changes how option @code{delays} is interpreted.
769 Delay first channel by 1.5 seconds, the third channel by 0.5 seconds and leave
770 the second channel (and any other channels that may be present) unchanged.
776 Delay second channel by 500 samples, the third channel by 700 samples and leave
777 the first channel (and any other channels that may be present) unchanged.
783 Delay all channels by same number of samples:
785 adelay=delays=64S:all=1
790 Remedy denormals in audio by adding extremely low-level noise.
792 This filter shall be placed before any filter that can produce denormals.
794 A description of the accepted parameters follows.
798 Set level of added noise in dB. Default is @code{-351}.
799 Allowed range is from -451 to -90.
802 Set type of added noise.
815 Default is @code{dc}.
820 This filter supports the all above options as @ref{commands}.
822 @section aderivative, aintegral
824 Compute derivative/integral of audio stream.
826 Applying both filters one after another produces original audio.
830 Apply echoing to the input audio.
832 Echoes are reflected sound and can occur naturally amongst mountains
833 (and sometimes large buildings) when talking or shouting; digital echo
834 effects emulate this behaviour and are often used to help fill out the
835 sound of a single instrument or vocal. The time difference between the
836 original signal and the reflection is the @code{delay}, and the
837 loudness of the reflected signal is the @code{decay}.
838 Multiple echoes can have different delays and decays.
840 A description of the accepted parameters follows.
844 Set input gain of reflected signal. Default is @code{0.6}.
847 Set output gain of reflected signal. Default is @code{0.3}.
850 Set list of time intervals in milliseconds between original signal and reflections
851 separated by '|'. Allowed range for each @code{delay} is @code{(0 - 90000.0]}.
852 Default is @code{1000}.
855 Set list of loudness of reflected signals separated by '|'.
856 Allowed range for each @code{decay} is @code{(0 - 1.0]}.
857 Default is @code{0.5}.
864 Make it sound as if there are twice as many instruments as are actually playing:
866 aecho=0.8:0.88:60:0.4
870 If delay is very short, then it sounds like a (metallic) robot playing music:
876 A longer delay will sound like an open air concert in the mountains:
878 aecho=0.8:0.9:1000:0.3
882 Same as above but with one more mountain:
884 aecho=0.8:0.9:1000|1800:0.3|0.25
889 Audio emphasis filter creates or restores material directly taken from LPs or
890 emphased CDs with different filter curves. E.g. to store music on vinyl the
891 signal has to be altered by a filter first to even out the disadvantages of
892 this recording medium.
893 Once the material is played back the inverse filter has to be applied to
894 restore the distortion of the frequency response.
896 The filter accepts the following options:
906 Set filter mode. For restoring material use @code{reproduction} mode, otherwise
907 use @code{production} mode. Default is @code{reproduction} mode.
910 Set filter type. Selects medium. Can be one of the following:
922 select Compact Disc (CD).
928 select 50µs (FM-KF).
930 select 75µs (FM-KF).
936 This filter supports the all above options as @ref{commands}.
940 Modify an audio signal according to the specified expressions.
942 This filter accepts one or more expressions (one for each channel),
943 which are evaluated and used to modify a corresponding audio signal.
945 It accepts the following parameters:
949 Set the '|'-separated expressions list for each separate channel. If
950 the number of input channels is greater than the number of
951 expressions, the last specified expression is used for the remaining
954 @item channel_layout, c
955 Set output channel layout. If not specified, the channel layout is
956 specified by the number of expressions. If set to @samp{same}, it will
957 use by default the same input channel layout.
960 Each expression in @var{exprs} can contain the following constants and functions:
964 channel number of the current expression
967 number of the evaluated sample, starting from 0
973 time of the evaluated sample expressed in seconds
976 @item nb_out_channels
977 input and output number of channels
980 the value of input channel with number @var{CH}
983 Note: this filter is slow. For faster processing you should use a
992 aeval=val(ch)/2:c=same
996 Invert phase of the second channel:
1005 Apply fade-in/out effect to input audio.
1007 A description of the accepted parameters follows.
1011 Specify the effect type, can be either @code{in} for fade-in, or
1012 @code{out} for a fade-out effect. Default is @code{in}.
1014 @item start_sample, ss
1015 Specify the number of the start sample for starting to apply the fade
1016 effect. Default is 0.
1018 @item nb_samples, ns
1019 Specify the number of samples for which the fade effect has to last. At
1020 the end of the fade-in effect the output audio will have the same
1021 volume as the input audio, at the end of the fade-out transition
1022 the output audio will be silence. Default is 44100.
1024 @item start_time, st
1025 Specify the start time of the fade effect. Default is 0.
1026 The value must be specified as a time duration; see
1027 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1028 for the accepted syntax.
1029 If set this option is used instead of @var{start_sample}.
1032 Specify the duration of the fade effect. See
1033 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1034 for the accepted syntax.
1035 At the end of the fade-in effect the output audio will have the same
1036 volume as the input audio, at the end of the fade-out transition
1037 the output audio will be silence.
1038 By default the duration is determined by @var{nb_samples}.
1039 If set this option is used instead of @var{nb_samples}.
1042 Set curve for fade transition.
1044 It accepts the following values:
1047 select triangular, linear slope (default)
1049 select quarter of sine wave
1051 select half of sine wave
1053 select exponential sine wave
1057 select inverted parabola
1071 select inverted quarter of sine wave
1073 select inverted half of sine wave
1075 select double-exponential seat
1077 select double-exponential sigmoid
1079 select logistic sigmoid
1081 select sine cardinal function
1083 select inverted sine cardinal function
1089 @subsection Commands
1091 This filter supports the all above options as @ref{commands}.
1093 @subsection Examples
1097 Fade in first 15 seconds of audio:
1099 afade=t=in:ss=0:d=15
1103 Fade out last 25 seconds of a 900 seconds audio:
1105 afade=t=out:st=875:d=25
1110 Denoise audio samples with FFT.
1112 A description of the accepted parameters follows.
1116 Set the noise reduction in dB, allowed range is 0.01 to 97.
1117 Default value is 12 dB.
1120 Set the noise floor in dB, allowed range is -80 to -20.
1121 Default value is -50 dB.
1126 It accepts the following values:
1135 Select shellac noise.
1138 Select custom noise, defined in @code{bn} option.
1140 Default value is white noise.
1144 Set custom band noise for every one of 15 bands.
1145 Bands are separated by ' ' or '|'.
1148 Set the residual floor in dB, allowed range is -80 to -20.
1149 Default value is -38 dB.
1152 Enable noise tracking. By default is disabled.
1153 With this enabled, noise floor is automatically adjusted.
1156 Enable residual tracking. By default is disabled.
1159 Set the output mode.
1161 It accepts the following values:
1164 Pass input unchanged.
1167 Pass noise filtered out.
1172 Default value is @var{o}.
1176 @subsection Commands
1178 This filter supports the following commands:
1180 @item sample_noise, sn
1181 Start or stop measuring noise profile.
1182 Syntax for the command is : "start" or "stop" string.
1183 After measuring noise profile is stopped it will be
1184 automatically applied in filtering.
1186 @item noise_reduction, nr
1187 Change noise reduction. Argument is single float number.
1188 Syntax for the command is : "@var{noise_reduction}"
1190 @item noise_floor, nf
1191 Change noise floor. Argument is single float number.
1192 Syntax for the command is : "@var{noise_floor}"
1194 @item output_mode, om
1195 Change output mode operation.
1196 Syntax for the command is : "i", "o" or "n" string.
1200 Apply arbitrary expressions to samples in frequency domain.
1204 Set frequency domain real expression for each separate channel separated
1205 by '|'. Default is "re".
1206 If the number of input channels is greater than the number of
1207 expressions, the last specified expression is used for the remaining
1211 Set frequency domain imaginary expression for each separate channel
1212 separated by '|'. Default is "im".
1214 Each expression in @var{real} and @var{imag} can contain the following
1215 constants and functions:
1222 current frequency bin number
1225 number of available bins
1228 channel number of the current expression
1237 current real part of frequency bin of current channel
1240 current imaginary part of frequency bin of current channel
1243 Return the value of real part of frequency bin at location (@var{bin},@var{channel})
1246 Return the value of imaginary part of frequency bin at location (@var{bin},@var{channel})
1250 Set window size. Allowed range is from 16 to 131072.
1251 Default is @code{4096}
1254 Set window function. Default is @code{hann}.
1257 Set window overlap. If set to 1, the recommended overlap for selected
1258 window function will be picked. Default is @code{0.75}.
1261 @subsection Examples
1265 Leave almost only low frequencies in audio:
1267 afftfilt="'real=re * (1-clip((b/nb)*b,0,1))':imag='im * (1-clip((b/nb)*b,0,1))'"
1271 Apply robotize effect:
1273 afftfilt="real='hypot(re,im)*sin(0)':imag='hypot(re,im)*cos(0)':win_size=512:overlap=0.75"
1277 Apply whisper effect:
1279 afftfilt="real='hypot(re,im)*cos((random(0)*2-1)*2*3.14)':imag='hypot(re,im)*sin((random(1)*2-1)*2*3.14)':win_size=128:overlap=0.8"
1286 Apply an arbitrary Finite Impulse Response filter.
1288 This filter is designed for applying long FIR filters,
1289 up to 60 seconds long.
1291 It can be used as component for digital crossover filters,
1292 room equalization, cross talk cancellation, wavefield synthesis,
1293 auralization, ambiophonics, ambisonics and spatialization.
1295 This filter uses the streams higher than first one as FIR coefficients.
1296 If the non-first stream holds a single channel, it will be used
1297 for all input channels in the first stream, otherwise
1298 the number of channels in the non-first stream must be same as
1299 the number of channels in the first stream.
1301 It accepts the following parameters:
1305 Set dry gain. This sets input gain.
1308 Set wet gain. This sets final output gain.
1311 Set Impulse Response filter length. Default is 1, which means whole IR is processed.
1314 Enable applying gain measured from power of IR.
1316 Set which approach to use for auto gain measurement.
1320 Do not apply any gain.
1323 select peak gain, very conservative approach. This is default value.
1326 select DC gain, limited application.
1329 select gain to noise approach, this is most popular one.
1333 Set gain to be applied to IR coefficients before filtering.
1334 Allowed range is 0 to 1. This gain is applied after any gain applied with @var{gtype} option.
1337 Set format of IR stream. Can be @code{mono} or @code{input}.
1338 Default is @code{input}.
1341 Set max allowed Impulse Response filter duration in seconds. Default is 30 seconds.
1342 Allowed range is 0.1 to 60 seconds.
1345 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1346 By default it is disabled.
1349 Set for which IR channel to display frequency response. By default is first channel
1350 displayed. This option is used only when @var{response} is enabled.
1353 Set video stream size. This option is used only when @var{response} is enabled.
1356 Set video stream frame rate. This option is used only when @var{response} is enabled.
1359 Set minimal partition size used for convolution. Default is @var{8192}.
1360 Allowed range is from @var{1} to @var{32768}.
1361 Lower values decreases latency at cost of higher CPU usage.
1364 Set maximal partition size used for convolution. Default is @var{8192}.
1365 Allowed range is from @var{8} to @var{32768}.
1366 Lower values may increase CPU usage.
1369 Set number of input impulse responses streams which will be switchable at runtime.
1370 Allowed range is from @var{1} to @var{32}. Default is @var{1}.
1373 Set IR stream which will be used for convolution, starting from @var{0}, should always be
1374 lower than supplied value by @code{nbirs} option. Default is @var{0}.
1375 This option can be changed at runtime via @ref{commands}.
1378 @subsection Examples
1382 Apply reverb to stream using mono IR file as second input, complete command using ffmpeg:
1384 ffmpeg -i input.wav -i middle_tunnel_1way_mono.wav -lavfi afir output.wav
1391 Set output format constraints for the input audio. The framework will
1392 negotiate the most appropriate format to minimize conversions.
1394 It accepts the following parameters:
1397 @item sample_fmts, f
1398 A '|'-separated list of requested sample formats.
1400 @item sample_rates, r
1401 A '|'-separated list of requested sample rates.
1403 @item channel_layouts, cl
1404 A '|'-separated list of requested channel layouts.
1406 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
1407 for the required syntax.
1410 If a parameter is omitted, all values are allowed.
1412 Force the output to either unsigned 8-bit or signed 16-bit stereo
1414 aformat=sample_fmts=u8|s16:channel_layouts=stereo
1418 Apply frequency shift to input audio samples.
1420 The filter accepts the following options:
1424 Specify frequency shift. Allowed range is -INT_MAX to INT_MAX.
1425 Default value is 0.0.
1428 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
1429 Default value is 1.0.
1432 @subsection Commands
1434 This filter supports the all above options as @ref{commands}.
1438 A gate is mainly used to reduce lower parts of a signal. This kind of signal
1439 processing reduces disturbing noise between useful signals.
1441 Gating is done by detecting the volume below a chosen level @var{threshold}
1442 and dividing it by the factor set with @var{ratio}. The bottom of the noise
1443 floor is set via @var{range}. Because an exact manipulation of the signal
1444 would cause distortion of the waveform the reduction can be levelled over
1445 time. This is done by setting @var{attack} and @var{release}.
1447 @var{attack} determines how long the signal has to fall below the threshold
1448 before any reduction will occur and @var{release} sets the time the signal
1449 has to rise above the threshold to reduce the reduction again.
1450 Shorter signals than the chosen attack time will be left untouched.
1454 Set input level before filtering.
1455 Default is 1. Allowed range is from 0.015625 to 64.
1458 Set the mode of operation. Can be @code{upward} or @code{downward}.
1459 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
1460 will be amplified, expanding dynamic range in upward direction.
1461 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
1464 Set the level of gain reduction when the signal is below the threshold.
1465 Default is 0.06125. Allowed range is from 0 to 1.
1466 Setting this to 0 disables reduction and then filter behaves like expander.
1469 If a signal rises above this level the gain reduction is released.
1470 Default is 0.125. Allowed range is from 0 to 1.
1473 Set a ratio by which the signal is reduced.
1474 Default is 2. Allowed range is from 1 to 9000.
1477 Amount of milliseconds the signal has to rise above the threshold before gain
1479 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
1482 Amount of milliseconds the signal has to fall below the threshold before the
1483 reduction is increased again. Default is 250 milliseconds.
1484 Allowed range is from 0.01 to 9000.
1487 Set amount of amplification of signal after processing.
1488 Default is 1. Allowed range is from 1 to 64.
1491 Curve the sharp knee around the threshold to enter gain reduction more softly.
1492 Default is 2.828427125. Allowed range is from 1 to 8.
1495 Choose if exact signal should be taken for detection or an RMS like one.
1496 Default is @code{rms}. Can be @code{peak} or @code{rms}.
1499 Choose if the average level between all channels or the louder channel affects
1501 Default is @code{average}. Can be @code{average} or @code{maximum}.
1504 @subsection Commands
1506 This filter supports the all above options as @ref{commands}.
1510 Apply an arbitrary Infinite Impulse Response filter.
1512 It accepts the following parameters:
1516 Set B/numerator/zeros/reflection coefficients.
1519 Set A/denominator/poles/ladder coefficients.
1531 Set coefficients format.
1535 lattice-ladder function
1537 analog transfer function
1539 digital transfer function
1541 Z-plane zeros/poles, cartesian (default)
1543 Z-plane zeros/poles, polar radians
1545 Z-plane zeros/poles, polar degrees
1551 Set type of processing.
1563 Set filtering precision.
1567 double-precision floating-point (default)
1569 single-precision floating-point
1577 Normalize filter coefficients, by default is enabled.
1578 Enabling it will normalize magnitude response at DC to 0dB.
1581 How much to use filtered signal in output. Default is 1.
1582 Range is between 0 and 1.
1585 Show IR frequency response, magnitude(magenta), phase(green) and group delay(yellow) in additional video stream.
1586 By default it is disabled.
1589 Set for which IR channel to display frequency response. By default is first channel
1590 displayed. This option is used only when @var{response} is enabled.
1593 Set video stream size. This option is used only when @var{response} is enabled.
1596 Coefficients in @code{tf} and @code{sf} format are separated by spaces and are in ascending
1599 Coefficients in @code{zp} format are separated by spaces and order of coefficients
1600 doesn't matter. Coefficients in @code{zp} format are complex numbers with @var{i}
1603 Different coefficients and gains can be provided for every channel, in such case
1604 use '|' to separate coefficients or gains. Last provided coefficients will be
1605 used for all remaining channels.
1607 @subsection Examples
1611 Apply 2 pole elliptic notch at around 5000Hz for 48000 Hz sample rate:
1613 aiir=k=1:z=7.957584807809675810E-1 -2.575128568908332300 3.674839853930788710 -2.57512875289799137 7.957586296317130880E-1:p=1 -2.86950072432325953 3.63022088054647218 -2.28075678147272232 6.361362326477423500E-1:f=tf:r=d
1617 Same as above but in @code{zp} format:
1619 aiir=k=0.79575848078096756:z=0.80918701+0.58773007i 0.80918701-0.58773007i 0.80884700+0.58784055i 0.80884700-0.58784055i:p=0.63892345+0.59951235i 0.63892345-0.59951235i 0.79582691+0.44198673i 0.79582691-0.44198673i:f=zp:r=s
1623 Apply 3-rd order analog normalized Butterworth low-pass filter, using analog transfer function format:
1625 aiir=z=1.3057 0 0 0:p=1.3057 2.3892 2.1860 1:f=sf:r=d
1631 The limiter prevents an input signal from rising over a desired threshold.
1632 This limiter uses lookahead technology to prevent your signal from distorting.
1633 It means that there is a small delay after the signal is processed. Keep in mind
1634 that the delay it produces is the attack time you set.
1636 The filter accepts the following options:
1640 Set input gain. Default is 1.
1643 Set output gain. Default is 1.
1646 Don't let signals above this level pass the limiter. Default is 1.
1649 The limiter will reach its attenuation level in this amount of time in
1650 milliseconds. Default is 5 milliseconds.
1653 Come back from limiting to attenuation 1.0 in this amount of milliseconds.
1654 Default is 50 milliseconds.
1657 When gain reduction is always needed ASC takes care of releasing to an
1658 average reduction level rather than reaching a reduction of 0 in the release
1662 Select how much the release time is affected by ASC, 0 means nearly no changes
1663 in release time while 1 produces higher release times.
1666 Auto level output signal. Default is enabled.
1667 This normalizes audio back to 0dB if enabled.
1670 Depending on picked setting it is recommended to upsample input 2x or 4x times
1671 with @ref{aresample} before applying this filter.
1675 Apply a two-pole all-pass filter with central frequency (in Hz)
1676 @var{frequency}, and filter-width @var{width}.
1677 An all-pass filter changes the audio's frequency to phase relationship
1678 without changing its frequency to amplitude relationship.
1680 The filter accepts the following options:
1684 Set frequency in Hz.
1687 Set method to specify band-width of filter.
1702 Specify the band-width of a filter in width_type units.
1705 How much to use filtered signal in output. Default is 1.
1706 Range is between 0 and 1.
1709 Specify which channels to filter, by default all available are filtered.
1712 Normalize biquad coefficients, by default is disabled.
1713 Enabling it will normalize magnitude response at DC to 0dB.
1716 Set the filter order, can be 1 or 2. Default is 2.
1719 Set transform type of IIR filter.
1728 Set precison of filtering.
1731 Pick automatic sample format depending on surround filters.
1733 Always use signed 16-bit.
1735 Always use signed 32-bit.
1737 Always use float 32-bit.
1739 Always use float 64-bit.
1743 @subsection Commands
1745 This filter supports the following commands:
1748 Change allpass frequency.
1749 Syntax for the command is : "@var{frequency}"
1752 Change allpass width_type.
1753 Syntax for the command is : "@var{width_type}"
1756 Change allpass width.
1757 Syntax for the command is : "@var{width}"
1761 Syntax for the command is : "@var{mix}"
1768 The filter accepts the following options:
1772 Set the number of loops. Setting this value to -1 will result in infinite loops.
1776 Set maximal number of samples. Default is 0.
1779 Set first sample of loop. Default is 0.
1785 Merge two or more audio streams into a single multi-channel stream.
1787 The filter accepts the following options:
1792 Set the number of inputs. Default is 2.
1796 If the channel layouts of the inputs are disjoint, and therefore compatible,
1797 the channel layout of the output will be set accordingly and the channels
1798 will be reordered as necessary. If the channel layouts of the inputs are not
1799 disjoint, the output will have all the channels of the first input then all
1800 the channels of the second input, in that order, and the channel layout of
1801 the output will be the default value corresponding to the total number of
1804 For example, if the first input is in 2.1 (FL+FR+LF) and the second input
1805 is FC+BL+BR, then the output will be in 5.1, with the channels in the
1806 following order: a1, a2, b1, a3, b2, b3 (a1 is the first channel of the
1807 first input, b1 is the first channel of the second input).
1809 On the other hand, if both input are in stereo, the output channels will be
1810 in the default order: a1, a2, b1, b2, and the channel layout will be
1811 arbitrarily set to 4.0, which may or may not be the expected value.
1813 All inputs must have the same sample rate, and format.
1815 If inputs do not have the same duration, the output will stop with the
1818 @subsection Examples
1822 Merge two mono files into a stereo stream:
1824 amovie=left.wav [l] ; amovie=right.mp3 [r] ; [l] [r] amerge
1828 Multiple merges assuming 1 video stream and 6 audio streams in @file{input.mkv}:
1830 ffmpeg -i input.mkv -filter_complex "[0:1][0:2][0:3][0:4][0:5][0:6] amerge=inputs=6" -c:a pcm_s16le output.mkv
1836 Mixes multiple audio inputs into a single output.
1838 Note that this filter only supports float samples (the @var{amerge}
1839 and @var{pan} audio filters support many formats). If the @var{amix}
1840 input has integer samples then @ref{aresample} will be automatically
1841 inserted to perform the conversion to float samples.
1845 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex amix=inputs=3:duration=first:dropout_transition=3 OUTPUT
1847 will mix 3 input audio streams to a single output with the same duration as the
1848 first input and a dropout transition time of 3 seconds.
1850 It accepts the following parameters:
1854 The number of inputs. If unspecified, it defaults to 2.
1857 How to determine the end-of-stream.
1861 The duration of the longest input. (default)
1864 The duration of the shortest input.
1867 The duration of the first input.
1871 @item dropout_transition
1872 The transition time, in seconds, for volume renormalization when an input
1873 stream ends. The default value is 2 seconds.
1876 Specify weight of each input audio stream as sequence.
1877 Each weight is separated by space. By default all inputs have same weight.
1880 @subsection Commands
1882 This filter supports the following commands:
1885 Syntax is same as option with same name.
1890 Multiply first audio stream with second audio stream and store result
1891 in output audio stream. Multiplication is done by multiplying each
1892 sample from first stream with sample at same position from second stream.
1894 With this element-wise multiplication one can create amplitude fades and
1895 amplitude modulations.
1897 @section anequalizer
1899 High-order parametric multiband equalizer for each channel.
1901 It accepts the following parameters:
1905 This option string is in format:
1906 "c@var{chn} f=@var{cf} w=@var{w} g=@var{g} t=@var{f} | ..."
1907 Each equalizer band is separated by '|'.
1911 Set channel number to which equalization will be applied.
1912 If input doesn't have that channel the entry is ignored.
1915 Set central frequency for band.
1916 If input doesn't have that frequency the entry is ignored.
1919 Set band width in Hertz.
1922 Set band gain in dB.
1925 Set filter type for band, optional, can be:
1929 Butterworth, this is default.
1940 With this option activated frequency response of anequalizer is displayed
1944 Set video stream size. Only useful if curves option is activated.
1947 Set max gain that will be displayed. Only useful if curves option is activated.
1948 Setting this to a reasonable value makes it possible to display gain which is derived from
1949 neighbour bands which are too close to each other and thus produce higher gain
1950 when both are activated.
1953 Set frequency scale used to draw frequency response in video output.
1954 Can be linear or logarithmic. Default is logarithmic.
1957 Set color for each channel curve which is going to be displayed in video stream.
1958 This is list of color names separated by space or by '|'.
1959 Unrecognised or missing colors will be replaced by white color.
1962 @subsection Examples
1966 Lower gain by 10 of central frequency 200Hz and width 100 Hz
1967 for first 2 channels using Chebyshev type 1 filter:
1969 anequalizer=c0 f=200 w=100 g=-10 t=1|c1 f=200 w=100 g=-10 t=1
1973 @subsection Commands
1975 This filter supports the following commands:
1978 Alter existing filter parameters.
1979 Syntax for the commands is : "@var{fN}|f=@var{freq}|w=@var{width}|g=@var{gain}"
1981 @var{fN} is existing filter number, starting from 0, if no such filter is available
1983 @var{freq} set new frequency parameter.
1984 @var{width} set new width parameter in Hertz.
1985 @var{gain} set new gain parameter in dB.
1987 Full filter invocation with asendcmd may look like this:
1988 asendcmd=c='4.0 anequalizer change 0|f=200|w=50|g=1',anequalizer=...
1993 Reduce broadband noise in audio samples using Non-Local Means algorithm.
1995 Each sample is adjusted by looking for other samples with similar contexts. This
1996 context similarity is defined by comparing their surrounding patches of size
1997 @option{p}. Patches are searched in an area of @option{r} around the sample.
1999 The filter accepts the following options:
2003 Set denoising strength. Allowed range is from 0.00001 to 10. Default value is 0.00001.
2006 Set patch radius duration. Allowed range is from 1 to 100 milliseconds.
2007 Default value is 2 milliseconds.
2010 Set research radius duration. Allowed range is from 2 to 300 milliseconds.
2011 Default value is 6 milliseconds.
2014 Set the output mode.
2016 It accepts the following values:
2019 Pass input unchanged.
2022 Pass noise filtered out.
2027 Default value is @var{o}.
2031 Set smooth factor. Default value is @var{11}. Allowed range is from @var{1} to @var{15}.
2034 @subsection Commands
2036 This filter supports the all above options as @ref{commands}.
2039 Apply Normalized Least-Mean-Squares algorithm to the first audio stream using the second audio stream.
2041 This adaptive filter is used to mimic a desired filter by finding the filter coefficients that
2042 relate to producing the least mean square of the error signal (difference between the desired,
2043 2nd input audio stream and the actual signal, the 1st input audio stream).
2045 A description of the accepted options follows.
2058 Set the filter leakage.
2061 It accepts the following values:
2070 Pass filtered samples.
2073 Pass difference between desired and filtered samples.
2075 Default value is @var{o}.
2079 @subsection Examples
2083 One of many usages of this filter is noise reduction, input audio is filtered
2084 with same samples that are delayed by fixed amount, one such example for stereo audio is:
2086 asplit[a][b],[a]adelay=32S|32S[a],[b][a]anlms=order=128:leakage=0.0005:mu=.5:out_mode=o
2090 @subsection Commands
2092 This filter supports the same commands as options, excluding option @code{order}.
2096 Pass the audio source unchanged to the output.
2100 Pad the end of an audio stream with silence.
2102 This can be used together with @command{ffmpeg} @option{-shortest} to
2103 extend audio streams to the same length as the video stream.
2105 A description of the accepted options follows.
2109 Set silence packet size. Default value is 4096.
2112 Set the number of samples of silence to add to the end. After the
2113 value is reached, the stream is terminated. This option is mutually
2114 exclusive with @option{whole_len}.
2117 Set the minimum total number of samples in the output audio stream. If
2118 the value is longer than the input audio length, silence is added to
2119 the end, until the value is reached. This option is mutually exclusive
2120 with @option{pad_len}.
2123 Specify the duration of samples of silence to add. See
2124 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2125 for the accepted syntax. Used only if set to non-zero value.
2128 Specify the minimum total duration in the output audio stream. See
2129 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
2130 for the accepted syntax. Used only if set to non-zero value. If the value is longer than
2131 the input audio length, silence is added to the end, until the value is reached.
2132 This option is mutually exclusive with @option{pad_dur}
2135 If neither the @option{pad_len} nor the @option{whole_len} nor @option{pad_dur}
2136 nor @option{whole_dur} option is set, the filter will add silence to the end of
2137 the input stream indefinitely.
2139 @subsection Examples
2143 Add 1024 samples of silence to the end of the input:
2149 Make sure the audio output will contain at least 10000 samples, pad
2150 the input with silence if required:
2152 apad=whole_len=10000
2156 Use @command{ffmpeg} to pad the audio input with silence, so that the
2157 video stream will always result the shortest and will be converted
2158 until the end in the output file when using the @option{shortest}
2161 ffmpeg -i VIDEO -i AUDIO -filter_complex "[1:0]apad" -shortest OUTPUT
2166 Add a phasing effect to the input audio.
2168 A phaser filter creates series of peaks and troughs in the frequency spectrum.
2169 The position of the peaks and troughs are modulated so that they vary over time, creating a sweeping effect.
2171 A description of the accepted parameters follows.
2175 Set input gain. Default is 0.4.
2178 Set output gain. Default is 0.74
2181 Set delay in milliseconds. Default is 3.0.
2184 Set decay. Default is 0.4.
2187 Set modulation speed in Hz. Default is 0.5.
2190 Set modulation type. Default is triangular.
2192 It accepts the following values:
2199 @section aphaseshift
2200 Apply phase shift to input audio samples.
2202 The filter accepts the following options:
2206 Specify phase shift. Allowed range is from -1.0 to 1.0.
2207 Default value is 0.0.
2210 Set output gain applied to final output. Allowed range is from 0.0 to 1.0.
2211 Default value is 1.0.
2214 @subsection Commands
2216 This filter supports the all above options as @ref{commands}.
2220 Audio pulsator is something between an autopanner and a tremolo.
2221 But it can produce funny stereo effects as well. Pulsator changes the volume
2222 of the left and right channel based on a LFO (low frequency oscillator) with
2223 different waveforms and shifted phases.
2224 This filter have the ability to define an offset between left and right
2225 channel. An offset of 0 means that both LFO shapes match each other.
2226 The left and right channel are altered equally - a conventional tremolo.
2227 An offset of 50% means that the shape of the right channel is exactly shifted
2228 in phase (or moved backwards about half of the frequency) - pulsator acts as
2229 an autopanner. At 1 both curves match again. Every setting in between moves the
2230 phase shift gapless between all stages and produces some "bypassing" sounds with
2231 sine and triangle waveforms. The more you set the offset near 1 (starting from
2232 the 0.5) the faster the signal passes from the left to the right speaker.
2234 The filter accepts the following options:
2238 Set input gain. By default it is 1. Range is [0.015625 - 64].
2241 Set output gain. By default it is 1. Range is [0.015625 - 64].
2244 Set waveform shape the LFO will use. Can be one of: sine, triangle, square,
2245 sawup or sawdown. Default is sine.
2248 Set modulation. Define how much of original signal is affected by the LFO.
2251 Set left channel offset. Default is 0. Allowed range is [0 - 1].
2254 Set right channel offset. Default is 0.5. Allowed range is [0 - 1].
2257 Set pulse width. Default is 1. Allowed range is [0 - 2].
2260 Set possible timing mode. Can be one of: bpm, ms or hz. Default is hz.
2263 Set bpm. Default is 120. Allowed range is [30 - 300]. Only used if timing
2267 Set ms. Default is 500. Allowed range is [10 - 2000]. Only used if timing
2271 Set frequency in Hz. Default is 2. Allowed range is [0.01 - 100]. Only used
2272 if timing is set to hz.
2278 Resample the input audio to the specified parameters, using the
2279 libswresample library. If none are specified then the filter will
2280 automatically convert between its input and output.
2282 This filter is also able to stretch/squeeze the audio data to make it match
2283 the timestamps or to inject silence / cut out audio to make it match the
2284 timestamps, do a combination of both or do neither.
2286 The filter accepts the syntax
2287 [@var{sample_rate}:]@var{resampler_options}, where @var{sample_rate}
2288 expresses a sample rate and @var{resampler_options} is a list of
2289 @var{key}=@var{value} pairs, separated by ":". See the
2290 @ref{Resampler Options,,"Resampler Options" section in the
2291 ffmpeg-resampler(1) manual,ffmpeg-resampler}
2292 for the complete list of supported options.
2294 @subsection Examples
2298 Resample the input audio to 44100Hz:
2304 Stretch/squeeze samples to the given timestamps, with a maximum of 1000
2305 samples per second compensation:
2307 aresample=async=1000
2313 Reverse an audio clip.
2315 Warning: This filter requires memory to buffer the entire clip, so trimming
2318 @subsection Examples
2322 Take the first 5 seconds of a clip, and reverse it.
2324 atrim=end=5,areverse
2330 Reduce noise from speech using Recurrent Neural Networks.
2332 This filter accepts the following options:
2336 Set train model file to load. This option is always required.
2339 Set how much to mix filtered samples into final output.
2340 Allowed range is from -1 to 1. Default value is 1.
2341 Negative values are special, they set how much to keep filtered noise
2342 in the final filter output. Set this option to -1 to hear actual
2343 noise removed from input signal.
2346 @section asetnsamples
2348 Set the number of samples per each output audio frame.
2350 The last output packet may contain a different number of samples, as
2351 the filter will flush all the remaining samples when the input audio
2354 The filter accepts the following options:
2358 @item nb_out_samples, n
2359 Set the number of frames per each output audio frame. The number is
2360 intended as the number of samples @emph{per each channel}.
2361 Default value is 1024.
2364 If set to 1, the filter will pad the last audio frame with zeroes, so
2365 that the last frame will contain the same number of samples as the
2366 previous ones. Default value is 1.
2369 For example, to set the number of per-frame samples to 1234 and
2370 disable padding for the last frame, use:
2372 asetnsamples=n=1234:p=0
2377 Set the sample rate without altering the PCM data.
2378 This will result in a change of speed and pitch.
2380 The filter accepts the following options:
2383 @item sample_rate, r
2384 Set the output sample rate. Default is 44100 Hz.
2389 Show a line containing various information for each input audio frame.
2390 The input audio is not modified.
2392 The shown line contains a sequence of key/value pairs of the form
2393 @var{key}:@var{value}.
2395 The following values are shown in the output:
2399 The (sequential) number of the input frame, starting from 0.
2402 The presentation timestamp of the input frame, in time base units; the time base
2403 depends on the filter input pad, and is usually 1/@var{sample_rate}.
2406 The presentation timestamp of the input frame in seconds.
2409 position of the frame in the input stream, -1 if this information in
2410 unavailable and/or meaningless (for example in case of synthetic audio)
2419 The sample rate for the audio frame.
2422 The number of samples (per channel) in the frame.
2425 The Adler-32 checksum (printed in hexadecimal) of the audio data. For planar
2426 audio, the data is treated as if all the planes were concatenated.
2428 @item plane_checksums
2429 A list of Adler-32 checksums for each data plane.
2433 Apply audio soft clipping.
2435 Soft clipping is a type of distortion effect where the amplitude of a signal is saturated
2436 along a smooth curve, rather than the abrupt shape of hard-clipping.
2438 This filter accepts the following options:
2442 Set type of soft-clipping.
2444 It accepts the following values:
2458 Set threshold from where to start clipping. Default value is 0dB or 1.
2461 Set gain applied to output. Default value is 0dB or 1.
2464 Set additional parameter which controls sigmoid function.
2467 Set oversampling factor.
2470 @subsection Commands
2472 This filter supports the all above options as @ref{commands}.
2475 Automatic Speech Recognition
2477 This filter uses PocketSphinx for speech recognition. To enable
2478 compilation of this filter, you need to configure FFmpeg with
2479 @code{--enable-pocketsphinx}.
2481 It accepts the following options:
2485 Set sampling rate of input audio. Defaults is @code{16000}.
2486 This need to match speech models, otherwise one will get poor results.
2489 Set dictionary containing acoustic model files.
2492 Set pronunciation dictionary.
2495 Set language model file.
2498 Set language model set.
2501 Set which language model to use.
2504 Set output for log messages.
2507 The filter exports recognized speech as the frame metadata @code{lavfi.asr.text}.
2512 Display time domain statistical information about the audio channels.
2513 Statistics are calculated and displayed for each audio channel and,
2514 where applicable, an overall figure is also given.
2516 It accepts the following option:
2519 Short window length in seconds, used for peak and trough RMS measurement.
2520 Default is @code{0.05} (50 milliseconds). Allowed range is @code{[0.01 - 10]}.
2524 Set metadata injection. All the metadata keys are prefixed with @code{lavfi.astats.X},
2525 where @code{X} is channel number starting from 1 or string @code{Overall}. Default is
2528 Available keys for each channel are:
2574 For example full key look like this @code{lavfi.astats.1.DC_offset} or
2575 this @code{lavfi.astats.Overall.Peak_count}.
2577 For description what each key means read below.
2580 Set number of frame after which stats are going to be recalculated.
2581 Default is disabled.
2583 @item measure_perchannel
2584 Select the entries which need to be measured per channel. The metadata keys can
2585 be used as flags, default is @option{all} which measures everything.
2586 @option{none} disables all per channel measurement.
2588 @item measure_overall
2589 Select the entries which need to be measured overall. The metadata keys can
2590 be used as flags, default is @option{all} which measures everything.
2591 @option{none} disables all overall measurement.
2595 A description of each shown parameter follows:
2599 Mean amplitude displacement from zero.
2602 Minimal sample level.
2605 Maximal sample level.
2607 @item Min difference
2608 Minimal difference between two consecutive samples.
2610 @item Max difference
2611 Maximal difference between two consecutive samples.
2613 @item Mean difference
2614 Mean difference between two consecutive samples.
2615 The average of each difference between two consecutive samples.
2617 @item RMS difference
2618 Root Mean Square difference between two consecutive samples.
2622 Standard peak and RMS level measured in dBFS.
2626 Peak and trough values for RMS level measured over a short window.
2629 Standard ratio of peak to RMS level (note: not in dB).
2632 Flatness (i.e. consecutive samples with the same value) of the signal at its peak levels
2633 (i.e. either @var{Min level} or @var{Max level}).
2636 Number of occasions (not the number of samples) that the signal attained either
2637 @var{Min level} or @var{Max level}.
2639 @item Noise floor dB
2640 Minimum local peak measured in dBFS over a short window.
2642 @item Noise floor count
2643 Number of occasions (not the number of samples) that the signal attained
2647 Overall bit depth of audio. Number of bits used for each sample.
2650 Measured dynamic range of audio in dB.
2652 @item Zero crossings
2653 Number of points where the waveform crosses the zero level axis.
2655 @item Zero crossings rate
2656 Rate of Zero crossings and number of audio samples.
2660 Boost subwoofer frequencies.
2662 The filter accepts the following options:
2666 Set dry gain, how much of original signal is kept. Allowed range is from 0 to 1.
2667 Default value is 0.7.
2670 Set wet gain, how much of filtered signal is kept. Allowed range is from 0 to 1.
2671 Default value is 0.7.
2674 Set delay line decay gain value. Allowed range is from 0 to 1.
2675 Default value is 0.7.
2678 Set delay line feedback gain value. Allowed range is from 0 to 1.
2679 Default value is 0.9.
2682 Set cutoff frequency in Hertz. Allowed range is 50 to 900.
2683 Default value is 100.
2686 Set slope amount for cutoff frequency. Allowed range is 0.0001 to 1.
2687 Default value is 0.5.
2690 Set delay. Allowed range is from 1 to 100.
2691 Default value is 20.
2694 @subsection Commands
2696 This filter supports the all above options as @ref{commands}.
2699 Cut subwoofer frequencies.
2701 This filter allows to set custom, steeper
2702 roll off than highpass filter, and thus is able to more attenuate
2703 frequency content in stop-band.
2705 The filter accepts the following options:
2709 Set cutoff frequency in Hertz. Allowed range is 2 to 200.
2710 Default value is 20.
2713 Set filter order. Available values are from 3 to 20.
2714 Default value is 10.
2717 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2720 @subsection Commands
2722 This filter supports the all above options as @ref{commands}.
2725 Cut super frequencies.
2727 The filter accepts the following options:
2731 Set cutoff frequency in Hertz. Allowed range is 20000 to 192000.
2732 Default value is 20000.
2735 Set filter order. Available values are from 3 to 20.
2736 Default value is 10.
2739 Set input gain level. Allowed range is from 0 to 1. Default value is 1.
2742 @subsection Commands
2744 This filter supports the all above options as @ref{commands}.
2747 Apply high order Butterworth band-pass filter.
2749 The filter accepts the following options:
2753 Set center frequency in Hertz. Allowed range is 2 to 999999.
2754 Default value is 1000.
2757 Set filter order. Available values are from 4 to 20.
2761 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2764 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2767 @subsection Commands
2769 This filter supports the all above options as @ref{commands}.
2772 Apply high order Butterworth band-stop filter.
2774 The filter accepts the following options:
2778 Set center frequency in Hertz. Allowed range is 2 to 999999.
2779 Default value is 1000.
2782 Set filter order. Available values are from 4 to 20.
2786 Set Q-factor. Allowed range is from 0.01 to 100. Default value is 1.
2789 Set input gain level. Allowed range is from 0 to 2. Default value is 1.
2792 @subsection Commands
2794 This filter supports the all above options as @ref{commands}.
2800 The filter accepts exactly one parameter, the audio tempo. If not
2801 specified then the filter will assume nominal 1.0 tempo. Tempo must
2802 be in the [0.5, 100.0] range.
2804 Note that tempo greater than 2 will skip some samples rather than
2805 blend them in. If for any reason this is a concern it is always
2806 possible to daisy-chain several instances of atempo to achieve the
2807 desired product tempo.
2809 @subsection Examples
2813 Slow down audio to 80% tempo:
2819 To speed up audio to 300% tempo:
2825 To speed up audio to 300% tempo by daisy-chaining two atempo instances:
2827 atempo=sqrt(3),atempo=sqrt(3)
2831 @subsection Commands
2833 This filter supports the following commands:
2836 Change filter tempo scale factor.
2837 Syntax for the command is : "@var{tempo}"
2842 Trim the input so that the output contains one continuous subpart of the input.
2844 It accepts the following parameters:
2847 Timestamp (in seconds) of the start of the section to keep. I.e. the audio
2848 sample with the timestamp @var{start} will be the first sample in the output.
2851 Specify time of the first audio sample that will be dropped, i.e. the
2852 audio sample immediately preceding the one with the timestamp @var{end} will be
2853 the last sample in the output.
2856 Same as @var{start}, except this option sets the start timestamp in samples
2860 Same as @var{end}, except this option sets the end timestamp in samples instead
2864 The maximum duration of the output in seconds.
2867 The number of the first sample that should be output.
2870 The number of the first sample that should be dropped.
2873 @option{start}, @option{end}, and @option{duration} are expressed as time
2874 duration specifications; see
2875 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}.
2877 Note that the first two sets of the start/end options and the @option{duration}
2878 option look at the frame timestamp, while the _sample options simply count the
2879 samples that pass through the filter. So start/end_pts and start/end_sample will
2880 give different results when the timestamps are wrong, inexact or do not start at
2881 zero. Also note that this filter does not modify the timestamps. If you wish
2882 to have the output timestamps start at zero, insert the asetpts filter after the
2885 If multiple start or end options are set, this filter tries to be greedy and
2886 keep all samples that match at least one of the specified constraints. To keep
2887 only the part that matches all the constraints at once, chain multiple atrim
2890 The defaults are such that all the input is kept. So it is possible to set e.g.
2891 just the end values to keep everything before the specified time.
2896 Drop everything except the second minute of input:
2898 ffmpeg -i INPUT -af atrim=60:120
2902 Keep only the first 1000 samples:
2904 ffmpeg -i INPUT -af atrim=end_sample=1000
2909 @section axcorrelate
2910 Calculate normalized cross-correlation between two input audio streams.
2912 Resulted samples are always between -1 and 1 inclusive.
2913 If result is 1 it means two input samples are highly correlated in that selected segment.
2914 Result 0 means they are not correlated at all.
2915 If result is -1 it means two input samples are out of phase, which means they cancel each
2918 The filter accepts the following options:
2922 Set size of segment over which cross-correlation is calculated.
2923 Default is 256. Allowed range is from 2 to 131072.
2926 Set algorithm for cross-correlation. Can be @code{slow} or @code{fast}.
2927 Default is @code{slow}. Fast algorithm assumes mean values over any given segment
2928 are always zero and thus need much less calculations to make.
2929 This is generally not true, but is valid for typical audio streams.
2932 @subsection Examples
2936 Calculate correlation between channels in stereo audio stream:
2938 ffmpeg -i stereo.wav -af channelsplit,axcorrelate=size=1024:algo=fast correlation.wav
2944 Apply a two-pole Butterworth band-pass filter with central
2945 frequency @var{frequency}, and (3dB-point) band-width width.
2946 The @var{csg} option selects a constant skirt gain (peak gain = Q)
2947 instead of the default: constant 0dB peak gain.
2948 The filter roll off at 6dB per octave (20dB per decade).
2950 The filter accepts the following options:
2954 Set the filter's central frequency. Default is @code{3000}.
2957 Constant skirt gain if set to 1. Defaults to 0.
2960 Set method to specify band-width of filter.
2975 Specify the band-width of a filter in width_type units.
2978 How much to use filtered signal in output. Default is 1.
2979 Range is between 0 and 1.
2982 Specify which channels to filter, by default all available are filtered.
2985 Normalize biquad coefficients, by default is disabled.
2986 Enabling it will normalize magnitude response at DC to 0dB.
2989 Set transform type of IIR filter.
2998 Set precison of filtering.
3001 Pick automatic sample format depending on surround filters.
3003 Always use signed 16-bit.
3005 Always use signed 32-bit.
3007 Always use float 32-bit.
3009 Always use float 64-bit.
3013 @subsection Commands
3015 This filter supports the following commands:
3018 Change bandpass frequency.
3019 Syntax for the command is : "@var{frequency}"
3022 Change bandpass width_type.
3023 Syntax for the command is : "@var{width_type}"
3026 Change bandpass width.
3027 Syntax for the command is : "@var{width}"
3030 Change bandpass mix.
3031 Syntax for the command is : "@var{mix}"
3036 Apply a two-pole Butterworth band-reject filter with central
3037 frequency @var{frequency}, and (3dB-point) band-width @var{width}.
3038 The filter roll off at 6dB per octave (20dB per decade).
3040 The filter accepts the following options:
3044 Set the filter's central frequency. Default is @code{3000}.
3047 Set method to specify band-width of filter.
3062 Specify the band-width of a filter in width_type units.
3065 How much to use filtered signal in output. Default is 1.
3066 Range is between 0 and 1.
3069 Specify which channels to filter, by default all available are filtered.
3072 Normalize biquad coefficients, by default is disabled.
3073 Enabling it will normalize magnitude response at DC to 0dB.
3076 Set transform type of IIR filter.
3085 Set precison of filtering.
3088 Pick automatic sample format depending on surround filters.
3090 Always use signed 16-bit.
3092 Always use signed 32-bit.
3094 Always use float 32-bit.
3096 Always use float 64-bit.
3100 @subsection Commands
3102 This filter supports the following commands:
3105 Change bandreject frequency.
3106 Syntax for the command is : "@var{frequency}"
3109 Change bandreject width_type.
3110 Syntax for the command is : "@var{width_type}"
3113 Change bandreject width.
3114 Syntax for the command is : "@var{width}"
3117 Change bandreject mix.
3118 Syntax for the command is : "@var{mix}"
3121 @section bass, lowshelf
3123 Boost or cut the bass (lower) frequencies of the audio using a two-pole
3124 shelving filter with a response similar to that of a standard
3125 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
3127 The filter accepts the following options:
3131 Give the gain at 0 Hz. Its useful range is about -20
3132 (for a large cut) to +20 (for a large boost).
3133 Beware of clipping when using a positive gain.
3136 Set the filter's central frequency and so can be used
3137 to extend or reduce the frequency range to be boosted or cut.
3138 The default value is @code{100} Hz.
3141 Set method to specify band-width of filter.
3156 Determine how steep is the filter's shelf transition.
3159 Set number of poles. Default is 2.
3162 How much to use filtered signal in output. Default is 1.
3163 Range is between 0 and 1.
3166 Specify which channels to filter, by default all available are filtered.
3169 Normalize biquad coefficients, by default is disabled.
3170 Enabling it will normalize magnitude response at DC to 0dB.
3173 Set transform type of IIR filter.
3182 Set precison of filtering.
3185 Pick automatic sample format depending on surround filters.
3187 Always use signed 16-bit.
3189 Always use signed 32-bit.
3191 Always use float 32-bit.
3193 Always use float 64-bit.
3197 @subsection Commands
3199 This filter supports the following commands:
3202 Change bass frequency.
3203 Syntax for the command is : "@var{frequency}"
3206 Change bass width_type.
3207 Syntax for the command is : "@var{width_type}"
3211 Syntax for the command is : "@var{width}"
3215 Syntax for the command is : "@var{gain}"
3219 Syntax for the command is : "@var{mix}"
3224 Apply a biquad IIR filter with the given coefficients.
3225 Where @var{b0}, @var{b1}, @var{b2} and @var{a0}, @var{a1}, @var{a2}
3226 are the numerator and denominator coefficients respectively.
3227 and @var{channels}, @var{c} specify which channels to filter, by default all
3228 available are filtered.
3230 @subsection Commands
3232 This filter supports the following commands:
3240 Change biquad parameter.
3241 Syntax for the command is : "@var{value}"
3244 How much to use filtered signal in output. Default is 1.
3245 Range is between 0 and 1.
3248 Specify which channels to filter, by default all available are filtered.
3251 Normalize biquad coefficients, by default is disabled.
3252 Enabling it will normalize magnitude response at DC to 0dB.
3255 Set transform type of IIR filter.
3264 Set precison of filtering.
3267 Pick automatic sample format depending on surround filters.
3269 Always use signed 16-bit.
3271 Always use signed 32-bit.
3273 Always use float 32-bit.
3275 Always use float 64-bit.
3280 Bauer stereo to binaural transformation, which improves headphone listening of
3281 stereo audio records.
3283 To enable compilation of this filter you need to configure FFmpeg with
3284 @code{--enable-libbs2b}.
3286 It accepts the following parameters:
3290 Pre-defined crossfeed level.
3294 Default level (fcut=700, feed=50).
3297 Chu Moy circuit (fcut=700, feed=60).
3300 Jan Meier circuit (fcut=650, feed=95).
3305 Cut frequency (in Hz).
3314 Remap input channels to new locations.
3316 It accepts the following parameters:
3319 Map channels from input to output. The argument is a '|'-separated list of
3320 mappings, each in the @code{@var{in_channel}-@var{out_channel}} or
3321 @var{in_channel} form. @var{in_channel} can be either the name of the input
3322 channel (e.g. FL for front left) or its index in the input channel layout.
3323 @var{out_channel} is the name of the output channel or its index in the output
3324 channel layout. If @var{out_channel} is not given then it is implicitly an
3325 index, starting with zero and increasing by one for each mapping.
3327 @item channel_layout
3328 The channel layout of the output stream.
3331 If no mapping is present, the filter will implicitly map input channels to
3332 output channels, preserving indices.
3334 @subsection Examples
3338 For example, assuming a 5.1+downmix input MOV file,
3340 ffmpeg -i in.mov -filter 'channelmap=map=DL-FL|DR-FR' out.wav
3342 will create an output WAV file tagged as stereo from the downmix channels of
3346 To fix a 5.1 WAV improperly encoded in AAC's native channel order
3348 ffmpeg -i in.wav -filter 'channelmap=1|2|0|5|3|4:5.1' out.wav
3352 @section channelsplit
3354 Split each channel from an input audio stream into a separate output stream.
3356 It accepts the following parameters:
3358 @item channel_layout
3359 The channel layout of the input stream. The default is "stereo".
3361 A channel layout describing the channels to be extracted as separate output streams
3362 or "all" to extract each input channel as a separate stream. The default is "all".
3364 Choosing channels not present in channel layout in the input will result in an error.
3367 @subsection Examples
3371 For example, assuming a stereo input MP3 file,
3373 ffmpeg -i in.mp3 -filter_complex channelsplit out.mkv
3375 will create an output Matroska file with two audio streams, one containing only
3376 the left channel and the other the right channel.
3379 Split a 5.1 WAV file into per-channel files:
3381 ffmpeg -i in.wav -filter_complex
3382 'channelsplit=channel_layout=5.1[FL][FR][FC][LFE][SL][SR]'
3383 -map '[FL]' front_left.wav -map '[FR]' front_right.wav -map '[FC]'
3384 front_center.wav -map '[LFE]' lfe.wav -map '[SL]' side_left.wav -map '[SR]'
3389 Extract only LFE from a 5.1 WAV file:
3391 ffmpeg -i in.wav -filter_complex 'channelsplit=channel_layout=5.1:channels=LFE[LFE]'
3392 -map '[LFE]' lfe.wav
3397 Add a chorus effect to the audio.
3399 Can make a single vocal sound like a chorus, but can also be applied to instrumentation.
3401 Chorus resembles an echo effect with a short delay, but whereas with echo the delay is
3402 constant, with chorus, it is varied using using sinusoidal or triangular modulation.
3403 The modulation depth defines the range the modulated delay is played before or after
3404 the delay. Hence the delayed sound will sound slower or faster, that is the delayed
3405 sound tuned around the original one, like in a chorus where some vocals are slightly
3408 It accepts the following parameters:
3411 Set input gain. Default is 0.4.
3414 Set output gain. Default is 0.4.
3417 Set delays. A typical delay is around 40ms to 60ms.
3429 @subsection Examples
3435 chorus=0.7:0.9:55:0.4:0.25:2
3441 chorus=0.6:0.9:50|60:0.4|0.32:0.25|0.4:2|1.3
3445 Fuller sounding chorus with three delays:
3447 chorus=0.5:0.9:50|60|40:0.4|0.32|0.3:0.25|0.4|0.3:2|2.3|1.3
3452 Compress or expand the audio's dynamic range.
3454 It accepts the following parameters:
3460 A list of times in seconds for each channel over which the instantaneous level
3461 of the input signal is averaged to determine its volume. @var{attacks} refers to
3462 increase of volume and @var{decays} refers to decrease of volume. For most
3463 situations, the attack time (response to the audio getting louder) should be
3464 shorter than the decay time, because the human ear is more sensitive to sudden
3465 loud audio than sudden soft audio. A typical value for attack is 0.3 seconds and
3466 a typical value for decay is 0.8 seconds.
3467 If specified number of attacks & decays is lower than number of channels, the last
3468 set attack/decay will be used for all remaining channels.
3471 A list of points for the transfer function, specified in dB relative to the
3472 maximum possible signal amplitude. Each key points list must be defined using
3473 the following syntax: @code{x0/y0|x1/y1|x2/y2|....} or
3474 @code{x0/y0 x1/y1 x2/y2 ....}
3476 The input values must be in strictly increasing order but the transfer function
3477 does not have to be monotonically rising. The point @code{0/0} is assumed but
3478 may be overridden (by @code{0/out-dBn}). Typical values for the transfer
3479 function are @code{-70/-70|-60/-20|1/0}.
3482 Set the curve radius in dB for all joints. It defaults to 0.01.
3485 Set the additional gain in dB to be applied at all points on the transfer
3486 function. This allows for easy adjustment of the overall gain.
3490 Set an initial volume, in dB, to be assumed for each channel when filtering
3491 starts. This permits the user to supply a nominal level initially, so that, for
3492 example, a very large gain is not applied to initial signal levels before the
3493 companding has begun to operate. A typical value for audio which is initially
3494 quiet is -90 dB. It defaults to 0.
3497 Set a delay, in seconds. The input audio is analyzed immediately, but audio is
3498 delayed before being fed to the volume adjuster. Specifying a delay
3499 approximately equal to the attack/decay times allows the filter to effectively
3500 operate in predictive rather than reactive mode. It defaults to 0.
3504 @subsection Examples
3508 Make music with both quiet and loud passages suitable for listening to in a
3511 compand=.3|.3:1|1:-90/-60|-60/-40|-40/-30|-20/-20:6:0:-90:0.2
3514 Another example for audio with whisper and explosion parts:
3516 compand=0|0:1|1:-90/-900|-70/-70|-30/-9|0/-3:6:0:0:0
3520 A noise gate for when the noise is at a lower level than the signal:
3522 compand=.1|.1:.2|.2:-900/-900|-50.1/-900|-50/-50:.01:0:-90:.1
3526 Here is another noise gate, this time for when the noise is at a higher level
3527 than the signal (making it, in some ways, similar to squelch):
3529 compand=.1|.1:.1|.1:-45.1/-45.1|-45/-900|0/-900:.01:45:-90:.1
3533 2:1 compression starting at -6dB:
3535 compand=points=-80/-80|-6/-6|0/-3.8|20/3.5
3539 2:1 compression starting at -9dB:
3541 compand=points=-80/-80|-9/-9|0/-5.3|20/2.9
3545 2:1 compression starting at -12dB:
3547 compand=points=-80/-80|-12/-12|0/-6.8|20/1.9
3551 2:1 compression starting at -18dB:
3553 compand=points=-80/-80|-18/-18|0/-9.8|20/0.7
3557 3:1 compression starting at -15dB:
3559 compand=points=-80/-80|-15/-15|0/-10.8|20/-5.2
3565 compand=points=-80/-105|-62/-80|-15.4/-15.4|0/-12|20/-7.6
3571 compand=attacks=0:points=-80/-169|-54/-80|-49.5/-64.6|-41.1/-41.1|-25.8/-15|-10.8/-4.5|0/0|20/8.3
3575 Hard limiter at -6dB:
3577 compand=attacks=0:points=-80/-80|-6/-6|20/-6
3581 Hard limiter at -12dB:
3583 compand=attacks=0:points=-80/-80|-12/-12|20/-12
3587 Hard noise gate at -35 dB:
3589 compand=attacks=0:points=-80/-115|-35.1/-80|-35/-35|20/20
3595 compand=attacks=0:points=-80/-80|-12.4/-12.4|-6/-8|0/-6.8|20/-2.8
3599 @section compensationdelay
3601 Compensation Delay Line is a metric based delay to compensate differing
3602 positions of microphones or speakers.
3604 For example, you have recorded guitar with two microphones placed in
3605 different locations. Because the front of sound wave has fixed speed in
3606 normal conditions, the phasing of microphones can vary and depends on
3607 their location and interposition. The best sound mix can be achieved when
3608 these microphones are in phase (synchronized). Note that a distance of
3609 ~30 cm between microphones makes one microphone capture the signal in
3610 antiphase to the other microphone. That makes the final mix sound moody.
3611 This filter helps to solve phasing problems by adding different delays
3612 to each microphone track and make them synchronized.
3614 The best result can be reached when you take one track as base and
3615 synchronize other tracks one by one with it.
3616 Remember that synchronization/delay tolerance depends on sample rate, too.
3617 Higher sample rates will give more tolerance.
3619 The filter accepts the following parameters:
3623 Set millimeters distance. This is compensation distance for fine tuning.
3627 Set cm distance. This is compensation distance for tightening distance setup.
3631 Set meters distance. This is compensation distance for hard distance setup.
3635 Set dry amount. Amount of unprocessed (dry) signal.
3639 Set wet amount. Amount of processed (wet) signal.
3643 Set temperature in degrees Celsius. This is the temperature of the environment.
3648 Apply headphone crossfeed filter.
3650 Crossfeed is the process of blending the left and right channels of stereo
3652 It is mainly used to reduce extreme stereo separation of low frequencies.
3654 The intent is to produce more speaker like sound to the listener.
3656 The filter accepts the following options:
3660 Set strength of crossfeed. Default is 0.2. Allowed range is from 0 to 1.
3661 This sets gain of low shelf filter for side part of stereo image.
3662 Default is -6dB. Max allowed is -30db when strength is set to 1.
3665 Set soundstage wideness. Default is 0.5. Allowed range is from 0 to 1.
3666 This sets cut off frequency of low shelf filter. Default is cut off near
3667 1550 Hz. With range set to 1 cut off frequency is set to 2100 Hz.
3670 Set curve slope of low shelf filter. Default is 0.5.
3671 Allowed range is from 0.01 to 1.
3674 Set input gain. Default is 0.9.
3677 Set output gain. Default is 1.
3680 @subsection Commands
3682 This filter supports the all above options as @ref{commands}.
3684 @section crystalizer
3685 Simple algorithm to expand audio dynamic range.
3687 The filter accepts the following options:
3691 Sets the intensity of effect (default: 2.0). Must be in range between 0.0
3692 (unchanged sound) to 10.0 (maximum effect).
3695 Enable clipping. By default is enabled.
3698 @subsection Commands
3700 This filter supports the all above options as @ref{commands}.
3703 Apply a DC shift to the audio.
3705 This can be useful to remove a DC offset (caused perhaps by a hardware problem
3706 in the recording chain) from the audio. The effect of a DC offset is reduced
3707 headroom and hence volume. The @ref{astats} filter can be used to determine if
3708 a signal has a DC offset.
3712 Set the DC shift, allowed range is [-1, 1]. It indicates the amount to shift
3716 Optional. It should have a value much less than 1 (e.g. 0.05 or 0.02) and is
3717 used to prevent clipping.
3722 Apply de-essing to the audio samples.
3726 Set intensity for triggering de-essing. Allowed range is from 0 to 1.
3730 Set amount of ducking on treble part of sound. Allowed range is from 0 to 1.
3734 How much of original frequency content to keep when de-essing. Allowed range is from 0 to 1.
3738 Set the output mode.
3740 It accepts the following values:
3743 Pass input unchanged.
3746 Pass ess filtered out.
3751 Default value is @var{o}.
3757 Measure audio dynamic range.
3759 DR values of 14 and higher is found in very dynamic material. DR of 8 to 13
3760 is found in transition material. And anything less that 8 have very poor dynamics
3761 and is very compressed.
3763 The filter accepts the following options:
3767 Set window length in seconds used to split audio into segments of equal length.
3768 Default is 3 seconds.
3772 Dynamic Audio Normalizer.
3774 This filter applies a certain amount of gain to the input audio in order
3775 to bring its peak magnitude to a target level (e.g. 0 dBFS). However, in
3776 contrast to more "simple" normalization algorithms, the Dynamic Audio
3777 Normalizer *dynamically* re-adjusts the gain factor to the input audio.
3778 This allows for applying extra gain to the "quiet" sections of the audio
3779 while avoiding distortions or clipping the "loud" sections. In other words:
3780 The Dynamic Audio Normalizer will "even out" the volume of quiet and loud
3781 sections, in the sense that the volume of each section is brought to the
3782 same target level. Note, however, that the Dynamic Audio Normalizer achieves
3783 this goal *without* applying "dynamic range compressing". It will retain 100%
3784 of the dynamic range *within* each section of the audio file.
3788 Set the frame length in milliseconds. In range from 10 to 8000 milliseconds.
3789 Default is 500 milliseconds.
3790 The Dynamic Audio Normalizer processes the input audio in small chunks,
3791 referred to as frames. This is required, because a peak magnitude has no
3792 meaning for just a single sample value. Instead, we need to determine the
3793 peak magnitude for a contiguous sequence of sample values. While a "standard"
3794 normalizer would simply use the peak magnitude of the complete file, the
3795 Dynamic Audio Normalizer determines the peak magnitude individually for each
3796 frame. The length of a frame is specified in milliseconds. By default, the
3797 Dynamic Audio Normalizer uses a frame length of 500 milliseconds, which has
3798 been found to give good results with most files.
3799 Note that the exact frame length, in number of samples, will be determined
3800 automatically, based on the sampling rate of the individual input audio file.
3803 Set the Gaussian filter window size. In range from 3 to 301, must be odd
3804 number. Default is 31.
3805 Probably the most important parameter of the Dynamic Audio Normalizer is the
3806 @code{window size} of the Gaussian smoothing filter. The filter's window size
3807 is specified in frames, centered around the current frame. For the sake of
3808 simplicity, this must be an odd number. Consequently, the default value of 31
3809 takes into account the current frame, as well as the 15 preceding frames and
3810 the 15 subsequent frames. Using a larger window results in a stronger
3811 smoothing effect and thus in less gain variation, i.e. slower gain
3812 adaptation. Conversely, using a smaller window results in a weaker smoothing
3813 effect and thus in more gain variation, i.e. faster gain adaptation.
3814 In other words, the more you increase this value, the more the Dynamic Audio
3815 Normalizer will behave like a "traditional" normalization filter. On the
3816 contrary, the more you decrease this value, the more the Dynamic Audio
3817 Normalizer will behave like a dynamic range compressor.
3820 Set the target peak value. This specifies the highest permissible magnitude
3821 level for the normalized audio input. This filter will try to approach the
3822 target peak magnitude as closely as possible, but at the same time it also
3823 makes sure that the normalized signal will never exceed the peak magnitude.
3824 A frame's maximum local gain factor is imposed directly by the target peak
3825 magnitude. The default value is 0.95 and thus leaves a headroom of 5%*.
3826 It is not recommended to go above this value.
3829 Set the maximum gain factor. In range from 1.0 to 100.0. Default is 10.0.
3830 The Dynamic Audio Normalizer determines the maximum possible (local) gain
3831 factor for each input frame, i.e. the maximum gain factor that does not
3832 result in clipping or distortion. The maximum gain factor is determined by
3833 the frame's highest magnitude sample. However, the Dynamic Audio Normalizer
3834 additionally bounds the frame's maximum gain factor by a predetermined
3835 (global) maximum gain factor. This is done in order to avoid excessive gain
3836 factors in "silent" or almost silent frames. By default, the maximum gain
3837 factor is 10.0, For most inputs the default value should be sufficient and
3838 it usually is not recommended to increase this value. Though, for input
3839 with an extremely low overall volume level, it may be necessary to allow even
3840 higher gain factors. Note, however, that the Dynamic Audio Normalizer does
3841 not simply apply a "hard" threshold (i.e. cut off values above the threshold).
3842 Instead, a "sigmoid" threshold function will be applied. This way, the
3843 gain factors will smoothly approach the threshold value, but never exceed that
3847 Set the target RMS. In range from 0.0 to 1.0. Default is 0.0 - disabled.
3848 By default, the Dynamic Audio Normalizer performs "peak" normalization.
3849 This means that the maximum local gain factor for each frame is defined
3850 (only) by the frame's highest magnitude sample. This way, the samples can
3851 be amplified as much as possible without exceeding the maximum signal
3852 level, i.e. without clipping. Optionally, however, the Dynamic Audio
3853 Normalizer can also take into account the frame's root mean square,
3854 abbreviated RMS. In electrical engineering, the RMS is commonly used to
3855 determine the power of a time-varying signal. It is therefore considered
3856 that the RMS is a better approximation of the "perceived loudness" than
3857 just looking at the signal's peak magnitude. Consequently, by adjusting all
3858 frames to a constant RMS value, a uniform "perceived loudness" can be
3859 established. If a target RMS value has been specified, a frame's local gain
3860 factor is defined as the factor that would result in exactly that RMS value.
3861 Note, however, that the maximum local gain factor is still restricted by the
3862 frame's highest magnitude sample, in order to prevent clipping.
3865 Enable channels coupling. By default is enabled.
3866 By default, the Dynamic Audio Normalizer will amplify all channels by the same
3867 amount. This means the same gain factor will be applied to all channels, i.e.
3868 the maximum possible gain factor is determined by the "loudest" channel.
3869 However, in some recordings, it may happen that the volume of the different
3870 channels is uneven, e.g. one channel may be "quieter" than the other one(s).
3871 In this case, this option can be used to disable the channel coupling. This way,
3872 the gain factor will be determined independently for each channel, depending
3873 only on the individual channel's highest magnitude sample. This allows for
3874 harmonizing the volume of the different channels.
3877 Enable DC bias correction. By default is disabled.
3878 An audio signal (in the time domain) is a sequence of sample values.
3879 In the Dynamic Audio Normalizer these sample values are represented in the
3880 -1.0 to 1.0 range, regardless of the original input format. Normally, the
3881 audio signal, or "waveform", should be centered around the zero point.
3882 That means if we calculate the mean value of all samples in a file, or in a
3883 single frame, then the result should be 0.0 or at least very close to that
3884 value. If, however, there is a significant deviation of the mean value from
3885 0.0, in either positive or negative direction, this is referred to as a
3886 DC bias or DC offset. Since a DC bias is clearly undesirable, the Dynamic
3887 Audio Normalizer provides optional DC bias correction.
3888 With DC bias correction enabled, the Dynamic Audio Normalizer will determine
3889 the mean value, or "DC correction" offset, of each input frame and subtract
3890 that value from all of the frame's sample values which ensures those samples
3891 are centered around 0.0 again. Also, in order to avoid "gaps" at the frame
3892 boundaries, the DC correction offset values will be interpolated smoothly
3893 between neighbouring frames.
3895 @item altboundary, b
3896 Enable alternative boundary mode. By default is disabled.
3897 The Dynamic Audio Normalizer takes into account a certain neighbourhood
3898 around each frame. This includes the preceding frames as well as the
3899 subsequent frames. However, for the "boundary" frames, located at the very
3900 beginning and at the very end of the audio file, not all neighbouring
3901 frames are available. In particular, for the first few frames in the audio
3902 file, the preceding frames are not known. And, similarly, for the last few
3903 frames in the audio file, the subsequent frames are not known. Thus, the
3904 question arises which gain factors should be assumed for the missing frames
3905 in the "boundary" region. The Dynamic Audio Normalizer implements two modes
3906 to deal with this situation. The default boundary mode assumes a gain factor
3907 of exactly 1.0 for the missing frames, resulting in a smooth "fade in" and
3908 "fade out" at the beginning and at the end of the input, respectively.
3911 Set the compress factor. In range from 0.0 to 30.0. Default is 0.0.
3912 By default, the Dynamic Audio Normalizer does not apply "traditional"
3913 compression. This means that signal peaks will not be pruned and thus the
3914 full dynamic range will be retained within each local neighbourhood. However,
3915 in some cases it may be desirable to combine the Dynamic Audio Normalizer's
3916 normalization algorithm with a more "traditional" compression.
3917 For this purpose, the Dynamic Audio Normalizer provides an optional compression
3918 (thresholding) function. If (and only if) the compression feature is enabled,
3919 all input frames will be processed by a soft knee thresholding function prior
3920 to the actual normalization process. Put simply, the thresholding function is
3921 going to prune all samples whose magnitude exceeds a certain threshold value.
3922 However, the Dynamic Audio Normalizer does not simply apply a fixed threshold
3923 value. Instead, the threshold value will be adjusted for each individual
3925 In general, smaller parameters result in stronger compression, and vice versa.
3926 Values below 3.0 are not recommended, because audible distortion may appear.
3929 Set the target threshold value. This specifies the lowest permissible
3930 magnitude level for the audio input which will be normalized.
3931 If input frame volume is above this value frame will be normalized.
3932 Otherwise frame may not be normalized at all. The default value is set
3933 to 0, which means all input frames will be normalized.
3934 This option is mostly useful if digital noise is not wanted to be amplified.
3937 @subsection Commands
3939 This filter supports the all above options as @ref{commands}.
3943 Make audio easier to listen to on headphones.
3945 This filter adds `cues' to 44.1kHz stereo (i.e. audio CD format) audio
3946 so that when listened to on headphones the stereo image is moved from
3947 inside your head (standard for headphones) to outside and in front of
3948 the listener (standard for speakers).
3954 Apply a two-pole peaking equalisation (EQ) filter. With this
3955 filter, the signal-level at and around a selected frequency can
3956 be increased or decreased, whilst (unlike bandpass and bandreject
3957 filters) that at all other frequencies is unchanged.
3959 In order to produce complex equalisation curves, this filter can
3960 be given several times, each with a different central frequency.
3962 The filter accepts the following options:
3966 Set the filter's central frequency in Hz.
3969 Set method to specify band-width of filter.
3984 Specify the band-width of a filter in width_type units.
3987 Set the required gain or attenuation in dB.
3988 Beware of clipping when using a positive gain.
3991 How much to use filtered signal in output. Default is 1.
3992 Range is between 0 and 1.
3995 Specify which channels to filter, by default all available are filtered.
3998 Normalize biquad coefficients, by default is disabled.
3999 Enabling it will normalize magnitude response at DC to 0dB.
4002 Set transform type of IIR filter.
4011 Set precison of filtering.
4014 Pick automatic sample format depending on surround filters.
4016 Always use signed 16-bit.
4018 Always use signed 32-bit.
4020 Always use float 32-bit.
4022 Always use float 64-bit.
4026 @subsection Examples
4029 Attenuate 10 dB at 1000 Hz, with a bandwidth of 200 Hz:
4031 equalizer=f=1000:t=h:width=200:g=-10
4035 Apply 2 dB gain at 1000 Hz with Q 1 and attenuate 5 dB at 100 Hz with Q 2:
4037 equalizer=f=1000:t=q:w=1:g=2,equalizer=f=100:t=q:w=2:g=-5
4041 @subsection Commands
4043 This filter supports the following commands:
4046 Change equalizer frequency.
4047 Syntax for the command is : "@var{frequency}"
4050 Change equalizer width_type.
4051 Syntax for the command is : "@var{width_type}"
4054 Change equalizer width.
4055 Syntax for the command is : "@var{width}"
4058 Change equalizer gain.
4059 Syntax for the command is : "@var{gain}"
4062 Change equalizer mix.
4063 Syntax for the command is : "@var{mix}"
4066 @section extrastereo
4068 Linearly increases the difference between left and right channels which
4069 adds some sort of "live" effect to playback.
4071 The filter accepts the following options:
4075 Sets the difference coefficient (default: 2.5). 0.0 means mono sound
4076 (average of both channels), with 1.0 sound will be unchanged, with
4077 -1.0 left and right channels will be swapped.
4080 Enable clipping. By default is enabled.
4083 @subsection Commands
4085 This filter supports the all above options as @ref{commands}.
4087 @section firequalizer
4088 Apply FIR Equalization using arbitrary frequency response.
4090 The filter accepts the following option:
4094 Set gain curve equation (in dB). The expression can contain variables:
4097 the evaluated frequency
4101 channel number, set to 0 when multichannels evaluation is disabled
4103 channel id, see libavutil/channel_layout.h, set to the first channel id when
4104 multichannels evaluation is disabled
4108 channel_layout, see libavutil/channel_layout.h
4113 @item gain_interpolate(f)
4114 interpolate gain on frequency f based on gain_entry
4115 @item cubic_interpolate(f)
4116 same as gain_interpolate, but smoother
4118 This option is also available as command. Default is @code{gain_interpolate(f)}.
4121 Set gain entry for gain_interpolate function. The expression can
4125 store gain entry at frequency f with value g
4127 This option is also available as command.
4130 Set filter delay in seconds. Higher value means more accurate.
4131 Default is @code{0.01}.
4134 Set filter accuracy in Hz. Lower value means more accurate.
4135 Default is @code{5}.
4138 Set window function. Acceptable values are:
4141 rectangular window, useful when gain curve is already smooth
4143 hann window (default)
4149 3-terms continuous 1st derivative nuttall window
4151 minimum 3-terms discontinuous nuttall window
4153 4-terms continuous 1st derivative nuttall window
4155 minimum 4-terms discontinuous nuttall (blackman-nuttall) window
4157 blackman-harris window
4163 If enabled, use fixed number of audio samples. This improves speed when
4164 filtering with large delay. Default is disabled.
4167 Enable multichannels evaluation on gain. Default is disabled.
4170 Enable zero phase mode by subtracting timestamp to compensate delay.
4171 Default is disabled.
4174 Set scale used by gain. Acceptable values are:
4177 linear frequency, linear gain
4179 linear frequency, logarithmic (in dB) gain (default)
4181 logarithmic (in octave scale where 20 Hz is 0) frequency, linear gain
4183 logarithmic frequency, logarithmic gain
4187 Set file for dumping, suitable for gnuplot.
4190 Set scale for dumpfile. Acceptable values are same with scale option.
4194 Enable 2-channel convolution using complex FFT. This improves speed significantly.
4195 Default is disabled.
4198 Enable minimum phase impulse response. Default is disabled.
4201 @subsection Examples
4206 firequalizer=gain='if(lt(f,1000), 0, -INF)'
4209 lowpass at 1000 Hz with gain_entry:
4211 firequalizer=gain_entry='entry(1000,0); entry(1001, -INF)'
4214 custom equalization:
4216 firequalizer=gain_entry='entry(100,0); entry(400, -4); entry(1000, -6); entry(2000, 0)'
4219 higher delay with zero phase to compensate delay:
4221 firequalizer=delay=0.1:fixed=on:zero_phase=on
4224 lowpass on left channel, highpass on right channel:
4226 firequalizer=gain='if(eq(chid,1), gain_interpolate(f), if(eq(chid,2), gain_interpolate(1e6+f), 0))'
4227 :gain_entry='entry(1000, 0); entry(1001,-INF); entry(1e6+1000,0)':multi=on
4232 Apply a flanging effect to the audio.
4234 The filter accepts the following options:
4238 Set base delay in milliseconds. Range from 0 to 30. Default value is 0.
4241 Set added sweep delay in milliseconds. Range from 0 to 10. Default value is 2.
4244 Set percentage regeneration (delayed signal feedback). Range from -95 to 95.
4248 Set percentage of delayed signal mixed with original. Range from 0 to 100.
4249 Default value is 71.
4252 Set sweeps per second (Hz). Range from 0.1 to 10. Default value is 0.5.
4255 Set swept wave shape, can be @var{triangular} or @var{sinusoidal}.
4256 Default value is @var{sinusoidal}.
4259 Set swept wave percentage-shift for multi channel. Range from 0 to 100.
4260 Default value is 25.
4263 Set delay-line interpolation, @var{linear} or @var{quadratic}.
4264 Default is @var{linear}.
4268 Apply Haas effect to audio.
4270 Note that this makes most sense to apply on mono signals.
4271 With this filter applied to mono signals it give some directionality and
4272 stretches its stereo image.
4274 The filter accepts the following options:
4278 Set input level. By default is @var{1}, or 0dB
4281 Set output level. By default is @var{1}, or 0dB.
4284 Set gain applied to side part of signal. By default is @var{1}.
4287 Set kind of middle source. Can be one of the following:
4297 Pick middle part signal of stereo image.
4300 Pick side part signal of stereo image.
4304 Change middle phase. By default is disabled.
4307 Set left channel delay. By default is @var{2.05} milliseconds.
4310 Set left channel balance. By default is @var{-1}.
4313 Set left channel gain. By default is @var{1}.
4316 Change left phase. By default is disabled.
4319 Set right channel delay. By defaults is @var{2.12} milliseconds.
4322 Set right channel balance. By default is @var{1}.
4325 Set right channel gain. By default is @var{1}.
4328 Change right phase. By default is enabled.
4333 Decodes High Definition Compatible Digital (HDCD) data. A 16-bit PCM stream with
4334 embedded HDCD codes is expanded into a 20-bit PCM stream.
4336 The filter supports the Peak Extend and Low-level Gain Adjustment features
4337 of HDCD, and detects the Transient Filter flag.
4340 ffmpeg -i HDCD16.flac -af hdcd OUT24.flac
4343 When using the filter with wav, note the default encoding for wav is 16-bit,
4344 so the resulting 20-bit stream will be truncated back to 16-bit. Use something
4345 like @command{-acodec pcm_s24le} after the filter to get 24-bit PCM output.
4347 ffmpeg -i HDCD16.wav -af hdcd OUT16.wav
4348 ffmpeg -i HDCD16.wav -af hdcd -c:a pcm_s24le OUT24.wav
4351 The filter accepts the following options:
4354 @item disable_autoconvert
4355 Disable any automatic format conversion or resampling in the filter graph.
4357 @item process_stereo
4358 Process the stereo channels together. If target_gain does not match between
4359 channels, consider it invalid and use the last valid target_gain.
4362 Set the code detect timer period in ms.
4365 Always extend peaks above -3dBFS even if PE isn't signaled.
4368 Replace audio with a solid tone and adjust the amplitude to signal some
4369 specific aspect of the decoding process. The output file can be loaded in
4370 an audio editor alongside the original to aid analysis.
4372 @code{analyze_mode=pe:force_pe=true} can be used to see all samples above the PE level.
4379 Gain adjustment level at each sample
4381 Samples where peak extend occurs
4383 Samples where the code detect timer is active
4385 Samples where the target gain does not match between channels
4391 Apply head-related transfer functions (HRTFs) to create virtual
4392 loudspeakers around the user for binaural listening via headphones.
4393 The HRIRs are provided via additional streams, for each channel
4394 one stereo input stream is needed.
4396 The filter accepts the following options:
4400 Set mapping of input streams for convolution.
4401 The argument is a '|'-separated list of channel names in order as they
4402 are given as additional stream inputs for filter.
4403 This also specify number of input streams. Number of input streams
4404 must be not less than number of channels in first stream plus one.
4407 Set gain applied to audio. Value is in dB. Default is 0.
4410 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
4411 processing audio in time domain which is slow.
4412 @var{freq} is processing audio in frequency domain which is fast.
4413 Default is @var{freq}.
4416 Set custom gain for LFE channels. Value is in dB. Default is 0.
4419 Set size of frame in number of samples which will be processed at once.
4420 Default value is @var{1024}. Allowed range is from 1024 to 96000.
4423 Set format of hrir stream.
4424 Default value is @var{stereo}. Alternative value is @var{multich}.
4425 If value is set to @var{stereo}, number of additional streams should
4426 be greater or equal to number of input channels in first input stream.
4427 Also each additional stream should have stereo number of channels.
4428 If value is set to @var{multich}, number of additional streams should
4429 be exactly one. Also number of input channels of additional stream
4430 should be equal or greater than twice number of channels of first input
4434 @subsection Examples
4438 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4439 each amovie filter use stereo file with IR coefficients as input.
4440 The files give coefficients for each position of virtual loudspeaker:
4443 -filter_complex "amovie=azi_270_ele_0_DFC.wav[sr];amovie=azi_90_ele_0_DFC.wav[sl];amovie=azi_225_ele_0_DFC.wav[br];amovie=azi_135_ele_0_DFC.wav[bl];amovie=azi_0_ele_0_DFC.wav,asplit[fc][lfe];amovie=azi_35_ele_0_DFC.wav[fl];amovie=azi_325_ele_0_DFC.wav[fr];[0:a][fl][fr][fc][lfe][bl][br][sl][sr]headphone=FL|FR|FC|LFE|BL|BR|SL|SR"
4448 Full example using wav files as coefficients with amovie filters for 7.1 downmix,
4449 but now in @var{multich} @var{hrir} format.
4451 ffmpeg -i input.wav -filter_complex "amovie=minp.wav[hrirs];[0:a][hrirs]headphone=map=FL|FR|FC|LFE|BL|BR|SL|SR:hrir=multich"
4458 Apply a high-pass filter with 3dB point frequency.
4459 The filter can be either single-pole, or double-pole (the default).
4460 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4462 The filter accepts the following options:
4466 Set frequency in Hz. Default is 3000.
4469 Set number of poles. Default is 2.
4472 Set method to specify band-width of filter.
4487 Specify the band-width of a filter in width_type units.
4488 Applies only to double-pole filter.
4489 The default is 0.707q and gives a Butterworth response.
4492 How much to use filtered signal in output. Default is 1.
4493 Range is between 0 and 1.
4496 Specify which channels to filter, by default all available are filtered.
4499 Normalize biquad coefficients, by default is disabled.
4500 Enabling it will normalize magnitude response at DC to 0dB.
4503 Set transform type of IIR filter.
4512 Set precison of filtering.
4515 Pick automatic sample format depending on surround filters.
4517 Always use signed 16-bit.
4519 Always use signed 32-bit.
4521 Always use float 32-bit.
4523 Always use float 64-bit.
4527 @subsection Commands
4529 This filter supports the following commands:
4532 Change highpass frequency.
4533 Syntax for the command is : "@var{frequency}"
4536 Change highpass width_type.
4537 Syntax for the command is : "@var{width_type}"
4540 Change highpass width.
4541 Syntax for the command is : "@var{width}"
4544 Change highpass mix.
4545 Syntax for the command is : "@var{mix}"
4550 Join multiple input streams into one multi-channel stream.
4552 It accepts the following parameters:
4556 The number of input streams. It defaults to 2.
4558 @item channel_layout
4559 The desired output channel layout. It defaults to stereo.
4562 Map channels from inputs to output. The argument is a '|'-separated list of
4563 mappings, each in the @code{@var{input_idx}.@var{in_channel}-@var{out_channel}}
4564 form. @var{input_idx} is the 0-based index of the input stream. @var{in_channel}
4565 can be either the name of the input channel (e.g. FL for front left) or its
4566 index in the specified input stream. @var{out_channel} is the name of the output
4570 The filter will attempt to guess the mappings when they are not specified
4571 explicitly. It does so by first trying to find an unused matching input channel
4572 and if that fails it picks the first unused input channel.
4574 Join 3 inputs (with properly set channel layouts):
4576 ffmpeg -i INPUT1 -i INPUT2 -i INPUT3 -filter_complex join=inputs=3 OUTPUT
4579 Build a 5.1 output from 6 single-channel streams:
4581 ffmpeg -i fl -i fr -i fc -i sl -i sr -i lfe -filter_complex
4582 'join=inputs=6:channel_layout=5.1:map=0.0-FL|1.0-FR|2.0-FC|3.0-SL|4.0-SR|5.0-LFE'
4588 Load a LADSPA (Linux Audio Developer's Simple Plugin API) plugin.
4590 To enable compilation of this filter you need to configure FFmpeg with
4591 @code{--enable-ladspa}.
4595 Specifies the name of LADSPA plugin library to load. If the environment
4596 variable @env{LADSPA_PATH} is defined, the LADSPA plugin is searched in
4597 each one of the directories specified by the colon separated list in
4598 @env{LADSPA_PATH}, otherwise in the standard LADSPA paths, which are in
4599 this order: @file{HOME/.ladspa/lib/}, @file{/usr/local/lib/ladspa/},
4600 @file{/usr/lib/ladspa/}.
4603 Specifies the plugin within the library. Some libraries contain only
4604 one plugin, but others contain many of them. If this is not set filter
4605 will list all available plugins within the specified library.
4608 Set the '|' separated list of controls which are zero or more floating point
4609 values that determine the behavior of the loaded plugin (for example delay,
4611 Controls need to be defined using the following syntax:
4612 c0=@var{value0}|c1=@var{value1}|c2=@var{value2}|..., where
4613 @var{valuei} is the value set on the @var{i}-th control.
4614 Alternatively they can be also defined using the following syntax:
4615 @var{value0}|@var{value1}|@var{value2}|..., where
4616 @var{valuei} is the value set on the @var{i}-th control.
4617 If @option{controls} is set to @code{help}, all available controls and
4618 their valid ranges are printed.
4620 @item sample_rate, s
4621 Specify the sample rate, default to 44100. Only used if plugin have
4625 Set the number of samples per channel per each output frame, default
4626 is 1024. Only used if plugin have zero inputs.
4629 Set the minimum duration of the sourced audio. See
4630 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4631 for the accepted syntax.
4632 Note that the resulting duration may be greater than the specified duration,
4633 as the generated audio is always cut at the end of a complete frame.
4634 If not specified, or the expressed duration is negative, the audio is
4635 supposed to be generated forever.
4636 Only used if plugin have zero inputs.
4639 Enable latency compensation, by default is disabled.
4640 Only used if plugin have inputs.
4643 @subsection Examples
4647 List all available plugins within amp (LADSPA example plugin) library:
4653 List all available controls and their valid ranges for @code{vcf_notch}
4654 plugin from @code{VCF} library:
4656 ladspa=f=vcf:p=vcf_notch:c=help
4660 Simulate low quality audio equipment using @code{Computer Music Toolkit} (CMT)
4663 ladspa=file=cmt:plugin=lofi:controls=c0=22|c1=12|c2=12
4667 Add reverberation to the audio using TAP-plugins
4668 (Tom's Audio Processing plugins):
4670 ladspa=file=tap_reverb:tap_reverb
4674 Generate white noise, with 0.2 amplitude:
4676 ladspa=file=cmt:noise_source_white:c=c0=.2
4680 Generate 20 bpm clicks using plugin @code{C* Click - Metronome} from the
4681 @code{C* Audio Plugin Suite} (CAPS) library:
4683 ladspa=file=caps:Click:c=c1=20'
4687 Apply @code{C* Eq10X2 - Stereo 10-band equaliser} effect:
4689 ladspa=caps:Eq10X2:c=c0=-48|c9=-24|c3=12|c4=2
4693 Increase volume by 20dB using fast lookahead limiter from Steve Harris
4694 @code{SWH Plugins} collection:
4696 ladspa=fast_lookahead_limiter_1913:fastLookaheadLimiter:20|0|2
4700 Attenuate low frequencies using Multiband EQ from Steve Harris
4701 @code{SWH Plugins} collection:
4703 ladspa=mbeq_1197:mbeq:-24|-24|-24|0|0|0|0|0|0|0|0|0|0|0|0
4707 Reduce stereo image using @code{Narrower} from the @code{C* Audio Plugin Suite}
4710 ladspa=caps:Narrower
4714 Another white noise, now using @code{C* Audio Plugin Suite} (CAPS) library:
4716 ladspa=caps:White:.2
4720 Some fractal noise, using @code{C* Audio Plugin Suite} (CAPS) library:
4722 ladspa=caps:Fractal:c=c1=1
4726 Dynamic volume normalization using @code{VLevel} plugin:
4728 ladspa=vlevel-ladspa:vlevel_mono
4732 @subsection Commands
4734 This filter supports the following commands:
4737 Modify the @var{N}-th control value.
4739 If the specified value is not valid, it is ignored and prior one is kept.
4744 EBU R128 loudness normalization. Includes both dynamic and linear normalization modes.
4745 Support for both single pass (livestreams, files) and double pass (files) modes.
4746 This algorithm can target IL, LRA, and maximum true peak. In dynamic mode, to accurately
4747 detect true peaks, the audio stream will be upsampled to 192 kHz.
4748 Use the @code{-ar} option or @code{aresample} filter to explicitly set an output sample rate.
4750 The filter accepts the following options:
4754 Set integrated loudness target.
4755 Range is -70.0 - -5.0. Default value is -24.0.
4758 Set loudness range target.
4759 Range is 1.0 - 20.0. Default value is 7.0.
4762 Set maximum true peak.
4763 Range is -9.0 - +0.0. Default value is -2.0.
4765 @item measured_I, measured_i
4766 Measured IL of input file.
4767 Range is -99.0 - +0.0.
4769 @item measured_LRA, measured_lra
4770 Measured LRA of input file.
4771 Range is 0.0 - 99.0.
4773 @item measured_TP, measured_tp
4774 Measured true peak of input file.
4775 Range is -99.0 - +99.0.
4777 @item measured_thresh
4778 Measured threshold of input file.
4779 Range is -99.0 - +0.0.
4782 Set offset gain. Gain is applied before the true-peak limiter.
4783 Range is -99.0 - +99.0. Default is +0.0.
4786 Normalize by linearly scaling the source audio.
4787 @code{measured_I}, @code{measured_LRA}, @code{measured_TP},
4788 and @code{measured_thresh} must all be specified. Target LRA shouldn't
4789 be lower than source LRA and the change in integrated loudness shouldn't
4790 result in a true peak which exceeds the target TP. If any of these
4791 conditions aren't met, normalization mode will revert to @var{dynamic}.
4792 Options are @code{true} or @code{false}. Default is @code{true}.
4795 Treat mono input files as "dual-mono". If a mono file is intended for playback
4796 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
4797 If set to @code{true}, this option will compensate for this effect.
4798 Multi-channel input files are not affected by this option.
4799 Options are true or false. Default is false.
4802 Set print format for stats. Options are summary, json, or none.
4803 Default value is none.
4808 Apply a low-pass filter with 3dB point frequency.
4809 The filter can be either single-pole or double-pole (the default).
4810 The filter roll off at 6dB per pole per octave (20dB per pole per decade).
4812 The filter accepts the following options:
4816 Set frequency in Hz. Default is 500.
4819 Set number of poles. Default is 2.
4822 Set method to specify band-width of filter.
4837 Specify the band-width of a filter in width_type units.
4838 Applies only to double-pole filter.
4839 The default is 0.707q and gives a Butterworth response.
4842 How much to use filtered signal in output. Default is 1.
4843 Range is between 0 and 1.
4846 Specify which channels to filter, by default all available are filtered.
4849 Normalize biquad coefficients, by default is disabled.
4850 Enabling it will normalize magnitude response at DC to 0dB.
4853 Set transform type of IIR filter.
4862 Set precison of filtering.
4865 Pick automatic sample format depending on surround filters.
4867 Always use signed 16-bit.
4869 Always use signed 32-bit.
4871 Always use float 32-bit.
4873 Always use float 64-bit.
4877 @subsection Examples
4880 Lowpass only LFE channel, it LFE is not present it does nothing:
4886 @subsection Commands
4888 This filter supports the following commands:
4891 Change lowpass frequency.
4892 Syntax for the command is : "@var{frequency}"
4895 Change lowpass width_type.
4896 Syntax for the command is : "@var{width_type}"
4899 Change lowpass width.
4900 Syntax for the command is : "@var{width}"
4904 Syntax for the command is : "@var{mix}"
4909 Load a LV2 (LADSPA Version 2) plugin.
4911 To enable compilation of this filter you need to configure FFmpeg with
4912 @code{--enable-lv2}.
4916 Specifies the plugin URI. You may need to escape ':'.
4919 Set the '|' separated list of controls which are zero or more floating point
4920 values that determine the behavior of the loaded plugin (for example delay,
4922 If @option{controls} is set to @code{help}, all available controls and
4923 their valid ranges are printed.
4925 @item sample_rate, s
4926 Specify the sample rate, default to 44100. Only used if plugin have
4930 Set the number of samples per channel per each output frame, default
4931 is 1024. Only used if plugin have zero inputs.
4934 Set the minimum duration of the sourced audio. See
4935 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
4936 for the accepted syntax.
4937 Note that the resulting duration may be greater than the specified duration,
4938 as the generated audio is always cut at the end of a complete frame.
4939 If not specified, or the expressed duration is negative, the audio is
4940 supposed to be generated forever.
4941 Only used if plugin have zero inputs.
4944 @subsection Examples
4948 Apply bass enhancer plugin from Calf:
4950 lv2=p=http\\\\://calf.sourceforge.net/plugins/BassEnhancer:c=amount=2
4954 Apply vinyl plugin from Calf:
4956 lv2=p=http\\\\://calf.sourceforge.net/plugins/Vinyl:c=drone=0.2|aging=0.5
4960 Apply bit crusher plugin from ArtyFX:
4962 lv2=p=http\\\\://www.openavproductions.com/artyfx#bitta:c=crush=0.3
4967 Multiband Compress or expand the audio's dynamic range.
4969 The input audio is divided into bands using 4th order Linkwitz-Riley IIRs.
4970 This is akin to the crossover of a loudspeaker, and results in flat frequency
4971 response when absent compander action.
4973 It accepts the following parameters:
4977 This option syntax is:
4978 attack,decay,[attack,decay..] soft-knee points crossover_frequency [delay [initial_volume [gain]]] | attack,decay ...
4979 For explanation of each item refer to compand filter documentation.
4985 Mix channels with specific gain levels. The filter accepts the output
4986 channel layout followed by a set of channels definitions.
4988 This filter is also designed to efficiently remap the channels of an audio
4991 The filter accepts parameters of the form:
4992 "@var{l}|@var{outdef}|@var{outdef}|..."
4996 output channel layout or number of channels
4999 output channel specification, of the form:
5000 "@var{out_name}=[@var{gain}*]@var{in_name}[(+-)[@var{gain}*]@var{in_name}...]"
5003 output channel to define, either a channel name (FL, FR, etc.) or a channel
5004 number (c0, c1, etc.)
5007 multiplicative coefficient for the channel, 1 leaving the volume unchanged
5010 input channel to use, see out_name for details; it is not possible to mix
5011 named and numbered input channels
5014 If the `=' in a channel specification is replaced by `<', then the gains for
5015 that specification will be renormalized so that the total is 1, thus
5016 avoiding clipping noise.
5018 @subsection Mixing examples
5020 For example, if you want to down-mix from stereo to mono, but with a bigger
5021 factor for the left channel:
5023 pan=1c|c0=0.9*c0+0.1*c1
5026 A customized down-mix to stereo that works automatically for 3-, 4-, 5- and
5027 7-channels surround:
5029 pan=stereo| FL < FL + 0.5*FC + 0.6*BL + 0.6*SL | FR < FR + 0.5*FC + 0.6*BR + 0.6*SR
5032 Note that @command{ffmpeg} integrates a default down-mix (and up-mix) system
5033 that should be preferred (see "-ac" option) unless you have very specific
5036 @subsection Remapping examples
5038 The channel remapping will be effective if, and only if:
5041 @item gain coefficients are zeroes or ones,
5042 @item only one input per channel output,
5045 If all these conditions are satisfied, the filter will notify the user ("Pure
5046 channel mapping detected"), and use an optimized and lossless method to do the
5049 For example, if you have a 5.1 source and want a stereo audio stream by
5050 dropping the extra channels:
5052 pan="stereo| c0=FL | c1=FR"
5055 Given the same source, you can also switch front left and front right channels
5056 and keep the input channel layout:
5058 pan="5.1| c0=c1 | c1=c0 | c2=c2 | c3=c3 | c4=c4 | c5=c5"
5061 If the input is a stereo audio stream, you can mute the front left channel (and
5062 still keep the stereo channel layout) with:
5067 Still with a stereo audio stream input, you can copy the right channel in both
5068 front left and right:
5070 pan="stereo| c0=FR | c1=FR"
5075 ReplayGain scanner filter. This filter takes an audio stream as an input and
5076 outputs it unchanged.
5077 At end of filtering it displays @code{track_gain} and @code{track_peak}.
5081 Convert the audio sample format, sample rate and channel layout. It is
5082 not meant to be used directly.
5085 Apply time-stretching and pitch-shifting with librubberband.
5087 To enable compilation of this filter, you need to configure FFmpeg with
5088 @code{--enable-librubberband}.
5090 The filter accepts the following options:
5094 Set tempo scale factor.
5097 Set pitch scale factor.
5100 Set transients detector.
5101 Possible values are:
5110 Possible values are:
5119 Possible values are:
5126 Set processing window size.
5127 Possible values are:
5136 Possible values are:
5143 Enable formant preservation when shift pitching.
5144 Possible values are:
5152 Possible values are:
5161 Possible values are:
5168 @subsection Commands
5170 This filter supports the following commands:
5173 Change filter tempo scale factor.
5174 Syntax for the command is : "@var{tempo}"
5177 Change filter pitch scale factor.
5178 Syntax for the command is : "@var{pitch}"
5181 @section sidechaincompress
5183 This filter acts like normal compressor but has the ability to compress
5184 detected signal using second input signal.
5185 It needs two input streams and returns one output stream.
5186 First input stream will be processed depending on second stream signal.
5187 The filtered signal then can be filtered with other filters in later stages of
5188 processing. See @ref{pan} and @ref{amerge} filter.
5190 The filter accepts the following options:
5194 Set input gain. Default is 1. Range is between 0.015625 and 64.
5197 Set mode of compressor operation. Can be @code{upward} or @code{downward}.
5198 Default is @code{downward}.
5201 If a signal of second stream raises above this level it will affect the gain
5202 reduction of first stream.
5203 By default is 0.125. Range is between 0.00097563 and 1.
5206 Set a ratio about which the signal is reduced. 1:2 means that if the level
5207 raised 4dB above the threshold, it will be only 2dB above after the reduction.
5208 Default is 2. Range is between 1 and 20.
5211 Amount of milliseconds the signal has to rise above the threshold before gain
5212 reduction starts. Default is 20. Range is between 0.01 and 2000.
5215 Amount of milliseconds the signal has to fall below the threshold before
5216 reduction is decreased again. Default is 250. Range is between 0.01 and 9000.
5219 Set the amount by how much signal will be amplified after processing.
5220 Default is 1. Range is from 1 to 64.
5223 Curve the sharp knee around the threshold to enter gain reduction more softly.
5224 Default is 2.82843. Range is between 1 and 8.
5227 Choose if the @code{average} level between all channels of side-chain stream
5228 or the louder(@code{maximum}) channel of side-chain stream affects the
5229 reduction. Default is @code{average}.
5232 Should the exact signal be taken in case of @code{peak} or an RMS one in case
5233 of @code{rms}. Default is @code{rms} which is mainly smoother.
5236 Set sidechain gain. Default is 1. Range is between 0.015625 and 64.
5239 How much to use compressed signal in output. Default is 1.
5240 Range is between 0 and 1.
5243 @subsection Commands
5245 This filter supports the all above options as @ref{commands}.
5247 @subsection Examples
5251 Full ffmpeg example taking 2 audio inputs, 1st input to be compressed
5252 depending on the signal of 2nd input and later compressed signal to be
5253 merged with 2nd input:
5255 ffmpeg -i main.flac -i sidechain.flac -filter_complex "[1:a]asplit=2[sc][mix];[0:a][sc]sidechaincompress[compr];[compr][mix]amerge"
5259 @section sidechaingate
5261 A sidechain gate acts like a normal (wideband) gate but has the ability to
5262 filter the detected signal before sending it to the gain reduction stage.
5263 Normally a gate uses the full range signal to detect a level above the
5265 For example: If you cut all lower frequencies from your sidechain signal
5266 the gate will decrease the volume of your track only if not enough highs
5267 appear. With this technique you are able to reduce the resonation of a
5268 natural drum or remove "rumbling" of muted strokes from a heavily distorted
5270 It needs two input streams and returns one output stream.
5271 First input stream will be processed depending on second stream signal.
5273 The filter accepts the following options:
5277 Set input level before filtering.
5278 Default is 1. Allowed range is from 0.015625 to 64.
5281 Set the mode of operation. Can be @code{upward} or @code{downward}.
5282 Default is @code{downward}. If set to @code{upward} mode, higher parts of signal
5283 will be amplified, expanding dynamic range in upward direction.
5284 Otherwise, in case of @code{downward} lower parts of signal will be reduced.
5287 Set the level of gain reduction when the signal is below the threshold.
5288 Default is 0.06125. Allowed range is from 0 to 1.
5289 Setting this to 0 disables reduction and then filter behaves like expander.
5292 If a signal rises above this level the gain reduction is released.
5293 Default is 0.125. Allowed range is from 0 to 1.
5296 Set a ratio about which the signal is reduced.
5297 Default is 2. Allowed range is from 1 to 9000.
5300 Amount of milliseconds the signal has to rise above the threshold before gain
5302 Default is 20 milliseconds. Allowed range is from 0.01 to 9000.
5305 Amount of milliseconds the signal has to fall below the threshold before the
5306 reduction is increased again. Default is 250 milliseconds.
5307 Allowed range is from 0.01 to 9000.
5310 Set amount of amplification of signal after processing.
5311 Default is 1. Allowed range is from 1 to 64.
5314 Curve the sharp knee around the threshold to enter gain reduction more softly.
5315 Default is 2.828427125. Allowed range is from 1 to 8.
5318 Choose if exact signal should be taken for detection or an RMS like one.
5319 Default is rms. Can be peak or rms.
5322 Choose if the average level between all channels or the louder channel affects
5324 Default is average. Can be average or maximum.
5327 Set sidechain gain. Default is 1. Range is from 0.015625 to 64.
5330 @subsection Commands
5332 This filter supports the all above options as @ref{commands}.
5334 @section silencedetect
5336 Detect silence in an audio stream.
5338 This filter logs a message when it detects that the input audio volume is less
5339 or equal to a noise tolerance value for a duration greater or equal to the
5340 minimum detected noise duration.
5342 The printed times and duration are expressed in seconds. The
5343 @code{lavfi.silence_start} or @code{lavfi.silence_start.X} metadata key
5344 is set on the first frame whose timestamp equals or exceeds the detection
5345 duration and it contains the timestamp of the first frame of the silence.
5347 The @code{lavfi.silence_duration} or @code{lavfi.silence_duration.X}
5348 and @code{lavfi.silence_end} or @code{lavfi.silence_end.X} metadata
5349 keys are set on the first frame after the silence. If @option{mono} is
5350 enabled, and each channel is evaluated separately, the @code{.X}
5351 suffixed keys are used, and @code{X} corresponds to the channel number.
5353 The filter accepts the following options:
5357 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
5358 specified value) or amplitude ratio. Default is -60dB, or 0.001.
5361 Set silence duration until notification (default is 2 seconds). See
5362 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5363 for the accepted syntax.
5366 Process each channel separately, instead of combined. By default is disabled.
5369 @subsection Examples
5373 Detect 5 seconds of silence with -50dB noise tolerance:
5375 silencedetect=n=-50dB:d=5
5379 Complete example with @command{ffmpeg} to detect silence with 0.0001 noise
5380 tolerance in @file{silence.mp3}:
5382 ffmpeg -i silence.mp3 -af silencedetect=noise=0.0001 -f null -
5386 @section silenceremove
5388 Remove silence from the beginning, middle or end of the audio.
5390 The filter accepts the following options:
5394 This value is used to indicate if audio should be trimmed at beginning of
5395 the audio. A value of zero indicates no silence should be trimmed from the
5396 beginning. When specifying a non-zero value, it trims audio up until it
5397 finds non-silence. Normally, when trimming silence from beginning of audio
5398 the @var{start_periods} will be @code{1} but it can be increased to higher
5399 values to trim all audio up to specific count of non-silence periods.
5400 Default value is @code{0}.
5402 @item start_duration
5403 Specify the amount of time that non-silence must be detected before it stops
5404 trimming audio. By increasing the duration, bursts of noises can be treated
5405 as silence and trimmed off. Default value is @code{0}.
5407 @item start_threshold
5408 This indicates what sample value should be treated as silence. For digital
5409 audio, a value of @code{0} may be fine but for audio recorded from analog,
5410 you may wish to increase the value to account for background noise.
5411 Can be specified in dB (in case "dB" is appended to the specified value)
5412 or amplitude ratio. Default value is @code{0}.
5415 Specify max duration of silence at beginning that will be kept after
5416 trimming. Default is 0, which is equal to trimming all samples detected
5420 Specify mode of detection of silence end in start of multi-channel audio.
5421 Can be @var{any} or @var{all}. Default is @var{any}.
5422 With @var{any}, any sample that is detected as non-silence will cause
5423 stopped trimming of silence.
5424 With @var{all}, only if all channels are detected as non-silence will cause
5425 stopped trimming of silence.
5428 Set the count for trimming silence from the end of audio.
5429 To remove silence from the middle of a file, specify a @var{stop_periods}
5430 that is negative. This value is then treated as a positive value and is
5431 used to indicate the effect should restart processing as specified by
5432 @var{start_periods}, making it suitable for removing periods of silence
5433 in the middle of the audio.
5434 Default value is @code{0}.
5437 Specify a duration of silence that must exist before audio is not copied any
5438 more. By specifying a higher duration, silence that is wanted can be left in
5440 Default value is @code{0}.
5442 @item stop_threshold
5443 This is the same as @option{start_threshold} but for trimming silence from
5445 Can be specified in dB (in case "dB" is appended to the specified value)
5446 or amplitude ratio. Default value is @code{0}.
5449 Specify max duration of silence at end that will be kept after
5450 trimming. Default is 0, which is equal to trimming all samples detected
5454 Specify mode of detection of silence start in end of multi-channel audio.
5455 Can be @var{any} or @var{all}. Default is @var{any}.
5456 With @var{any}, any sample that is detected as non-silence will cause
5457 stopped trimming of silence.
5458 With @var{all}, only if all channels are detected as non-silence will cause
5459 stopped trimming of silence.
5462 Set how is silence detected. Can be @code{rms} or @code{peak}. Second is faster
5463 and works better with digital silence which is exactly 0.
5464 Default value is @code{rms}.
5467 Set duration in number of seconds used to calculate size of window in number
5468 of samples for detecting silence.
5469 Default value is @code{0.02}. Allowed range is from @code{0} to @code{10}.
5472 @subsection Examples
5476 The following example shows how this filter can be used to start a recording
5477 that does not contain the delay at the start which usually occurs between
5478 pressing the record button and the start of the performance:
5480 silenceremove=start_periods=1:start_duration=5:start_threshold=0.02
5484 Trim all silence encountered from beginning to end where there is more than 1
5485 second of silence in audio:
5487 silenceremove=stop_periods=-1:stop_duration=1:stop_threshold=-90dB
5491 Trim all digital silence samples, using peak detection, from beginning to end
5492 where there is more than 0 samples of digital silence in audio and digital
5493 silence is detected in all channels at same positions in stream:
5495 silenceremove=window=0:detection=peak:stop_mode=all:start_mode=all:stop_periods=-1:stop_threshold=0
5501 SOFAlizer uses head-related transfer functions (HRTFs) to create virtual
5502 loudspeakers around the user for binaural listening via headphones (audio
5503 formats up to 9 channels supported).
5504 The HRTFs are stored in SOFA files (see @url{http://www.sofacoustics.org/} for a database).
5505 SOFAlizer is developed at the Acoustics Research Institute (ARI) of the
5506 Austrian Academy of Sciences.
5508 To enable compilation of this filter you need to configure FFmpeg with
5509 @code{--enable-libmysofa}.
5511 The filter accepts the following options:
5515 Set the SOFA file used for rendering.
5518 Set gain applied to audio. Value is in dB. Default is 0.
5521 Set rotation of virtual loudspeakers in deg. Default is 0.
5524 Set elevation of virtual speakers in deg. Default is 0.
5527 Set distance in meters between loudspeakers and the listener with near-field
5528 HRTFs. Default is 1.
5531 Set processing type. Can be @var{time} or @var{freq}. @var{time} is
5532 processing audio in time domain which is slow.
5533 @var{freq} is processing audio in frequency domain which is fast.
5534 Default is @var{freq}.
5537 Set custom positions of virtual loudspeakers. Syntax for this option is:
5538 <CH> <AZIM> <ELEV>[|<CH> <AZIM> <ELEV>|...].
5539 Each virtual loudspeaker is described with short channel name following with
5540 azimuth and elevation in degrees.
5541 Each virtual loudspeaker description is separated by '|'.
5542 For example to override front left and front right channel positions use:
5543 'speakers=FL 45 15|FR 345 15'.
5544 Descriptions with unrecognised channel names are ignored.
5547 Set custom gain for LFE channels. Value is in dB. Default is 0.
5550 Set custom frame size in number of samples. Default is 1024.
5551 Allowed range is from 1024 to 96000. Only used if option @samp{type}
5552 is set to @var{freq}.
5555 Should all IRs be normalized upon importing SOFA file.
5556 By default is enabled.
5559 Should nearest IRs be interpolated with neighbor IRs if exact position
5560 does not match. By default is disabled.
5563 Minphase all IRs upon loading of SOFA file. By default is disabled.
5566 Set neighbor search angle step. Only used if option @var{interpolate} is enabled.
5569 Set neighbor search radius step. Only used if option @var{interpolate} is enabled.
5572 @subsection Examples
5576 Using ClubFritz6 sofa file:
5578 sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=1
5582 Using ClubFritz12 sofa file and bigger radius with small rotation:
5584 sofalizer=sofa=/path/to/ClubFritz12.sofa:type=freq:radius=2:rotation=5
5588 Similar as above but with custom speaker positions for front left, front right, back left and back right
5589 and also with custom gain:
5591 "sofalizer=sofa=/path/to/ClubFritz6.sofa:type=freq:radius=2:speakers=FL 45|FR 315|BL 135|BR 225:gain=28"
5598 This filter expands or compresses each half-cycle of audio samples
5599 (local set of samples all above or all below zero and between two nearest zero crossings) depending
5600 on threshold value, so audio reaches target peak value under conditions controlled by below options.
5602 The filter accepts the following options:
5606 Set the expansion target peak value. This specifies the highest allowed absolute amplitude
5607 level for the normalized audio input. Default value is 0.95. Allowed range is from 0.0 to 1.0.
5610 Set the maximum expansion factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5611 This option controls maximum local half-cycle of samples expansion. The maximum expansion
5612 would be such that local peak value reaches target peak value but never to surpass it and that
5613 ratio between new and previous peak value does not surpass this option value.
5615 @item compression, c
5616 Set the maximum compression factor. Allowed range is from 1.0 to 50.0. Default value is 2.0.
5617 This option controls maximum local half-cycle of samples compression. This option is used
5618 only if @option{threshold} option is set to value greater than 0.0, then in such cases
5619 when local peak is lower or same as value set by @option{threshold} all samples belonging to
5620 that peak's half-cycle will be compressed by current compression factor.
5623 Set the threshold value. Default value is 0.0. Allowed range is from 0.0 to 1.0.
5624 This option specifies which half-cycles of samples will be compressed and which will be expanded.
5625 Any half-cycle samples with their local peak value below or same as this option value will be
5626 compressed by current compression factor, otherwise, if greater than threshold value they will be
5627 expanded with expansion factor so that it could reach peak target value but never surpass it.
5630 Set the expansion raising amount per each half-cycle of samples. Default value is 0.001.
5631 Allowed range is from 0.0 to 1.0. This controls how fast expansion factor is raised per
5632 each new half-cycle until it reaches @option{expansion} value.
5633 Setting this options too high may lead to distortions.
5636 Set the compression raising amount per each half-cycle of samples. Default value is 0.001.
5637 Allowed range is from 0.0 to 1.0. This controls how fast compression factor is raised per
5638 each new half-cycle until it reaches @option{compression} value.
5641 Specify which channels to filter, by default all available channels are filtered.
5644 Enable inverted filtering, by default is disabled. This inverts interpretation of @option{threshold}
5645 option. When enabled any half-cycle of samples with their local peak value below or same as
5646 @option{threshold} option will be expanded otherwise it will be compressed.
5649 Link channels when calculating gain applied to each filtered channel sample, by default is disabled.
5650 When disabled each filtered channel gain calculation is independent, otherwise when this option
5651 is enabled the minimum of all possible gains for each filtered channel is used.
5654 @subsection Commands
5656 This filter supports the all above options as @ref{commands}.
5658 @section stereotools
5660 This filter has some handy utilities to manage stereo signals, for converting
5661 M/S stereo recordings to L/R signal while having control over the parameters
5662 or spreading the stereo image of master track.
5664 The filter accepts the following options:
5668 Set input level before filtering for both channels. Defaults is 1.
5669 Allowed range is from 0.015625 to 64.
5672 Set output level after filtering for both channels. Defaults is 1.
5673 Allowed range is from 0.015625 to 64.
5676 Set input balance between both channels. Default is 0.
5677 Allowed range is from -1 to 1.
5680 Set output balance between both channels. Default is 0.
5681 Allowed range is from -1 to 1.
5684 Enable softclipping. Results in analog distortion instead of harsh digital 0dB
5685 clipping. Disabled by default.
5688 Mute the left channel. Disabled by default.
5691 Mute the right channel. Disabled by default.
5694 Change the phase of the left channel. Disabled by default.
5697 Change the phase of the right channel. Disabled by default.
5700 Set stereo mode. Available values are:
5704 Left/Right to Left/Right, this is default.
5707 Left/Right to Mid/Side.
5710 Mid/Side to Left/Right.
5713 Left/Right to Left/Left.
5716 Left/Right to Right/Right.
5719 Left/Right to Left + Right.
5722 Left/Right to Right/Left.
5725 Mid/Side to Left/Left.
5728 Mid/Side to Right/Right.
5731 Mid/Side to Right/Left.
5734 Left/Right to Left - Right.
5738 Set level of side signal. Default is 1.
5739 Allowed range is from 0.015625 to 64.
5742 Set balance of side signal. Default is 0.
5743 Allowed range is from -1 to 1.
5746 Set level of the middle signal. Default is 1.
5747 Allowed range is from 0.015625 to 64.
5750 Set middle signal pan. Default is 0. Allowed range is from -1 to 1.
5753 Set stereo base between mono and inversed channels. Default is 0.
5754 Allowed range is from -1 to 1.
5757 Set delay in milliseconds how much to delay left from right channel and
5758 vice versa. Default is 0. Allowed range is from -20 to 20.
5761 Set S/C level. Default is 1. Allowed range is from 1 to 100.
5764 Set the stereo phase in degrees. Default is 0. Allowed range is from 0 to 360.
5766 @item bmode_in, bmode_out
5767 Set balance mode for balance_in/balance_out option.
5769 Can be one of the following:
5773 Classic balance mode. Attenuate one channel at time.
5774 Gain is raised up to 1.
5777 Similar as classic mode above but gain is raised up to 2.
5780 Equal power distribution, from -6dB to +6dB range.
5784 @subsection Commands
5786 This filter supports the all above options as @ref{commands}.
5788 @subsection Examples
5792 Apply karaoke like effect:
5794 stereotools=mlev=0.015625
5798 Convert M/S signal to L/R:
5800 "stereotools=mode=ms>lr"
5804 @section stereowiden
5806 This filter enhance the stereo effect by suppressing signal common to both
5807 channels and by delaying the signal of left into right and vice versa,
5808 thereby widening the stereo effect.
5810 The filter accepts the following options:
5814 Time in milliseconds of the delay of left signal into right and vice versa.
5815 Default is 20 milliseconds.
5818 Amount of gain in delayed signal into right and vice versa. Gives a delay
5819 effect of left signal in right output and vice versa which gives widening
5820 effect. Default is 0.3.
5823 Cross feed of left into right with inverted phase. This helps in suppressing
5824 the mono. If the value is 1 it will cancel all the signal common to both
5825 channels. Default is 0.3.
5828 Set level of input signal of original channel. Default is 0.8.
5831 @subsection Commands
5833 This filter supports the all above options except @code{delay} as @ref{commands}.
5835 @section superequalizer
5836 Apply 18 band equalizer.
5838 The filter accepts the following options:
5845 Set 131Hz band gain.
5847 Set 185Hz band gain.
5849 Set 262Hz band gain.
5851 Set 370Hz band gain.
5853 Set 523Hz band gain.
5855 Set 740Hz band gain.
5857 Set 1047Hz band gain.
5859 Set 1480Hz band gain.
5861 Set 2093Hz band gain.
5863 Set 2960Hz band gain.
5865 Set 4186Hz band gain.
5867 Set 5920Hz band gain.
5869 Set 8372Hz band gain.
5871 Set 11840Hz band gain.
5873 Set 16744Hz band gain.
5875 Set 20000Hz band gain.
5879 Apply audio surround upmix filter.
5881 This filter allows to produce multichannel output from audio stream.
5883 The filter accepts the following options:
5887 Set output channel layout. By default, this is @var{5.1}.
5889 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5890 for the required syntax.
5893 Set input channel layout. By default, this is @var{stereo}.
5895 See @ref{channel layout syntax,,the Channel Layout section in the ffmpeg-utils(1) manual,ffmpeg-utils}
5896 for the required syntax.
5899 Set input volume level. By default, this is @var{1}.
5902 Set output volume level. By default, this is @var{1}.
5905 Enable LFE channel output if output channel layout has it. By default, this is enabled.
5908 Set LFE low cut off frequency. By default, this is @var{128} Hz.
5911 Set LFE high cut off frequency. By default, this is @var{256} Hz.
5914 Set LFE mode, can be @var{add} or @var{sub}. Default is @var{add}.
5915 In @var{add} mode, LFE channel is created from input audio and added to output.
5916 In @var{sub} mode, LFE channel is created from input audio and added to output but
5917 also all non-LFE output channels are subtracted with output LFE channel.
5920 Set angle of stereo surround transform, Allowed range is from @var{0} to @var{360}.
5921 Default is @var{90}.
5924 Set front center input volume. By default, this is @var{1}.
5927 Set front center output volume. By default, this is @var{1}.
5930 Set front left input volume. By default, this is @var{1}.
5933 Set front left output volume. By default, this is @var{1}.
5936 Set front right input volume. By default, this is @var{1}.
5939 Set front right output volume. By default, this is @var{1}.
5942 Set side left input volume. By default, this is @var{1}.
5945 Set side left output volume. By default, this is @var{1}.
5948 Set side right input volume. By default, this is @var{1}.
5951 Set side right output volume. By default, this is @var{1}.
5954 Set back left input volume. By default, this is @var{1}.
5957 Set back left output volume. By default, this is @var{1}.
5960 Set back right input volume. By default, this is @var{1}.
5963 Set back right output volume. By default, this is @var{1}.
5966 Set back center input volume. By default, this is @var{1}.
5969 Set back center output volume. By default, this is @var{1}.
5972 Set LFE input volume. By default, this is @var{1}.
5975 Set LFE output volume. By default, this is @var{1}.
5978 Set spread usage of stereo image across X axis for all channels.
5981 Set spread usage of stereo image across Y axis for all channels.
5983 @item fcx, flx, frx, blx, brx, slx, srx, bcx
5984 Set spread usage of stereo image across X axis for each channel.
5986 @item fcy, fly, fry, bly, bry, sly, sry, bcy
5987 Set spread usage of stereo image across Y axis for each channel.
5990 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
5993 Set window function.
5995 It accepts the following values:
6018 Default is @code{hann}.
6021 Set window overlap. If set to 1, the recommended overlap for selected
6022 window function will be picked. Default is @code{0.5}.
6025 @section treble, highshelf
6027 Boost or cut treble (upper) frequencies of the audio using a two-pole
6028 shelving filter with a response similar to that of a standard
6029 hi-fi's tone-controls. This is also known as shelving equalisation (EQ).
6031 The filter accepts the following options:
6035 Give the gain at whichever is the lower of ~22 kHz and the
6036 Nyquist frequency. Its useful range is about -20 (for a large cut)
6037 to +20 (for a large boost). Beware of clipping when using a positive gain.
6040 Set the filter's central frequency and so can be used
6041 to extend or reduce the frequency range to be boosted or cut.
6042 The default value is @code{3000} Hz.
6045 Set method to specify band-width of filter.
6060 Determine how steep is the filter's shelf transition.
6063 Set number of poles. Default is 2.
6066 How much to use filtered signal in output. Default is 1.
6067 Range is between 0 and 1.
6070 Specify which channels to filter, by default all available are filtered.
6073 Normalize biquad coefficients, by default is disabled.
6074 Enabling it will normalize magnitude response at DC to 0dB.
6077 Set transform type of IIR filter.
6086 Set precison of filtering.
6089 Pick automatic sample format depending on surround filters.
6091 Always use signed 16-bit.
6093 Always use signed 32-bit.
6095 Always use float 32-bit.
6097 Always use float 64-bit.
6101 @subsection Commands
6103 This filter supports the following commands:
6106 Change treble frequency.
6107 Syntax for the command is : "@var{frequency}"
6110 Change treble width_type.
6111 Syntax for the command is : "@var{width_type}"
6114 Change treble width.
6115 Syntax for the command is : "@var{width}"
6119 Syntax for the command is : "@var{gain}"
6123 Syntax for the command is : "@var{mix}"
6128 Sinusoidal amplitude modulation.
6130 The filter accepts the following options:
6134 Modulation frequency in Hertz. Modulation frequencies in the subharmonic range
6135 (20 Hz or lower) will result in a tremolo effect.
6136 This filter may also be used as a ring modulator by specifying
6137 a modulation frequency higher than 20 Hz.
6138 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6141 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6142 Default value is 0.5.
6147 Sinusoidal phase modulation.
6149 The filter accepts the following options:
6153 Modulation frequency in Hertz.
6154 Range is 0.1 - 20000.0. Default value is 5.0 Hz.
6157 Depth of modulation as a percentage. Range is 0.0 - 1.0.
6158 Default value is 0.5.
6163 Adjust the input audio volume.
6165 It accepts the following parameters:
6169 Set audio volume expression.
6171 Output values are clipped to the maximum value.
6173 The output audio volume is given by the relation:
6175 @var{output_volume} = @var{volume} * @var{input_volume}
6178 The default value for @var{volume} is "1.0".
6181 This parameter represents the mathematical precision.
6183 It determines which input sample formats will be allowed, which affects the
6184 precision of the volume scaling.
6188 8-bit fixed-point; this limits input sample format to U8, S16, and S32.
6190 32-bit floating-point; this limits input sample format to FLT. (default)
6192 64-bit floating-point; this limits input sample format to DBL.
6196 Choose the behaviour on encountering ReplayGain side data in input frames.
6200 Remove ReplayGain side data, ignoring its contents (the default).
6203 Ignore ReplayGain side data, but leave it in the frame.
6206 Prefer the track gain, if present.
6209 Prefer the album gain, if present.
6212 @item replaygain_preamp
6213 Pre-amplification gain in dB to apply to the selected replaygain gain.
6215 Default value for @var{replaygain_preamp} is 0.0.
6217 @item replaygain_noclip
6218 Prevent clipping by limiting the gain applied.
6220 Default value for @var{replaygain_noclip} is 1.
6223 Set when the volume expression is evaluated.
6225 It accepts the following values:
6228 only evaluate expression once during the filter initialization, or
6229 when the @samp{volume} command is sent
6232 evaluate expression for each incoming frame
6235 Default value is @samp{once}.
6238 The volume expression can contain the following parameters.
6242 frame number (starting at zero)
6245 @item nb_consumed_samples
6246 number of samples consumed by the filter
6248 number of samples in the current frame
6250 original frame position in the file
6256 PTS at start of stream
6258 time at start of stream
6264 last set volume value
6267 Note that when @option{eval} is set to @samp{once} only the
6268 @var{sample_rate} and @var{tb} variables are available, all other
6269 variables will evaluate to NAN.
6271 @subsection Commands
6273 This filter supports the following commands:
6276 Modify the volume expression.
6277 The command accepts the same syntax of the corresponding option.
6279 If the specified expression is not valid, it is kept at its current
6283 @subsection Examples
6287 Halve the input audio volume:
6291 volume=volume=-6.0206dB
6294 In all the above example the named key for @option{volume} can be
6295 omitted, for example like in:
6301 Increase input audio power by 6 decibels using fixed-point precision:
6303 volume=volume=6dB:precision=fixed
6307 Fade volume after time 10 with an annihilation period of 5 seconds:
6309 volume='if(lt(t,10),1,max(1-(t-10)/5,0))':eval=frame
6313 @section volumedetect
6315 Detect the volume of the input video.
6317 The filter has no parameters. The input is not modified. Statistics about
6318 the volume will be printed in the log when the input stream end is reached.
6320 In particular it will show the mean volume (root mean square), maximum
6321 volume (on a per-sample basis), and the beginning of a histogram of the
6322 registered volume values (from the maximum value to a cumulated 1/1000 of
6325 All volumes are in decibels relative to the maximum PCM value.
6327 @subsection Examples
6329 Here is an excerpt of the output:
6331 [Parsed_volumedetect_0 @ 0xa23120] mean_volume: -27 dB
6332 [Parsed_volumedetect_0 @ 0xa23120] max_volume: -4 dB
6333 [Parsed_volumedetect_0 @ 0xa23120] histogram_4db: 6
6334 [Parsed_volumedetect_0 @ 0xa23120] histogram_5db: 62
6335 [Parsed_volumedetect_0 @ 0xa23120] histogram_6db: 286
6336 [Parsed_volumedetect_0 @ 0xa23120] histogram_7db: 1042
6337 [Parsed_volumedetect_0 @ 0xa23120] histogram_8db: 2551
6338 [Parsed_volumedetect_0 @ 0xa23120] histogram_9db: 4609
6339 [Parsed_volumedetect_0 @ 0xa23120] histogram_10db: 8409
6345 The mean square energy is approximately -27 dB, or 10^-2.7.
6347 The largest sample is at -4 dB, or more precisely between -4 dB and -5 dB.
6349 There are 6 samples at -4 dB, 62 at -5 dB, 286 at -6 dB, etc.
6352 In other words, raising the volume by +4 dB does not cause any clipping,
6353 raising it by +5 dB causes clipping for 6 samples, etc.
6355 @c man end AUDIO FILTERS
6357 @chapter Audio Sources
6358 @c man begin AUDIO SOURCES
6360 Below is a description of the currently available audio sources.
6364 Buffer audio frames, and make them available to the filter chain.
6366 This source is mainly intended for a programmatic use, in particular
6367 through the interface defined in @file{libavfilter/buffersrc.h}.
6369 It accepts the following parameters:
6373 The timebase which will be used for timestamps of submitted frames. It must be
6374 either a floating-point number or in @var{numerator}/@var{denominator} form.
6377 The sample rate of the incoming audio buffers.
6380 The sample format of the incoming audio buffers.
6381 Either a sample format name or its corresponding integer representation from
6382 the enum AVSampleFormat in @file{libavutil/samplefmt.h}
6384 @item channel_layout
6385 The channel layout of the incoming audio buffers.
6386 Either a channel layout name from channel_layout_map in
6387 @file{libavutil/channel_layout.c} or its corresponding integer representation
6388 from the AV_CH_LAYOUT_* macros in @file{libavutil/channel_layout.h}
6391 The number of channels of the incoming audio buffers.
6392 If both @var{channels} and @var{channel_layout} are specified, then they
6397 @subsection Examples
6400 abuffer=sample_rate=44100:sample_fmt=s16p:channel_layout=stereo
6403 will instruct the source to accept planar 16bit signed stereo at 44100Hz.
6404 Since the sample format with name "s16p" corresponds to the number
6405 6 and the "stereo" channel layout corresponds to the value 0x3, this is
6408 abuffer=sample_rate=44100:sample_fmt=6:channel_layout=0x3
6413 Generate an audio signal specified by an expression.
6415 This source accepts in input one or more expressions (one for each
6416 channel), which are evaluated and used to generate a corresponding
6419 This source accepts the following options:
6423 Set the '|'-separated expressions list for each separate channel. In case the
6424 @option{channel_layout} option is not specified, the selected channel layout
6425 depends on the number of provided expressions. Otherwise the last
6426 specified expression is applied to the remaining output channels.
6428 @item channel_layout, c
6429 Set the channel layout. The number of channels in the specified layout
6430 must be equal to the number of specified expressions.
6433 Set the minimum duration of the sourced audio. See
6434 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6435 for the accepted syntax.
6436 Note that the resulting duration may be greater than the specified
6437 duration, as the generated audio is always cut at the end of a
6440 If not specified, or the expressed duration is negative, the audio is
6441 supposed to be generated forever.
6444 Set the number of samples per channel per each output frame,
6447 @item sample_rate, s
6448 Specify the sample rate, default to 44100.
6451 Each expression in @var{exprs} can contain the following constants:
6455 number of the evaluated sample, starting from 0
6458 time of the evaluated sample expressed in seconds, starting from 0
6465 @subsection Examples
6475 Generate a sin signal with frequency of 440 Hz, set sample rate to
6478 aevalsrc="sin(440*2*PI*t):s=8000"
6482 Generate a two channels signal, specify the channel layout (Front
6483 Center + Back Center) explicitly:
6485 aevalsrc="sin(420*2*PI*t)|cos(430*2*PI*t):c=FC|BC"
6489 Generate white noise:
6491 aevalsrc="-2+random(0)"
6495 Generate an amplitude modulated signal:
6497 aevalsrc="sin(10*2*PI*t)*sin(880*2*PI*t)"
6501 Generate 2.5 Hz binaural beats on a 360 Hz carrier:
6503 aevalsrc="0.1*sin(2*PI*(360-2.5/2)*t) | 0.1*sin(2*PI*(360+2.5/2)*t)"
6510 Generate a FIR coefficients using frequency sampling method.
6512 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6514 The filter accepts the following options:
6518 Set number of filter coefficents in output audio stream.
6519 Default value is 1025.
6522 Set frequency points from where magnitude and phase are set.
6523 This must be in non decreasing order, and first element must be 0, while last element
6524 must be 1. Elements are separated by white spaces.
6527 Set magnitude value for every frequency point set by @option{frequency}.
6528 Number of values must be same as number of frequency points.
6529 Values are separated by white spaces.
6532 Set phase value for every frequency point set by @option{frequency}.
6533 Number of values must be same as number of frequency points.
6534 Values are separated by white spaces.
6536 @item sample_rate, r
6537 Set sample rate, default is 44100.
6540 Set number of samples per each frame. Default is 1024.
6543 Set window function. Default is blackman.
6548 The null audio source, return unprocessed audio frames. It is mainly useful
6549 as a template and to be employed in analysis / debugging tools, or as
6550 the source for filters which ignore the input data (for example the sox
6553 This source accepts the following options:
6557 @item channel_layout, cl
6559 Specifies the channel layout, and can be either an integer or a string
6560 representing a channel layout. The default value of @var{channel_layout}
6563 Check the channel_layout_map definition in
6564 @file{libavutil/channel_layout.c} for the mapping between strings and
6565 channel layout values.
6567 @item sample_rate, r
6568 Specifies the sample rate, and defaults to 44100.
6571 Set the number of samples per requested frames.
6574 Set the duration of the sourced audio. See
6575 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
6576 for the accepted syntax.
6578 If not specified, or the expressed duration is negative, the audio is
6579 supposed to be generated forever.
6582 @subsection Examples
6586 Set the sample rate to 48000 Hz and the channel layout to AV_CH_LAYOUT_MONO.
6588 anullsrc=r=48000:cl=4
6592 Do the same operation with a more obvious syntax:
6594 anullsrc=r=48000:cl=mono
6598 All the parameters need to be explicitly defined.
6602 Synthesize a voice utterance using the libflite library.
6604 To enable compilation of this filter you need to configure FFmpeg with
6605 @code{--enable-libflite}.
6607 Note that versions of the flite library prior to 2.0 are not thread-safe.
6609 The filter accepts the following options:
6614 If set to 1, list the names of the available voices and exit
6615 immediately. Default value is 0.
6618 Set the maximum number of samples per frame. Default value is 512.
6621 Set the filename containing the text to speak.
6624 Set the text to speak.
6627 Set the voice to use for the speech synthesis. Default value is
6628 @code{kal}. See also the @var{list_voices} option.
6631 @subsection Examples
6635 Read from file @file{speech.txt}, and synthesize the text using the
6636 standard flite voice:
6638 flite=textfile=speech.txt
6642 Read the specified text selecting the @code{slt} voice:
6644 flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6648 Input text to ffmpeg:
6650 ffmpeg -f lavfi -i flite=text='So fare thee well, poor devil of a Sub-Sub, whose commentator I am':voice=slt
6654 Make @file{ffplay} speak the specified text, using @code{flite} and
6655 the @code{lavfi} device:
6657 ffplay -f lavfi flite=text='No more be grieved for which that thou hast done.'
6661 For more information about libflite, check:
6662 @url{http://www.festvox.org/flite/}
6666 Generate a noise audio signal.
6668 The filter accepts the following options:
6671 @item sample_rate, r
6672 Specify the sample rate. Default value is 48000 Hz.
6675 Specify the amplitude (0.0 - 1.0) of the generated audio stream. Default value
6679 Specify the duration of the generated audio stream. Not specifying this option
6680 results in noise with an infinite length.
6682 @item color, colour, c
6683 Specify the color of noise. Available noise colors are white, pink, brown,
6684 blue, violet and velvet. Default color is white.
6687 Specify a value used to seed the PRNG.
6690 Set the number of samples per each output frame, default is 1024.
6693 @subsection Examples
6698 Generate 60 seconds of pink noise, with a 44.1 kHz sampling rate and an amplitude of 0.5:
6700 anoisesrc=d=60:c=pink:r=44100:a=0.5
6706 Generate odd-tap Hilbert transform FIR coefficients.
6708 The resulting stream can be used with @ref{afir} filter for phase-shifting
6709 the signal by 90 degrees.
6711 This is used in many matrix coding schemes and for analytic signal generation.
6712 The process is often written as a multiplication by i (or j), the imaginary unit.
6714 The filter accepts the following options:
6718 @item sample_rate, s
6719 Set sample rate, default is 44100.
6722 Set length of FIR filter, default is 22051.
6725 Set number of samples per each frame.
6728 Set window function to be used when generating FIR coefficients.
6733 Generate a sinc kaiser-windowed low-pass, high-pass, band-pass, or band-reject FIR coefficients.
6735 The resulting stream can be used with @ref{afir} filter for filtering the audio signal.
6737 The filter accepts the following options:
6740 @item sample_rate, r
6741 Set sample rate, default is 44100.
6744 Set number of samples per each frame. Default is 1024.
6747 Set high-pass frequency. Default is 0.
6750 Set low-pass frequency. Default is 0.
6751 If high-pass frequency is lower than low-pass frequency and low-pass frequency
6752 is higher than 0 then filter will create band-pass filter coefficients,
6753 otherwise band-reject filter coefficients.
6756 Set filter phase response. Default is 50. Allowed range is from 0 to 100.
6759 Set Kaiser window beta.
6762 Set stop-band attenuation. Default is 120dB, allowed range is from 40 to 180 dB.
6765 Enable rounding, by default is disabled.
6768 Set number of taps for high-pass filter.
6771 Set number of taps for low-pass filter.
6776 Generate an audio signal made of a sine wave with amplitude 1/8.
6778 The audio signal is bit-exact.
6780 The filter accepts the following options:
6785 Set the carrier frequency. Default is 440 Hz.
6787 @item beep_factor, b
6788 Enable a periodic beep every second with frequency @var{beep_factor} times
6789 the carrier frequency. Default is 0, meaning the beep is disabled.
6791 @item sample_rate, r
6792 Specify the sample rate, default is 44100.
6795 Specify the duration of the generated audio stream.
6797 @item samples_per_frame
6798 Set the number of samples per output frame.
6800 The expression can contain the following constants:
6804 The (sequential) number of the output audio frame, starting from 0.
6807 The PTS (Presentation TimeStamp) of the output audio frame,
6808 expressed in @var{TB} units.
6811 The PTS of the output audio frame, expressed in seconds.
6814 The timebase of the output audio frames.
6817 Default is @code{1024}.
6820 @subsection Examples
6825 Generate a simple 440 Hz sine wave:
6831 Generate a 220 Hz sine wave with a 880 Hz beep each second, for 5 seconds:
6835 sine=frequency=220:beep_factor=4:duration=5
6839 Generate a 1 kHz sine wave following @code{1602,1601,1602,1601,1602} NTSC
6842 sine=1000:samples_per_frame='st(0,mod(n,5)); 1602-not(not(eq(ld(0),1)+eq(ld(0),3)))'
6846 @c man end AUDIO SOURCES
6848 @chapter Audio Sinks
6849 @c man begin AUDIO SINKS
6851 Below is a description of the currently available audio sinks.
6853 @section abuffersink
6855 Buffer audio frames, and make them available to the end of filter chain.
6857 This sink is mainly intended for programmatic use, in particular
6858 through the interface defined in @file{libavfilter/buffersink.h}
6859 or the options system.
6861 It accepts a pointer to an AVABufferSinkContext structure, which
6862 defines the incoming buffers' formats, to be passed as the opaque
6863 parameter to @code{avfilter_init_filter} for initialization.
6866 Null audio sink; do absolutely nothing with the input audio. It is
6867 mainly useful as a template and for use in analysis / debugging
6870 @c man end AUDIO SINKS
6872 @chapter Video Filters
6873 @c man begin VIDEO FILTERS
6875 When you configure your FFmpeg build, you can disable any of the
6876 existing filters using @code{--disable-filters}.
6877 The configure output will show the video filters included in your
6880 Below is a description of the currently available video filters.
6884 Mark a region of interest in a video frame.
6886 The frame data is passed through unchanged, but metadata is attached
6887 to the frame indicating regions of interest which can affect the
6888 behaviour of later encoding. Multiple regions can be marked by
6889 applying the filter multiple times.
6893 Region distance in pixels from the left edge of the frame.
6895 Region distance in pixels from the top edge of the frame.
6897 Region width in pixels.
6899 Region height in pixels.
6901 The parameters @var{x}, @var{y}, @var{w} and @var{h} are expressions,
6902 and may contain the following variables:
6905 Width of the input frame.
6907 Height of the input frame.
6911 Quantisation offset to apply within the region.
6913 This must be a real value in the range -1 to +1. A value of zero
6914 indicates no quality change. A negative value asks for better quality
6915 (less quantisation), while a positive value asks for worse quality
6916 (greater quantisation).
6918 The range is calibrated so that the extreme values indicate the
6919 largest possible offset - if the rest of the frame is encoded with the
6920 worst possible quality, an offset of -1 indicates that this region
6921 should be encoded with the best possible quality anyway. Intermediate
6922 values are then interpolated in some codec-dependent way.
6924 For example, in 10-bit H.264 the quantisation parameter varies between
6925 -12 and 51. A typical qoffset value of -1/10 therefore indicates that
6926 this region should be encoded with a QP around one-tenth of the full
6927 range better than the rest of the frame. So, if most of the frame
6928 were to be encoded with a QP of around 30, this region would get a QP
6929 of around 24 (an offset of approximately -1/10 * (51 - -12) = -6.3).
6930 An extreme value of -1 would indicate that this region should be
6931 encoded with the best possible quality regardless of the treatment of
6932 the rest of the frame - that is, should be encoded at a QP of -12.
6934 If set to true, remove any existing regions of interest marked on the
6935 frame before adding the new one.
6938 @subsection Examples
6942 Mark the centre quarter of the frame as interesting.
6944 addroi=iw/4:ih/4:iw/2:ih/2:-1/10
6947 Mark the 100-pixel-wide region on the left edge of the frame as very
6948 uninteresting (to be encoded at much lower quality than the rest of
6951 addroi=0:0:100:ih:+1/5
6955 @section alphaextract
6957 Extract the alpha component from the input as a grayscale video. This
6958 is especially useful with the @var{alphamerge} filter.
6962 Add or replace the alpha component of the primary input with the
6963 grayscale value of a second input. This is intended for use with
6964 @var{alphaextract} to allow the transmission or storage of frame
6965 sequences that have alpha in a format that doesn't support an alpha
6968 For example, to reconstruct full frames from a normal YUV-encoded video
6969 and a separate video created with @var{alphaextract}, you might use:
6971 movie=in_alpha.mkv [alpha]; [in][alpha] alphamerge [out]
6976 Amplify differences between current pixel and pixels of adjacent frames in
6977 same pixel location.
6979 This filter accepts the following options:
6983 Set frame radius. Default is 2. Allowed range is from 1 to 63.
6984 For example radius of 3 will instruct filter to calculate average of 7 frames.
6987 Set factor to amplify difference. Default is 2. Allowed range is from 0 to 65535.
6990 Set threshold for difference amplification. Any difference greater or equal to
6991 this value will not alter source pixel. Default is 10.
6992 Allowed range is from 0 to 65535.
6995 Set tolerance for difference amplification. Any difference lower to
6996 this value will not alter source pixel. Default is 0.
6997 Allowed range is from 0 to 65535.
7000 Set lower limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7001 This option controls maximum possible value that will decrease source pixel value.
7004 Set high limit for changing source pixel. Default is 65535. Allowed range is from 0 to 65535.
7005 This option controls maximum possible value that will increase source pixel value.
7008 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
7011 @subsection Commands
7013 This filter supports the following @ref{commands} that corresponds to option of same name:
7025 Same as the @ref{subtitles} filter, except that it doesn't require libavcodec
7026 and libavformat to work. On the other hand, it is limited to ASS (Advanced
7027 Substation Alpha) subtitles files.
7029 This filter accepts the following option in addition to the common options from
7030 the @ref{subtitles} filter:
7034 Set the shaping engine
7036 Available values are:
7039 The default libass shaping engine, which is the best available.
7041 Fast, font-agnostic shaper that can do only substitutions
7043 Slower shaper using OpenType for substitutions and positioning
7046 The default is @code{auto}.
7050 Apply an Adaptive Temporal Averaging Denoiser to the video input.
7052 The filter accepts the following options:
7056 Set threshold A for 1st plane. Default is 0.02.
7057 Valid range is 0 to 0.3.
7060 Set threshold B for 1st plane. Default is 0.04.
7061 Valid range is 0 to 5.
7064 Set threshold A for 2nd plane. Default is 0.02.
7065 Valid range is 0 to 0.3.
7068 Set threshold B for 2nd plane. Default is 0.04.
7069 Valid range is 0 to 5.
7072 Set threshold A for 3rd plane. Default is 0.02.
7073 Valid range is 0 to 0.3.
7076 Set threshold B for 3rd plane. Default is 0.04.
7077 Valid range is 0 to 5.
7079 Threshold A is designed to react on abrupt changes in the input signal and
7080 threshold B is designed to react on continuous changes in the input signal.
7083 Set number of frames filter will use for averaging. Default is 9. Must be odd
7084 number in range [5, 129].
7087 Set what planes of frame filter will use for averaging. Default is all.
7090 Set what variant of algorithm filter will use for averaging. Default is @code{p} parallel.
7091 Alternatively can be set to @code{s} serial.
7093 Parallel can be faster then serial, while other way around is never true.
7094 Parallel will abort early on first change being greater then thresholds, while serial
7095 will continue processing other side of frames if they are equal or below thresholds.
7098 @subsection Commands
7099 This filter supports same @ref{commands} as options except option @code{s}.
7100 The command accepts the same syntax of the corresponding option.
7104 Apply average blur filter.
7106 The filter accepts the following options:
7110 Set horizontal radius size.
7113 Set which planes to filter. By default all planes are filtered.
7116 Set vertical radius size, if zero it will be same as @code{sizeX}.
7117 Default is @code{0}.
7120 @subsection Commands
7121 This filter supports same commands as options.
7122 The command accepts the same syntax of the corresponding option.
7124 If the specified expression is not valid, it is kept at its current
7129 Compute the bounding box for the non-black pixels in the input frame
7132 This filter computes the bounding box containing all the pixels with a
7133 luminance value greater than the minimum allowed value.
7134 The parameters describing the bounding box are printed on the filter
7137 The filter accepts the following option:
7141 Set the minimal luminance value. Default is @code{16}.
7145 Apply bilateral filter, spatial smoothing while preserving edges.
7147 The filter accepts the following options:
7150 Set sigma of gaussian function to calculate spatial weight.
7151 Allowed range is 0 to 512. Default is 0.1.
7154 Set sigma of gaussian function to calculate range weight.
7155 Allowed range is 0 to 1. Default is 0.1.
7158 Set planes to filter. Default is first only.
7161 @subsection Commands
7163 This filter supports the all above options as @ref{commands}.
7165 @section bitplanenoise
7167 Show and measure bit plane noise.
7169 The filter accepts the following options:
7173 Set which plane to analyze. Default is @code{1}.
7176 Filter out noisy pixels from @code{bitplane} set above.
7177 Default is disabled.
7180 @section blackdetect
7182 Detect video intervals that are (almost) completely black. Can be
7183 useful to detect chapter transitions, commercials, or invalid
7186 The filter outputs its detection analysis to both the log as well as
7187 frame metadata. If a black segment of at least the specified minimum
7188 duration is found, a line with the start and end timestamps as well
7189 as duration is printed to the log with level @code{info}. In addition,
7190 a log line with level @code{debug} is printed per frame showing the
7191 black amount detected for that frame.
7193 The filter also attaches metadata to the first frame of a black
7194 segment with key @code{lavfi.black_start} and to the first frame
7195 after the black segment ends with key @code{lavfi.black_end}. The
7196 value is the frame's timestamp. This metadata is added regardless
7197 of the minimum duration specified.
7199 The filter accepts the following options:
7202 @item black_min_duration, d
7203 Set the minimum detected black duration expressed in seconds. It must
7204 be a non-negative floating point number.
7206 Default value is 2.0.
7208 @item picture_black_ratio_th, pic_th
7209 Set the threshold for considering a picture "black".
7210 Express the minimum value for the ratio:
7212 @var{nb_black_pixels} / @var{nb_pixels}
7215 for which a picture is considered black.
7216 Default value is 0.98.
7218 @item pixel_black_th, pix_th
7219 Set the threshold for considering a pixel "black".
7221 The threshold expresses the maximum pixel luminance value for which a
7222 pixel is considered "black". The provided value is scaled according to
7223 the following equation:
7225 @var{absolute_threshold} = @var{luminance_minimum_value} + @var{pixel_black_th} * @var{luminance_range_size}
7228 @var{luminance_range_size} and @var{luminance_minimum_value} depend on
7229 the input video format, the range is [0-255] for YUV full-range
7230 formats and [16-235] for YUV non full-range formats.
7232 Default value is 0.10.
7235 The following example sets the maximum pixel threshold to the minimum
7236 value, and detects only black intervals of 2 or more seconds:
7238 blackdetect=d=2:pix_th=0.00
7243 Detect frames that are (almost) completely black. Can be useful to
7244 detect chapter transitions or commercials. Output lines consist of
7245 the frame number of the detected frame, the percentage of blackness,
7246 the position in the file if known or -1 and the timestamp in seconds.
7248 In order to display the output lines, you need to set the loglevel at
7249 least to the AV_LOG_INFO value.
7251 This filter exports frame metadata @code{lavfi.blackframe.pblack}.
7252 The value represents the percentage of pixels in the picture that
7253 are below the threshold value.
7255 It accepts the following parameters:
7260 The percentage of the pixels that have to be below the threshold; it defaults to
7263 @item threshold, thresh
7264 The threshold below which a pixel value is considered black; it defaults to
7272 Blend two video frames into each other.
7274 The @code{blend} filter takes two input streams and outputs one
7275 stream, the first input is the "top" layer and second input is
7276 "bottom" layer. By default, the output terminates when the longest input terminates.
7278 The @code{tblend} (time blend) filter takes two consecutive frames
7279 from one single stream, and outputs the result obtained by blending
7280 the new frame on top of the old frame.
7282 A description of the accepted options follows.
7290 Set blend mode for specific pixel component or all pixel components in case
7291 of @var{all_mode}. Default value is @code{normal}.
7293 Available values for component modes are:
7335 Set blend opacity for specific pixel component or all pixel components in case
7336 of @var{all_opacity}. Only used in combination with pixel component blend modes.
7343 Set blend expression for specific pixel component or all pixel components in case
7344 of @var{all_expr}. Note that related mode options will be ignored if those are set.
7346 The expressions can use the following variables:
7350 The sequential number of the filtered frame, starting from @code{0}.
7354 the coordinates of the current sample
7358 the width and height of currently filtered plane
7362 Width and height scale for the plane being filtered. It is the
7363 ratio between the dimensions of the current plane to the luma plane,
7364 e.g. for a @code{yuv420p} frame, the values are @code{1,1} for
7365 the luma plane and @code{0.5,0.5} for the chroma planes.
7368 Time of the current frame, expressed in seconds.
7371 Value of pixel component at current location for first video frame (top layer).
7374 Value of pixel component at current location for second video frame (bottom layer).
7378 The @code{blend} filter also supports the @ref{framesync} options.
7380 @subsection Examples
7384 Apply transition from bottom layer to top layer in first 10 seconds:
7386 blend=all_expr='A*(if(gte(T,10),1,T/10))+B*(1-(if(gte(T,10),1,T/10)))'
7390 Apply linear horizontal transition from top layer to bottom layer:
7392 blend=all_expr='A*(X/W)+B*(1-X/W)'
7396 Apply 1x1 checkerboard effect:
7398 blend=all_expr='if(eq(mod(X,2),mod(Y,2)),A,B)'
7402 Apply uncover left effect:
7404 blend=all_expr='if(gte(N*SW+X,W),A,B)'
7408 Apply uncover down effect:
7410 blend=all_expr='if(gte(Y-N*SH,0),A,B)'
7414 Apply uncover up-left effect:
7416 blend=all_expr='if(gte(T*SH*40+Y,H)*gte((T*40*SW+X)*W/H,W),A,B)'
7420 Split diagonally video and shows top and bottom layer on each side:
7422 blend=all_expr='if(gt(X,Y*(W/H)),A,B)'
7426 Display differences between the current and the previous frame:
7428 tblend=all_mode=grainextract
7434 Denoise frames using Block-Matching 3D algorithm.
7436 The filter accepts the following options.
7440 Set denoising strength. Default value is 1.
7441 Allowed range is from 0 to 999.9.
7442 The denoising algorithm is very sensitive to sigma, so adjust it
7443 according to the source.
7446 Set local patch size. This sets dimensions in 2D.
7449 Set sliding step for processing blocks. Default value is 4.
7450 Allowed range is from 1 to 64.
7451 Smaller values allows processing more reference blocks and is slower.
7454 Set maximal number of similar blocks for 3rd dimension. Default value is 1.
7455 When set to 1, no block matching is done. Larger values allows more blocks
7457 Allowed range is from 1 to 256.
7460 Set radius for search block matching. Default is 9.
7461 Allowed range is from 1 to INT32_MAX.
7464 Set step between two search locations for block matching. Default is 1.
7465 Allowed range is from 1 to 64. Smaller is slower.
7468 Set threshold of mean square error for block matching. Valid range is 0 to
7472 Set thresholding parameter for hard thresholding in 3D transformed domain.
7473 Larger values results in stronger hard-thresholding filtering in frequency
7477 Set filtering estimation mode. Can be @code{basic} or @code{final}.
7478 Default is @code{basic}.
7481 If enabled, filter will use 2nd stream for block matching.
7482 Default is disabled for @code{basic} value of @var{estim} option,
7483 and always enabled if value of @var{estim} is @code{final}.
7486 Set planes to filter. Default is all available except alpha.
7489 @subsection Examples
7493 Basic filtering with bm3d:
7495 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic
7499 Same as above, but filtering only luma:
7501 bm3d=sigma=3:block=4:bstep=2:group=1:estim=basic:planes=1
7505 Same as above, but with both estimation modes:
7507 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
7511 Same as above, but prefilter with @ref{nlmeans} filter instead:
7513 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
7519 Apply a boxblur algorithm to the input video.
7521 It accepts the following parameters:
7525 @item luma_radius, lr
7526 @item luma_power, lp
7527 @item chroma_radius, cr
7528 @item chroma_power, cp
7529 @item alpha_radius, ar
7530 @item alpha_power, ap
7534 A description of the accepted options follows.
7537 @item luma_radius, lr
7538 @item chroma_radius, cr
7539 @item alpha_radius, ar
7540 Set an expression for the box radius in pixels used for blurring the
7541 corresponding input plane.
7543 The radius value must be a non-negative number, and must not be
7544 greater than the value of the expression @code{min(w,h)/2} for the
7545 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
7548 Default value for @option{luma_radius} is "2". If not specified,
7549 @option{chroma_radius} and @option{alpha_radius} default to the
7550 corresponding value set for @option{luma_radius}.
7552 The expressions can contain the following constants:
7556 The input width and height in pixels.
7560 The input chroma image width and height in pixels.
7564 The horizontal and vertical chroma subsample values. For example, for the
7565 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
7568 @item luma_power, lp
7569 @item chroma_power, cp
7570 @item alpha_power, ap
7571 Specify how many times the boxblur filter is applied to the
7572 corresponding plane.
7574 Default value for @option{luma_power} is 2. If not specified,
7575 @option{chroma_power} and @option{alpha_power} default to the
7576 corresponding value set for @option{luma_power}.
7578 A value of 0 will disable the effect.
7581 @subsection Examples
7585 Apply a boxblur filter with the luma, chroma, and alpha radii
7588 boxblur=luma_radius=2:luma_power=1
7593 Set the luma radius to 2, and alpha and chroma radius to 0:
7595 boxblur=2:1:cr=0:ar=0
7599 Set the luma and chroma radii to a fraction of the video dimension:
7601 boxblur=luma_radius=min(h\,w)/10:luma_power=1:chroma_radius=min(cw\,ch)/10:chroma_power=1
7607 Deinterlace the input video ("bwdif" stands for "Bob Weaver
7608 Deinterlacing Filter").
7610 Motion adaptive deinterlacing based on yadif with the use of w3fdif and cubic
7611 interpolation algorithms.
7612 It accepts the following parameters:
7616 The interlacing mode to adopt. It accepts one of the following values:
7620 Output one frame for each frame.
7622 Output one frame for each field.
7625 The default value is @code{send_field}.
7628 The picture field parity assumed for the input interlaced video. It accepts one
7629 of the following values:
7633 Assume the top field is first.
7635 Assume the bottom field is first.
7637 Enable automatic detection of field parity.
7640 The default value is @code{auto}.
7641 If the interlacing is unknown or the decoder does not export this information,
7642 top field first will be assumed.
7645 Specify which frames to deinterlace. Accepts one of the following
7650 Deinterlace all frames.
7652 Only deinterlace frames marked as interlaced.
7655 The default value is @code{all}.
7660 Apply Contrast Adaptive Sharpen filter to video stream.
7662 The filter accepts the following options:
7666 Set the sharpening strength. Default value is 0.
7669 Set planes to filter. Default value is to filter all
7670 planes except alpha plane.
7673 @subsection Commands
7674 This filter supports same @ref{commands} as options.
7677 Remove all color information for all colors except for certain one.
7679 The filter accepts the following options:
7683 The color which will not be replaced with neutral chroma.
7686 Similarity percentage with the above color.
7687 0.01 matches only the exact key color, while 1.0 matches everything.
7691 0.0 makes pixels either fully gray, or not gray at all.
7692 Higher values result in more preserved color.
7695 Signals that the color passed is already in YUV instead of RGB.
7697 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7698 This can be used to pass exact YUV values as hexadecimal numbers.
7701 @subsection Commands
7702 This filter supports same @ref{commands} as options.
7703 The command accepts the same syntax of the corresponding option.
7705 If the specified expression is not valid, it is kept at its current
7709 YUV colorspace color/chroma keying.
7711 The filter accepts the following options:
7715 The color which will be replaced with transparency.
7718 Similarity percentage with the key color.
7720 0.01 matches only the exact key color, while 1.0 matches everything.
7725 0.0 makes pixels either fully transparent, or not transparent at all.
7727 Higher values result in semi-transparent pixels, with a higher transparency
7728 the more similar the pixels color is to the key color.
7731 Signals that the color passed is already in YUV instead of RGB.
7733 Literal colors like "green" or "red" don't make sense with this enabled anymore.
7734 This can be used to pass exact YUV values as hexadecimal numbers.
7737 @subsection Commands
7738 This filter supports same @ref{commands} as options.
7739 The command accepts the same syntax of the corresponding option.
7741 If the specified expression is not valid, it is kept at its current
7744 @subsection Examples
7748 Make every green pixel in the input image transparent:
7750 ffmpeg -i input.png -vf chromakey=green out.png
7754 Overlay a greenscreen-video on top of a static black background.
7756 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
7761 Reduce chrominance noise.
7763 The filter accepts the following options:
7767 Set threshold for averaging chrominance values.
7768 Sum of absolute difference of Y, U and V pixel components of current
7769 pixel and neighbour pixels lower than this threshold will be used in
7770 averaging. Luma component is left unchanged and is copied to output.
7771 Default value is 30. Allowed range is from 1 to 200.
7774 Set horizontal radius of rectangle used for averaging.
7775 Allowed range is from 1 to 100. Default value is 5.
7778 Set vertical radius of rectangle used for averaging.
7779 Allowed range is from 1 to 100. Default value is 5.
7782 Set horizontal step when averaging. Default value is 1.
7783 Allowed range is from 1 to 50.
7784 Mostly useful to speed-up filtering.
7787 Set vertical step when averaging. Default value is 1.
7788 Allowed range is from 1 to 50.
7789 Mostly useful to speed-up filtering.
7792 Set Y threshold for averaging chrominance values.
7793 Set finer control for max allowed difference between Y components
7794 of current pixel and neigbour pixels.
7795 Default value is 200. Allowed range is from 1 to 200.
7798 Set U threshold for averaging chrominance values.
7799 Set finer control for max allowed difference between U components
7800 of current pixel and neigbour pixels.
7801 Default value is 200. Allowed range is from 1 to 200.
7804 Set V threshold for averaging chrominance values.
7805 Set finer control for max allowed difference between V components
7806 of current pixel and neigbour pixels.
7807 Default value is 200. Allowed range is from 1 to 200.
7810 @subsection Commands
7811 This filter supports same @ref{commands} as options.
7812 The command accepts the same syntax of the corresponding option.
7814 @section chromashift
7815 Shift chroma pixels horizontally and/or vertically.
7817 The filter accepts the following options:
7820 Set amount to shift chroma-blue horizontally.
7822 Set amount to shift chroma-blue vertically.
7824 Set amount to shift chroma-red horizontally.
7826 Set amount to shift chroma-red vertically.
7828 Set edge mode, can be @var{smear}, default, or @var{warp}.
7831 @subsection Commands
7833 This filter supports the all above options as @ref{commands}.
7837 Display CIE color diagram with pixels overlaid onto it.
7839 The filter accepts the following options:
7854 @item uhdtv, rec2020
7868 Set what gamuts to draw.
7870 See @code{system} option for available values.
7873 Set ciescope size, by default set to 512.
7876 Set intensity used to map input pixel values to CIE diagram.
7879 Set contrast used to draw tongue colors that are out of active color system gamut.
7882 Correct gamma displayed on scope, by default enabled.
7885 Show white point on CIE diagram, by default disabled.
7888 Set input gamma. Used only with XYZ input color space.
7893 Visualize information exported by some codecs.
7895 Some codecs can export information through frames using side-data or other
7896 means. For example, some MPEG based codecs export motion vectors through the
7897 @var{export_mvs} flag in the codec @option{flags2} option.
7899 The filter accepts the following option:
7903 Set motion vectors to visualize.
7905 Available flags for @var{mv} are:
7909 forward predicted MVs of P-frames
7911 forward predicted MVs of B-frames
7913 backward predicted MVs of B-frames
7917 Display quantization parameters using the chroma planes.
7920 Set motion vectors type to visualize. Includes MVs from all frames unless specified by @var{frame_type} option.
7922 Available flags for @var{mv_type} are:
7926 forward predicted MVs
7928 backward predicted MVs
7931 @item frame_type, ft
7932 Set frame type to visualize motion vectors of.
7934 Available flags for @var{frame_type} are:
7938 intra-coded frames (I-frames)
7940 predicted frames (P-frames)
7942 bi-directionally predicted frames (B-frames)
7946 @subsection Examples
7950 Visualize forward predicted MVs of all frames using @command{ffplay}:
7952 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv_type=fp
7956 Visualize multi-directionals MVs of P and B-Frames using @command{ffplay}:
7958 ffplay -flags2 +export_mvs input.mp4 -vf codecview=mv=pf+bf+bb
7962 @section colorbalance
7963 Modify intensity of primary colors (red, green and blue) of input frames.
7965 The filter allows an input frame to be adjusted in the shadows, midtones or highlights
7966 regions for the red-cyan, green-magenta or blue-yellow balance.
7968 A positive adjustment value shifts the balance towards the primary color, a negative
7969 value towards the complementary color.
7971 The filter accepts the following options:
7977 Adjust red, green and blue shadows (darkest pixels).
7982 Adjust red, green and blue midtones (medium pixels).
7987 Adjust red, green and blue highlights (brightest pixels).
7989 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
7992 Preserve lightness when changing color balance. Default is disabled.
7995 @subsection Examples
7999 Add red color cast to shadows:
8005 @subsection Commands
8007 This filter supports the all above options as @ref{commands}.
8009 @section colorchannelmixer
8011 Adjust video input frames by re-mixing color channels.
8013 This filter modifies a color channel by adding the values associated to
8014 the other channels of the same pixels. For example if the value to
8015 modify is red, the output value will be:
8017 @var{red}=@var{red}*@var{rr} + @var{blue}*@var{rb} + @var{green}*@var{rg} + @var{alpha}*@var{ra}
8020 The filter accepts the following options:
8027 Adjust contribution of input red, green, blue and alpha channels for output red channel.
8028 Default is @code{1} for @var{rr}, and @code{0} for @var{rg}, @var{rb} and @var{ra}.
8034 Adjust contribution of input red, green, blue and alpha channels for output green channel.
8035 Default is @code{1} for @var{gg}, and @code{0} for @var{gr}, @var{gb} and @var{ga}.
8041 Adjust contribution of input red, green, blue and alpha channels for output blue channel.
8042 Default is @code{1} for @var{bb}, and @code{0} for @var{br}, @var{bg} and @var{ba}.
8048 Adjust contribution of input red, green, blue and alpha channels for output alpha channel.
8049 Default is @code{1} for @var{aa}, and @code{0} for @var{ar}, @var{ag} and @var{ab}.
8051 Allowed ranges for options are @code{[-2.0, 2.0]}.
8054 @subsection Examples
8058 Convert source to grayscale:
8060 colorchannelmixer=.3:.4:.3:0:.3:.4:.3:0:.3:.4:.3
8063 Simulate sepia tones:
8065 colorchannelmixer=.393:.769:.189:0:.349:.686:.168:0:.272:.534:.131
8069 @subsection Commands
8071 This filter supports the all above options as @ref{commands}.
8074 RGB colorspace color keying.
8076 The filter accepts the following options:
8080 The color which will be replaced with transparency.
8083 Similarity percentage with the key color.
8085 0.01 matches only the exact key color, while 1.0 matches everything.
8090 0.0 makes pixels either fully transparent, or not transparent at all.
8092 Higher values result in semi-transparent pixels, with a higher transparency
8093 the more similar the pixels color is to the key color.
8096 @subsection Examples
8100 Make every green pixel in the input image transparent:
8102 ffmpeg -i input.png -vf colorkey=green out.png
8106 Overlay a greenscreen-video on top of a static background image.
8108 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
8112 @subsection Commands
8113 This filter supports same @ref{commands} as options.
8114 The command accepts the same syntax of the corresponding option.
8116 If the specified expression is not valid, it is kept at its current
8120 Remove all color information for all RGB colors except for certain one.
8122 The filter accepts the following options:
8126 The color which will not be replaced with neutral gray.
8129 Similarity percentage with the above color.
8130 0.01 matches only the exact key color, while 1.0 matches everything.
8133 Blend percentage. 0.0 makes pixels fully gray.
8134 Higher values result in more preserved color.
8137 @subsection Commands
8138 This filter supports same @ref{commands} as options.
8139 The command accepts the same syntax of the corresponding option.
8141 If the specified expression is not valid, it is kept at its current
8144 @section colorlevels
8146 Adjust video input frames using levels.
8148 The filter accepts the following options:
8155 Adjust red, green, blue and alpha input black point.
8156 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{0}.
8162 Adjust red, green, blue and alpha input white point.
8163 Allowed ranges for options are @code{[-1.0, 1.0]}. Defaults are @code{1}.
8165 Input levels are used to lighten highlights (bright tones), darken shadows
8166 (dark tones), change the balance of bright and dark tones.
8172 Adjust red, green, blue and alpha output black point.
8173 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{0}.
8179 Adjust red, green, blue and alpha output white point.
8180 Allowed ranges for options are @code{[0, 1.0]}. Defaults are @code{1}.
8182 Output levels allows manual selection of a constrained output level range.
8185 @subsection Examples
8189 Make video output darker:
8191 colorlevels=rimin=0.058:gimin=0.058:bimin=0.058
8197 colorlevels=rimin=0.039:gimin=0.039:bimin=0.039:rimax=0.96:gimax=0.96:bimax=0.96
8201 Make video output lighter:
8203 colorlevels=rimax=0.902:gimax=0.902:bimax=0.902
8207 Increase brightness:
8209 colorlevels=romin=0.5:gomin=0.5:bomin=0.5
8213 @subsection Commands
8215 This filter supports the all above options as @ref{commands}.
8217 @section colormatrix
8219 Convert color matrix.
8221 The filter accepts the following options:
8226 Specify the source and destination color matrix. Both values must be
8229 The accepted values are:
8257 For example to convert from BT.601 to SMPTE-240M, use the command:
8259 colormatrix=bt601:smpte240m
8264 Convert colorspace, transfer characteristics or color primaries.
8265 Input video needs to have an even size.
8267 The filter accepts the following options:
8272 Specify all color properties at once.
8274 The accepted values are:
8304 Specify output colorspace.
8306 The accepted values are:
8315 BT.470BG or BT.601-6 625
8318 SMPTE-170M or BT.601-6 525
8327 BT.2020 with non-constant luminance
8333 Specify output transfer characteristics.
8335 The accepted values are:
8347 Constant gamma of 2.2
8350 Constant gamma of 2.8
8353 SMPTE-170M, BT.601-6 625 or BT.601-6 525
8371 BT.2020 for 10-bits content
8374 BT.2020 for 12-bits content
8380 Specify output color primaries.
8382 The accepted values are:
8391 BT.470BG or BT.601-6 625
8394 SMPTE-170M or BT.601-6 525
8418 Specify output color range.
8420 The accepted values are:
8423 TV (restricted) range
8426 MPEG (restricted) range
8437 Specify output color format.
8439 The accepted values are:
8442 YUV 4:2:0 planar 8-bits
8445 YUV 4:2:0 planar 10-bits
8448 YUV 4:2:0 planar 12-bits
8451 YUV 4:2:2 planar 8-bits
8454 YUV 4:2:2 planar 10-bits
8457 YUV 4:2:2 planar 12-bits
8460 YUV 4:4:4 planar 8-bits
8463 YUV 4:4:4 planar 10-bits
8466 YUV 4:4:4 planar 12-bits
8471 Do a fast conversion, which skips gamma/primary correction. This will take
8472 significantly less CPU, but will be mathematically incorrect. To get output
8473 compatible with that produced by the colormatrix filter, use fast=1.
8476 Specify dithering mode.
8478 The accepted values are:
8484 Floyd-Steinberg dithering
8488 Whitepoint adaptation mode.
8490 The accepted values are:
8493 Bradford whitepoint adaptation
8496 von Kries whitepoint adaptation
8499 identity whitepoint adaptation (i.e. no whitepoint adaptation)
8503 Override all input properties at once. Same accepted values as @ref{all}.
8506 Override input colorspace. Same accepted values as @ref{space}.
8509 Override input color primaries. Same accepted values as @ref{primaries}.
8512 Override input transfer characteristics. Same accepted values as @ref{trc}.
8515 Override input color range. Same accepted values as @ref{range}.
8519 The filter converts the transfer characteristics, color space and color
8520 primaries to the specified user values. The output value, if not specified,
8521 is set to a default value based on the "all" property. If that property is
8522 also not specified, the filter will log an error. The output color range and
8523 format default to the same value as the input color range and format. The
8524 input transfer characteristics, color space, color primaries and color range
8525 should be set on the input data. If any of these are missing, the filter will
8526 log an error and no conversion will take place.
8528 For example to convert the input to SMPTE-240M, use the command:
8530 colorspace=smpte240m
8533 @section convolution
8535 Apply convolution of 3x3, 5x5, 7x7 or horizontal/vertical up to 49 elements.
8537 The filter accepts the following options:
8544 Set matrix for each plane.
8545 Matrix is sequence of 9, 25 or 49 signed integers in @var{square} mode,
8546 and from 1 to 49 odd number of signed integers in @var{row} mode.
8552 Set multiplier for calculated value for each plane.
8553 If unset or 0, it will be sum of all matrix elements.
8559 Set bias for each plane. This value is added to the result of the multiplication.
8560 Useful for making the overall image brighter or darker. Default is 0.0.
8566 Set matrix mode for each plane. Can be @var{square}, @var{row} or @var{column}.
8567 Default is @var{square}.
8570 @subsection Commands
8572 This filter supports the all above options as @ref{commands}.
8574 @subsection Examples
8580 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"
8586 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"
8592 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"
8598 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"
8602 Apply laplacian edge detector which includes diagonals:
8604 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"
8610 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"
8616 Apply 2D convolution of video stream in frequency domain using second stream
8619 The filter accepts the following options:
8623 Set which planes to process.
8626 Set which impulse video frames will be processed, can be @var{first}
8627 or @var{all}. Default is @var{all}.
8630 The @code{convolve} filter also supports the @ref{framesync} options.
8634 Copy the input video source unchanged to the output. This is mainly useful for
8639 Video filtering on GPU using Apple's CoreImage API on OSX.
8641 Hardware acceleration is based on an OpenGL context. Usually, this means it is
8642 processed by video hardware. However, software-based OpenGL implementations
8643 exist which means there is no guarantee for hardware processing. It depends on
8646 There are many filters and image generators provided by Apple that come with a
8647 large variety of options. The filter has to be referenced by its name along
8650 The coreimage filter accepts the following options:
8653 List all available filters and generators along with all their respective
8654 options as well as possible minimum and maximum values along with the default
8661 Specify all filters by their respective name and options.
8662 Use @var{list_filters} to determine all valid filter names and options.
8663 Numerical options are specified by a float value and are automatically clamped
8664 to their respective value range. Vector and color options have to be specified
8665 by a list of space separated float values. Character escaping has to be done.
8666 A special option name @code{default} is available to use default options for a
8669 It is required to specify either @code{default} or at least one of the filter options.
8670 All omitted options are used with their default values.
8671 The syntax of the filter string is as follows:
8673 filter=<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...][#<NAME>@@<OPTION>=<VALUE>[@@<OPTION>=<VALUE>][@@...]][#...]
8677 Specify a rectangle where the output of the filter chain is copied into the
8678 input image. It is given by a list of space separated float values:
8680 output_rect=x\ y\ width\ height
8682 If not given, the output rectangle equals the dimensions of the input image.
8683 The output rectangle is automatically cropped at the borders of the input
8684 image. Negative values are valid for each component.
8686 output_rect=25\ 25\ 100\ 100
8690 Several filters can be chained for successive processing without GPU-HOST
8691 transfers allowing for fast processing of complex filter chains.
8692 Currently, only filters with zero (generators) or exactly one (filters) input
8693 image and one output image are supported. Also, transition filters are not yet
8696 Some filters generate output images with additional padding depending on the
8697 respective filter kernel. The padding is automatically removed to ensure the
8698 filter output has the same size as the input image.
8700 For image generators, the size of the output image is determined by the
8701 previous output image of the filter chain or the input image of the whole
8702 filterchain, respectively. The generators do not use the pixel information of
8703 this image to generate their output. However, the generated output is
8704 blended onto this image, resulting in partial or complete coverage of the
8707 The @ref{coreimagesrc} video source can be used for generating input images
8708 which are directly fed into the filter chain. By using it, providing input
8709 images by another video source or an input video is not required.
8711 @subsection Examples
8716 List all filters available:
8718 coreimage=list_filters=true
8722 Use the CIBoxBlur filter with default options to blur an image:
8724 coreimage=filter=CIBoxBlur@@default
8728 Use a filter chain with CISepiaTone at default values and CIVignetteEffect with
8729 its center at 100x100 and a radius of 50 pixels:
8731 coreimage=filter=CIBoxBlur@@default#CIVignetteEffect@@inputCenter=100\ 100@@inputRadius=50
8735 Use nullsrc and CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
8736 given as complete and escaped command-line for Apple's standard bash shell:
8738 ffmpeg -f lavfi -i nullsrc=s=100x100,coreimage=filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
8744 Cover a rectangular object
8746 It accepts the following options:
8750 Filepath of the optional cover image, needs to be in yuv420.
8755 It accepts the following values:
8758 cover it by the supplied image
8760 cover it by interpolating the surrounding pixels
8763 Default value is @var{blur}.
8766 @subsection Examples
8770 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
8772 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
8778 Crop the input video to given dimensions.
8780 It accepts the following parameters:
8784 The width of the output video. It defaults to @code{iw}.
8785 This expression is evaluated only once during the filter
8786 configuration, or when the @samp{w} or @samp{out_w} command is sent.
8789 The height of the output video. It defaults to @code{ih}.
8790 This expression is evaluated only once during the filter
8791 configuration, or when the @samp{h} or @samp{out_h} command is sent.
8794 The horizontal position, in the input video, of the left edge of the output
8795 video. It defaults to @code{(in_w-out_w)/2}.
8796 This expression is evaluated per-frame.
8799 The vertical position, in the input video, of the top edge of the output video.
8800 It defaults to @code{(in_h-out_h)/2}.
8801 This expression is evaluated per-frame.
8804 If set to 1 will force the output display aspect ratio
8805 to be the same of the input, by changing the output sample aspect
8806 ratio. It defaults to 0.
8809 Enable exact cropping. If enabled, subsampled videos will be cropped at exact
8810 width/height/x/y as specified and will not be rounded to nearest smaller value.
8814 The @var{out_w}, @var{out_h}, @var{x}, @var{y} parameters are
8815 expressions containing the following constants:
8820 The computed values for @var{x} and @var{y}. They are evaluated for
8825 The input width and height.
8829 These are the same as @var{in_w} and @var{in_h}.
8833 The output (cropped) width and height.
8837 These are the same as @var{out_w} and @var{out_h}.
8840 same as @var{iw} / @var{ih}
8843 input sample aspect ratio
8846 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
8850 horizontal and vertical chroma subsample values. For example for the
8851 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
8854 The number of the input frame, starting from 0.
8857 the position in the file of the input frame, NAN if unknown
8860 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
8864 The expression for @var{out_w} may depend on the value of @var{out_h},
8865 and the expression for @var{out_h} may depend on @var{out_w}, but they
8866 cannot depend on @var{x} and @var{y}, as @var{x} and @var{y} are
8867 evaluated after @var{out_w} and @var{out_h}.
8869 The @var{x} and @var{y} parameters specify the expressions for the
8870 position of the top-left corner of the output (non-cropped) area. They
8871 are evaluated for each frame. If the evaluated value is not valid, it
8872 is approximated to the nearest valid value.
8874 The expression for @var{x} may depend on @var{y}, and the expression
8875 for @var{y} may depend on @var{x}.
8877 @subsection Examples
8881 Crop area with size 100x100 at position (12,34).
8886 Using named options, the example above becomes:
8888 crop=w=100:h=100:x=12:y=34
8892 Crop the central input area with size 100x100:
8898 Crop the central input area with size 2/3 of the input video:
8900 crop=2/3*in_w:2/3*in_h
8904 Crop the input video central square:
8911 Delimit the rectangle with the top-left corner placed at position
8912 100:100 and the right-bottom corner corresponding to the right-bottom
8913 corner of the input image.
8915 crop=in_w-100:in_h-100:100:100
8919 Crop 10 pixels from the left and right borders, and 20 pixels from
8920 the top and bottom borders
8922 crop=in_w-2*10:in_h-2*20
8926 Keep only the bottom right quarter of the input image:
8928 crop=in_w/2:in_h/2:in_w/2:in_h/2
8932 Crop height for getting Greek harmony:
8934 crop=in_w:1/PHI*in_w
8938 Apply trembling effect:
8940 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)
8944 Apply erratic camera effect depending on timestamp:
8946 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)"
8950 Set x depending on the value of y:
8952 crop=in_w/2:in_h/2:y:10+10*sin(n/10)
8956 @subsection Commands
8958 This filter supports the following commands:
8964 Set width/height of the output video and the horizontal/vertical position
8966 The command accepts the same syntax of the corresponding option.
8968 If the specified expression is not valid, it is kept at its current
8974 Auto-detect the crop size.
8976 It calculates the necessary cropping parameters and prints the
8977 recommended parameters via the logging system. The detected dimensions
8978 correspond to the non-black area of the input video.
8980 It accepts the following parameters:
8985 Set higher black value threshold, which can be optionally specified
8986 from nothing (0) to everything (255 for 8-bit based formats). An intensity
8987 value greater to the set value is considered non-black. It defaults to 24.
8988 You can also specify a value between 0.0 and 1.0 which will be scaled depending
8989 on the bitdepth of the pixel format.
8992 The value which the width/height should be divisible by. It defaults to
8993 16. The offset is automatically adjusted to center the video. Use 2 to
8994 get only even dimensions (needed for 4:2:2 video). 16 is best when
8995 encoding to most video codecs.
8998 Set the number of initial frames for which evaluation is skipped.
8999 Default is 2. Range is 0 to INT_MAX.
9001 @item reset_count, reset
9002 Set the counter that determines after how many frames cropdetect will
9003 reset the previously detected largest video area and start over to
9004 detect the current optimal crop area. Default value is 0.
9006 This can be useful when channel logos distort the video area. 0
9007 indicates 'never reset', and returns the largest area encountered during
9014 Delay video filtering until a given wallclock timestamp. The filter first
9015 passes on @option{preroll} amount of frames, then it buffers at most
9016 @option{buffer} amount of frames and waits for the cue. After reaching the cue
9017 it forwards the buffered frames and also any subsequent frames coming in its
9020 The filter can be used synchronize the output of multiple ffmpeg processes for
9021 realtime output devices like decklink. By putting the delay in the filtering
9022 chain and pre-buffering frames the process can pass on data to output almost
9023 immediately after the target wallclock timestamp is reached.
9025 Perfect frame accuracy cannot be guaranteed, but the result is good enough for
9031 The cue timestamp expressed in a UNIX timestamp in microseconds. Default is 0.
9034 The duration of content to pass on as preroll expressed in seconds. Default is 0.
9037 The maximum duration of content to buffer before waiting for the cue expressed
9038 in seconds. Default is 0.
9045 Apply color adjustments using curves.
9047 This filter is similar to the Adobe Photoshop and GIMP curves tools. Each
9048 component (red, green and blue) has its values defined by @var{N} key points
9049 tied from each other using a smooth curve. The x-axis represents the pixel
9050 values from the input frame, and the y-axis the new pixel values to be set for
9053 By default, a component curve is defined by the two points @var{(0;0)} and
9054 @var{(1;1)}. This creates a straight line where each original pixel value is
9055 "adjusted" to its own value, which means no change to the image.
9057 The filter allows you to redefine these two points and add some more. A new
9058 curve (using a natural cubic spline interpolation) will be define to pass
9059 smoothly through all these new coordinates. The new defined points needs to be
9060 strictly increasing over the x-axis, and their @var{x} and @var{y} values must
9061 be in the @var{[0;1]} interval. If the computed curves happened to go outside
9062 the vector spaces, the values will be clipped accordingly.
9064 The filter accepts the following options:
9068 Select one of the available color presets. This option can be used in addition
9069 to the @option{r}, @option{g}, @option{b} parameters; in this case, the later
9070 options takes priority on the preset values.
9071 Available presets are:
9074 @item color_negative
9077 @item increase_contrast
9079 @item linear_contrast
9080 @item medium_contrast
9082 @item strong_contrast
9085 Default is @code{none}.
9087 Set the master key points. These points will define a second pass mapping. It
9088 is sometimes called a "luminance" or "value" mapping. It can be used with
9089 @option{r}, @option{g}, @option{b} or @option{all} since it acts like a
9090 post-processing LUT.
9092 Set the key points for the red component.
9094 Set the key points for the green component.
9096 Set the key points for the blue component.
9098 Set the key points for all components (not including master).
9099 Can be used in addition to the other key points component
9100 options. In this case, the unset component(s) will fallback on this
9101 @option{all} setting.
9103 Specify a Photoshop curves file (@code{.acv}) to import the settings from.
9105 Save Gnuplot script of the curves in specified file.
9108 To avoid some filtergraph syntax conflicts, each key points list need to be
9109 defined using the following syntax: @code{x0/y0 x1/y1 x2/y2 ...}.
9111 @subsection Examples
9115 Increase slightly the middle level of blue:
9117 curves=blue='0/0 0.5/0.58 1/1'
9123 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'
9125 Here we obtain the following coordinates for each components:
9128 @code{(0;0.11) (0.42;0.51) (1;0.95)}
9130 @code{(0;0) (0.50;0.48) (1;1)}
9132 @code{(0;0.22) (0.49;0.44) (1;0.80)}
9136 The previous example can also be achieved with the associated built-in preset:
9138 curves=preset=vintage
9148 Use a Photoshop preset and redefine the points of the green component:
9150 curves=psfile='MyCurvesPresets/purple.acv':green='0/0 0.45/0.53 1/1'
9154 Check out the curves of the @code{cross_process} profile using @command{ffmpeg}
9155 and @command{gnuplot}:
9157 ffmpeg -f lavfi -i color -vf curves=cross_process:plot=/tmp/curves.plt -frames:v 1 -f null -
9158 gnuplot -p /tmp/curves.plt
9164 Video data analysis filter.
9166 This filter shows hexadecimal pixel values of part of video.
9168 The filter accepts the following options:
9172 Set output video size.
9175 Set x offset from where to pick pixels.
9178 Set y offset from where to pick pixels.
9181 Set scope mode, can be one of the following:
9184 Draw hexadecimal pixel values with white color on black background.
9187 Draw hexadecimal pixel values with input video pixel color on black
9191 Draw hexadecimal pixel values on color background picked from input video,
9192 the text color is picked in such way so its always visible.
9196 Draw rows and columns numbers on left and top of video.
9199 Set background opacity.
9202 Set display number format. Can be @code{hex}, or @code{dec}. Default is @code{hex}.
9206 Apply Directional blur filter.
9208 The filter accepts the following options:
9212 Set angle of directional blur. Default is @code{45}.
9215 Set radius of directional blur. Default is @code{5}.
9218 Set which planes to filter. By default all planes are filtered.
9221 @subsection Commands
9222 This filter supports same @ref{commands} as options.
9223 The command accepts the same syntax of the corresponding option.
9225 If the specified expression is not valid, it is kept at its current
9230 Denoise frames using 2D DCT (frequency domain filtering).
9232 This filter is not designed for real time.
9234 The filter accepts the following options:
9238 Set the noise sigma constant.
9240 This @var{sigma} defines a hard threshold of @code{3 * sigma}; every DCT
9241 coefficient (absolute value) below this threshold with be dropped.
9243 If you need a more advanced filtering, see @option{expr}.
9245 Default is @code{0}.
9248 Set number overlapping pixels for each block. Since the filter can be slow, you
9249 may want to reduce this value, at the cost of a less effective filter and the
9250 risk of various artefacts.
9252 If the overlapping value doesn't permit processing the whole input width or
9253 height, a warning will be displayed and according borders won't be denoised.
9255 Default value is @var{blocksize}-1, which is the best possible setting.
9258 Set the coefficient factor expression.
9260 For each coefficient of a DCT block, this expression will be evaluated as a
9261 multiplier value for the coefficient.
9263 If this is option is set, the @option{sigma} option will be ignored.
9265 The absolute value of the coefficient can be accessed through the @var{c}
9269 Set the @var{blocksize} using the number of bits. @code{1<<@var{n}} defines the
9270 @var{blocksize}, which is the width and height of the processed blocks.
9272 The default value is @var{3} (8x8) and can be raised to @var{4} for a
9273 @var{blocksize} of 16x16. Note that changing this setting has huge consequences
9274 on the speed processing. Also, a larger block size does not necessarily means a
9278 @subsection Examples
9280 Apply a denoise with a @option{sigma} of @code{4.5}:
9285 The same operation can be achieved using the expression system:
9287 dctdnoiz=e='gte(c, 4.5*3)'
9290 Violent denoise using a block size of @code{16x16}:
9297 Remove banding artifacts from input video.
9298 It works by replacing banded pixels with average value of referenced pixels.
9300 The filter accepts the following options:
9307 Set banding detection threshold for each plane. Default is 0.02.
9308 Valid range is 0.00003 to 0.5.
9309 If difference between current pixel and reference pixel is less than threshold,
9310 it will be considered as banded.
9313 Banding detection range in pixels. Default is 16. If positive, random number
9314 in range 0 to set value will be used. If negative, exact absolute value
9316 The range defines square of four pixels around current pixel.
9319 Set direction in radians from which four pixel will be compared. If positive,
9320 random direction from 0 to set direction will be picked. If negative, exact of
9321 absolute value will be picked. For example direction 0, -PI or -2*PI radians
9322 will pick only pixels on same row and -PI/2 will pick only pixels on same
9326 If enabled, current pixel is compared with average value of all four
9327 surrounding pixels. The default is enabled. If disabled current pixel is
9328 compared with all four surrounding pixels. The pixel is considered banded
9329 if only all four differences with surrounding pixels are less than threshold.
9332 If enabled, current pixel is changed if and only if all pixel components are banded,
9333 e.g. banding detection threshold is triggered for all color components.
9334 The default is disabled.
9339 Remove blocking artifacts from input video.
9341 The filter accepts the following options:
9345 Set filter type, can be @var{weak} or @var{strong}. Default is @var{strong}.
9346 This controls what kind of deblocking is applied.
9349 Set size of block, allowed range is from 4 to 512. Default is @var{8}.
9355 Set blocking detection thresholds. Allowed range is 0 to 1.
9356 Defaults are: @var{0.098} for @var{alpha} and @var{0.05} for the rest.
9357 Using higher threshold gives more deblocking strength.
9358 Setting @var{alpha} controls threshold detection at exact edge of block.
9359 Remaining options controls threshold detection near the edge. Each one for
9360 below/above or left/right. Setting any of those to @var{0} disables
9364 Set planes to filter. Default is to filter all available planes.
9367 @subsection Examples
9371 Deblock using weak filter and block size of 4 pixels.
9373 deblock=filter=weak:block=4
9377 Deblock using strong filter, block size of 4 pixels and custom thresholds for
9378 deblocking more edges.
9380 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05
9384 Similar as above, but filter only first plane.
9386 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=1
9390 Similar as above, but filter only second and third plane.
9392 deblock=filter=strong:block=4:alpha=0.12:beta=0.07:gamma=0.06:delta=0.05:planes=6
9399 Drop duplicated frames at regular intervals.
9401 The filter accepts the following options:
9405 Set the number of frames from which one will be dropped. Setting this to
9406 @var{N} means one frame in every batch of @var{N} frames will be dropped.
9407 Default is @code{5}.
9410 Set the threshold for duplicate detection. If the difference metric for a frame
9411 is less than or equal to this value, then it is declared as duplicate. Default
9415 Set scene change threshold. Default is @code{15}.
9419 Set the size of the x and y-axis blocks used during metric calculations.
9420 Larger blocks give better noise suppression, but also give worse detection of
9421 small movements. Must be a power of two. Default is @code{32}.
9424 Mark main input as a pre-processed input and activate clean source input
9425 stream. This allows the input to be pre-processed with various filters to help
9426 the metrics calculation while keeping the frame selection lossless. When set to
9427 @code{1}, the first stream is for the pre-processed input, and the second
9428 stream is the clean source from where the kept frames are chosen. Default is
9432 Set whether or not chroma is considered in the metric calculations. Default is
9438 Apply 2D deconvolution of video stream in frequency domain using second stream
9441 The filter accepts the following options:
9445 Set which planes to process.
9448 Set which impulse video frames will be processed, can be @var{first}
9449 or @var{all}. Default is @var{all}.
9452 Set noise when doing divisions. Default is @var{0.0000001}. Useful when width
9453 and height are not same and not power of 2 or if stream prior to convolving
9457 The @code{deconvolve} filter also supports the @ref{framesync} options.
9461 Reduce cross-luminance (dot-crawl) and cross-color (rainbows) from video.
9463 It accepts the following options:
9467 Set mode of operation. Can be combination of @var{dotcrawl} for cross-luminance reduction and/or
9468 @var{rainbows} for cross-color reduction.
9471 Set spatial luma threshold. Lower values increases reduction of cross-luminance.
9474 Set tolerance for temporal luma. Higher values increases reduction of cross-luminance.
9477 Set tolerance for chroma temporal variation. Higher values increases reduction of cross-color.
9480 Set temporal chroma threshold. Lower values increases reduction of cross-color.
9485 Apply deflate effect to the video.
9487 This filter replaces the pixel by the local(3x3) average by taking into account
9488 only values lower than the pixel.
9490 It accepts the following options:
9497 Limit the maximum change for each plane, default is 65535.
9498 If 0, plane will remain unchanged.
9501 @subsection Commands
9503 This filter supports the all above options as @ref{commands}.
9507 Remove temporal frame luminance variations.
9509 It accepts the following options:
9513 Set moving-average filter size in frames. Default is 5. Allowed range is 2 - 129.
9516 Set averaging mode to smooth temporal luminance variations.
9518 Available values are:
9543 Do not actually modify frame. Useful when one only wants metadata.
9548 Remove judder produced by partially interlaced telecined content.
9550 Judder can be introduced, for instance, by @ref{pullup} filter. If the original
9551 source was partially telecined content then the output of @code{pullup,dejudder}
9552 will have a variable frame rate. May change the recorded frame rate of the
9553 container. Aside from that change, this filter will not affect constant frame
9556 The option available in this filter is:
9560 Specify the length of the window over which the judder repeats.
9562 Accepts any integer greater than 1. Useful values are:
9566 If the original was telecined from 24 to 30 fps (Film to NTSC).
9569 If the original was telecined from 25 to 30 fps (PAL to NTSC).
9572 If a mixture of the two.
9575 The default is @samp{4}.
9580 Suppress a TV station logo by a simple interpolation of the surrounding
9581 pixels. Just set a rectangle covering the logo and watch it disappear
9582 (and sometimes something even uglier appear - your mileage may vary).
9584 It accepts the following parameters:
9589 Specify the top left corner coordinates of the logo. They must be
9594 Specify the width and height of the logo to clear. They must be
9598 Specify the thickness of the fuzzy edge of the rectangle (added to
9599 @var{w} and @var{h}). The default value is 1. This option is
9600 deprecated, setting higher values should no longer be necessary and
9604 When set to 1, a green rectangle is drawn on the screen to simplify
9605 finding the right @var{x}, @var{y}, @var{w}, and @var{h} parameters.
9606 The default value is 0.
9608 The rectangle is drawn on the outermost pixels which will be (partly)
9609 replaced with interpolated values. The values of the next pixels
9610 immediately outside this rectangle in each direction will be used to
9611 compute the interpolated pixel values inside the rectangle.
9615 @subsection Examples
9619 Set a rectangle covering the area with top left corner coordinates 0,0
9620 and size 100x77, and a band of size 10:
9622 delogo=x=0:y=0:w=100:h=77:band=10
9630 Remove the rain in the input image/video by applying the derain methods based on
9631 convolutional neural networks. Supported models:
9635 Recurrent Squeeze-and-Excitation Context Aggregation Net (RESCAN).
9636 See @url{http://openaccess.thecvf.com/content_ECCV_2018/papers/Xia_Li_Recurrent_Squeeze-and-Excitation_Context_ECCV_2018_paper.pdf}.
9639 Training as well as model generation scripts are provided in
9640 the repository at @url{https://github.com/XueweiMeng/derain_filter.git}.
9642 Native model files (.model) can be generated from TensorFlow model
9643 files (.pb) by using tools/python/convert.py
9645 The filter accepts the following options:
9649 Specify which filter to use. This option accepts the following values:
9653 Derain filter. To conduct derain filter, you need to use a derain model.
9656 Dehaze filter. To conduct dehaze filter, you need to use a dehaze model.
9658 Default value is @samp{derain}.
9661 Specify which DNN backend to use for model loading and execution. This option accepts
9662 the following values:
9666 Native implementation of DNN loading and execution.
9669 TensorFlow backend. To enable this backend you
9670 need to install the TensorFlow for C library (see
9671 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9672 @code{--enable-libtensorflow}
9674 Default value is @samp{native}.
9677 Set path to model file specifying network architecture and its parameters.
9678 Note that different backends use different file formats. TensorFlow and native
9679 backend can load files for only its format.
9682 It can also be finished with @ref{dnn_processing} filter.
9686 Attempt to fix small changes in horizontal and/or vertical shift. This
9687 filter helps remove camera shake from hand-holding a camera, bumping a
9688 tripod, moving on a vehicle, etc.
9690 The filter accepts the following options:
9698 Specify a rectangular area where to limit the search for motion
9700 If desired the search for motion vectors can be limited to a
9701 rectangular area of the frame defined by its top left corner, width
9702 and height. These parameters have the same meaning as the drawbox
9703 filter which can be used to visualise the position of the bounding
9706 This is useful when simultaneous movement of subjects within the frame
9707 might be confused for camera motion by the motion vector search.
9709 If any or all of @var{x}, @var{y}, @var{w} and @var{h} are set to -1
9710 then the full frame is used. This allows later options to be set
9711 without specifying the bounding box for the motion vector search.
9713 Default - search the whole frame.
9717 Specify the maximum extent of movement in x and y directions in the
9718 range 0-64 pixels. Default 16.
9721 Specify how to generate pixels to fill blanks at the edge of the
9722 frame. Available values are:
9725 Fill zeroes at blank locations
9727 Original image at blank locations
9729 Extruded edge value at blank locations
9731 Mirrored edge at blank locations
9733 Default value is @samp{mirror}.
9736 Specify the blocksize to use for motion search. Range 4-128 pixels,
9740 Specify the contrast threshold for blocks. Only blocks with more than
9741 the specified contrast (difference between darkest and lightest
9742 pixels) will be considered. Range 1-255, default 125.
9745 Specify the search strategy. Available values are:
9748 Set exhaustive search
9750 Set less exhaustive search.
9752 Default value is @samp{exhaustive}.
9755 If set then a detailed log of the motion search is written to the
9762 Remove unwanted contamination of foreground colors, caused by reflected color of
9763 greenscreen or bluescreen.
9765 This filter accepts the following options:
9769 Set what type of despill to use.
9772 Set how spillmap will be generated.
9775 Set how much to get rid of still remaining spill.
9778 Controls amount of red in spill area.
9781 Controls amount of green in spill area.
9782 Should be -1 for greenscreen.
9785 Controls amount of blue in spill area.
9786 Should be -1 for bluescreen.
9789 Controls brightness of spill area, preserving colors.
9792 Modify alpha from generated spillmap.
9795 @subsection Commands
9797 This filter supports the all above options as @ref{commands}.
9801 Apply an exact inverse of the telecine operation. It requires a predefined
9802 pattern specified using the pattern option which must be the same as that passed
9803 to the telecine filter.
9805 This filter accepts the following options:
9814 The default value is @code{top}.
9818 A string of numbers representing the pulldown pattern you wish to apply.
9819 The default value is @code{23}.
9822 A number representing position of the first frame with respect to the telecine
9823 pattern. This is to be used if the stream is cut. The default value is @code{0}.
9828 Apply dilation effect to the video.
9830 This filter replaces the pixel by the local(3x3) maximum.
9832 It accepts the following options:
9839 Limit the maximum change for each plane, default is 65535.
9840 If 0, plane will remain unchanged.
9843 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
9846 Flags to local 3x3 coordinates maps like this:
9853 @subsection Commands
9855 This filter supports the all above options as @ref{commands}.
9859 Displace pixels as indicated by second and third input stream.
9861 It takes three input streams and outputs one stream, the first input is the
9862 source, and second and third input are displacement maps.
9864 The second input specifies how much to displace pixels along the
9865 x-axis, while the third input specifies how much to displace pixels
9867 If one of displacement map streams terminates, last frame from that
9868 displacement map will be used.
9870 Note that once generated, displacements maps can be reused over and over again.
9872 A description of the accepted options follows.
9876 Set displace behavior for pixels that are out of range.
9878 Available values are:
9881 Missing pixels are replaced by black pixels.
9884 Adjacent pixels will spread out to replace missing pixels.
9887 Out of range pixels are wrapped so they point to pixels of other side.
9890 Out of range pixels will be replaced with mirrored pixels.
9892 Default is @samp{smear}.
9896 @subsection Examples
9900 Add ripple effect to rgb input of video size hd720:
9902 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
9906 Add wave effect to rgb input of video size hd720:
9908 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
9912 @anchor{dnn_processing}
9913 @section dnn_processing
9915 Do image processing with deep neural networks. It works together with another filter
9916 which converts the pixel format of the Frame to what the dnn network requires.
9918 The filter accepts the following options:
9922 Specify which DNN backend to use for model loading and execution. This option accepts
9923 the following values:
9927 Native implementation of DNN loading and execution.
9930 TensorFlow backend. To enable this backend you
9931 need to install the TensorFlow for C library (see
9932 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
9933 @code{--enable-libtensorflow}
9936 OpenVINO backend. To enable this backend you
9937 need to build and install the OpenVINO for C library (see
9938 @url{https://github.com/openvinotoolkit/openvino/blob/master/build-instruction.md}) and configure FFmpeg with
9939 @code{--enable-libopenvino} (--extra-cflags=-I... --extra-ldflags=-L... might
9940 be needed if the header files and libraries are not installed into system path)
9944 Default value is @samp{native}.
9947 Set path to model file specifying network architecture and its parameters.
9948 Note that different backends use different file formats. TensorFlow, OpenVINO and native
9949 backend can load files for only its format.
9951 Native model file (.model) can be generated from TensorFlow model file (.pb) by using tools/python/convert.py
9954 Set the input name of the dnn network.
9957 Set the output name of the dnn network.
9961 @subsection Examples
9965 Remove rain in rgb24 frame with can.pb (see @ref{derain} filter):
9967 ./ffmpeg -i rain.jpg -vf format=rgb24,dnn_processing=dnn_backend=tensorflow:model=can.pb:input=x:output=y derain.jpg
9971 Halve the pixel value of the frame with format gray32f:
9973 ffmpeg -i input.jpg -vf format=grayf32,dnn_processing=model=halve_gray_float.model:input=dnn_in:output=dnn_out:dnn_backend=native -y out.native.png
9977 Handle the Y channel with srcnn.pb (see @ref{sr} filter) for frame with yuv420p (planar YUV formats supported):
9979 ./ffmpeg -i 480p.jpg -vf format=yuv420p,scale=w=iw*2:h=ih*2,dnn_processing=dnn_backend=tensorflow:model=srcnn.pb:input=x:output=y -y srcnn.jpg
9983 Handle the Y channel with espcn.pb (see @ref{sr} filter), which changes frame size, for format yuv420p (planar YUV formats supported):
9985 ./ffmpeg -i 480p.jpg -vf format=yuv420p,dnn_processing=dnn_backend=tensorflow:model=espcn.pb:input=x:output=y -y tmp.espcn.jpg
9992 Draw a colored box on the input image.
9994 It accepts the following parameters:
9999 The expressions which specify the top left corner coordinates of the box. It defaults to 0.
10003 The expressions which specify the width and height of the box; if 0 they are interpreted as
10004 the input width and height. It defaults to 0.
10007 Specify the color of the box to write. For the general syntax of this option,
10008 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10009 value @code{invert} is used, the box edge color is the same as the
10010 video with inverted luma.
10013 The expression which sets the thickness of the box edge.
10014 A value of @code{fill} will create a filled box. Default value is @code{3}.
10016 See below for the list of accepted constants.
10019 Applicable if the input has alpha. With value @code{1}, the pixels of the painted box
10020 will overwrite the video's color and alpha pixels.
10021 Default is @code{0}, which composites the box onto the input, leaving the video's alpha intact.
10024 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10025 following constants:
10029 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10033 horizontal and vertical chroma subsample values. For example for the
10034 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10038 The input width and height.
10041 The input sample aspect ratio.
10045 The x and y offset coordinates where the box is drawn.
10049 The width and height of the drawn box.
10052 The thickness of the drawn box.
10054 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10055 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10059 @subsection Examples
10063 Draw a black box around the edge of the input image:
10069 Draw a box with color red and an opacity of 50%:
10071 drawbox=10:20:200:60:red@@0.5
10074 The previous example can be specified as:
10076 drawbox=x=10:y=20:w=200:h=60:color=red@@0.5
10080 Fill the box with pink color:
10082 drawbox=x=10:y=10:w=100:h=100:color=pink@@0.5:t=fill
10086 Draw a 2-pixel red 2.40:1 mask:
10088 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
10092 @subsection Commands
10093 This filter supports same commands as options.
10094 The command accepts the same syntax of the corresponding option.
10096 If the specified expression is not valid, it is kept at its current
10101 Draw a graph using input video metadata.
10103 It accepts the following parameters:
10107 Set 1st frame metadata key from which metadata values will be used to draw a graph.
10110 Set 1st foreground color expression.
10113 Set 2nd frame metadata key from which metadata values will be used to draw a graph.
10116 Set 2nd foreground color expression.
10119 Set 3rd frame metadata key from which metadata values will be used to draw a graph.
10122 Set 3rd foreground color expression.
10125 Set 4th frame metadata key from which metadata values will be used to draw a graph.
10128 Set 4th foreground color expression.
10131 Set minimal value of metadata value.
10134 Set maximal value of metadata value.
10137 Set graph background color. Default is white.
10142 Available values for mode is:
10149 Default is @code{line}.
10154 Available values for slide is:
10157 Draw new frame when right border is reached.
10160 Replace old columns with new ones.
10163 Scroll from right to left.
10166 Scroll from left to right.
10169 Draw single picture.
10172 Default is @code{frame}.
10175 Set size of graph video. For the syntax of this option, check the
10176 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
10177 The default value is @code{900x256}.
10180 Set the output frame rate. Default value is @code{25}.
10182 The foreground color expressions can use the following variables:
10185 Minimal value of metadata value.
10188 Maximal value of metadata value.
10191 Current metadata key value.
10194 The color is defined as 0xAABBGGRR.
10197 Example using metadata from @ref{signalstats} filter:
10199 signalstats,drawgraph=lavfi.signalstats.YAVG:min=0:max=255
10202 Example using metadata from @ref{ebur128} filter:
10204 ebur128=metadata=1,adrawgraph=lavfi.r128.M:min=-120:max=5
10209 Draw a grid on the input image.
10211 It accepts the following parameters:
10216 The expressions which specify the coordinates of some point of grid intersection (meant to configure offset). Both default to 0.
10220 The expressions which specify the width and height of the grid cell, if 0 they are interpreted as the
10221 input width and height, respectively, minus @code{thickness}, so image gets
10222 framed. Default to 0.
10225 Specify the color of the grid. For the general syntax of this option,
10226 check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}. If the special
10227 value @code{invert} is used, the grid color is the same as the
10228 video with inverted luma.
10231 The expression which sets the thickness of the grid line. Default value is @code{1}.
10233 See below for the list of accepted constants.
10236 Applicable if the input has alpha. With @code{1} the pixels of the painted grid
10237 will overwrite the video's color and alpha pixels.
10238 Default is @code{0}, which composites the grid onto the input, leaving the video's alpha intact.
10241 The parameters for @var{x}, @var{y}, @var{w} and @var{h} and @var{t} are expressions containing the
10242 following constants:
10246 The input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}.
10250 horizontal and vertical chroma subsample values. For example for the
10251 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10255 The input grid cell width and height.
10258 The input sample aspect ratio.
10262 The x and y coordinates of some point of grid intersection (meant to configure offset).
10266 The width and height of the drawn cell.
10269 The thickness of the drawn cell.
10271 These constants allow the @var{x}, @var{y}, @var{w}, @var{h} and @var{t} expressions to refer to
10272 each other, so you may for example specify @code{y=x/dar} or @code{h=w/dar}.
10276 @subsection Examples
10280 Draw a grid with cell 100x100 pixels, thickness 2 pixels, with color red and an opacity of 50%:
10282 drawgrid=width=100:height=100:thickness=2:color=red@@0.5
10286 Draw a white 3x3 grid with an opacity of 50%:
10288 drawgrid=w=iw/3:h=ih/3:t=2:c=white@@0.5
10292 @subsection Commands
10293 This filter supports same commands as options.
10294 The command accepts the same syntax of the corresponding option.
10296 If the specified expression is not valid, it is kept at its current
10302 Draw a text string or text from a specified file on top of a video, using the
10303 libfreetype library.
10305 To enable compilation of this filter, you need to configure FFmpeg with
10306 @code{--enable-libfreetype}.
10307 To enable default font fallback and the @var{font} option you need to
10308 configure FFmpeg with @code{--enable-libfontconfig}.
10309 To enable the @var{text_shaping} option, you need to configure FFmpeg with
10310 @code{--enable-libfribidi}.
10314 It accepts the following parameters:
10319 Used to draw a box around text using the background color.
10320 The value must be either 1 (enable) or 0 (disable).
10321 The default value of @var{box} is 0.
10324 Set the width of the border to be drawn around the box using @var{boxcolor}.
10325 The default value of @var{boxborderw} is 0.
10328 The color to be used for drawing box around text. For the syntax of this
10329 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10331 The default value of @var{boxcolor} is "white".
10334 Set the line spacing in pixels of the border to be drawn around the box using @var{box}.
10335 The default value of @var{line_spacing} is 0.
10338 Set the width of the border to be drawn around the text using @var{bordercolor}.
10339 The default value of @var{borderw} is 0.
10342 Set the color to be used for drawing border around text. For the syntax of this
10343 option, check the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10345 The default value of @var{bordercolor} is "black".
10348 Select how the @var{text} is expanded. Can be either @code{none},
10349 @code{strftime} (deprecated) or
10350 @code{normal} (default). See the @ref{drawtext_expansion, Text expansion} section
10354 Set a start time for the count. Value is in microseconds. Only applied
10355 in the deprecated strftime expansion mode. To emulate in normal expansion
10356 mode use the @code{pts} function, supplying the start time (in seconds)
10357 as the second argument.
10360 If true, check and fix text coords to avoid clipping.
10363 The color to be used for drawing fonts. For the syntax of this option, check
10364 the @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
10366 The default value of @var{fontcolor} is "black".
10368 @item fontcolor_expr
10369 String which is expanded the same way as @var{text} to obtain dynamic
10370 @var{fontcolor} value. By default this option has empty value and is not
10371 processed. When this option is set, it overrides @var{fontcolor} option.
10374 The font family to be used for drawing text. By default Sans.
10377 The font file to be used for drawing text. The path must be included.
10378 This parameter is mandatory if the fontconfig support is disabled.
10381 Draw the text applying alpha blending. The value can
10382 be a number between 0.0 and 1.0.
10383 The expression accepts the same variables @var{x, y} as well.
10384 The default value is 1.
10385 Please see @var{fontcolor_expr}.
10388 The font size to be used for drawing text.
10389 The default value of @var{fontsize} is 16.
10392 If set to 1, attempt to shape the text (for example, reverse the order of
10393 right-to-left text and join Arabic characters) before drawing it.
10394 Otherwise, just draw the text exactly as given.
10395 By default 1 (if supported).
10397 @item ft_load_flags
10398 The flags to be used for loading the fonts.
10400 The flags map the corresponding flags supported by libfreetype, and are
10401 a combination of the following values:
10408 @item vertical_layout
10409 @item force_autohint
10412 @item ignore_global_advance_width
10414 @item ignore_transform
10416 @item linear_design
10420 Default value is "default".
10422 For more information consult the documentation for the FT_LOAD_*
10426 The color to be used for drawing a shadow behind the drawn text. For the
10427 syntax of this option, check the @ref{color syntax,,"Color" section in the
10428 ffmpeg-utils manual,ffmpeg-utils}.
10430 The default value of @var{shadowcolor} is "black".
10434 The x and y offsets for the text shadow position with respect to the
10435 position of the text. They can be either positive or negative
10436 values. The default value for both is "0".
10439 The starting frame number for the n/frame_num variable. The default value
10443 The size in number of spaces to use for rendering the tab.
10444 Default value is 4.
10447 Set the initial timecode representation in "hh:mm:ss[:;.]ff"
10448 format. It can be used with or without text parameter. @var{timecode_rate}
10449 option must be specified.
10451 @item timecode_rate, rate, r
10452 Set the timecode frame rate (timecode only). Value will be rounded to nearest
10453 integer. Minimum value is "1".
10454 Drop-frame timecode is supported for frame rates 30 & 60.
10457 If set to 1, the output of the timecode option will wrap around at 24 hours.
10458 Default is 0 (disabled).
10461 The text string to be drawn. The text must be a sequence of UTF-8
10462 encoded characters.
10463 This parameter is mandatory if no file is specified with the parameter
10467 A text file containing text to be drawn. The text must be a sequence
10468 of UTF-8 encoded characters.
10470 This parameter is mandatory if no text string is specified with the
10471 parameter @var{text}.
10473 If both @var{text} and @var{textfile} are specified, an error is thrown.
10476 If set to 1, the @var{textfile} will be reloaded before each frame.
10477 Be sure to update it atomically, or it may be read partially, or even fail.
10481 The expressions which specify the offsets where text will be drawn
10482 within the video frame. They are relative to the top/left border of the
10485 The default value of @var{x} and @var{y} is "0".
10487 See below for the list of accepted constants and functions.
10490 The parameters for @var{x} and @var{y} are expressions containing the
10491 following constants and functions:
10495 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
10499 horizontal and vertical chroma subsample values. For example for the
10500 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
10503 the height of each text line
10511 @item max_glyph_a, ascent
10512 the maximum distance from the baseline to the highest/upper grid
10513 coordinate used to place a glyph outline point, for all the rendered
10515 It is a positive value, due to the grid's orientation with the Y axis
10518 @item max_glyph_d, descent
10519 the maximum distance from the baseline to the lowest grid coordinate
10520 used to place a glyph outline point, for all the rendered glyphs.
10521 This is a negative value, due to the grid's orientation, with the Y axis
10525 maximum glyph height, that is the maximum height for all the glyphs
10526 contained in the rendered text, it is equivalent to @var{ascent} -
10530 maximum glyph width, that is the maximum width for all the glyphs
10531 contained in the rendered text
10534 the number of input frame, starting from 0
10536 @item rand(min, max)
10537 return a random number included between @var{min} and @var{max}
10540 The input sample aspect ratio.
10543 timestamp expressed in seconds, NAN if the input timestamp is unknown
10546 the height of the rendered text
10549 the width of the rendered text
10553 the x and y offset coordinates where the text is drawn.
10555 These parameters allow the @var{x} and @var{y} expressions to refer
10556 to each other, so you can for example specify @code{y=x/dar}.
10559 A one character description of the current frame's picture type.
10562 The current packet's position in the input file or stream
10563 (in bytes, from the start of the input). A value of -1 indicates
10564 this info is not available.
10567 The current packet's duration, in seconds.
10570 The current packet's size (in bytes).
10573 @anchor{drawtext_expansion}
10574 @subsection Text expansion
10576 If @option{expansion} is set to @code{strftime},
10577 the filter recognizes strftime() sequences in the provided text and
10578 expands them accordingly. Check the documentation of strftime(). This
10579 feature is deprecated.
10581 If @option{expansion} is set to @code{none}, the text is printed verbatim.
10583 If @option{expansion} is set to @code{normal} (which is the default),
10584 the following expansion mechanism is used.
10586 The backslash character @samp{\}, followed by any character, always expands to
10587 the second character.
10589 Sequences of the form @code{%@{...@}} are expanded. The text between the
10590 braces is a function name, possibly followed by arguments separated by ':'.
10591 If the arguments contain special characters or delimiters (':' or '@}'),
10592 they should be escaped.
10594 Note that they probably must also be escaped as the value for the
10595 @option{text} option in the filter argument string and as the filter
10596 argument in the filtergraph description, and possibly also for the shell,
10597 that makes up to four levels of escaping; using a text file avoids these
10600 The following functions are available:
10605 The expression evaluation result.
10607 It must take one argument specifying the expression to be evaluated,
10608 which accepts the same constants and functions as the @var{x} and
10609 @var{y} values. Note that not all constants should be used, for
10610 example the text size is not known when evaluating the expression, so
10611 the constants @var{text_w} and @var{text_h} will have an undefined
10614 @item expr_int_format, eif
10615 Evaluate the expression's value and output as formatted integer.
10617 The first argument is the expression to be evaluated, just as for the @var{expr} function.
10618 The second argument specifies the output format. Allowed values are @samp{x},
10619 @samp{X}, @samp{d} and @samp{u}. They are treated exactly as in the
10620 @code{printf} function.
10621 The third parameter is optional and sets the number of positions taken by the output.
10622 It can be used to add padding with zeros from the left.
10625 The time at which the filter is running, expressed in UTC.
10626 It can accept an argument: a strftime() format string.
10629 The time at which the filter is running, expressed in the local time zone.
10630 It can accept an argument: a strftime() format string.
10633 Frame metadata. Takes one or two arguments.
10635 The first argument is mandatory and specifies the metadata key.
10637 The second argument is optional and specifies a default value, used when the
10638 metadata key is not found or empty.
10640 Available metadata can be identified by inspecting entries
10641 starting with TAG included within each frame section
10642 printed by running @code{ffprobe -show_frames}.
10644 String metadata generated in filters leading to
10645 the drawtext filter are also available.
10648 The frame number, starting from 0.
10651 A one character description of the current picture type.
10654 The timestamp of the current frame.
10655 It can take up to three arguments.
10657 The first argument is the format of the timestamp; it defaults to @code{flt}
10658 for seconds as a decimal number with microsecond accuracy; @code{hms} stands
10659 for a formatted @var{[-]HH:MM:SS.mmm} timestamp with millisecond accuracy.
10660 @code{gmtime} stands for the timestamp of the frame formatted as UTC time;
10661 @code{localtime} stands for the timestamp of the frame formatted as
10662 local time zone time.
10664 The second argument is an offset added to the timestamp.
10666 If the format is set to @code{hms}, a third argument @code{24HH} may be
10667 supplied to present the hour part of the formatted timestamp in 24h format
10670 If the format is set to @code{localtime} or @code{gmtime},
10671 a third argument may be supplied: a strftime() format string.
10672 By default, @var{YYYY-MM-DD HH:MM:SS} format will be used.
10675 @subsection Commands
10677 This filter supports altering parameters via commands:
10680 Alter existing filter parameters.
10682 Syntax for the argument is the same as for filter invocation, e.g.
10685 fontsize=56:fontcolor=green:text='Hello World'
10688 Full filter invocation with sendcmd would look like this:
10691 sendcmd=c='56.0 drawtext reinit fontsize=56\:fontcolor=green\:text=Hello\\ World'
10695 If the entire argument can't be parsed or applied as valid values then the filter will
10696 continue with its existing parameters.
10698 @subsection Examples
10702 Draw "Test Text" with font FreeSerif, using the default values for the
10703 optional parameters.
10706 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text'"
10710 Draw 'Test Text' with font FreeSerif of size 24 at position x=100
10711 and y=50 (counting from the top-left corner of the screen), text is
10712 yellow with a red box around it. Both the text and the box have an
10716 drawtext="fontfile=/usr/share/fonts/truetype/freefont/FreeSerif.ttf: text='Test Text':\
10717 x=100: y=50: fontsize=24: fontcolor=yellow@@0.2: box=1: boxcolor=red@@0.2"
10720 Note that the double quotes are not necessary if spaces are not used
10721 within the parameter list.
10724 Show the text at the center of the video frame:
10726 drawtext="fontsize=30:fontfile=FreeSerif.ttf:text='hello world':x=(w-text_w)/2:y=(h-text_h)/2"
10730 Show the text at a random position, switching to a new position every 30 seconds:
10732 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)"
10736 Show a text line sliding from right to left in the last row of the video
10737 frame. The file @file{LONG_LINE} is assumed to contain a single line
10740 drawtext="fontsize=15:fontfile=FreeSerif.ttf:text=LONG_LINE:y=h-line_h:x=-50*t"
10744 Show the content of file @file{CREDITS} off the bottom of the frame and scroll up.
10746 drawtext="fontsize=20:fontfile=FreeSerif.ttf:textfile=CREDITS:y=h-20*t"
10750 Draw a single green letter "g", at the center of the input video.
10751 The glyph baseline is placed at half screen height.
10753 drawtext="fontsize=60:fontfile=FreeSerif.ttf:fontcolor=green:text=g:x=(w-max_glyph_w)/2:y=h/2-ascent"
10757 Show text for 1 second every 3 seconds:
10759 drawtext="fontfile=FreeSerif.ttf:fontcolor=white:x=100:y=x/dar:enable=lt(mod(t\,3)\,1):text='blink'"
10763 Use fontconfig to set the font. Note that the colons need to be escaped.
10765 drawtext='fontfile=Linux Libertine O-40\:style=Semibold:text=FFmpeg'
10769 Draw "Test Text" with font size dependent on height of the video.
10771 drawtext="text='Test Text': fontsize=h/30: x=(w-text_w)/2: y=(h-text_h*2)"
10775 Print the date of a real-time encoding (see strftime(3)):
10777 drawtext='fontfile=FreeSans.ttf:text=%@{localtime\:%a %b %d %Y@}'
10781 Show text fading in and out (appearing/disappearing):
10784 DS=1.0 # display start
10785 DE=10.0 # display end
10786 FID=1.5 # fade in duration
10787 FOD=5 # fade out duration
10788 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 @}"
10792 Horizontally align multiple separate texts. Note that @option{max_glyph_a}
10793 and the @option{fontsize} value are included in the @option{y} offset.
10795 drawtext=fontfile=FreeSans.ttf:text=DOG:fontsize=24:x=10:y=20+24-max_glyph_a,
10796 drawtext=fontfile=FreeSans.ttf:text=cow:fontsize=24:x=80:y=20+24-max_glyph_a
10800 Plot special @var{lavf.image2dec.source_basename} metadata onto each frame if
10801 such metadata exists. Otherwise, plot the string "NA". Note that image2 demuxer
10802 must have option @option{-export_path_metadata 1} for the special metadata fields
10803 to be available for filters.
10805 drawtext="fontsize=20:fontcolor=white:fontfile=FreeSans.ttf:text='%@{metadata\:lavf.image2dec.source_basename\:NA@}':x=10:y=10"
10810 For more information about libfreetype, check:
10811 @url{http://www.freetype.org/}.
10813 For more information about fontconfig, check:
10814 @url{http://freedesktop.org/software/fontconfig/fontconfig-user.html}.
10816 For more information about libfribidi, check:
10817 @url{http://fribidi.org/}.
10819 @section edgedetect
10821 Detect and draw edges. The filter uses the Canny Edge Detection algorithm.
10823 The filter accepts the following options:
10828 Set low and high threshold values used by the Canny thresholding
10831 The high threshold selects the "strong" edge pixels, which are then
10832 connected through 8-connectivity with the "weak" edge pixels selected
10833 by the low threshold.
10835 @var{low} and @var{high} threshold values must be chosen in the range
10836 [0,1], and @var{low} should be lesser or equal to @var{high}.
10838 Default value for @var{low} is @code{20/255}, and default value for @var{high}
10842 Define the drawing mode.
10846 Draw white/gray wires on black background.
10849 Mix the colors to create a paint/cartoon effect.
10852 Apply Canny edge detector on all selected planes.
10854 Default value is @var{wires}.
10857 Select planes for filtering. By default all available planes are filtered.
10860 @subsection Examples
10864 Standard edge detection with custom values for the hysteresis thresholding:
10866 edgedetect=low=0.1:high=0.4
10870 Painting effect without thresholding:
10872 edgedetect=mode=colormix:high=0
10878 Apply a posterize effect using the ELBG (Enhanced LBG) algorithm.
10880 For each input image, the filter will compute the optimal mapping from
10881 the input to the output given the codebook length, that is the number
10882 of distinct output colors.
10884 This filter accepts the following options.
10887 @item codebook_length, l
10888 Set codebook length. The value must be a positive integer, and
10889 represents the number of distinct output colors. Default value is 256.
10892 Set the maximum number of iterations to apply for computing the optimal
10893 mapping. The higher the value the better the result and the higher the
10894 computation time. Default value is 1.
10897 Set a random seed, must be an integer included between 0 and
10898 UINT32_MAX. If not specified, or if explicitly set to -1, the filter
10899 will try to use a good random seed on a best effort basis.
10902 Set pal8 output pixel format. This option does not work with codebook
10903 length greater than 256.
10908 Measure graylevel entropy in histogram of color channels of video frames.
10910 It accepts the following parameters:
10914 Can be either @var{normal} or @var{diff}. Default is @var{normal}.
10916 @var{diff} mode measures entropy of histogram delta values, absolute differences
10917 between neighbour histogram values.
10921 Set brightness, contrast, saturation and approximate gamma adjustment.
10923 The filter accepts the following options:
10927 Set the contrast expression. The value must be a float value in range
10928 @code{-1000.0} to @code{1000.0}. The default value is "1".
10931 Set the brightness expression. The value must be a float value in
10932 range @code{-1.0} to @code{1.0}. The default value is "0".
10935 Set the saturation expression. The value must be a float in
10936 range @code{0.0} to @code{3.0}. The default value is "1".
10939 Set the gamma expression. The value must be a float in range
10940 @code{0.1} to @code{10.0}. The default value is "1".
10943 Set the gamma expression for red. The value must be a float in
10944 range @code{0.1} to @code{10.0}. The default value is "1".
10947 Set the gamma expression for green. The value must be a float in range
10948 @code{0.1} to @code{10.0}. The default value is "1".
10951 Set the gamma expression for blue. The value must be a float in range
10952 @code{0.1} to @code{10.0}. The default value is "1".
10955 Set the gamma weight expression. It can be used to reduce the effect
10956 of a high gamma value on bright image areas, e.g. keep them from
10957 getting overamplified and just plain white. The value must be a float
10958 in range @code{0.0} to @code{1.0}. A value of @code{0.0} turns the
10959 gamma correction all the way down while @code{1.0} leaves it at its
10960 full strength. Default is "1".
10963 Set when the expressions for brightness, contrast, saturation and
10964 gamma expressions are evaluated.
10966 It accepts the following values:
10969 only evaluate expressions once during the filter initialization or
10970 when a command is processed
10973 evaluate expressions for each incoming frame
10976 Default value is @samp{init}.
10979 The expressions accept the following parameters:
10982 frame count of the input frame starting from 0
10985 byte position of the corresponding packet in the input file, NAN if
10989 frame rate of the input video, NAN if the input frame rate is unknown
10992 timestamp expressed in seconds, NAN if the input timestamp is unknown
10995 @subsection Commands
10996 The filter supports the following commands:
11000 Set the contrast expression.
11003 Set the brightness expression.
11006 Set the saturation expression.
11009 Set the gamma expression.
11012 Set the gamma_r expression.
11015 Set gamma_g expression.
11018 Set gamma_b expression.
11021 Set gamma_weight expression.
11023 The command accepts the same syntax of the corresponding option.
11025 If the specified expression is not valid, it is kept at its current
11032 Apply erosion effect to the video.
11034 This filter replaces the pixel by the local(3x3) minimum.
11036 It accepts the following options:
11043 Limit the maximum change for each plane, default is 65535.
11044 If 0, plane will remain unchanged.
11047 Flag which specifies the pixel to refer to. Default is 255 i.e. all eight
11050 Flags to local 3x3 coordinates maps like this:
11057 @subsection Commands
11059 This filter supports the all above options as @ref{commands}.
11061 @section extractplanes
11063 Extract color channel components from input video stream into
11064 separate grayscale video streams.
11066 The filter accepts the following option:
11070 Set plane(s) to extract.
11072 Available values for planes are:
11083 Choosing planes not available in the input will result in an error.
11084 That means you cannot select @code{r}, @code{g}, @code{b} planes
11085 with @code{y}, @code{u}, @code{v} planes at same time.
11088 @subsection Examples
11092 Extract luma, u and v color channel component from input video frame
11093 into 3 grayscale outputs:
11095 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
11101 Apply a fade-in/out effect to the input video.
11103 It accepts the following parameters:
11107 The effect type can be either "in" for a fade-in, or "out" for a fade-out
11109 Default is @code{in}.
11111 @item start_frame, s
11112 Specify the number of the frame to start applying the fade
11113 effect at. Default is 0.
11116 The number of frames that the fade effect lasts. At the end of the
11117 fade-in effect, the output video will have the same intensity as the input video.
11118 At the end of the fade-out transition, the output video will be filled with the
11119 selected @option{color}.
11123 If set to 1, fade only alpha channel, if one exists on the input.
11124 Default value is 0.
11126 @item start_time, st
11127 Specify the timestamp (in seconds) of the frame to start to apply the fade
11128 effect. If both start_frame and start_time are specified, the fade will start at
11129 whichever comes last. Default is 0.
11132 The number of seconds for which the fade effect has to last. At the end of the
11133 fade-in effect the output video will have the same intensity as the input video,
11134 at the end of the fade-out transition the output video will be filled with the
11135 selected @option{color}.
11136 If both duration and nb_frames are specified, duration is used. Default is 0
11137 (nb_frames is used by default).
11140 Specify the color of the fade. Default is "black".
11143 @subsection Examples
11147 Fade in the first 30 frames of video:
11152 The command above is equivalent to:
11158 Fade out the last 45 frames of a 200-frame video:
11161 fade=type=out:start_frame=155:nb_frames=45
11165 Fade in the first 25 frames and fade out the last 25 frames of a 1000-frame video:
11167 fade=in:0:25, fade=out:975:25
11171 Make the first 5 frames yellow, then fade in from frame 5-24:
11173 fade=in:5:20:color=yellow
11177 Fade in alpha over first 25 frames of video:
11179 fade=in:0:25:alpha=1
11183 Make the first 5.5 seconds black, then fade in for 0.5 seconds:
11185 fade=t=in:st=5.5:d=0.5
11191 Denoise frames using 3D FFT (frequency domain filtering).
11193 The filter accepts the following options:
11197 Set the noise sigma constant. This sets denoising strength.
11198 Default value is 1. Allowed range is from 0 to 30.
11199 Using very high sigma with low overlap may give blocking artifacts.
11202 Set amount of denoising. By default all detected noise is reduced.
11203 Default value is 1. Allowed range is from 0 to 1.
11206 Set size of block, Default is 4, can be 3, 4, 5 or 6.
11207 Actual size of block in pixels is 2 to power of @var{block}, so by default
11208 block size in pixels is 2^4 which is 16.
11211 Set block overlap. Default is 0.5. Allowed range is from 0.2 to 0.8.
11214 Set number of previous frames to use for denoising. By default is set to 0.
11217 Set number of next frames to to use for denoising. By default is set to 0.
11220 Set planes which will be filtered, by default are all available filtered
11225 Apply arbitrary expressions to samples in frequency domain
11229 Adjust the dc value (gain) of the luma plane of the image. The filter
11230 accepts an integer value in range @code{0} to @code{1000}. The default
11231 value is set to @code{0}.
11234 Adjust the dc value (gain) of the 1st chroma plane of the image. The
11235 filter accepts an integer value in range @code{0} to @code{1000}. The
11236 default value is set to @code{0}.
11239 Adjust the dc value (gain) of the 2nd chroma plane of the image. The
11240 filter accepts an integer value in range @code{0} to @code{1000}. The
11241 default value is set to @code{0}.
11244 Set the frequency domain weight expression for the luma plane.
11247 Set the frequency domain weight expression for the 1st chroma plane.
11250 Set the frequency domain weight expression for the 2nd chroma plane.
11253 Set when the expressions are evaluated.
11255 It accepts the following values:
11258 Only evaluate expressions once during the filter initialization.
11261 Evaluate expressions for each incoming frame.
11264 Default value is @samp{init}.
11266 The filter accepts the following variables:
11269 The coordinates of the current sample.
11273 The width and height of the image.
11276 The number of input frame, starting from 0.
11279 @subsection Examples
11285 fftfilt=dc_Y=128:weight_Y='squish(1-(Y+X)/100)'
11291 fftfilt=dc_Y=0:weight_Y='squish((Y+X)/100-1)'
11297 fftfilt=dc_Y=0:weight_Y='1+squish(1-(Y+X)/100)'
11303 fftfilt=dc_Y=0:weight_Y='exp(-4 * ((Y+X)/(W+H)))'
11310 Extract a single field from an interlaced image using stride
11311 arithmetic to avoid wasting CPU time. The output frames are marked as
11314 The filter accepts the following options:
11318 Specify whether to extract the top (if the value is @code{0} or
11319 @code{top}) or the bottom field (if the value is @code{1} or
11325 Create new frames by copying the top and bottom fields from surrounding frames
11326 supplied as numbers by the hint file.
11330 Set file containing hints: absolute/relative frame numbers.
11332 There must be one line for each frame in a clip. Each line must contain two
11333 numbers separated by the comma, optionally followed by @code{-} or @code{+}.
11334 Numbers supplied on each line of file can not be out of [N-1,N+1] where N
11335 is current frame number for @code{absolute} mode or out of [-1, 1] range
11336 for @code{relative} mode. First number tells from which frame to pick up top
11337 field and second number tells from which frame to pick up bottom field.
11339 If optionally followed by @code{+} output frame will be marked as interlaced,
11340 else if followed by @code{-} output frame will be marked as progressive, else
11341 it will be marked same as input frame.
11342 If optionally followed by @code{t} output frame will use only top field, or in
11343 case of @code{b} it will use only bottom field.
11344 If line starts with @code{#} or @code{;} that line is skipped.
11347 Can be item @code{absolute} or @code{relative}. Default is @code{absolute}.
11350 Example of first several lines of @code{hint} file for @code{relative} mode:
11352 0,0 - # first frame
11353 1,0 - # second frame, use third's frame top field and second's frame bottom field
11354 1,0 - # third frame, use fourth's frame top field and third's frame bottom field
11369 @section fieldmatch
11371 Field matching filter for inverse telecine. It is meant to reconstruct the
11372 progressive frames from a telecined stream. The filter does not drop duplicated
11373 frames, so to achieve a complete inverse telecine @code{fieldmatch} needs to be
11374 followed by a decimation filter such as @ref{decimate} in the filtergraph.
11376 The separation of the field matching and the decimation is notably motivated by
11377 the possibility of inserting a de-interlacing filter fallback between the two.
11378 If the source has mixed telecined and real interlaced content,
11379 @code{fieldmatch} will not be able to match fields for the interlaced parts.
11380 But these remaining combed frames will be marked as interlaced, and thus can be
11381 de-interlaced by a later filter such as @ref{yadif} before decimation.
11383 In addition to the various configuration options, @code{fieldmatch} can take an
11384 optional second stream, activated through the @option{ppsrc} option. If
11385 enabled, the frames reconstruction will be based on the fields and frames from
11386 this second stream. This allows the first input to be pre-processed in order to
11387 help the various algorithms of the filter, while keeping the output lossless
11388 (assuming the fields are matched properly). Typically, a field-aware denoiser,
11389 or brightness/contrast adjustments can help.
11391 Note that this filter uses the same algorithms as TIVTC/TFM (AviSynth project)
11392 and VIVTC/VFM (VapourSynth project). The later is a light clone of TFM from
11393 which @code{fieldmatch} is based on. While the semantic and usage are very
11394 close, some behaviour and options names can differ.
11396 The @ref{decimate} filter currently only works for constant frame rate input.
11397 If your input has mixed telecined (30fps) and progressive content with a lower
11398 framerate like 24fps use the following filterchain to produce the necessary cfr
11399 stream: @code{dejudder,fps=30000/1001,fieldmatch,decimate}.
11401 The filter accepts the following options:
11405 Specify the assumed field order of the input stream. Available values are:
11409 Auto detect parity (use FFmpeg's internal parity value).
11411 Assume bottom field first.
11413 Assume top field first.
11416 Note that it is sometimes recommended not to trust the parity announced by the
11419 Default value is @var{auto}.
11422 Set the matching mode or strategy to use. @option{pc} mode is the safest in the
11423 sense that it won't risk creating jerkiness due to duplicate frames when
11424 possible, but if there are bad edits or blended fields it will end up
11425 outputting combed frames when a good match might actually exist. On the other
11426 hand, @option{pcn_ub} mode is the most risky in terms of creating jerkiness,
11427 but will almost always find a good frame if there is one. The other values are
11428 all somewhere in between @option{pc} and @option{pcn_ub} in terms of risking
11429 jerkiness and creating duplicate frames versus finding good matches in sections
11430 with bad edits, orphaned fields, blended fields, etc.
11432 More details about p/c/n/u/b are available in @ref{p/c/n/u/b meaning} section.
11434 Available values are:
11438 2-way matching (p/c)
11440 2-way matching, and trying 3rd match if still combed (p/c + n)
11442 2-way matching, and trying 3rd match (same order) if still combed (p/c + u)
11444 2-way matching, trying 3rd match if still combed, and trying 4th/5th matches if
11445 still combed (p/c + n + u/b)
11447 3-way matching (p/c/n)
11449 3-way matching, and trying 4th/5th matches if all 3 of the original matches are
11450 detected as combed (p/c/n + u/b)
11453 The parenthesis at the end indicate the matches that would be used for that
11454 mode assuming @option{order}=@var{tff} (and @option{field} on @var{auto} or
11457 In terms of speed @option{pc} mode is by far the fastest and @option{pcn_ub} is
11460 Default value is @var{pc_n}.
11463 Mark the main input stream as a pre-processed input, and enable the secondary
11464 input stream as the clean source to pick the fields from. See the filter
11465 introduction for more details. It is similar to the @option{clip2} feature from
11468 Default value is @code{0} (disabled).
11471 Set the field to match from. It is recommended to set this to the same value as
11472 @option{order} unless you experience matching failures with that setting. In
11473 certain circumstances changing the field that is used to match from can have a
11474 large impact on matching performance. Available values are:
11478 Automatic (same value as @option{order}).
11480 Match from the bottom field.
11482 Match from the top field.
11485 Default value is @var{auto}.
11488 Set whether or not chroma is included during the match comparisons. In most
11489 cases it is recommended to leave this enabled. You should set this to @code{0}
11490 only if your clip has bad chroma problems such as heavy rainbowing or other
11491 artifacts. Setting this to @code{0} could also be used to speed things up at
11492 the cost of some accuracy.
11494 Default value is @code{1}.
11498 These define an exclusion band which excludes the lines between @option{y0} and
11499 @option{y1} from being included in the field matching decision. An exclusion
11500 band can be used to ignore subtitles, a logo, or other things that may
11501 interfere with the matching. @option{y0} sets the starting scan line and
11502 @option{y1} sets the ending line; all lines in between @option{y0} and
11503 @option{y1} (including @option{y0} and @option{y1}) will be ignored. Setting
11504 @option{y0} and @option{y1} to the same value will disable the feature.
11505 @option{y0} and @option{y1} defaults to @code{0}.
11508 Set the scene change detection threshold as a percentage of maximum change on
11509 the luma plane. Good values are in the @code{[8.0, 14.0]} range. Scene change
11510 detection is only relevant in case @option{combmatch}=@var{sc}. The range for
11511 @option{scthresh} is @code{[0.0, 100.0]}.
11513 Default value is @code{12.0}.
11516 When @option{combatch} is not @var{none}, @code{fieldmatch} will take into
11517 account the combed scores of matches when deciding what match to use as the
11518 final match. Available values are:
11522 No final matching based on combed scores.
11524 Combed scores are only used when a scene change is detected.
11526 Use combed scores all the time.
11529 Default is @var{sc}.
11532 Force @code{fieldmatch} to calculate the combed metrics for certain matches and
11533 print them. This setting is known as @option{micout} in TFM/VFM vocabulary.
11534 Available values are:
11538 No forced calculation.
11540 Force p/c/n calculations.
11542 Force p/c/n/u/b calculations.
11545 Default value is @var{none}.
11548 This is the area combing threshold used for combed frame detection. This
11549 essentially controls how "strong" or "visible" combing must be to be detected.
11550 Larger values mean combing must be more visible and smaller values mean combing
11551 can be less visible or strong and still be detected. Valid settings are from
11552 @code{-1} (every pixel will be detected as combed) to @code{255} (no pixel will
11553 be detected as combed). This is basically a pixel difference value. A good
11554 range is @code{[8, 12]}.
11556 Default value is @code{9}.
11559 Sets whether or not chroma is considered in the combed frame decision. Only
11560 disable this if your source has chroma problems (rainbowing, etc.) that are
11561 causing problems for the combed frame detection with chroma enabled. Actually,
11562 using @option{chroma}=@var{0} is usually more reliable, except for the case
11563 where there is chroma only combing in the source.
11565 Default value is @code{0}.
11569 Respectively set the x-axis and y-axis size of the window used during combed
11570 frame detection. This has to do with the size of the area in which
11571 @option{combpel} pixels are required to be detected as combed for a frame to be
11572 declared combed. See the @option{combpel} parameter description for more info.
11573 Possible values are any number that is a power of 2 starting at 4 and going up
11576 Default value is @code{16}.
11579 The number of combed pixels inside any of the @option{blocky} by
11580 @option{blockx} size blocks on the frame for the frame to be detected as
11581 combed. While @option{cthresh} controls how "visible" the combing must be, this
11582 setting controls "how much" combing there must be in any localized area (a
11583 window defined by the @option{blockx} and @option{blocky} settings) on the
11584 frame. Minimum value is @code{0} and maximum is @code{blocky x blockx} (at
11585 which point no frames will ever be detected as combed). This setting is known
11586 as @option{MI} in TFM/VFM vocabulary.
11588 Default value is @code{80}.
11591 @anchor{p/c/n/u/b meaning}
11592 @subsection p/c/n/u/b meaning
11594 @subsubsection p/c/n
11596 We assume the following telecined stream:
11599 Top fields: 1 2 2 3 4
11600 Bottom fields: 1 2 3 4 4
11603 The numbers correspond to the progressive frame the fields relate to. Here, the
11604 first two frames are progressive, the 3rd and 4th are combed, and so on.
11606 When @code{fieldmatch} is configured to run a matching from bottom
11607 (@option{field}=@var{bottom}) this is how this input stream get transformed:
11612 B 1 2 3 4 4 <-- matching reference
11621 As a result of the field matching, we can see that some frames get duplicated.
11622 To perform a complete inverse telecine, you need to rely on a decimation filter
11623 after this operation. See for instance the @ref{decimate} filter.
11625 The same operation now matching from top fields (@option{field}=@var{top})
11630 T 1 2 2 3 4 <-- matching reference
11640 In these examples, we can see what @var{p}, @var{c} and @var{n} mean;
11641 basically, they refer to the frame and field of the opposite parity:
11644 @item @var{p} matches the field of the opposite parity in the previous frame
11645 @item @var{c} matches the field of the opposite parity in the current frame
11646 @item @var{n} matches the field of the opposite parity in the next frame
11651 The @var{u} and @var{b} matching are a bit special in the sense that they match
11652 from the opposite parity flag. In the following examples, we assume that we are
11653 currently matching the 2nd frame (Top:2, bottom:2). According to the match, a
11654 'x' is placed above and below each matched fields.
11656 With bottom matching (@option{field}=@var{bottom}):
11661 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11662 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11670 With top matching (@option{field}=@var{top}):
11675 Top 1 2 2 1 2 2 1 2 2 1 2 2 1 2 2
11676 Bottom 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
11684 @subsection Examples
11686 Simple IVTC of a top field first telecined stream:
11688 fieldmatch=order=tff:combmatch=none, decimate
11691 Advanced IVTC, with fallback on @ref{yadif} for still combed frames:
11693 fieldmatch=order=tff:combmatch=full, yadif=deint=interlaced, decimate
11696 @section fieldorder
11698 Transform the field order of the input video.
11700 It accepts the following parameters:
11705 The output field order. Valid values are @var{tff} for top field first or @var{bff}
11706 for bottom field first.
11709 The default value is @samp{tff}.
11711 The transformation is done by shifting the picture content up or down
11712 by one line, and filling the remaining line with appropriate picture content.
11713 This method is consistent with most broadcast field order converters.
11715 If the input video is not flagged as being interlaced, or it is already
11716 flagged as being of the required output field order, then this filter does
11717 not alter the incoming video.
11719 It is very useful when converting to or from PAL DV material,
11720 which is bottom field first.
11724 ffmpeg -i in.vob -vf "fieldorder=bff" out.dv
11727 @section fifo, afifo
11729 Buffer input images and send them when they are requested.
11731 It is mainly useful when auto-inserted by the libavfilter
11734 It does not take parameters.
11736 @section fillborders
11738 Fill borders of the input video, without changing video stream dimensions.
11739 Sometimes video can have garbage at the four edges and you may not want to
11740 crop video input to keep size multiple of some number.
11742 This filter accepts the following options:
11746 Number of pixels to fill from left border.
11749 Number of pixels to fill from right border.
11752 Number of pixels to fill from top border.
11755 Number of pixels to fill from bottom border.
11760 It accepts the following values:
11763 fill pixels using outermost pixels
11766 fill pixels using mirroring (half sample symmetric)
11769 fill pixels with constant value
11772 fill pixels using reflecting (whole sample symmetric)
11775 fill pixels using wrapping
11778 fade pixels to constant value
11781 Default is @var{smear}.
11784 Set color for pixels in fixed or fade mode. Default is @var{black}.
11787 @subsection Commands
11788 This filter supports same @ref{commands} as options.
11789 The command accepts the same syntax of the corresponding option.
11791 If the specified expression is not valid, it is kept at its current
11796 Find a rectangular object
11798 It accepts the following options:
11802 Filepath of the object image, needs to be in gray8.
11805 Detection threshold, default is 0.5.
11808 Number of mipmaps, default is 3.
11810 @item xmin, ymin, xmax, ymax
11811 Specifies the rectangle in which to search.
11814 @subsection Examples
11818 Cover a rectangular object by the supplied image of a given video using @command{ffmpeg}:
11820 ffmpeg -i file.ts -vf find_rect=newref.pgm,cover_rect=cover.jpg:mode=cover new.mkv
11826 Flood area with values of same pixel components with another values.
11828 It accepts the following options:
11831 Set pixel x coordinate.
11834 Set pixel y coordinate.
11837 Set source #0 component value.
11840 Set source #1 component value.
11843 Set source #2 component value.
11846 Set source #3 component value.
11849 Set destination #0 component value.
11852 Set destination #1 component value.
11855 Set destination #2 component value.
11858 Set destination #3 component value.
11864 Convert the input video to one of the specified pixel formats.
11865 Libavfilter will try to pick one that is suitable as input to
11868 It accepts the following parameters:
11872 A '|'-separated list of pixel format names, such as
11873 "pix_fmts=yuv420p|monow|rgb24".
11877 @subsection Examples
11881 Convert the input video to the @var{yuv420p} format
11883 format=pix_fmts=yuv420p
11886 Convert the input video to any of the formats in the list
11888 format=pix_fmts=yuv420p|yuv444p|yuv410p
11895 Convert the video to specified constant frame rate by duplicating or dropping
11896 frames as necessary.
11898 It accepts the following parameters:
11902 The desired output frame rate. The default is @code{25}.
11905 Assume the first PTS should be the given value, in seconds. This allows for
11906 padding/trimming at the start of stream. By default, no assumption is made
11907 about the first frame's expected PTS, so no padding or trimming is done.
11908 For example, this could be set to 0 to pad the beginning with duplicates of
11909 the first frame if a video stream starts after the audio stream or to trim any
11910 frames with a negative PTS.
11913 Timestamp (PTS) rounding method.
11915 Possible values are:
11922 round towards -infinity
11924 round towards +infinity
11928 The default is @code{near}.
11931 Action performed when reading the last frame.
11933 Possible values are:
11936 Use same timestamp rounding method as used for other frames.
11938 Pass through last frame if input duration has not been reached yet.
11940 The default is @code{round}.
11944 Alternatively, the options can be specified as a flat string:
11945 @var{fps}[:@var{start_time}[:@var{round}]].
11947 See also the @ref{setpts} filter.
11949 @subsection Examples
11953 A typical usage in order to set the fps to 25:
11959 Sets the fps to 24, using abbreviation and rounding method to round to nearest:
11961 fps=fps=film:round=near
11967 Pack two different video streams into a stereoscopic video, setting proper
11968 metadata on supported codecs. The two views should have the same size and
11969 framerate and processing will stop when the shorter video ends. Please note
11970 that you may conveniently adjust view properties with the @ref{scale} and
11973 It accepts the following parameters:
11977 The desired packing format. Supported values are:
11982 The views are next to each other (default).
11985 The views are on top of each other.
11988 The views are packed by line.
11991 The views are packed by column.
11994 The views are temporally interleaved.
12003 # Convert left and right views into a frame-sequential video
12004 ffmpeg -i LEFT -i RIGHT -filter_complex framepack=frameseq OUTPUT
12006 # Convert views into a side-by-side video with the same output resolution as the input
12007 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
12012 Change the frame rate by interpolating new video output frames from the source
12015 This filter is not designed to function correctly with interlaced media. If
12016 you wish to change the frame rate of interlaced media then you are required
12017 to deinterlace before this filter and re-interlace after this filter.
12019 A description of the accepted options follows.
12023 Specify the output frames per second. This option can also be specified
12024 as a value alone. The default is @code{50}.
12027 Specify the start of a range where the output frame will be created as a
12028 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12029 the default is @code{15}.
12032 Specify the end of a range where the output frame will be created as a
12033 linear interpolation of two frames. The range is [@code{0}-@code{255}],
12034 the default is @code{240}.
12037 Specify the level at which a scene change is detected as a value between
12038 0 and 100 to indicate a new scene; a low value reflects a low
12039 probability for the current frame to introduce a new scene, while a higher
12040 value means the current frame is more likely to be one.
12041 The default is @code{8.2}.
12044 Specify flags influencing the filter process.
12046 Available value for @var{flags} is:
12049 @item scene_change_detect, scd
12050 Enable scene change detection using the value of the option @var{scene}.
12051 This flag is enabled by default.
12057 Select one frame every N-th frame.
12059 This filter accepts the following option:
12062 Select frame after every @code{step} frames.
12063 Allowed values are positive integers higher than 0. Default value is @code{1}.
12066 @section freezedetect
12068 Detect frozen video.
12070 This filter logs a message and sets frame metadata when it detects that the
12071 input video has no significant change in content during a specified duration.
12072 Video freeze detection calculates the mean average absolute difference of all
12073 the components of video frames and compares it to a noise floor.
12075 The printed times and duration are expressed in seconds. The
12076 @code{lavfi.freezedetect.freeze_start} metadata key is set on the first frame
12077 whose timestamp equals or exceeds the detection duration and it contains the
12078 timestamp of the first frame of the freeze. The
12079 @code{lavfi.freezedetect.freeze_duration} and
12080 @code{lavfi.freezedetect.freeze_end} metadata keys are set on the first frame
12083 The filter accepts the following options:
12087 Set noise tolerance. Can be specified in dB (in case "dB" is appended to the
12088 specified value) or as a difference ratio between 0 and 1. Default is -60dB, or
12092 Set freeze duration until notification (default is 2 seconds).
12095 @section freezeframes
12097 Freeze video frames.
12099 This filter freezes video frames using frame from 2nd input.
12101 The filter accepts the following options:
12105 Set number of first frame from which to start freeze.
12108 Set number of last frame from which to end freeze.
12111 Set number of frame from 2nd input which will be used instead of replaced frames.
12117 Apply a frei0r effect to the input video.
12119 To enable the compilation of this filter, you need to install the frei0r
12120 header and configure FFmpeg with @code{--enable-frei0r}.
12122 It accepts the following parameters:
12127 The name of the frei0r effect to load. If the environment variable
12128 @env{FREI0R_PATH} is defined, the frei0r effect is searched for in each of the
12129 directories specified by the colon-separated list in @env{FREI0R_PATH}.
12130 Otherwise, the standard frei0r paths are searched, in this order:
12131 @file{HOME/.frei0r-1/lib/}, @file{/usr/local/lib/frei0r-1/},
12132 @file{/usr/lib/frei0r-1/}.
12134 @item filter_params
12135 A '|'-separated list of parameters to pass to the frei0r effect.
12139 A frei0r effect parameter can be a boolean (its value is either
12140 "y" or "n"), a double, a color (specified as
12141 @var{R}/@var{G}/@var{B}, where @var{R}, @var{G}, and @var{B} are floating point
12142 numbers between 0.0 and 1.0, inclusive) or a color description as specified in the
12143 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils},
12144 a position (specified as @var{X}/@var{Y}, where
12145 @var{X} and @var{Y} are floating point numbers) and/or a string.
12147 The number and types of parameters depend on the loaded effect. If an
12148 effect parameter is not specified, the default value is set.
12150 @subsection Examples
12154 Apply the distort0r effect, setting the first two double parameters:
12156 frei0r=filter_name=distort0r:filter_params=0.5|0.01
12160 Apply the colordistance effect, taking a color as the first parameter:
12162 frei0r=colordistance:0.2/0.3/0.4
12163 frei0r=colordistance:violet
12164 frei0r=colordistance:0x112233
12168 Apply the perspective effect, specifying the top left and top right image
12171 frei0r=perspective:0.2/0.2|0.8/0.2
12175 For more information, see
12176 @url{http://frei0r.dyne.org}
12178 @subsection Commands
12180 This filter supports the @option{filter_params} option as @ref{commands}.
12184 Apply fast and simple postprocessing. It is a faster version of @ref{spp}.
12186 It splits (I)DCT into horizontal/vertical passes. Unlike the simple post-
12187 processing filter, one of them is performed once per block, not per pixel.
12188 This allows for much higher speed.
12190 The filter accepts the following options:
12194 Set quality. This option defines the number of levels for averaging. It accepts
12195 an integer in the range 4-5. Default value is @code{4}.
12198 Force a constant quantization parameter. It accepts an integer in range 0-63.
12199 If not set, the filter will use the QP from the video stream (if available).
12202 Set filter strength. It accepts an integer in range -15 to 32. Lower values mean
12203 more details but also more artifacts, while higher values make the image smoother
12204 but also blurrier. Default value is @code{0} − PSNR optimal.
12206 @item use_bframe_qp
12207 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
12208 option may cause flicker since the B-Frames have often larger QP. Default is
12209 @code{0} (not enabled).
12215 Apply Gaussian blur filter.
12217 The filter accepts the following options:
12221 Set horizontal sigma, standard deviation of Gaussian blur. Default is @code{0.5}.
12224 Set number of steps for Gaussian approximation. Default is @code{1}.
12227 Set which planes to filter. By default all planes are filtered.
12230 Set vertical sigma, if negative it will be same as @code{sigma}.
12231 Default is @code{-1}.
12234 @subsection Commands
12235 This filter supports same commands as options.
12236 The command accepts the same syntax of the corresponding option.
12238 If the specified expression is not valid, it is kept at its current
12243 Apply generic equation to each pixel.
12245 The filter accepts the following options:
12248 @item lum_expr, lum
12249 Set the luminance expression.
12251 Set the chrominance blue expression.
12253 Set the chrominance red expression.
12254 @item alpha_expr, a
12255 Set the alpha expression.
12257 Set the red expression.
12258 @item green_expr, g
12259 Set the green expression.
12261 Set the blue expression.
12264 The colorspace is selected according to the specified options. If one
12265 of the @option{lum_expr}, @option{cb_expr}, or @option{cr_expr}
12266 options is specified, the filter will automatically select a YCbCr
12267 colorspace. If one of the @option{red_expr}, @option{green_expr}, or
12268 @option{blue_expr} options is specified, it will select an RGB
12271 If one of the chrominance expression is not defined, it falls back on the other
12272 one. If no alpha expression is specified it will evaluate to opaque value.
12273 If none of chrominance expressions are specified, they will evaluate
12274 to the luminance expression.
12276 The expressions can use the following variables and functions:
12280 The sequential number of the filtered frame, starting from @code{0}.
12284 The coordinates of the current sample.
12288 The width and height of the image.
12292 Width and height scale depending on the currently filtered plane. It is the
12293 ratio between the corresponding luma plane number of pixels and the current
12294 plane ones. E.g. for YUV4:2:0 the values are @code{1,1} for the luma plane, and
12295 @code{0.5,0.5} for chroma planes.
12298 Time of the current frame, expressed in seconds.
12301 Return the value of the pixel at location (@var{x},@var{y}) of the current
12305 Return the value of the pixel at location (@var{x},@var{y}) of the luminance
12309 Return the value of the pixel at location (@var{x},@var{y}) of the
12310 blue-difference chroma plane. Return 0 if there is no such plane.
12313 Return the value of the pixel at location (@var{x},@var{y}) of the
12314 red-difference chroma plane. Return 0 if there is no such plane.
12319 Return the value of the pixel at location (@var{x},@var{y}) of the
12320 red/green/blue component. Return 0 if there is no such component.
12323 Return the value of the pixel at location (@var{x},@var{y}) of the alpha
12324 plane. Return 0 if there is no such plane.
12326 @item psum(x,y), lumsum(x, y), cbsum(x,y), crsum(x,y), rsum(x,y), gsum(x,y), bsum(x,y), alphasum(x,y)
12327 Sum of sample values in the rectangle from (0,0) to (x,y), this allows obtaining
12328 sums of samples within a rectangle. See the functions without the sum postfix.
12330 @item interpolation
12331 Set one of interpolation methods:
12336 Default is bilinear.
12339 For functions, if @var{x} and @var{y} are outside the area, the value will be
12340 automatically clipped to the closer edge.
12342 Please note that this filter can use multiple threads in which case each slice
12343 will have its own expression state. If you want to use only a single expression
12344 state because your expressions depend on previous state then you should limit
12345 the number of filter threads to 1.
12347 @subsection Examples
12351 Flip the image horizontally:
12357 Generate a bidimensional sine wave, with angle @code{PI/3} and a
12358 wavelength of 100 pixels:
12360 geq=128 + 100*sin(2*(PI/100)*(cos(PI/3)*(X-50*T) + sin(PI/3)*Y)):128:128
12364 Generate a fancy enigmatic moving light:
12366 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
12370 Generate a quick emboss effect:
12372 format=gray,geq=lum_expr='(p(X,Y)+(256-p(X-4,Y-4)))/2'
12376 Modify RGB components depending on pixel position:
12378 geq=r='X/W*r(X,Y)':g='(1-X/W)*g(X,Y)':b='(H-Y)/H*b(X,Y)'
12382 Create a radial gradient that is the same size as the input (also see
12383 the @ref{vignette} filter):
12385 geq=lum=255*gauss((X/W-0.5)*3)*gauss((Y/H-0.5)*3)/gauss(0)/gauss(0),format=gray
12391 Fix the banding artifacts that are sometimes introduced into nearly flat
12392 regions by truncation to 8-bit color depth.
12393 Interpolate the gradients that should go where the bands are, and
12396 It is designed for playback only. Do not use it prior to
12397 lossy compression, because compression tends to lose the dither and
12398 bring back the bands.
12400 It accepts the following parameters:
12405 The maximum amount by which the filter will change any one pixel. This is also
12406 the threshold for detecting nearly flat regions. Acceptable values range from
12407 .51 to 64; the default value is 1.2. Out-of-range values will be clipped to the
12411 The neighborhood to fit the gradient to. A larger radius makes for smoother
12412 gradients, but also prevents the filter from modifying the pixels near detailed
12413 regions. Acceptable values are 8-32; the default value is 16. Out-of-range
12414 values will be clipped to the valid range.
12418 Alternatively, the options can be specified as a flat string:
12419 @var{strength}[:@var{radius}]
12421 @subsection Examples
12425 Apply the filter with a @code{3.5} strength and radius of @code{8}:
12431 Specify radius, omitting the strength (which will fall-back to the default
12439 @anchor{graphmonitor}
12440 @section graphmonitor
12441 Show various filtergraph stats.
12443 With this filter one can debug complete filtergraph.
12444 Especially issues with links filling with queued frames.
12446 The filter accepts the following options:
12450 Set video output size. Default is @var{hd720}.
12453 Set video opacity. Default is @var{0.9}. Allowed range is from @var{0} to @var{1}.
12456 Set output mode, can be @var{fulll} or @var{compact}.
12457 In @var{compact} mode only filters with some queued frames have displayed stats.
12460 Set flags which enable which stats are shown in video.
12462 Available values for flags are:
12465 Display number of queued frames in each link.
12467 @item frame_count_in
12468 Display number of frames taken from filter.
12470 @item frame_count_out
12471 Display number of frames given out from filter.
12474 Display current filtered frame pts.
12477 Display current filtered frame time.
12480 Display time base for filter link.
12483 Display used format for filter link.
12486 Display video size or number of audio channels in case of audio used by filter link.
12489 Display video frame rate or sample rate in case of audio used by filter link.
12492 Display link output status.
12496 Set upper limit for video rate of output stream, Default value is @var{25}.
12497 This guarantee that output video frame rate will not be higher than this value.
12501 A color constancy variation filter which estimates scene illumination via grey edge algorithm
12502 and corrects the scene colors accordingly.
12504 See: @url{https://staff.science.uva.nl/th.gevers/pub/GeversTIP07.pdf}
12506 The filter accepts the following options:
12510 The order of differentiation to be applied on the scene. Must be chosen in the range
12511 [0,2] and default value is 1.
12514 The Minkowski parameter to be used for calculating the Minkowski distance. Must
12515 be chosen in the range [0,20] and default value is 1. Set to 0 for getting
12516 max value instead of calculating Minkowski distance.
12519 The standard deviation of Gaussian blur to be applied on the scene. Must be
12520 chosen in the range [0,1024.0] and default value = 1. floor( @var{sigma} * break_off_sigma(3) )
12521 can't be equal to 0 if @var{difford} is greater than 0.
12524 @subsection Examples
12530 greyedge=difford=1:minknorm=5:sigma=2
12536 greyedge=difford=1:minknorm=0:sigma=2
12544 Apply a Hald CLUT to a video stream.
12546 First input is the video stream to process, and second one is the Hald CLUT.
12547 The Hald CLUT input can be a simple picture or a complete video stream.
12549 The filter accepts the following options:
12553 Force termination when the shortest input terminates. Default is @code{0}.
12555 Continue applying the last CLUT after the end of the stream. A value of
12556 @code{0} disable the filter after the last frame of the CLUT is reached.
12557 Default is @code{1}.
12560 @code{haldclut} also has the same interpolation options as @ref{lut3d} (both
12561 filters share the same internals).
12563 This filter also supports the @ref{framesync} options.
12565 More information about the Hald CLUT can be found on Eskil Steenberg's website
12566 (Hald CLUT author) at @url{http://www.quelsolaar.com/technology/clut.html}.
12568 @subsection Workflow examples
12570 @subsubsection Hald CLUT video stream
12572 Generate an identity Hald CLUT stream altered with various effects:
12574 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
12577 Note: make sure you use a lossless codec.
12579 Then use it with @code{haldclut} to apply it on some random stream:
12581 ffmpeg -f lavfi -i mandelbrot -i clut.nut -filter_complex '[0][1] haldclut' -t 20 mandelclut.mkv
12584 The Hald CLUT will be applied to the 10 first seconds (duration of
12585 @file{clut.nut}), then the latest picture of that CLUT stream will be applied
12586 to the remaining frames of the @code{mandelbrot} stream.
12588 @subsubsection Hald CLUT with preview
12590 A Hald CLUT is supposed to be a squared image of @code{Level*Level*Level} by
12591 @code{Level*Level*Level} pixels. For a given Hald CLUT, FFmpeg will select the
12592 biggest possible square starting at the top left of the picture. The remaining
12593 padding pixels (bottom or right) will be ignored. This area can be used to add
12594 a preview of the Hald CLUT.
12596 Typically, the following generated Hald CLUT will be supported by the
12597 @code{haldclut} filter:
12600 ffmpeg -f lavfi -i @ref{haldclutsrc}=8 -vf "
12601 pad=iw+320 [padded_clut];
12602 smptebars=s=320x256, split [a][b];
12603 [padded_clut][a] overlay=W-320:h, curves=color_negative [main];
12604 [main][b] overlay=W-320" -frames:v 1 clut.png
12607 It contains the original and a preview of the effect of the CLUT: SMPTE color
12608 bars are displayed on the right-top, and below the same color bars processed by
12611 Then, the effect of this Hald CLUT can be visualized with:
12613 ffplay input.mkv -vf "movie=clut.png, [in] haldclut"
12618 Flip the input video horizontally.
12620 For example, to horizontally flip the input video with @command{ffmpeg}:
12622 ffmpeg -i in.avi -vf "hflip" out.avi
12626 This filter applies a global color histogram equalization on a
12629 It can be used to correct video that has a compressed range of pixel
12630 intensities. The filter redistributes the pixel intensities to
12631 equalize their distribution across the intensity range. It may be
12632 viewed as an "automatically adjusting contrast filter". This filter is
12633 useful only for correcting degraded or poorly captured source
12636 The filter accepts the following options:
12640 Determine the amount of equalization to be applied. As the strength
12641 is reduced, the distribution of pixel intensities more-and-more
12642 approaches that of the input frame. The value must be a float number
12643 in the range [0,1] and defaults to 0.200.
12646 Set the maximum intensity that can generated and scale the output
12647 values appropriately. The strength should be set as desired and then
12648 the intensity can be limited if needed to avoid washing-out. The value
12649 must be a float number in the range [0,1] and defaults to 0.210.
12652 Set the antibanding level. If enabled the filter will randomly vary
12653 the luminance of output pixels by a small amount to avoid banding of
12654 the histogram. Possible values are @code{none}, @code{weak} or
12655 @code{strong}. It defaults to @code{none}.
12661 Compute and draw a color distribution histogram for the input video.
12663 The computed histogram is a representation of the color component
12664 distribution in an image.
12666 Standard histogram displays the color components distribution in an image.
12667 Displays color graph for each color component. Shows distribution of
12668 the Y, U, V, A or R, G, B components, depending on input format, in the
12669 current frame. Below each graph a color component scale meter is shown.
12671 The filter accepts the following options:
12675 Set height of level. Default value is @code{200}.
12676 Allowed range is [50, 2048].
12679 Set height of color scale. Default value is @code{12}.
12680 Allowed range is [0, 40].
12684 It accepts the following values:
12687 Per color component graphs are placed below each other.
12690 Per color component graphs are placed side by side.
12693 Presents information identical to that in the @code{parade}, except
12694 that the graphs representing color components are superimposed directly
12697 Default is @code{stack}.
12700 Set mode. Can be either @code{linear}, or @code{logarithmic}.
12701 Default is @code{linear}.
12704 Set what color components to display.
12705 Default is @code{7}.
12708 Set foreground opacity. Default is @code{0.7}.
12711 Set background opacity. Default is @code{0.5}.
12714 @subsection Examples
12719 Calculate and draw histogram:
12721 ffplay -i input -vf histogram
12729 This is a high precision/quality 3d denoise filter. It aims to reduce
12730 image noise, producing smooth images and making still images really
12731 still. It should enhance compressibility.
12733 It accepts the following optional parameters:
12737 A non-negative floating point number which specifies spatial luma strength.
12738 It defaults to 4.0.
12740 @item chroma_spatial
12741 A non-negative floating point number which specifies spatial chroma strength.
12742 It defaults to 3.0*@var{luma_spatial}/4.0.
12745 A floating point number which specifies luma temporal strength. It defaults to
12746 6.0*@var{luma_spatial}/4.0.
12749 A floating point number which specifies chroma temporal strength. It defaults to
12750 @var{luma_tmp}*@var{chroma_spatial}/@var{luma_spatial}.
12753 @subsection Commands
12754 This filter supports same @ref{commands} as options.
12755 The command accepts the same syntax of the corresponding option.
12757 If the specified expression is not valid, it is kept at its current
12760 @anchor{hwdownload}
12761 @section hwdownload
12763 Download hardware frames to system memory.
12765 The input must be in hardware frames, and the output a non-hardware format.
12766 Not all formats will be supported on the output - it may be necessary to insert
12767 an additional @option{format} filter immediately following in the graph to get
12768 the output in a supported format.
12772 Map hardware frames to system memory or to another device.
12774 This filter has several different modes of operation; which one is used depends
12775 on the input and output formats:
12778 Hardware frame input, normal frame output
12780 Map the input frames to system memory and pass them to the output. If the
12781 original hardware frame is later required (for example, after overlaying
12782 something else on part of it), the @option{hwmap} filter can be used again
12783 in the next mode to retrieve it.
12785 Normal frame input, hardware frame output
12787 If the input is actually a software-mapped hardware frame, then unmap it -
12788 that is, return the original hardware frame.
12790 Otherwise, a device must be provided. Create new hardware surfaces on that
12791 device for the output, then map them back to the software format at the input
12792 and give those frames to the preceding filter. This will then act like the
12793 @option{hwupload} filter, but may be able to avoid an additional copy when
12794 the input is already in a compatible format.
12796 Hardware frame input and output
12798 A device must be supplied for the output, either directly or with the
12799 @option{derive_device} option. The input and output devices must be of
12800 different types and compatible - the exact meaning of this is
12801 system-dependent, but typically it means that they must refer to the same
12802 underlying hardware context (for example, refer to the same graphics card).
12804 If the input frames were originally created on the output device, then unmap
12805 to retrieve the original frames.
12807 Otherwise, map the frames to the output device - create new hardware frames
12808 on the output corresponding to the frames on the input.
12811 The following additional parameters are accepted:
12815 Set the frame mapping mode. Some combination of:
12818 The mapped frame should be readable.
12820 The mapped frame should be writeable.
12822 The mapping will always overwrite the entire frame.
12824 This may improve performance in some cases, as the original contents of the
12825 frame need not be loaded.
12827 The mapping must not involve any copying.
12829 Indirect mappings to copies of frames are created in some cases where either
12830 direct mapping is not possible or it would have unexpected properties.
12831 Setting this flag ensures that the mapping is direct and will fail if that is
12834 Defaults to @var{read+write} if not specified.
12836 @item derive_device @var{type}
12837 Rather than using the device supplied at initialisation, instead derive a new
12838 device of type @var{type} from the device the input frames exist on.
12841 In a hardware to hardware mapping, map in reverse - create frames in the sink
12842 and map them back to the source. This may be necessary in some cases where
12843 a mapping in one direction is required but only the opposite direction is
12844 supported by the devices being used.
12846 This option is dangerous - it may break the preceding filter in undefined
12847 ways if there are any additional constraints on that filter's output.
12848 Do not use it without fully understanding the implications of its use.
12854 Upload system memory frames to hardware surfaces.
12856 The device to upload to must be supplied when the filter is initialised. If
12857 using ffmpeg, select the appropriate device with the @option{-filter_hw_device}
12858 option or with the @option{derive_device} option. The input and output devices
12859 must be of different types and compatible - the exact meaning of this is
12860 system-dependent, but typically it means that they must refer to the same
12861 underlying hardware context (for example, refer to the same graphics card).
12863 The following additional parameters are accepted:
12866 @item derive_device @var{type}
12867 Rather than using the device supplied at initialisation, instead derive a new
12868 device of type @var{type} from the device the input frames exist on.
12871 @anchor{hwupload_cuda}
12872 @section hwupload_cuda
12874 Upload system memory frames to a CUDA device.
12876 It accepts the following optional parameters:
12880 The number of the CUDA device to use
12885 Apply a high-quality magnification filter designed for pixel art. This filter
12886 was originally created by Maxim Stepin.
12888 It accepts the following option:
12892 Set the scaling dimension: @code{2} for @code{hq2x}, @code{3} for
12893 @code{hq3x} and @code{4} for @code{hq4x}.
12894 Default is @code{3}.
12898 Stack input videos horizontally.
12900 All streams must be of same pixel format and of same height.
12902 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
12903 to create same output.
12905 The filter accepts the following option:
12909 Set number of input streams. Default is 2.
12912 If set to 1, force the output to terminate when the shortest input
12913 terminates. Default value is 0.
12918 Modify the hue and/or the saturation of the input.
12920 It accepts the following parameters:
12924 Specify the hue angle as a number of degrees. It accepts an expression,
12925 and defaults to "0".
12928 Specify the saturation in the [-10,10] range. It accepts an expression and
12932 Specify the hue angle as a number of radians. It accepts an
12933 expression, and defaults to "0".
12936 Specify the brightness in the [-10,10] range. It accepts an expression and
12940 @option{h} and @option{H} are mutually exclusive, and can't be
12941 specified at the same time.
12943 The @option{b}, @option{h}, @option{H} and @option{s} option values are
12944 expressions containing the following constants:
12948 frame count of the input frame starting from 0
12951 presentation timestamp of the input frame expressed in time base units
12954 frame rate of the input video, NAN if the input frame rate is unknown
12957 timestamp expressed in seconds, NAN if the input timestamp is unknown
12960 time base of the input video
12963 @subsection Examples
12967 Set the hue to 90 degrees and the saturation to 1.0:
12973 Same command but expressing the hue in radians:
12979 Rotate hue and make the saturation swing between 0
12980 and 2 over a period of 1 second:
12982 hue="H=2*PI*t: s=sin(2*PI*t)+1"
12986 Apply a 3 seconds saturation fade-in effect starting at 0:
12988 hue="s=min(t/3\,1)"
12991 The general fade-in expression can be written as:
12993 hue="s=min(0\, max((t-START)/DURATION\, 1))"
12997 Apply a 3 seconds saturation fade-out effect starting at 5 seconds:
12999 hue="s=max(0\, min(1\, (8-t)/3))"
13002 The general fade-out expression can be written as:
13004 hue="s=max(0\, min(1\, (START+DURATION-t)/DURATION))"
13009 @subsection Commands
13011 This filter supports the following commands:
13017 Modify the hue and/or the saturation and/or brightness of the input video.
13018 The command accepts the same syntax of the corresponding option.
13020 If the specified expression is not valid, it is kept at its current
13024 @section hysteresis
13026 Grow first stream into second stream by connecting components.
13027 This makes it possible to build more robust edge masks.
13029 This filter accepts the following options:
13033 Set which planes will be processed as bitmap, unprocessed planes will be
13034 copied from first stream.
13035 By default value 0xf, all planes will be processed.
13038 Set threshold which is used in filtering. If pixel component value is higher than
13039 this value filter algorithm for connecting components is activated.
13040 By default value is 0.
13043 The @code{hysteresis} filter also supports the @ref{framesync} options.
13047 Detect video interlacing type.
13049 This filter tries to detect if the input frames are interlaced, progressive,
13050 top or bottom field first. It will also try to detect fields that are
13051 repeated between adjacent frames (a sign of telecine).
13053 Single frame detection considers only immediately adjacent frames when classifying each frame.
13054 Multiple frame detection incorporates the classification history of previous frames.
13056 The filter will log these metadata values:
13059 @item single.current_frame
13060 Detected type of current frame using single-frame detection. One of:
13061 ``tff'' (top field first), ``bff'' (bottom field first),
13062 ``progressive'', or ``undetermined''
13065 Cumulative number of frames detected as top field first using single-frame detection.
13068 Cumulative number of frames detected as top field first using multiple-frame detection.
13071 Cumulative number of frames detected as bottom field first using single-frame detection.
13073 @item multiple.current_frame
13074 Detected type of current frame using multiple-frame detection. One of:
13075 ``tff'' (top field first), ``bff'' (bottom field first),
13076 ``progressive'', or ``undetermined''
13079 Cumulative number of frames detected as bottom field first using multiple-frame detection.
13081 @item single.progressive
13082 Cumulative number of frames detected as progressive using single-frame detection.
13084 @item multiple.progressive
13085 Cumulative number of frames detected as progressive using multiple-frame detection.
13087 @item single.undetermined
13088 Cumulative number of frames that could not be classified using single-frame detection.
13090 @item multiple.undetermined
13091 Cumulative number of frames that could not be classified using multiple-frame detection.
13093 @item repeated.current_frame
13094 Which field in the current frame is repeated from the last. One of ``neither'', ``top'', or ``bottom''.
13096 @item repeated.neither
13097 Cumulative number of frames with no repeated field.
13100 Cumulative number of frames with the top field repeated from the previous frame's top field.
13102 @item repeated.bottom
13103 Cumulative number of frames with the bottom field repeated from the previous frame's bottom field.
13106 The filter accepts the following options:
13110 Set interlacing threshold.
13112 Set progressive threshold.
13114 Threshold for repeated field detection.
13116 Number of frames after which a given frame's contribution to the
13117 statistics is halved (i.e., it contributes only 0.5 to its
13118 classification). The default of 0 means that all frames seen are given
13119 full weight of 1.0 forever.
13120 @item analyze_interlaced_flag
13121 When this is not 0 then idet will use the specified number of frames to determine
13122 if the interlaced flag is accurate, it will not count undetermined frames.
13123 If the flag is found to be accurate it will be used without any further
13124 computations, if it is found to be inaccurate it will be cleared without any
13125 further computations. This allows inserting the idet filter as a low computational
13126 method to clean up the interlaced flag
13131 Deinterleave or interleave fields.
13133 This filter allows one to process interlaced images fields without
13134 deinterlacing them. Deinterleaving splits the input frame into 2
13135 fields (so called half pictures). Odd lines are moved to the top
13136 half of the output image, even lines to the bottom half.
13137 You can process (filter) them independently and then re-interleave them.
13139 The filter accepts the following options:
13143 @item chroma_mode, c
13144 @item alpha_mode, a
13145 Available values for @var{luma_mode}, @var{chroma_mode} and
13146 @var{alpha_mode} are:
13152 @item deinterleave, d
13153 Deinterleave fields, placing one above the other.
13155 @item interleave, i
13156 Interleave fields. Reverse the effect of deinterleaving.
13158 Default value is @code{none}.
13160 @item luma_swap, ls
13161 @item chroma_swap, cs
13162 @item alpha_swap, as
13163 Swap luma/chroma/alpha fields. Exchange even & odd lines. Default value is @code{0}.
13166 @subsection Commands
13168 This filter supports the all above options as @ref{commands}.
13172 Apply inflate effect to the video.
13174 This filter replaces the pixel by the local(3x3) average by taking into account
13175 only values higher than the pixel.
13177 It accepts the following options:
13184 Limit the maximum change for each plane, default is 65535.
13185 If 0, plane will remain unchanged.
13188 @subsection Commands
13190 This filter supports the all above options as @ref{commands}.
13194 Simple interlacing filter from progressive contents. This interleaves upper (or
13195 lower) lines from odd frames with lower (or upper) lines from even frames,
13196 halving the frame rate and preserving image height.
13199 Original Original New Frame
13200 Frame 'j' Frame 'j+1' (tff)
13201 ========== =========== ==================
13202 Line 0 --------------------> Frame 'j' Line 0
13203 Line 1 Line 1 ----> Frame 'j+1' Line 1
13204 Line 2 ---------------------> Frame 'j' Line 2
13205 Line 3 Line 3 ----> Frame 'j+1' Line 3
13207 New Frame + 1 will be generated by Frame 'j+2' and Frame 'j+3' and so on
13210 It accepts the following optional parameters:
13214 This determines whether the interlaced frame is taken from the even
13215 (tff - default) or odd (bff) lines of the progressive frame.
13218 Vertical lowpass filter to avoid twitter interlacing and
13219 reduce moire patterns.
13223 Disable vertical lowpass filter
13226 Enable linear filter (default)
13229 Enable complex filter. This will slightly less reduce twitter and moire
13230 but better retain detail and subjective sharpness impression.
13237 Deinterlace input video by applying Donald Graft's adaptive kernel
13238 deinterling. Work on interlaced parts of a video to produce
13239 progressive frames.
13241 The description of the accepted parameters follows.
13245 Set the threshold which affects the filter's tolerance when
13246 determining if a pixel line must be processed. It must be an integer
13247 in the range [0,255] and defaults to 10. A value of 0 will result in
13248 applying the process on every pixels.
13251 Paint pixels exceeding the threshold value to white if set to 1.
13255 Set the fields order. Swap fields if set to 1, leave fields alone if
13259 Enable additional sharpening if set to 1. Default is 0.
13262 Enable twoway sharpening if set to 1. Default is 0.
13265 @subsection Examples
13269 Apply default values:
13271 kerndeint=thresh=10:map=0:order=0:sharp=0:twoway=0
13275 Enable additional sharpening:
13281 Paint processed pixels in white:
13289 Slowly update darker pixels.
13291 This filter makes short flashes of light appear longer.
13292 This filter accepts the following options:
13296 Set factor for decaying. Default is .95. Allowed range is from 0 to 1.
13299 Set which planes to filter. Default is all. Allowed range is from 0 to 15.
13302 @section lenscorrection
13304 Correct radial lens distortion
13306 This filter can be used to correct for radial distortion as can result from the use
13307 of wide angle lenses, and thereby re-rectify the image. To find the right parameters
13308 one can use tools available for example as part of opencv or simply trial-and-error.
13309 To use opencv use the calibration sample (under samples/cpp) from the opencv sources
13310 and extract the k1 and k2 coefficients from the resulting matrix.
13312 Note that effectively the same filter is available in the open-source tools Krita and
13313 Digikam from the KDE project.
13315 In contrast to the @ref{vignette} filter, which can also be used to compensate lens errors,
13316 this filter corrects the distortion of the image, whereas @ref{vignette} corrects the
13317 brightness distribution, so you may want to use both filters together in certain
13318 cases, though you will have to take care of ordering, i.e. whether vignetting should
13319 be applied before or after lens correction.
13321 @subsection Options
13323 The filter accepts the following options:
13327 Relative x-coordinate of the focal point of the image, and thereby the center of the
13328 distortion. This value has a range [0,1] and is expressed as fractions of the image
13329 width. Default is 0.5.
13331 Relative y-coordinate of the focal point of the image, and thereby the center of the
13332 distortion. This value has a range [0,1] and is expressed as fractions of the image
13333 height. Default is 0.5.
13335 Coefficient of the quadratic correction term. This value has a range [-1,1]. 0 means
13336 no correction. Default is 0.
13338 Coefficient of the double quadratic correction term. This value has a range [-1,1].
13339 0 means no correction. Default is 0.
13342 The formula that generates the correction is:
13344 @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)
13346 where @var{r_0} is halve of the image diagonal and @var{r_src} and @var{r_tgt} are the
13347 distances from the focal point in the source and target images, respectively.
13351 Apply lens correction via the lensfun library (@url{http://lensfun.sourceforge.net/}).
13353 The @code{lensfun} filter requires the camera make, camera model, and lens model
13354 to apply the lens correction. The filter will load the lensfun database and
13355 query it to find the corresponding camera and lens entries in the database. As
13356 long as these entries can be found with the given options, the filter can
13357 perform corrections on frames. Note that incomplete strings will result in the
13358 filter choosing the best match with the given options, and the filter will
13359 output the chosen camera and lens models (logged with level "info"). You must
13360 provide the make, camera model, and lens model as they are required.
13362 The filter accepts the following options:
13366 The make of the camera (for example, "Canon"). This option is required.
13369 The model of the camera (for example, "Canon EOS 100D"). This option is
13373 The model of the lens (for example, "Canon EF-S 18-55mm f/3.5-5.6 IS STM"). This
13374 option is required.
13377 The type of correction to apply. The following values are valid options:
13381 Enables fixing lens vignetting.
13384 Enables fixing lens geometry. This is the default.
13387 Enables fixing chromatic aberrations.
13390 Enables fixing lens vignetting and lens geometry.
13393 Enables fixing lens vignetting and chromatic aberrations.
13396 Enables fixing both lens geometry and chromatic aberrations.
13399 Enables all possible corrections.
13403 The focal length of the image/video (zoom; expected constant for video). For
13404 example, a 18--55mm lens has focal length range of [18--55], so a value in that
13405 range should be chosen when using that lens. Default 18.
13408 The aperture of the image/video (expected constant for video). Note that
13409 aperture is only used for vignetting correction. Default 3.5.
13411 @item focus_distance
13412 The focus distance of the image/video (expected constant for video). Note that
13413 focus distance is only used for vignetting and only slightly affects the
13414 vignetting correction process. If unknown, leave it at the default value (which
13418 The scale factor which is applied after transformation. After correction the
13419 video is no longer necessarily rectangular. This parameter controls how much of
13420 the resulting image is visible. The value 0 means that a value will be chosen
13421 automatically such that there is little or no unmapped area in the output
13422 image. 1.0 means that no additional scaling is done. Lower values may result
13423 in more of the corrected image being visible, while higher values may avoid
13424 unmapped areas in the output.
13426 @item target_geometry
13427 The target geometry of the output image/video. The following values are valid
13431 @item rectilinear (default)
13434 @item equirectangular
13435 @item fisheye_orthographic
13436 @item fisheye_stereographic
13437 @item fisheye_equisolid
13438 @item fisheye_thoby
13441 Apply the reverse of image correction (instead of correcting distortion, apply
13444 @item interpolation
13445 The type of interpolation used when correcting distortion. The following values
13450 @item linear (default)
13455 @subsection Examples
13459 Apply lens correction with make "Canon", camera model "Canon EOS 100D", and lens
13460 model "Canon EF-S 18-55mm f/3.5-5.6 IS STM" with focal length of "18" and
13464 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
13468 Apply the same as before, but only for the first 5 seconds of video.
13471 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
13478 Obtain the VMAF (Video Multi-Method Assessment Fusion)
13479 score between two input videos.
13481 The obtained VMAF score is printed through the logging system.
13483 It requires Netflix's vmaf library (libvmaf) as a pre-requisite.
13484 After installing the library it can be enabled using:
13485 @code{./configure --enable-libvmaf}.
13486 If no model path is specified it uses the default model: @code{vmaf_v0.6.1.pkl}.
13488 The filter has following options:
13492 Set the model path which is to be used for SVM.
13493 Default value: @code{"/usr/local/share/model/vmaf_v0.6.1.pkl"}
13496 Set the file path to be used to store logs.
13499 Set the format of the log file (csv, json or xml).
13501 @item enable_transform
13502 This option can enable/disable the @code{score_transform} applied to the final predicted VMAF score,
13503 if you have specified score_transform option in the input parameter file passed to @code{run_vmaf_training.py}
13504 Default value: @code{false}
13507 Invokes the phone model which will generate VMAF scores higher than in the
13508 regular model, which is more suitable for laptop, TV, etc. viewing conditions.
13509 Default value: @code{false}
13512 Enables computing psnr along with vmaf.
13513 Default value: @code{false}
13516 Enables computing ssim along with vmaf.
13517 Default value: @code{false}
13520 Enables computing ms_ssim along with vmaf.
13521 Default value: @code{false}
13524 Set the pool method to be used for computing vmaf.
13525 Options are @code{min}, @code{harmonic_mean} or @code{mean} (default).
13528 Set number of threads to be used when computing vmaf.
13529 Default value: @code{0}, which makes use of all available logical processors.
13532 Set interval for frame subsampling used when computing vmaf.
13533 Default value: @code{1}
13535 @item enable_conf_interval
13536 Enables confidence interval.
13537 Default value: @code{false}
13540 This filter also supports the @ref{framesync} options.
13542 @subsection Examples
13545 On the below examples the input file @file{main.mpg} being processed is
13546 compared with the reference file @file{ref.mpg}.
13549 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf -f null -
13553 Example with options:
13555 ffmpeg -i main.mpg -i ref.mpg -lavfi libvmaf="psnr=1:log_fmt=json" -f null -
13559 Example with options and different containers:
13561 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 -
13567 Limits the pixel components values to the specified range [min, max].
13569 The filter accepts the following options:
13573 Lower bound. Defaults to the lowest allowed value for the input.
13576 Upper bound. Defaults to the highest allowed value for the input.
13579 Specify which planes will be processed. Defaults to all available.
13582 @subsection Commands
13584 This filter supports the all above options as @ref{commands}.
13590 The filter accepts the following options:
13594 Set the number of loops. Setting this value to -1 will result in infinite loops.
13598 Set maximal size in number of frames. Default is 0.
13601 Set first frame of loop. Default is 0.
13604 @subsection Examples
13608 Loop single first frame infinitely:
13610 loop=loop=-1:size=1:start=0
13614 Loop single first frame 10 times:
13616 loop=loop=10:size=1:start=0
13620 Loop 10 first frames 5 times:
13622 loop=loop=5:size=10:start=0
13628 Apply a 1D LUT to an input video.
13630 The filter accepts the following options:
13634 Set the 1D LUT file name.
13636 Currently supported formats:
13645 Select interpolation mode.
13647 Available values are:
13651 Use values from the nearest defined point.
13653 Interpolate values using the linear interpolation.
13655 Interpolate values using the cosine interpolation.
13657 Interpolate values using the cubic interpolation.
13659 Interpolate values using the spline interpolation.
13666 Apply a 3D LUT to an input video.
13668 The filter accepts the following options:
13672 Set the 3D LUT file name.
13674 Currently supported formats:
13688 Select interpolation mode.
13690 Available values are:
13694 Use values from the nearest defined point.
13696 Interpolate values using the 8 points defining a cube.
13698 Interpolate values using a tetrahedron.
13704 Turn certain luma values into transparency.
13706 The filter accepts the following options:
13710 Set the luma which will be used as base for transparency.
13711 Default value is @code{0}.
13714 Set the range of luma values to be keyed out.
13715 Default value is @code{0.01}.
13718 Set the range of softness. Default value is @code{0}.
13719 Use this to control gradual transition from zero to full transparency.
13722 @subsection Commands
13723 This filter supports same @ref{commands} as options.
13724 The command accepts the same syntax of the corresponding option.
13726 If the specified expression is not valid, it is kept at its current
13729 @section lut, lutrgb, lutyuv
13731 Compute a look-up table for binding each pixel component input value
13732 to an output value, and apply it to the input video.
13734 @var{lutyuv} applies a lookup table to a YUV input video, @var{lutrgb}
13735 to an RGB input video.
13737 These filters accept the following parameters:
13740 set first pixel component expression
13742 set second pixel component expression
13744 set third pixel component expression
13746 set fourth pixel component expression, corresponds to the alpha component
13749 set red component expression
13751 set green component expression
13753 set blue component expression
13755 alpha component expression
13758 set Y/luminance component expression
13760 set U/Cb component expression
13762 set V/Cr component expression
13765 Each of them specifies the expression to use for computing the lookup table for
13766 the corresponding pixel component values.
13768 The exact component associated to each of the @var{c*} options depends on the
13771 The @var{lut} filter requires either YUV or RGB pixel formats in input,
13772 @var{lutrgb} requires RGB pixel formats in input, and @var{lutyuv} requires YUV.
13774 The expressions can contain the following constants and functions:
13779 The input width and height.
13782 The input value for the pixel component.
13785 The input value, clipped to the @var{minval}-@var{maxval} range.
13788 The maximum value for the pixel component.
13791 The minimum value for the pixel component.
13794 The negated value for the pixel component value, clipped to the
13795 @var{minval}-@var{maxval} range; it corresponds to the expression
13796 "maxval-clipval+minval".
13799 The computed value in @var{val}, clipped to the
13800 @var{minval}-@var{maxval} range.
13802 @item gammaval(gamma)
13803 The computed gamma correction value of the pixel component value,
13804 clipped to the @var{minval}-@var{maxval} range. It corresponds to the
13806 "pow((clipval-minval)/(maxval-minval)\,@var{gamma})*(maxval-minval)+minval"
13810 All expressions default to "val".
13812 @subsection Examples
13816 Negate input video:
13818 lutrgb="r=maxval+minval-val:g=maxval+minval-val:b=maxval+minval-val"
13819 lutyuv="y=maxval+minval-val:u=maxval+minval-val:v=maxval+minval-val"
13822 The above is the same as:
13824 lutrgb="r=negval:g=negval:b=negval"
13825 lutyuv="y=negval:u=negval:v=negval"
13835 Remove chroma components, turning the video into a graytone image:
13837 lutyuv="u=128:v=128"
13841 Apply a luma burning effect:
13847 Remove green and blue components:
13853 Set a constant alpha channel value on input:
13855 format=rgba,lutrgb=a="maxval-minval/2"
13859 Correct luminance gamma by a factor of 0.5:
13861 lutyuv=y=gammaval(0.5)
13865 Discard least significant bits of luma:
13867 lutyuv=y='bitand(val, 128+64+32)'
13871 Technicolor like effect:
13873 lutyuv=u='(val-maxval/2)*2+maxval/2':v='(val-maxval/2)*2+maxval/2'
13877 @section lut2, tlut2
13879 The @code{lut2} filter takes two input streams and outputs one
13882 The @code{tlut2} (time lut2) filter takes two consecutive frames
13883 from one single stream.
13885 This filter accepts the following parameters:
13888 set first pixel component expression
13890 set second pixel component expression
13892 set third pixel component expression
13894 set fourth pixel component expression, corresponds to the alpha component
13897 set output bit depth, only available for @code{lut2} filter. By default is 0,
13898 which means bit depth is automatically picked from first input format.
13901 The @code{lut2} filter also supports the @ref{framesync} options.
13903 Each of them specifies the expression to use for computing the lookup table for
13904 the corresponding pixel component values.
13906 The exact component associated to each of the @var{c*} options depends on the
13909 The expressions can contain the following constants:
13914 The input width and height.
13917 The first input value for the pixel component.
13920 The second input value for the pixel component.
13923 The first input video bit depth.
13926 The second input video bit depth.
13929 All expressions default to "x".
13931 @subsection Examples
13935 Highlight differences between two RGB video streams:
13937 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)'
13941 Highlight differences between two YUV video streams:
13943 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)'
13947 Show max difference between two video streams:
13949 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)))'
13953 @section maskedclamp
13955 Clamp the first input stream with the second input and third input stream.
13957 Returns the value of first stream to be between second input
13958 stream - @code{undershoot} and third input stream + @code{overshoot}.
13960 This filter accepts the following options:
13963 Default value is @code{0}.
13966 Default value is @code{0}.
13969 Set which planes will be processed as bitmap, unprocessed planes will be
13970 copied from first stream.
13971 By default value 0xf, all planes will be processed.
13974 @subsection Commands
13976 This filter supports the all above options as @ref{commands}.
13980 Merge the second and third input stream into output stream using absolute differences
13981 between second input stream and first input stream and absolute difference between
13982 third input stream and first input stream. The picked value will be from second input
13983 stream if second absolute difference is greater than first one or from third input stream
13986 This filter accepts the following options:
13989 Set which planes will be processed as bitmap, unprocessed planes will be
13990 copied from first stream.
13991 By default value 0xf, all planes will be processed.
13994 @section maskedmerge
13996 Merge the first input stream with the second input stream using per pixel
13997 weights in the third input stream.
13999 A value of 0 in the third stream pixel component means that pixel component
14000 from first stream is returned unchanged, while maximum value (eg. 255 for
14001 8-bit videos) means that pixel component from second stream is returned
14002 unchanged. Intermediate values define the amount of merging between both
14003 input stream's pixel components.
14005 This filter accepts the following options:
14008 Set which planes will be processed as bitmap, unprocessed planes will be
14009 copied from first stream.
14010 By default value 0xf, all planes will be processed.
14015 Merge the second and third input stream into output stream using absolute differences
14016 between second input stream and first input stream and absolute difference between
14017 third input stream and first input stream. The picked value will be from second input
14018 stream if second absolute difference is less than first one or from third input stream
14021 This filter accepts the following options:
14024 Set which planes will be processed as bitmap, unprocessed planes will be
14025 copied from first stream.
14026 By default value 0xf, all planes will be processed.
14029 @section maskedthreshold
14030 Pick pixels comparing absolute difference of two video streams with fixed
14033 If absolute difference between pixel component of first and second video
14034 stream is equal or lower than user supplied threshold than pixel component
14035 from first video stream is picked, otherwise pixel component from second
14036 video stream is picked.
14038 This filter accepts the following options:
14041 Set threshold used when picking pixels from absolute difference from two input
14045 Set which planes will be processed as bitmap, unprocessed planes will be
14046 copied from second stream.
14047 By default value 0xf, all planes will be processed.
14051 Create mask from input video.
14053 For example it is useful to create motion masks after @code{tblend} filter.
14055 This filter accepts the following options:
14059 Set low threshold. Any pixel component lower or exact than this value will be set to 0.
14062 Set high threshold. Any pixel component higher than this value will be set to max value
14063 allowed for current pixel format.
14066 Set planes to filter, by default all available planes are filtered.
14069 Fill all frame pixels with this value.
14072 Set max average pixel value for frame. If sum of all pixel components is higher that this
14073 average, output frame will be completely filled with value set by @var{fill} option.
14074 Typically useful for scene changes when used in combination with @code{tblend} filter.
14079 Apply motion-compensation deinterlacing.
14081 It needs one field per frame as input and must thus be used together
14082 with yadif=1/3 or equivalent.
14084 This filter accepts the following options:
14087 Set the deinterlacing mode.
14089 It accepts one of the following values:
14094 use iterative motion estimation
14096 like @samp{slow}, but use multiple reference frames.
14098 Default value is @samp{fast}.
14101 Set the picture field parity assumed for the input video. It must be
14102 one of the following values:
14106 assume top field first
14108 assume bottom field first
14111 Default value is @samp{bff}.
14114 Set per-block quantization parameter (QP) used by the internal
14117 Higher values should result in a smoother motion vector field but less
14118 optimal individual vectors. Default value is 1.
14123 Pick median pixel from certain rectangle defined by radius.
14125 This filter accepts the following options:
14129 Set horizontal radius size. Default value is @code{1}.
14130 Allowed range is integer from 1 to 127.
14133 Set which planes to process. Default is @code{15}, which is all available planes.
14136 Set vertical radius size. Default value is @code{0}.
14137 Allowed range is integer from 0 to 127.
14138 If it is 0, value will be picked from horizontal @code{radius} option.
14141 Set median percentile. Default value is @code{0.5}.
14142 Default value of @code{0.5} will pick always median values, while @code{0} will pick
14143 minimum values, and @code{1} maximum values.
14146 @subsection Commands
14147 This filter supports same @ref{commands} as options.
14148 The command accepts the same syntax of the corresponding option.
14150 If the specified expression is not valid, it is kept at its current
14153 @section mergeplanes
14155 Merge color channel components from several video streams.
14157 The filter accepts up to 4 input streams, and merge selected input
14158 planes to the output video.
14160 This filter accepts the following options:
14163 Set input to output plane mapping. Default is @code{0}.
14165 The mappings is specified as a bitmap. It should be specified as a
14166 hexadecimal number in the form 0xAa[Bb[Cc[Dd]]]. 'Aa' describes the
14167 mapping for the first plane of the output stream. 'A' sets the number of
14168 the input stream to use (from 0 to 3), and 'a' the plane number of the
14169 corresponding input to use (from 0 to 3). The rest of the mappings is
14170 similar, 'Bb' describes the mapping for the output stream second
14171 plane, 'Cc' describes the mapping for the output stream third plane and
14172 'Dd' describes the mapping for the output stream fourth plane.
14175 Set output pixel format. Default is @code{yuva444p}.
14178 @subsection Examples
14182 Merge three gray video streams of same width and height into single video stream:
14184 [a0][a1][a2]mergeplanes=0x001020:yuv444p
14188 Merge 1st yuv444p stream and 2nd gray video stream into yuva444p video stream:
14190 [a0][a1]mergeplanes=0x00010210:yuva444p
14194 Swap Y and A plane in yuva444p stream:
14196 format=yuva444p,mergeplanes=0x03010200:yuva444p
14200 Swap U and V plane in yuv420p stream:
14202 format=yuv420p,mergeplanes=0x000201:yuv420p
14206 Cast a rgb24 clip to yuv444p:
14208 format=rgb24,mergeplanes=0x000102:yuv444p
14214 Estimate and export motion vectors using block matching algorithms.
14215 Motion vectors are stored in frame side data to be used by other filters.
14217 This filter accepts the following options:
14220 Specify the motion estimation method. Accepts one of the following values:
14224 Exhaustive search algorithm.
14226 Three step search algorithm.
14228 Two dimensional logarithmic search algorithm.
14230 New three step search algorithm.
14232 Four step search algorithm.
14234 Diamond search algorithm.
14236 Hexagon-based search algorithm.
14238 Enhanced predictive zonal search algorithm.
14240 Uneven multi-hexagon search algorithm.
14242 Default value is @samp{esa}.
14245 Macroblock size. Default @code{16}.
14248 Search parameter. Default @code{7}.
14251 @section midequalizer
14253 Apply Midway Image Equalization effect using two video streams.
14255 Midway Image Equalization adjusts a pair of images to have the same
14256 histogram, while maintaining their dynamics as much as possible. It's
14257 useful for e.g. matching exposures from a pair of stereo cameras.
14259 This filter has two inputs and one output, which must be of same pixel format, but
14260 may be of different sizes. The output of filter is first input adjusted with
14261 midway histogram of both inputs.
14263 This filter accepts the following option:
14267 Set which planes to process. Default is @code{15}, which is all available planes.
14270 @section minterpolate
14272 Convert the video to specified frame rate using motion interpolation.
14274 This filter accepts the following options:
14277 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}.
14280 Motion interpolation mode. Following values are accepted:
14283 Duplicate previous or next frame for interpolating new ones.
14285 Blend source frames. Interpolated frame is mean of previous and next frames.
14287 Motion compensated interpolation. Following options are effective when this mode is selected:
14291 Motion compensation mode. Following values are accepted:
14294 Overlapped block motion compensation.
14296 Adaptive overlapped block motion compensation. Window weighting coefficients are controlled adaptively according to the reliabilities of the neighboring motion vectors to reduce oversmoothing.
14298 Default mode is @samp{obmc}.
14301 Motion estimation mode. Following values are accepted:
14304 Bidirectional motion estimation. Motion vectors are estimated for each source frame in both forward and backward directions.
14306 Bilateral motion estimation. Motion vectors are estimated directly for interpolated frame.
14308 Default mode is @samp{bilat}.
14311 The algorithm to be used for motion estimation. Following values are accepted:
14314 Exhaustive search algorithm.
14316 Three step search algorithm.
14318 Two dimensional logarithmic search algorithm.
14320 New three step search algorithm.
14322 Four step search algorithm.
14324 Diamond search algorithm.
14326 Hexagon-based search algorithm.
14328 Enhanced predictive zonal search algorithm.
14330 Uneven multi-hexagon search algorithm.
14332 Default algorithm is @samp{epzs}.
14335 Macroblock size. Default @code{16}.
14338 Motion estimation search parameter. Default @code{32}.
14341 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).
14346 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:
14349 Disable scene change detection.
14351 Frame difference. Corresponding pixel values are compared and if it satisfies @var{scd_threshold} scene change is detected.
14353 Default method is @samp{fdiff}.
14355 @item scd_threshold
14356 Scene change detection threshold. Default is @code{10.}.
14361 Mix several video input streams into one video stream.
14363 A description of the accepted options follows.
14367 The number of inputs. If unspecified, it defaults to 2.
14370 Specify weight of each input video stream as sequence.
14371 Each weight is separated by space. If number of weights
14372 is smaller than number of @var{frames} last specified
14373 weight will be used for all remaining unset weights.
14376 Specify scale, if it is set it will be multiplied with sum
14377 of each weight multiplied with pixel values to give final destination
14378 pixel value. By default @var{scale} is auto scaled to sum of weights.
14381 Specify how end of stream is determined.
14384 The duration of the longest input. (default)
14387 The duration of the shortest input.
14390 The duration of the first input.
14394 @section mpdecimate
14396 Drop frames that do not differ greatly from the previous frame in
14397 order to reduce frame rate.
14399 The main use of this filter is for very-low-bitrate encoding
14400 (e.g. streaming over dialup modem), but it could in theory be used for
14401 fixing movies that were inverse-telecined incorrectly.
14403 A description of the accepted options follows.
14407 Set the maximum number of consecutive frames which can be dropped (if
14408 positive), or the minimum interval between dropped frames (if
14409 negative). If the value is 0, the frame is dropped disregarding the
14410 number of previous sequentially dropped frames.
14412 Default value is 0.
14417 Set the dropping threshold values.
14419 Values for @option{hi} and @option{lo} are for 8x8 pixel blocks and
14420 represent actual pixel value differences, so a threshold of 64
14421 corresponds to 1 unit of difference for each pixel, or the same spread
14422 out differently over the block.
14424 A frame is a candidate for dropping if no 8x8 blocks differ by more
14425 than a threshold of @option{hi}, and if no more than @option{frac} blocks (1
14426 meaning the whole image) differ by more than a threshold of @option{lo}.
14428 Default value for @option{hi} is 64*12, default value for @option{lo} is
14429 64*5, and default value for @option{frac} is 0.33.
14435 Negate (invert) the input video.
14437 It accepts the following option:
14442 With value 1, it negates the alpha component, if present. Default value is 0.
14448 Denoise frames using Non-Local Means algorithm.
14450 Each pixel is adjusted by looking for other pixels with similar contexts. This
14451 context similarity is defined by comparing their surrounding patches of size
14452 @option{p}x@option{p}. Patches are searched in an area of @option{r}x@option{r}
14455 Note that the research area defines centers for patches, which means some
14456 patches will be made of pixels outside that research area.
14458 The filter accepts the following options.
14462 Set denoising strength. Default is 1.0. Must be in range [1.0, 30.0].
14465 Set patch size. Default is 7. Must be odd number in range [0, 99].
14468 Same as @option{p} but for chroma planes.
14470 The default value is @var{0} and means automatic.
14473 Set research size. Default is 15. Must be odd number in range [0, 99].
14476 Same as @option{r} but for chroma planes.
14478 The default value is @var{0} and means automatic.
14483 Deinterlace video using neural network edge directed interpolation.
14485 This filter accepts the following options:
14489 Mandatory option, without binary file filter can not work.
14490 Currently file can be found here:
14491 https://github.com/dubhater/vapoursynth-nnedi3/blob/master/src/nnedi3_weights.bin
14494 Set which frames to deinterlace, by default it is @code{all}.
14495 Can be @code{all} or @code{interlaced}.
14498 Set mode of operation.
14500 Can be one of the following:
14504 Use frame flags, both fields.
14506 Use frame flags, single field.
14508 Use top field only.
14510 Use bottom field only.
14512 Use both fields, top first.
14514 Use both fields, bottom first.
14518 Set which planes to process, by default filter process all frames.
14521 Set size of local neighborhood around each pixel, used by the predictor neural
14524 Can be one of the following:
14537 Set the number of neurons in predictor neural network.
14538 Can be one of the following:
14549 Controls the number of different neural network predictions that are blended
14550 together to compute the final output value. Can be @code{fast}, default or
14554 Set which set of weights to use in the predictor.
14555 Can be one of the following:
14559 weights trained to minimize absolute error
14561 weights trained to minimize squared error
14565 Controls whether or not the prescreener neural network is used to decide
14566 which pixels should be processed by the predictor neural network and which
14567 can be handled by simple cubic interpolation.
14568 The prescreener is trained to know whether cubic interpolation will be
14569 sufficient for a pixel or whether it should be predicted by the predictor nn.
14570 The computational complexity of the prescreener nn is much less than that of
14571 the predictor nn. Since most pixels can be handled by cubic interpolation,
14572 using the prescreener generally results in much faster processing.
14573 The prescreener is pretty accurate, so the difference between using it and not
14574 using it is almost always unnoticeable.
14576 Can be one of the following:
14584 Default is @code{new}.
14587 Set various debugging flags.
14592 Force libavfilter not to use any of the specified pixel formats for the
14593 input to the next filter.
14595 It accepts the following parameters:
14599 A '|'-separated list of pixel format names, such as
14600 pix_fmts=yuv420p|monow|rgb24".
14604 @subsection Examples
14608 Force libavfilter to use a format different from @var{yuv420p} for the
14609 input to the vflip filter:
14611 noformat=pix_fmts=yuv420p,vflip
14615 Convert the input video to any of the formats not contained in the list:
14617 noformat=yuv420p|yuv444p|yuv410p
14623 Add noise on video input frame.
14625 The filter accepts the following options:
14633 Set noise seed for specific pixel component or all pixel components in case
14634 of @var{all_seed}. Default value is @code{123457}.
14636 @item all_strength, alls
14637 @item c0_strength, c0s
14638 @item c1_strength, c1s
14639 @item c2_strength, c2s
14640 @item c3_strength, c3s
14641 Set noise strength for specific pixel component or all pixel components in case
14642 @var{all_strength}. Default value is @code{0}. Allowed range is [0, 100].
14644 @item all_flags, allf
14645 @item c0_flags, c0f
14646 @item c1_flags, c1f
14647 @item c2_flags, c2f
14648 @item c3_flags, c3f
14649 Set pixel component flags or set flags for all components if @var{all_flags}.
14650 Available values for component flags are:
14653 averaged temporal noise (smoother)
14655 mix random noise with a (semi)regular pattern
14657 temporal noise (noise pattern changes between frames)
14659 uniform noise (gaussian otherwise)
14663 @subsection Examples
14665 Add temporal and uniform noise to input video:
14667 noise=alls=20:allf=t+u
14672 Normalize RGB video (aka histogram stretching, contrast stretching).
14673 See: https://en.wikipedia.org/wiki/Normalization_(image_processing)
14675 For each channel of each frame, the filter computes the input range and maps
14676 it linearly to the user-specified output range. The output range defaults
14677 to the full dynamic range from pure black to pure white.
14679 Temporal smoothing can be used on the input range to reduce flickering (rapid
14680 changes in brightness) caused when small dark or bright objects enter or leave
14681 the scene. This is similar to the auto-exposure (automatic gain control) on a
14682 video camera, and, like a video camera, it may cause a period of over- or
14683 under-exposure of the video.
14685 The R,G,B channels can be normalized independently, which may cause some
14686 color shifting, or linked together as a single channel, which prevents
14687 color shifting. Linked normalization preserves hue. Independent normalization
14688 does not, so it can be used to remove some color casts. Independent and linked
14689 normalization can be combined in any ratio.
14691 The normalize filter accepts the following options:
14696 Colors which define the output range. The minimum input value is mapped to
14697 the @var{blackpt}. The maximum input value is mapped to the @var{whitept}.
14698 The defaults are black and white respectively. Specifying white for
14699 @var{blackpt} and black for @var{whitept} will give color-inverted,
14700 normalized video. Shades of grey can be used to reduce the dynamic range
14701 (contrast). Specifying saturated colors here can create some interesting
14705 The number of previous frames to use for temporal smoothing. The input range
14706 of each channel is smoothed using a rolling average over the current frame
14707 and the @var{smoothing} previous frames. The default is 0 (no temporal
14711 Controls the ratio of independent (color shifting) channel normalization to
14712 linked (color preserving) normalization. 0.0 is fully linked, 1.0 is fully
14713 independent. Defaults to 1.0 (fully independent).
14716 Overall strength of the filter. 1.0 is full strength. 0.0 is a rather
14717 expensive no-op. Defaults to 1.0 (full strength).
14721 @subsection Commands
14722 This filter supports same @ref{commands} as options, excluding @var{smoothing} option.
14723 The command accepts the same syntax of the corresponding option.
14725 If the specified expression is not valid, it is kept at its current
14728 @subsection Examples
14730 Stretch video contrast to use the full dynamic range, with no temporal
14731 smoothing; may flicker depending on the source content:
14733 normalize=blackpt=black:whitept=white:smoothing=0
14736 As above, but with 50 frames of temporal smoothing; flicker should be
14737 reduced, depending on the source content:
14739 normalize=blackpt=black:whitept=white:smoothing=50
14742 As above, but with hue-preserving linked channel normalization:
14744 normalize=blackpt=black:whitept=white:smoothing=50:independence=0
14747 As above, but with half strength:
14749 normalize=blackpt=black:whitept=white:smoothing=50:independence=0:strength=0.5
14752 Map the darkest input color to red, the brightest input color to cyan:
14754 normalize=blackpt=red:whitept=cyan
14759 Pass the video source unchanged to the output.
14762 Optical Character Recognition
14764 This filter uses Tesseract for optical character recognition. To enable
14765 compilation of this filter, you need to configure FFmpeg with
14766 @code{--enable-libtesseract}.
14768 It accepts the following options:
14772 Set datapath to tesseract data. Default is to use whatever was
14773 set at installation.
14776 Set language, default is "eng".
14779 Set character whitelist.
14782 Set character blacklist.
14785 The filter exports recognized text as the frame metadata @code{lavfi.ocr.text}.
14786 The filter exports confidence of recognized words as the frame metadata @code{lavfi.ocr.confidence}.
14790 Apply a video transform using libopencv.
14792 To enable this filter, install the libopencv library and headers and
14793 configure FFmpeg with @code{--enable-libopencv}.
14795 It accepts the following parameters:
14800 The name of the libopencv filter to apply.
14802 @item filter_params
14803 The parameters to pass to the libopencv filter. If not specified, the default
14804 values are assumed.
14808 Refer to the official libopencv documentation for more precise
14810 @url{http://docs.opencv.org/master/modules/imgproc/doc/filtering.html}
14812 Several libopencv filters are supported; see the following subsections.
14817 Dilate an image by using a specific structuring element.
14818 It corresponds to the libopencv function @code{cvDilate}.
14820 It accepts the parameters: @var{struct_el}|@var{nb_iterations}.
14822 @var{struct_el} represents a structuring element, and has the syntax:
14823 @var{cols}x@var{rows}+@var{anchor_x}x@var{anchor_y}/@var{shape}
14825 @var{cols} and @var{rows} represent the number of columns and rows of
14826 the structuring element, @var{anchor_x} and @var{anchor_y} the anchor
14827 point, and @var{shape} the shape for the structuring element. @var{shape}
14828 must be "rect", "cross", "ellipse", or "custom".
14830 If the value for @var{shape} is "custom", it must be followed by a
14831 string of the form "=@var{filename}". The file with name
14832 @var{filename} is assumed to represent a binary image, with each
14833 printable character corresponding to a bright pixel. When a custom
14834 @var{shape} is used, @var{cols} and @var{rows} are ignored, the number
14835 or columns and rows of the read file are assumed instead.
14837 The default value for @var{struct_el} is "3x3+0x0/rect".
14839 @var{nb_iterations} specifies the number of times the transform is
14840 applied to the image, and defaults to 1.
14844 # Use the default values
14847 # Dilate using a structuring element with a 5x5 cross, iterating two times
14848 ocv=filter_name=dilate:filter_params=5x5+2x2/cross|2
14850 # Read the shape from the file diamond.shape, iterating two times.
14851 # The file diamond.shape may contain a pattern of characters like this
14857 # The specified columns and rows are ignored
14858 # but the anchor point coordinates are not
14859 ocv=dilate:0x0+2x2/custom=diamond.shape|2
14864 Erode an image by using a specific structuring element.
14865 It corresponds to the libopencv function @code{cvErode}.
14867 It accepts the parameters: @var{struct_el}:@var{nb_iterations},
14868 with the same syntax and semantics as the @ref{dilate} filter.
14872 Smooth the input video.
14874 The filter takes the following parameters:
14875 @var{type}|@var{param1}|@var{param2}|@var{param3}|@var{param4}.
14877 @var{type} is the type of smooth filter to apply, and must be one of
14878 the following values: "blur", "blur_no_scale", "median", "gaussian",
14879 or "bilateral". The default value is "gaussian".
14881 The meaning of @var{param1}, @var{param2}, @var{param3}, and @var{param4}
14882 depends on the smooth type. @var{param1} and
14883 @var{param2} accept integer positive values or 0. @var{param3} and
14884 @var{param4} accept floating point values.
14886 The default value for @var{param1} is 3. The default value for the
14887 other parameters is 0.
14889 These parameters correspond to the parameters assigned to the
14890 libopencv function @code{cvSmooth}.
14892 @section oscilloscope
14894 2D Video Oscilloscope.
14896 Useful to measure spatial impulse, step responses, chroma delays, etc.
14898 It accepts the following parameters:
14902 Set scope center x position.
14905 Set scope center y position.
14908 Set scope size, relative to frame diagonal.
14911 Set scope tilt/rotation.
14917 Set trace center x position.
14920 Set trace center y position.
14923 Set trace width, relative to width of frame.
14926 Set trace height, relative to height of frame.
14929 Set which components to trace. By default it traces first three components.
14932 Draw trace grid. By default is enabled.
14935 Draw some statistics. By default is enabled.
14938 Draw scope. By default is enabled.
14941 @subsection Commands
14942 This filter supports same @ref{commands} as options.
14943 The command accepts the same syntax of the corresponding option.
14945 If the specified expression is not valid, it is kept at its current
14948 @subsection Examples
14952 Inspect full first row of video frame.
14954 oscilloscope=x=0.5:y=0:s=1
14958 Inspect full last row of video frame.
14960 oscilloscope=x=0.5:y=1:s=1
14964 Inspect full 5th line of video frame of height 1080.
14966 oscilloscope=x=0.5:y=5/1080:s=1
14970 Inspect full last column of video frame.
14972 oscilloscope=x=1:y=0.5:s=1:t=1
14980 Overlay one video on top of another.
14982 It takes two inputs and has one output. The first input is the "main"
14983 video on which the second input is overlaid.
14985 It accepts the following parameters:
14987 A description of the accepted options follows.
14992 Set the expression for the x and y coordinates of the overlaid video
14993 on the main video. Default value is "0" for both expressions. In case
14994 the expression is invalid, it is set to a huge value (meaning that the
14995 overlay will not be displayed within the output visible area).
14998 See @ref{framesync}.
15001 Set when the expressions for @option{x}, and @option{y} are evaluated.
15003 It accepts the following values:
15006 only evaluate expressions once during the filter initialization or
15007 when a command is processed
15010 evaluate expressions for each incoming frame
15013 Default value is @samp{frame}.
15016 See @ref{framesync}.
15019 Set the format for the output video.
15021 It accepts the following values:
15024 force YUV420 output
15027 force YUV420p10 output
15030 force YUV422 output
15033 force YUV422p10 output
15036 force YUV444 output
15039 force packed RGB output
15042 force planar RGB output
15045 automatically pick format
15048 Default value is @samp{yuv420}.
15051 See @ref{framesync}.
15054 Set format of alpha of the overlaid video, it can be @var{straight} or
15055 @var{premultiplied}. Default is @var{straight}.
15058 The @option{x}, and @option{y} expressions can contain the following
15064 The main input width and height.
15068 The overlay input width and height.
15072 The computed values for @var{x} and @var{y}. They are evaluated for
15077 horizontal and vertical chroma subsample values of the output
15078 format. For example for the pixel format "yuv422p" @var{hsub} is 2 and
15082 the number of input frame, starting from 0
15085 the position in the file of the input frame, NAN if unknown
15088 The timestamp, expressed in seconds. It's NAN if the input timestamp is unknown.
15092 This filter also supports the @ref{framesync} options.
15094 Note that the @var{n}, @var{pos}, @var{t} variables are available only
15095 when evaluation is done @emph{per frame}, and will evaluate to NAN
15096 when @option{eval} is set to @samp{init}.
15098 Be aware that frames are taken from each input video in timestamp
15099 order, hence, if their initial timestamps differ, it is a good idea
15100 to pass the two inputs through a @var{setpts=PTS-STARTPTS} filter to
15101 have them begin in the same zero timestamp, as the example for
15102 the @var{movie} filter does.
15104 You can chain together more overlays but you should test the
15105 efficiency of such approach.
15107 @subsection Commands
15109 This filter supports the following commands:
15113 Modify the x and y of the overlay input.
15114 The command accepts the same syntax of the corresponding option.
15116 If the specified expression is not valid, it is kept at its current
15120 @subsection Examples
15124 Draw the overlay at 10 pixels from the bottom right corner of the main
15127 overlay=main_w-overlay_w-10:main_h-overlay_h-10
15130 Using named options the example above becomes:
15132 overlay=x=main_w-overlay_w-10:y=main_h-overlay_h-10
15136 Insert a transparent PNG logo in the bottom left corner of the input,
15137 using the @command{ffmpeg} tool with the @code{-filter_complex} option:
15139 ffmpeg -i input -i logo -filter_complex 'overlay=10:main_h-overlay_h-10' output
15143 Insert 2 different transparent PNG logos (second logo on bottom
15144 right corner) using the @command{ffmpeg} tool:
15146 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
15150 Add a transparent color layer on top of the main video; @code{WxH}
15151 must specify the size of the main input to the overlay filter:
15153 color=color=red@@.3:size=WxH [over]; [in][over] overlay [out]
15157 Play an original video and a filtered version (here with the deshake
15158 filter) side by side using the @command{ffplay} tool:
15160 ffplay input.avi -vf 'split[a][b]; [a]pad=iw*2:ih[src]; [b]deshake[filt]; [src][filt]overlay=w'
15163 The above command is the same as:
15165 ffplay input.avi -vf 'split[b], pad=iw*2[src], [b]deshake, [src]overlay=w'
15169 Make a sliding overlay appearing from the left to the right top part of the
15170 screen starting since time 2:
15172 overlay=x='if(gte(t,2), -w+(t-2)*20, NAN)':y=0
15176 Compose output by putting two input videos side to side:
15178 ffmpeg -i left.avi -i right.avi -filter_complex "
15179 nullsrc=size=200x100 [background];
15180 [0:v] setpts=PTS-STARTPTS, scale=100x100 [left];
15181 [1:v] setpts=PTS-STARTPTS, scale=100x100 [right];
15182 [background][left] overlay=shortest=1 [background+left];
15183 [background+left][right] overlay=shortest=1:x=100 [left+right]
15188 Mask 10-20 seconds of a video by applying the delogo filter to a section
15190 ffmpeg -i test.avi -codec:v:0 wmv2 -ar 11025 -b:v 9000k
15191 -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]'
15196 Chain several overlays in cascade:
15198 nullsrc=s=200x200 [bg];
15199 testsrc=s=100x100, split=4 [in0][in1][in2][in3];
15200 [in0] lutrgb=r=0, [bg] overlay=0:0 [mid0];
15201 [in1] lutrgb=g=0, [mid0] overlay=100:0 [mid1];
15202 [in2] lutrgb=b=0, [mid1] overlay=0:100 [mid2];
15203 [in3] null, [mid2] overlay=100:100 [out0]
15208 @anchor{overlay_cuda}
15209 @section overlay_cuda
15211 Overlay one video on top of another.
15213 This is the CUDA variant of the @ref{overlay} filter.
15214 It only accepts CUDA frames. The underlying input pixel formats have to match.
15216 It takes two inputs and has one output. The first input is the "main"
15217 video on which the second input is overlaid.
15219 It accepts the following parameters:
15224 Set the x and y coordinates of the overlaid video on the main video.
15225 Default value is "0" for both expressions.
15228 See @ref{framesync}.
15231 See @ref{framesync}.
15234 See @ref{framesync}.
15238 This filter also supports the @ref{framesync} options.
15242 Apply Overcomplete Wavelet denoiser.
15244 The filter accepts the following options:
15250 Larger depth values will denoise lower frequency components more, but
15251 slow down filtering.
15253 Must be an int in the range 8-16, default is @code{8}.
15255 @item luma_strength, ls
15258 Must be a double value in the range 0-1000, default is @code{1.0}.
15260 @item chroma_strength, cs
15261 Set chroma strength.
15263 Must be a double value in the range 0-1000, default is @code{1.0}.
15269 Add paddings to the input image, and place the original input at the
15270 provided @var{x}, @var{y} coordinates.
15272 It accepts the following parameters:
15277 Specify an expression for the size of the output image with the
15278 paddings added. If the value for @var{width} or @var{height} is 0, the
15279 corresponding input size is used for the output.
15281 The @var{width} expression can reference the value set by the
15282 @var{height} expression, and vice versa.
15284 The default value of @var{width} and @var{height} is 0.
15288 Specify the offsets to place the input image at within the padded area,
15289 with respect to the top/left border of the output image.
15291 The @var{x} expression can reference the value set by the @var{y}
15292 expression, and vice versa.
15294 The default value of @var{x} and @var{y} is 0.
15296 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
15297 so the input image is centered on the padded area.
15300 Specify the color of the padded area. For the syntax of this option,
15301 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
15302 manual,ffmpeg-utils}.
15304 The default value of @var{color} is "black".
15307 Specify when to evaluate @var{width}, @var{height}, @var{x} and @var{y} expression.
15309 It accepts the following values:
15313 Only evaluate expressions once during the filter initialization or when
15314 a command is processed.
15317 Evaluate expressions for each incoming frame.
15321 Default value is @samp{init}.
15324 Pad to aspect instead to a resolution.
15328 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
15329 options are expressions containing the following constants:
15334 The input video width and height.
15338 These are the same as @var{in_w} and @var{in_h}.
15342 The output width and height (the size of the padded area), as
15343 specified by the @var{width} and @var{height} expressions.
15347 These are the same as @var{out_w} and @var{out_h}.
15351 The x and y offsets as specified by the @var{x} and @var{y}
15352 expressions, or NAN if not yet specified.
15355 same as @var{iw} / @var{ih}
15358 input sample aspect ratio
15361 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
15365 The horizontal and vertical chroma subsample values. For example for the
15366 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
15369 @subsection Examples
15373 Add paddings with the color "violet" to the input video. The output video
15374 size is 640x480, and the top-left corner of the input video is placed at
15377 pad=640:480:0:40:violet
15380 The example above is equivalent to the following command:
15382 pad=width=640:height=480:x=0:y=40:color=violet
15386 Pad the input to get an output with dimensions increased by 3/2,
15387 and put the input video at the center of the padded area:
15389 pad="3/2*iw:3/2*ih:(ow-iw)/2:(oh-ih)/2"
15393 Pad the input to get a squared output with size equal to the maximum
15394 value between the input width and height, and put the input video at
15395 the center of the padded area:
15397 pad="max(iw\,ih):ow:(ow-iw)/2:(oh-ih)/2"
15401 Pad the input to get a final w/h ratio of 16:9:
15403 pad="ih*16/9:ih:(ow-iw)/2:(oh-ih)/2"
15407 In case of anamorphic video, in order to set the output display aspect
15408 correctly, it is necessary to use @var{sar} in the expression,
15409 according to the relation:
15411 (ih * X / ih) * sar = output_dar
15412 X = output_dar / sar
15415 Thus the previous example needs to be modified to:
15417 pad="ih*16/9/sar:ih:(ow-iw)/2:(oh-ih)/2"
15421 Double the output size and put the input video in the bottom-right
15422 corner of the output padded area:
15424 pad="2*iw:2*ih:ow-iw:oh-ih"
15428 @anchor{palettegen}
15429 @section palettegen
15431 Generate one palette for a whole video stream.
15433 It accepts the following options:
15437 Set the maximum number of colors to quantize in the palette.
15438 Note: the palette will still contain 256 colors; the unused palette entries
15441 @item reserve_transparent
15442 Create a palette of 255 colors maximum and reserve the last one for
15443 transparency. Reserving the transparency color is useful for GIF optimization.
15444 If not set, the maximum of colors in the palette will be 256. You probably want
15445 to disable this option for a standalone image.
15448 @item transparency_color
15449 Set the color that will be used as background for transparency.
15452 Set statistics mode.
15454 It accepts the following values:
15457 Compute full frame histograms.
15459 Compute histograms only for the part that differs from previous frame. This
15460 might be relevant to give more importance to the moving part of your input if
15461 the background is static.
15463 Compute new histogram for each frame.
15466 Default value is @var{full}.
15469 The filter also exports the frame metadata @code{lavfi.color_quant_ratio}
15470 (@code{nb_color_in / nb_color_out}) which you can use to evaluate the degree of
15471 color quantization of the palette. This information is also visible at
15472 @var{info} logging level.
15474 @subsection Examples
15478 Generate a representative palette of a given video using @command{ffmpeg}:
15480 ffmpeg -i input.mkv -vf palettegen palette.png
15484 @section paletteuse
15486 Use a palette to downsample an input video stream.
15488 The filter takes two inputs: one video stream and a palette. The palette must
15489 be a 256 pixels image.
15491 It accepts the following options:
15495 Select dithering mode. Available algorithms are:
15498 Ordered 8x8 bayer dithering (deterministic)
15500 Dithering as defined by Paul Heckbert in 1982 (simple error diffusion).
15501 Note: this dithering is sometimes considered "wrong" and is included as a
15503 @item floyd_steinberg
15504 Floyd and Steingberg dithering (error diffusion)
15506 Frankie Sierra dithering v2 (error diffusion)
15508 Frankie Sierra dithering v2 "Lite" (error diffusion)
15511 Default is @var{sierra2_4a}.
15514 When @var{bayer} dithering is selected, this option defines the scale of the
15515 pattern (how much the crosshatch pattern is visible). A low value means more
15516 visible pattern for less banding, and higher value means less visible pattern
15517 at the cost of more banding.
15519 The option must be an integer value in the range [0,5]. Default is @var{2}.
15522 If set, define the zone to process
15526 Only the changing rectangle will be reprocessed. This is similar to GIF
15527 cropping/offsetting compression mechanism. This option can be useful for speed
15528 if only a part of the image is changing, and has use cases such as limiting the
15529 scope of the error diffusal @option{dither} to the rectangle that bounds the
15530 moving scene (it leads to more deterministic output if the scene doesn't change
15531 much, and as a result less moving noise and better GIF compression).
15534 Default is @var{none}.
15537 Take new palette for each output frame.
15539 @item alpha_threshold
15540 Sets the alpha threshold for transparency. Alpha values above this threshold
15541 will be treated as completely opaque, and values below this threshold will be
15542 treated as completely transparent.
15544 The option must be an integer value in the range [0,255]. Default is @var{128}.
15547 @subsection Examples
15551 Use a palette (generated for example with @ref{palettegen}) to encode a GIF
15552 using @command{ffmpeg}:
15554 ffmpeg -i input.mkv -i palette.png -lavfi paletteuse output.gif
15558 @section perspective
15560 Correct perspective of video not recorded perpendicular to the screen.
15562 A description of the accepted parameters follows.
15573 Set coordinates expression for top left, top right, bottom left and bottom right corners.
15574 Default values are @code{0:0:W:0:0:H:W:H} with which perspective will remain unchanged.
15575 If the @code{sense} option is set to @code{source}, then the specified points will be sent
15576 to the corners of the destination. If the @code{sense} option is set to @code{destination},
15577 then the corners of the source will be sent to the specified coordinates.
15579 The expressions can use the following variables:
15584 the width and height of video frame.
15588 Output frame count.
15591 @item interpolation
15592 Set interpolation for perspective correction.
15594 It accepts the following values:
15600 Default value is @samp{linear}.
15603 Set interpretation of coordinate options.
15605 It accepts the following values:
15609 Send point in the source specified by the given coordinates to
15610 the corners of the destination.
15612 @item 1, destination
15614 Send the corners of the source to the point in the destination specified
15615 by the given coordinates.
15617 Default value is @samp{source}.
15621 Set when the expressions for coordinates @option{x0,y0,...x3,y3} are evaluated.
15623 It accepts the following values:
15626 only evaluate expressions once during the filter initialization or
15627 when a command is processed
15630 evaluate expressions for each incoming frame
15633 Default value is @samp{init}.
15638 Delay interlaced video by one field time so that the field order changes.
15640 The intended use is to fix PAL movies that have been captured with the
15641 opposite field order to the film-to-video transfer.
15643 A description of the accepted parameters follows.
15649 It accepts the following values:
15652 Capture field order top-first, transfer bottom-first.
15653 Filter will delay the bottom field.
15656 Capture field order bottom-first, transfer top-first.
15657 Filter will delay the top field.
15660 Capture and transfer with the same field order. This mode only exists
15661 for the documentation of the other options to refer to, but if you
15662 actually select it, the filter will faithfully do nothing.
15665 Capture field order determined automatically by field flags, transfer
15667 Filter selects among @samp{t} and @samp{b} modes on a frame by frame
15668 basis using field flags. If no field information is available,
15669 then this works just like @samp{u}.
15672 Capture unknown or varying, transfer opposite.
15673 Filter selects among @samp{t} and @samp{b} on a frame by frame basis by
15674 analyzing the images and selecting the alternative that produces best
15675 match between the fields.
15678 Capture top-first, transfer unknown or varying.
15679 Filter selects among @samp{t} and @samp{p} using image analysis.
15682 Capture bottom-first, transfer unknown or varying.
15683 Filter selects among @samp{b} and @samp{p} using image analysis.
15686 Capture determined by field flags, transfer unknown or varying.
15687 Filter selects among @samp{t}, @samp{b} and @samp{p} using field flags and
15688 image analysis. If no field information is available, then this works just
15689 like @samp{U}. This is the default mode.
15692 Both capture and transfer unknown or varying.
15693 Filter selects among @samp{t}, @samp{b} and @samp{p} using image analysis only.
15697 @section photosensitivity
15698 Reduce various flashes in video, so to help users with epilepsy.
15700 It accepts the following options:
15703 Set how many frames to use when filtering. Default is 30.
15706 Set detection threshold factor. Default is 1.
15710 Set how many pixels to skip when sampling frames. Default is 1.
15711 Allowed range is from 1 to 1024.
15714 Leave frames unchanged. Default is disabled.
15717 @section pixdesctest
15719 Pixel format descriptor test filter, mainly useful for internal
15720 testing. The output video should be equal to the input video.
15724 format=monow, pixdesctest
15727 can be used to test the monowhite pixel format descriptor definition.
15731 Display sample values of color channels. Mainly useful for checking color
15732 and levels. Minimum supported resolution is 640x480.
15734 The filters accept the following options:
15738 Set scope X position, relative offset on X axis.
15741 Set scope Y position, relative offset on Y axis.
15750 Set window opacity. This window also holds statistics about pixel area.
15753 Set window X position, relative offset on X axis.
15756 Set window Y position, relative offset on Y axis.
15761 Enable the specified chain of postprocessing subfilters using libpostproc. This
15762 library should be automatically selected with a GPL build (@code{--enable-gpl}).
15763 Subfilters must be separated by '/' and can be disabled by prepending a '-'.
15764 Each subfilter and some options have a short and a long name that can be used
15765 interchangeably, i.e. dr/dering are the same.
15767 The filters accept the following options:
15771 Set postprocessing subfilters string.
15774 All subfilters share common options to determine their scope:
15778 Honor the quality commands for this subfilter.
15781 Do chrominance filtering, too (default).
15784 Do luminance filtering only (no chrominance).
15787 Do chrominance filtering only (no luminance).
15790 These options can be appended after the subfilter name, separated by a '|'.
15792 Available subfilters are:
15795 @item hb/hdeblock[|difference[|flatness]]
15796 Horizontal deblocking filter
15799 Difference factor where higher values mean more deblocking (default: @code{32}).
15801 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15804 @item vb/vdeblock[|difference[|flatness]]
15805 Vertical deblocking filter
15808 Difference factor where higher values mean more deblocking (default: @code{32}).
15810 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15813 @item ha/hadeblock[|difference[|flatness]]
15814 Accurate horizontal deblocking filter
15817 Difference factor where higher values mean more deblocking (default: @code{32}).
15819 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15822 @item va/vadeblock[|difference[|flatness]]
15823 Accurate vertical deblocking filter
15826 Difference factor where higher values mean more deblocking (default: @code{32}).
15828 Flatness threshold where lower values mean more deblocking (default: @code{39}).
15832 The horizontal and vertical deblocking filters share the difference and
15833 flatness values so you cannot set different horizontal and vertical
15837 @item h1/x1hdeblock
15838 Experimental horizontal deblocking filter
15840 @item v1/x1vdeblock
15841 Experimental vertical deblocking filter
15846 @item tn/tmpnoise[|threshold1[|threshold2[|threshold3]]], temporal noise reducer
15849 larger -> stronger filtering
15851 larger -> stronger filtering
15853 larger -> stronger filtering
15856 @item al/autolevels[:f/fullyrange], automatic brightness / contrast correction
15859 Stretch luminance to @code{0-255}.
15862 @item lb/linblenddeint
15863 Linear blend deinterlacing filter that deinterlaces the given block by
15864 filtering all lines with a @code{(1 2 1)} filter.
15866 @item li/linipoldeint
15867 Linear interpolating deinterlacing filter that deinterlaces the given block by
15868 linearly interpolating every second line.
15870 @item ci/cubicipoldeint
15871 Cubic interpolating deinterlacing filter deinterlaces the given block by
15872 cubically interpolating every second line.
15874 @item md/mediandeint
15875 Median deinterlacing filter that deinterlaces the given block by applying a
15876 median filter to every second line.
15878 @item fd/ffmpegdeint
15879 FFmpeg deinterlacing filter that deinterlaces the given block by filtering every
15880 second line with a @code{(-1 4 2 4 -1)} filter.
15883 Vertically applied FIR lowpass deinterlacing filter that deinterlaces the given
15884 block by filtering all lines with a @code{(-1 2 6 2 -1)} filter.
15886 @item fq/forceQuant[|quantizer]
15887 Overrides the quantizer table from the input with the constant quantizer you
15895 Default pp filter combination (@code{hb|a,vb|a,dr|a})
15898 Fast pp filter combination (@code{h1|a,v1|a,dr|a})
15901 High quality pp filter combination (@code{ha|a|128|7,va|a,dr|a})
15904 @subsection Examples
15908 Apply horizontal and vertical deblocking, deringing and automatic
15909 brightness/contrast:
15915 Apply default filters without brightness/contrast correction:
15921 Apply default filters and temporal denoiser:
15923 pp=default/tmpnoise|1|2|3
15927 Apply deblocking on luminance only, and switch vertical deblocking on or off
15928 automatically depending on available CPU time:
15935 Apply Postprocessing filter 7. It is variant of the @ref{spp} filter,
15936 similar to spp = 6 with 7 point DCT, where only the center sample is
15939 The filter accepts the following options:
15943 Force a constant quantization parameter. It accepts an integer in range
15944 0 to 63. If not set, the filter will use the QP from the video stream
15948 Set thresholding mode. Available modes are:
15952 Set hard thresholding.
15954 Set soft thresholding (better de-ringing effect, but likely blurrier).
15956 Set medium thresholding (good results, default).
15960 @section premultiply
15961 Apply alpha premultiply effect to input video stream using first plane
15962 of second stream as alpha.
15964 Both streams must have same dimensions and same pixel format.
15966 The filter accepts the following option:
15970 Set which planes will be processed, unprocessed planes will be copied.
15971 By default value 0xf, all planes will be processed.
15974 Do not require 2nd input for processing, instead use alpha plane from input stream.
15978 Apply prewitt operator to input video stream.
15980 The filter accepts the following option:
15984 Set which planes will be processed, unprocessed planes will be copied.
15985 By default value 0xf, all planes will be processed.
15988 Set value which will be multiplied with filtered result.
15991 Set value which will be added to filtered result.
15994 @subsection Commands
15996 This filter supports the all above options as @ref{commands}.
15998 @section pseudocolor
16000 Alter frame colors in video with pseudocolors.
16002 This filter accepts the following options:
16006 set pixel first component expression
16009 set pixel second component expression
16012 set pixel third component expression
16015 set pixel fourth component expression, corresponds to the alpha component
16018 set component to use as base for altering colors
16021 Each of them specifies the expression to use for computing the lookup table for
16022 the corresponding pixel component values.
16024 The expressions can contain the following constants and functions:
16029 The input width and height.
16032 The input value for the pixel component.
16034 @item ymin, umin, vmin, amin
16035 The minimum allowed component value.
16037 @item ymax, umax, vmax, amax
16038 The maximum allowed component value.
16041 All expressions default to "val".
16043 @subsection Examples
16047 Change too high luma values to gradient:
16049 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'"
16055 Obtain the average, maximum and minimum PSNR (Peak Signal to Noise
16056 Ratio) between two input videos.
16058 This filter takes in input two input videos, the first input is
16059 considered the "main" source and is passed unchanged to the
16060 output. The second input is used as a "reference" video for computing
16063 Both video inputs must have the same resolution and pixel format for
16064 this filter to work correctly. Also it assumes that both inputs
16065 have the same number of frames, which are compared one by one.
16067 The obtained average PSNR is printed through the logging system.
16069 The filter stores the accumulated MSE (mean squared error) of each
16070 frame, and at the end of the processing it is averaged across all frames
16071 equally, and the following formula is applied to obtain the PSNR:
16074 PSNR = 10*log10(MAX^2/MSE)
16077 Where MAX is the average of the maximum values of each component of the
16080 The description of the accepted parameters follows.
16083 @item stats_file, f
16084 If specified the filter will use the named file to save the PSNR of
16085 each individual frame. When filename equals "-" the data is sent to
16088 @item stats_version
16089 Specifies which version of the stats file format to use. Details of
16090 each format are written below.
16091 Default value is 1.
16093 @item stats_add_max
16094 Determines whether the max value is output to the stats log.
16095 Default value is 0.
16096 Requires stats_version >= 2. If this is set and stats_version < 2,
16097 the filter will return an error.
16100 This filter also supports the @ref{framesync} options.
16102 The file printed if @var{stats_file} is selected, contains a sequence of
16103 key/value pairs of the form @var{key}:@var{value} for each compared
16106 If a @var{stats_version} greater than 1 is specified, a header line precedes
16107 the list of per-frame-pair stats, with key value pairs following the frame
16108 format with the following parameters:
16111 @item psnr_log_version
16112 The version of the log file format. Will match @var{stats_version}.
16115 A comma separated list of the per-frame-pair parameters included in
16119 A description of each shown per-frame-pair parameter follows:
16123 sequential number of the input frame, starting from 1
16126 Mean Square Error pixel-by-pixel average difference of the compared
16127 frames, averaged over all the image components.
16129 @item mse_y, mse_u, mse_v, mse_r, mse_g, mse_b, mse_a
16130 Mean Square Error pixel-by-pixel average difference of the compared
16131 frames for the component specified by the suffix.
16133 @item psnr_y, psnr_u, psnr_v, psnr_r, psnr_g, psnr_b, psnr_a
16134 Peak Signal to Noise ratio of the compared frames for the component
16135 specified by the suffix.
16137 @item max_avg, max_y, max_u, max_v
16138 Maximum allowed value for each channel, and average over all
16142 @subsection Examples
16147 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
16148 [main][ref] psnr="stats_file=stats.log" [out]
16151 On this example the input file being processed is compared with the
16152 reference file @file{ref_movie.mpg}. The PSNR of each individual frame
16153 is stored in @file{stats.log}.
16156 Another example with different containers:
16158 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 -
16165 Pulldown reversal (inverse telecine) filter, capable of handling mixed
16166 hard-telecine, 24000/1001 fps progressive, and 30000/1001 fps progressive
16169 The pullup filter is designed to take advantage of future context in making
16170 its decisions. This filter is stateless in the sense that it does not lock
16171 onto a pattern to follow, but it instead looks forward to the following
16172 fields in order to identify matches and rebuild progressive frames.
16174 To produce content with an even framerate, insert the fps filter after
16175 pullup, use @code{fps=24000/1001} if the input frame rate is 29.97fps,
16176 @code{fps=24} for 30fps and the (rare) telecined 25fps input.
16178 The filter accepts the following options:
16185 These options set the amount of "junk" to ignore at the left, right, top, and
16186 bottom of the image, respectively. Left and right are in units of 8 pixels,
16187 while top and bottom are in units of 2 lines.
16188 The default is 8 pixels on each side.
16191 Set the strict breaks. Setting this option to 1 will reduce the chances of
16192 filter generating an occasional mismatched frame, but it may also cause an
16193 excessive number of frames to be dropped during high motion sequences.
16194 Conversely, setting it to -1 will make filter match fields more easily.
16195 This may help processing of video where there is slight blurring between
16196 the fields, but may also cause there to be interlaced frames in the output.
16197 Default value is @code{0}.
16200 Set the metric plane to use. It accepts the following values:
16206 Use chroma blue plane.
16209 Use chroma red plane.
16212 This option may be set to use chroma plane instead of the default luma plane
16213 for doing filter's computations. This may improve accuracy on very clean
16214 source material, but more likely will decrease accuracy, especially if there
16215 is chroma noise (rainbow effect) or any grayscale video.
16216 The main purpose of setting @option{mp} to a chroma plane is to reduce CPU
16217 load and make pullup usable in realtime on slow machines.
16220 For best results (without duplicated frames in the output file) it is
16221 necessary to change the output frame rate. For example, to inverse
16222 telecine NTSC input:
16224 ffmpeg -i input -vf pullup -r 24000/1001 ...
16229 Change video quantization parameters (QP).
16231 The filter accepts the following option:
16235 Set expression for quantization parameter.
16238 The expression is evaluated through the eval API and can contain, among others,
16239 the following constants:
16243 1 if index is not 129, 0 otherwise.
16246 Sequential index starting from -129 to 128.
16249 @subsection Examples
16253 Some equation like:
16261 Flush video frames from internal cache of frames into a random order.
16262 No frame is discarded.
16263 Inspired by @ref{frei0r} nervous filter.
16267 Set size in number of frames of internal cache, in range from @code{2} to
16268 @code{512}. Default is @code{30}.
16271 Set seed for random number generator, must be an integer included between
16272 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
16273 less than @code{0}, the filter will try to use a good random seed on a
16277 @section readeia608
16279 Read closed captioning (EIA-608) information from the top lines of a video frame.
16281 This filter adds frame metadata for @code{lavfi.readeia608.X.cc} and
16282 @code{lavfi.readeia608.X.line}, where @code{X} is the number of the identified line
16283 with EIA-608 data (starting from 0). A description of each metadata value follows:
16286 @item lavfi.readeia608.X.cc
16287 The two bytes stored as EIA-608 data (printed in hexadecimal).
16289 @item lavfi.readeia608.X.line
16290 The number of the line on which the EIA-608 data was identified and read.
16293 This filter accepts the following options:
16297 Set the line to start scanning for EIA-608 data. Default is @code{0}.
16300 Set the line to end scanning for EIA-608 data. Default is @code{29}.
16303 Set the ratio of width reserved for sync code detection.
16304 Default is @code{0.27}. Allowed range is @code{[0.1 - 0.7]}.
16307 Enable checking the parity bit. In the event of a parity error, the filter will output
16308 @code{0x00} for that character. Default is false.
16311 Lowpass lines prior to further processing. Default is enabled.
16314 @subsection Commands
16316 This filter supports the all above options as @ref{commands}.
16318 @subsection Examples
16322 Output a csv with presentation time and the first two lines of identified EIA-608 captioning data.
16324 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
16330 Read vertical interval timecode (VITC) information from the top lines of a
16333 The filter adds frame metadata key @code{lavfi.readvitc.tc_str} with the
16334 timecode value, if a valid timecode has been detected. Further metadata key
16335 @code{lavfi.readvitc.found} is set to 0/1 depending on whether
16336 timecode data has been found or not.
16338 This filter accepts the following options:
16342 Set the maximum number of lines to scan for VITC data. If the value is set to
16343 @code{-1} the full video frame is scanned. Default is @code{45}.
16346 Set the luma threshold for black. Accepts float numbers in the range [0.0,1.0],
16347 default value is @code{0.2}. The value must be equal or less than @code{thr_w}.
16350 Set the luma threshold for white. Accepts float numbers in the range [0.0,1.0],
16351 default value is @code{0.6}. The value must be equal or greater than @code{thr_b}.
16354 @subsection Examples
16358 Detect and draw VITC data onto the video frame; if no valid VITC is detected,
16359 draw @code{--:--:--:--} as a placeholder:
16361 ffmpeg -i input.avi -filter:v 'readvitc,drawtext=fontfile=FreeMono.ttf:text=%@{metadata\\:lavfi.readvitc.tc_str\\:--\\\\\\:--\\\\\\:--\\\\\\:--@}:x=(w-tw)/2:y=400-ascent'
16367 Remap pixels using 2nd: Xmap and 3rd: Ymap input video stream.
16369 Destination pixel at position (X, Y) will be picked from source (x, y) position
16370 where x = Xmap(X, Y) and y = Ymap(X, Y). If mapping values are out of range, zero
16371 value for pixel will be used for destination pixel.
16373 Xmap and Ymap input video streams must be of same dimensions. Output video stream
16374 will have Xmap/Ymap video stream dimensions.
16375 Xmap and Ymap input video streams are 16bit depth, single channel.
16379 Specify pixel format of output from this filter. Can be @code{color} or @code{gray}.
16380 Default is @code{color}.
16383 Specify the color of the unmapped pixels. For the syntax of this option,
16384 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
16385 manual,ffmpeg-utils}. Default color is @code{black}.
16388 @section removegrain
16390 The removegrain filter is a spatial denoiser for progressive video.
16394 Set mode for the first plane.
16397 Set mode for the second plane.
16400 Set mode for the third plane.
16403 Set mode for the fourth plane.
16406 Range of mode is from 0 to 24. Description of each mode follows:
16410 Leave input plane unchanged. Default.
16413 Clips the pixel with the minimum and maximum of the 8 neighbour pixels.
16416 Clips the pixel with the second minimum and maximum of the 8 neighbour pixels.
16419 Clips the pixel with the third minimum and maximum of the 8 neighbour pixels.
16422 Clips the pixel with the fourth minimum and maximum of the 8 neighbour pixels.
16423 This is equivalent to a median filter.
16426 Line-sensitive clipping giving the minimal change.
16429 Line-sensitive clipping, intermediate.
16432 Line-sensitive clipping, intermediate.
16435 Line-sensitive clipping, intermediate.
16438 Line-sensitive clipping on a line where the neighbours pixels are the closest.
16441 Replaces the target pixel with the closest neighbour.
16444 [1 2 1] horizontal and vertical kernel blur.
16450 Bob mode, interpolates top field from the line where the neighbours
16451 pixels are the closest.
16454 Bob mode, interpolates bottom field from the line where the neighbours
16455 pixels are the closest.
16458 Bob mode, interpolates top field. Same as 13 but with a more complicated
16459 interpolation formula.
16462 Bob mode, interpolates bottom field. Same as 14 but with a more complicated
16463 interpolation formula.
16466 Clips the pixel with the minimum and maximum of respectively the maximum and
16467 minimum of each pair of opposite neighbour pixels.
16470 Line-sensitive clipping using opposite neighbours whose greatest distance from
16471 the current pixel is minimal.
16474 Replaces the pixel with the average of its 8 neighbours.
16477 Averages the 9 pixels ([1 1 1] horizontal and vertical blur).
16480 Clips pixels using the averages of opposite neighbour.
16483 Same as mode 21 but simpler and faster.
16486 Small edge and halo removal, but reputed useless.
16492 @section removelogo
16494 Suppress a TV station logo, using an image file to determine which
16495 pixels comprise the logo. It works by filling in the pixels that
16496 comprise the logo with neighboring pixels.
16498 The filter accepts the following options:
16502 Set the filter bitmap file, which can be any image format supported by
16503 libavformat. The width and height of the image file must match those of the
16504 video stream being processed.
16507 Pixels in the provided bitmap image with a value of zero are not
16508 considered part of the logo, non-zero pixels are considered part of
16509 the logo. If you use white (255) for the logo and black (0) for the
16510 rest, you will be safe. For making the filter bitmap, it is
16511 recommended to take a screen capture of a black frame with the logo
16512 visible, and then using a threshold filter followed by the erode
16513 filter once or twice.
16515 If needed, little splotches can be fixed manually. Remember that if
16516 logo pixels are not covered, the filter quality will be much
16517 reduced. Marking too many pixels as part of the logo does not hurt as
16518 much, but it will increase the amount of blurring needed to cover over
16519 the image and will destroy more information than necessary, and extra
16520 pixels will slow things down on a large logo.
16522 @section repeatfields
16524 This filter uses the repeat_field flag from the Video ES headers and hard repeats
16525 fields based on its value.
16529 Reverse a video clip.
16531 Warning: This filter requires memory to buffer the entire clip, so trimming
16534 @subsection Examples
16538 Take the first 5 seconds of a clip, and reverse it.
16545 Shift R/G/B/A pixels horizontally and/or vertically.
16547 The filter accepts the following options:
16550 Set amount to shift red horizontally.
16552 Set amount to shift red vertically.
16554 Set amount to shift green horizontally.
16556 Set amount to shift green vertically.
16558 Set amount to shift blue horizontally.
16560 Set amount to shift blue vertically.
16562 Set amount to shift alpha horizontally.
16564 Set amount to shift alpha vertically.
16566 Set edge mode, can be @var{smear}, default, or @var{warp}.
16569 @subsection Commands
16571 This filter supports the all above options as @ref{commands}.
16574 Apply roberts cross operator to input video stream.
16576 The filter accepts the following option:
16580 Set which planes will be processed, unprocessed planes will be copied.
16581 By default value 0xf, all planes will be processed.
16584 Set value which will be multiplied with filtered result.
16587 Set value which will be added to filtered result.
16590 @subsection Commands
16592 This filter supports the all above options as @ref{commands}.
16596 Rotate video by an arbitrary angle expressed in radians.
16598 The filter accepts the following options:
16600 A description of the optional parameters follows.
16603 Set an expression for the angle by which to rotate the input video
16604 clockwise, expressed as a number of radians. A negative value will
16605 result in a counter-clockwise rotation. By default it is set to "0".
16607 This expression is evaluated for each frame.
16610 Set the output width expression, default value is "iw".
16611 This expression is evaluated just once during configuration.
16614 Set the output height expression, default value is "ih".
16615 This expression is evaluated just once during configuration.
16618 Enable bilinear interpolation if set to 1, a value of 0 disables
16619 it. Default value is 1.
16622 Set the color used to fill the output area not covered by the rotated
16623 image. For the general syntax of this option, check the
16624 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
16625 If the special value "none" is selected then no
16626 background is printed (useful for example if the background is never shown).
16628 Default value is "black".
16631 The expressions for the angle and the output size can contain the
16632 following constants and functions:
16636 sequential number of the input frame, starting from 0. It is always NAN
16637 before the first frame is filtered.
16640 time in seconds of the input frame, it is set to 0 when the filter is
16641 configured. It is always NAN before the first frame is filtered.
16645 horizontal and vertical chroma subsample values. For example for the
16646 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
16650 the input video width and height
16654 the output width and height, that is the size of the padded area as
16655 specified by the @var{width} and @var{height} expressions
16659 the minimal width/height required for completely containing the input
16660 video rotated by @var{a} radians.
16662 These are only available when computing the @option{out_w} and
16663 @option{out_h} expressions.
16666 @subsection Examples
16670 Rotate the input by PI/6 radians clockwise:
16676 Rotate the input by PI/6 radians counter-clockwise:
16682 Rotate the input by 45 degrees clockwise:
16688 Apply a constant rotation with period T, starting from an angle of PI/3:
16690 rotate=PI/3+2*PI*t/T
16694 Make the input video rotation oscillating with a period of T
16695 seconds and an amplitude of A radians:
16697 rotate=A*sin(2*PI/T*t)
16701 Rotate the video, output size is chosen so that the whole rotating
16702 input video is always completely contained in the output:
16704 rotate='2*PI*t:ow=hypot(iw,ih):oh=ow'
16708 Rotate the video, reduce the output size so that no background is ever
16711 rotate=2*PI*t:ow='min(iw,ih)/sqrt(2)':oh=ow:c=none
16715 @subsection Commands
16717 The filter supports the following commands:
16721 Set the angle expression.
16722 The command accepts the same syntax of the corresponding option.
16724 If the specified expression is not valid, it is kept at its current
16730 Apply Shape Adaptive Blur.
16732 The filter accepts the following options:
16735 @item luma_radius, lr
16736 Set luma blur filter strength, must be a value in range 0.1-4.0, default
16737 value is 1.0. A greater value will result in a more blurred image, and
16738 in slower processing.
16740 @item luma_pre_filter_radius, lpfr
16741 Set luma pre-filter radius, must be a value in the 0.1-2.0 range, default
16744 @item luma_strength, ls
16745 Set luma maximum difference between pixels to still be considered, must
16746 be a value in the 0.1-100.0 range, default value is 1.0.
16748 @item chroma_radius, cr
16749 Set chroma blur filter strength, must be a value in range -0.9-4.0. A
16750 greater value will result in a more blurred image, and in slower
16753 @item chroma_pre_filter_radius, cpfr
16754 Set chroma pre-filter radius, must be a value in the -0.9-2.0 range.
16756 @item chroma_strength, cs
16757 Set chroma maximum difference between pixels to still be considered,
16758 must be a value in the -0.9-100.0 range.
16761 Each chroma option value, if not explicitly specified, is set to the
16762 corresponding luma option value.
16767 Scale (resize) the input video, using the libswscale library.
16769 The scale filter forces the output display aspect ratio to be the same
16770 of the input, by changing the output sample aspect ratio.
16772 If the input image format is different from the format requested by
16773 the next filter, the scale filter will convert the input to the
16776 @subsection Options
16777 The filter accepts the following options, or any of the options
16778 supported by the libswscale scaler.
16780 See @ref{scaler_options,,the ffmpeg-scaler manual,ffmpeg-scaler} for
16781 the complete list of scaler options.
16786 Set the output video dimension expression. Default value is the input
16789 If the @var{width} or @var{w} value is 0, the input width is used for
16790 the output. If the @var{height} or @var{h} value is 0, the input height
16791 is used for the output.
16793 If one and only one of the values is -n with n >= 1, the scale filter
16794 will use a value that maintains the aspect ratio of the input image,
16795 calculated from the other specified dimension. After that it will,
16796 however, make sure that the calculated dimension is divisible by n and
16797 adjust the value if necessary.
16799 If both values are -n with n >= 1, the behavior will be identical to
16800 both values being set to 0 as previously detailed.
16802 See below for the list of accepted constants for use in the dimension
16806 Specify when to evaluate @var{width} and @var{height} expression. It accepts the following values:
16810 Only evaluate expressions once during the filter initialization or when a command is processed.
16813 Evaluate expressions for each incoming frame.
16817 Default value is @samp{init}.
16821 Set the interlacing mode. It accepts the following values:
16825 Force interlaced aware scaling.
16828 Do not apply interlaced scaling.
16831 Select interlaced aware scaling depending on whether the source frames
16832 are flagged as interlaced or not.
16835 Default value is @samp{0}.
16838 Set libswscale scaling flags. See
16839 @ref{sws_flags,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16840 complete list of values. If not explicitly specified the filter applies
16844 @item param0, param1
16845 Set libswscale input parameters for scaling algorithms that need them. See
16846 @ref{sws_params,,the ffmpeg-scaler manual,ffmpeg-scaler} for the
16847 complete documentation. If not explicitly specified the filter applies
16853 Set the video size. For the syntax of this option, check the
16854 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
16856 @item in_color_matrix
16857 @item out_color_matrix
16858 Set in/output YCbCr color space type.
16860 This allows the autodetected value to be overridden as well as allows forcing
16861 a specific value used for the output and encoder.
16863 If not specified, the color space type depends on the pixel format.
16869 Choose automatically.
16872 Format conforming to International Telecommunication Union (ITU)
16873 Recommendation BT.709.
16876 Set color space conforming to the United States Federal Communications
16877 Commission (FCC) Code of Federal Regulations (CFR) Title 47 (2003) 73.682 (a).
16882 Set color space conforming to:
16886 ITU Radiocommunication Sector (ITU-R) Recommendation BT.601
16889 ITU-R Rec. BT.470-6 (1998) Systems B, B1, and G
16892 Society of Motion Picture and Television Engineers (SMPTE) ST 170:2004
16897 Set color space conforming to SMPTE ST 240:1999.
16900 Set color space conforming to ITU-R BT.2020 non-constant luminance system.
16905 Set in/output YCbCr sample range.
16907 This allows the autodetected value to be overridden as well as allows forcing
16908 a specific value used for the output and encoder. If not specified, the
16909 range depends on the pixel format. Possible values:
16913 Choose automatically.
16916 Set full range (0-255 in case of 8-bit luma).
16918 @item mpeg/limited/tv
16919 Set "MPEG" range (16-235 in case of 8-bit luma).
16922 @item force_original_aspect_ratio
16923 Enable decreasing or increasing output video width or height if necessary to
16924 keep the original aspect ratio. Possible values:
16928 Scale the video as specified and disable this feature.
16931 The output video dimensions will automatically be decreased if needed.
16934 The output video dimensions will automatically be increased if needed.
16938 One useful instance of this option is that when you know a specific device's
16939 maximum allowed resolution, you can use this to limit the output video to
16940 that, while retaining the aspect ratio. For example, device A allows
16941 1280x720 playback, and your video is 1920x800. Using this option (set it to
16942 decrease) and specifying 1280x720 to the command line makes the output
16945 Please note that this is a different thing than specifying -1 for @option{w}
16946 or @option{h}, you still need to specify the output resolution for this option
16949 @item force_divisible_by
16950 Ensures that both the output dimensions, width and height, are divisible by the
16951 given integer when used together with @option{force_original_aspect_ratio}. This
16952 works similar to using @code{-n} in the @option{w} and @option{h} options.
16954 This option respects the value set for @option{force_original_aspect_ratio},
16955 increasing or decreasing the resolution accordingly. The video's aspect ratio
16956 may be slightly modified.
16958 This option can be handy if you need to have a video fit within or exceed
16959 a defined resolution using @option{force_original_aspect_ratio} but also have
16960 encoder restrictions on width or height divisibility.
16964 The values of the @option{w} and @option{h} options are expressions
16965 containing the following constants:
16970 The input width and height
16974 These are the same as @var{in_w} and @var{in_h}.
16978 The output (scaled) width and height
16982 These are the same as @var{out_w} and @var{out_h}
16985 The same as @var{iw} / @var{ih}
16988 input sample aspect ratio
16991 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
16995 horizontal and vertical input chroma subsample values. For example for the
16996 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17000 horizontal and vertical output chroma subsample values. For example for the
17001 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17004 The (sequential) number of the input frame, starting from 0.
17005 Only available with @code{eval=frame}.
17008 The presentation timestamp of the input frame, expressed as a number of
17009 seconds. Only available with @code{eval=frame}.
17012 The position (byte offset) of the frame in the input stream, or NaN if
17013 this information is unavailable and/or meaningless (for example in case of synthetic video).
17014 Only available with @code{eval=frame}.
17017 @subsection Examples
17021 Scale the input video to a size of 200x100
17026 This is equivalent to:
17037 Specify a size abbreviation for the output size:
17042 which can also be written as:
17048 Scale the input to 2x:
17050 scale=w=2*iw:h=2*ih
17054 The above is the same as:
17056 scale=2*in_w:2*in_h
17060 Scale the input to 2x with forced interlaced scaling:
17062 scale=2*iw:2*ih:interl=1
17066 Scale the input to half size:
17068 scale=w=iw/2:h=ih/2
17072 Increase the width, and set the height to the same size:
17078 Seek Greek harmony:
17085 Increase the height, and set the width to 3/2 of the height:
17087 scale=w=3/2*oh:h=3/5*ih
17091 Increase the size, making the size a multiple of the chroma
17094 scale="trunc(3/2*iw/hsub)*hsub:trunc(3/2*ih/vsub)*vsub"
17098 Increase the width to a maximum of 500 pixels,
17099 keeping the same aspect ratio as the input:
17101 scale=w='min(500\, iw*3/2):h=-1'
17105 Make pixels square by combining scale and setsar:
17107 scale='trunc(ih*dar):ih',setsar=1/1
17111 Make pixels square by combining scale and setsar,
17112 making sure the resulting resolution is even (required by some codecs):
17114 scale='trunc(ih*dar/2)*2:trunc(ih/2)*2',setsar=1/1
17118 @subsection Commands
17120 This filter supports the following commands:
17124 Set the output video dimension expression.
17125 The command accepts the same syntax of the corresponding option.
17127 If the specified expression is not valid, it is kept at its current
17133 Use the NVIDIA Performance Primitives (libnpp) to perform scaling and/or pixel
17134 format conversion on CUDA video frames. Setting the output width and height
17135 works in the same way as for the @var{scale} filter.
17137 The following additional options are accepted:
17140 The pixel format of the output CUDA frames. If set to the string "same" (the
17141 default), the input format will be kept. Note that automatic format negotiation
17142 and conversion is not yet supported for hardware frames
17145 The interpolation algorithm used for resizing. One of the following:
17152 @item cubic2p_bspline
17153 2-parameter cubic (B=1, C=0)
17155 @item cubic2p_catmullrom
17156 2-parameter cubic (B=0, C=1/2)
17158 @item cubic2p_b05c03
17159 2-parameter cubic (B=1/2, C=3/10)
17167 @item force_original_aspect_ratio
17168 Enable decreasing or increasing output video width or height if necessary to
17169 keep the original aspect ratio. Possible values:
17173 Scale the video as specified and disable this feature.
17176 The output video dimensions will automatically be decreased if needed.
17179 The output video dimensions will automatically be increased if needed.
17183 One useful instance of this option is that when you know a specific device's
17184 maximum allowed resolution, you can use this to limit the output video to
17185 that, while retaining the aspect ratio. For example, device A allows
17186 1280x720 playback, and your video is 1920x800. Using this option (set it to
17187 decrease) and specifying 1280x720 to the command line makes the output
17190 Please note that this is a different thing than specifying -1 for @option{w}
17191 or @option{h}, you still need to specify the output resolution for this option
17194 @item force_divisible_by
17195 Ensures that both the output dimensions, width and height, are divisible by the
17196 given integer when used together with @option{force_original_aspect_ratio}. This
17197 works similar to using @code{-n} in the @option{w} and @option{h} options.
17199 This option respects the value set for @option{force_original_aspect_ratio},
17200 increasing or decreasing the resolution accordingly. The video's aspect ratio
17201 may be slightly modified.
17203 This option can be handy if you need to have a video fit within or exceed
17204 a defined resolution using @option{force_original_aspect_ratio} but also have
17205 encoder restrictions on width or height divisibility.
17211 Scale (resize) the input video, based on a reference video.
17213 See the scale filter for available options, scale2ref supports the same but
17214 uses the reference video instead of the main input as basis. scale2ref also
17215 supports the following additional constants for the @option{w} and
17216 @option{h} options:
17221 The main input video's width and height
17224 The same as @var{main_w} / @var{main_h}
17227 The main input video's sample aspect ratio
17229 @item main_dar, mdar
17230 The main input video's display aspect ratio. Calculated from
17231 @code{(main_w / main_h) * main_sar}.
17235 The main input video's horizontal and vertical chroma subsample values.
17236 For example for the pixel format "yuv422p" @var{hsub} is 2 and @var{vsub}
17240 The (sequential) number of the main input frame, starting from 0.
17241 Only available with @code{eval=frame}.
17244 The presentation timestamp of the main input frame, expressed as a number of
17245 seconds. Only available with @code{eval=frame}.
17248 The position (byte offset) of the frame in the main input stream, or NaN if
17249 this information is unavailable and/or meaningless (for example in case of synthetic video).
17250 Only available with @code{eval=frame}.
17253 @subsection Examples
17257 Scale a subtitle stream (b) to match the main video (a) in size before overlaying
17259 'scale2ref[b][a];[a][b]overlay'
17263 Scale a logo to 1/10th the height of a video, while preserving its display aspect ratio.
17265 [logo-in][video-in]scale2ref=w=oh*mdar:h=ih/10[logo-out][video-out]
17269 @subsection Commands
17271 This filter supports the following commands:
17275 Set the output video dimension expression.
17276 The command accepts the same syntax of the corresponding option.
17278 If the specified expression is not valid, it is kept at its current
17283 Scroll input video horizontally and/or vertically by constant speed.
17285 The filter accepts the following options:
17287 @item horizontal, h
17288 Set the horizontal scrolling speed. Default is 0. Allowed range is from -1 to 1.
17289 Negative values changes scrolling direction.
17292 Set the vertical scrolling speed. Default is 0. Allowed range is from -1 to 1.
17293 Negative values changes scrolling direction.
17296 Set the initial horizontal scrolling position. Default is 0. Allowed range is from 0 to 1.
17299 Set the initial vertical scrolling position. Default is 0. Allowed range is from 0 to 1.
17302 @subsection Commands
17304 This filter supports the following @ref{commands}:
17306 @item horizontal, h
17307 Set the horizontal scrolling speed.
17309 Set the vertical scrolling speed.
17315 Detect video scene change.
17317 This filter sets frame metadata with mafd between frame, the scene score, and
17318 forward the frame to the next filter, so they can use these metadata to detect
17319 scene change or others.
17321 In addition, this filter logs a message and sets frame metadata when it detects
17322 a scene change by @option{threshold}.
17324 @code{lavfi.scd.mafd} metadata keys are set with mafd for every frame.
17326 @code{lavfi.scd.score} metadata keys are set with scene change score for every frame
17327 to detect scene change.
17329 @code{lavfi.scd.time} metadata keys are set with current filtered frame time which
17330 detect scene change with @option{threshold}.
17332 The filter accepts the following options:
17336 Set the scene change detection threshold as a percentage of maximum change. Good
17337 values are in the @code{[8.0, 14.0]} range. The range for @option{threshold} is
17340 Default value is @code{10.}.
17343 Set the flag to pass scene change frames to the next filter. Default value is @code{0}
17344 You can enable it if you want to get snapshot of scene change frames only.
17347 @anchor{selectivecolor}
17348 @section selectivecolor
17350 Adjust cyan, magenta, yellow and black (CMYK) to certain ranges of colors (such
17351 as "reds", "yellows", "greens", "cyans", ...). The adjustment range is defined
17352 by the "purity" of the color (that is, how saturated it already is).
17354 This filter is similar to the Adobe Photoshop Selective Color tool.
17356 The filter accepts the following options:
17359 @item correction_method
17360 Select color correction method.
17362 Available values are:
17365 Specified adjustments are applied "as-is" (added/subtracted to original pixel
17368 Specified adjustments are relative to the original component value.
17370 Default is @code{absolute}.
17372 Adjustments for red pixels (pixels where the red component is the maximum)
17374 Adjustments for yellow pixels (pixels where the blue component is the minimum)
17376 Adjustments for green pixels (pixels where the green component is the maximum)
17378 Adjustments for cyan pixels (pixels where the red component is the minimum)
17380 Adjustments for blue pixels (pixels where the blue component is the maximum)
17382 Adjustments for magenta pixels (pixels where the green component is the minimum)
17384 Adjustments for white pixels (pixels where all components are greater than 128)
17386 Adjustments for all pixels except pure black and pure white
17388 Adjustments for black pixels (pixels where all components are lesser than 128)
17390 Specify a Photoshop selective color file (@code{.asv}) to import the settings from.
17393 All the adjustment settings (@option{reds}, @option{yellows}, ...) accept up to
17394 4 space separated floating point adjustment values in the [-1,1] range,
17395 respectively to adjust the amount of cyan, magenta, yellow and black for the
17396 pixels of its range.
17398 @subsection Examples
17402 Increase cyan by 50% and reduce yellow by 33% in every green areas, and
17403 increase magenta by 27% in blue areas:
17405 selectivecolor=greens=.5 0 -.33 0:blues=0 .27
17409 Use a Photoshop selective color preset:
17411 selectivecolor=psfile=MySelectiveColorPresets/Misty.asv
17415 @anchor{separatefields}
17416 @section separatefields
17418 The @code{separatefields} takes a frame-based video input and splits
17419 each frame into its components fields, producing a new half height clip
17420 with twice the frame rate and twice the frame count.
17422 This filter use field-dominance information in frame to decide which
17423 of each pair of fields to place first in the output.
17424 If it gets it wrong use @ref{setfield} filter before @code{separatefields} filter.
17426 @section setdar, setsar
17428 The @code{setdar} filter sets the Display Aspect Ratio for the filter
17431 This is done by changing the specified Sample (aka Pixel) Aspect
17432 Ratio, according to the following equation:
17434 @var{DAR} = @var{HORIZONTAL_RESOLUTION} / @var{VERTICAL_RESOLUTION} * @var{SAR}
17437 Keep in mind that the @code{setdar} filter does not modify the pixel
17438 dimensions of the video frame. Also, the display aspect ratio set by
17439 this filter may be changed by later filters in the filterchain,
17440 e.g. in case of scaling or if another "setdar" or a "setsar" filter is
17443 The @code{setsar} filter sets the Sample (aka Pixel) Aspect Ratio for
17444 the filter output video.
17446 Note that as a consequence of the application of this filter, the
17447 output display aspect ratio will change according to the equation
17450 Keep in mind that the sample aspect ratio set by the @code{setsar}
17451 filter may be changed by later filters in the filterchain, e.g. if
17452 another "setsar" or a "setdar" filter is applied.
17454 It accepts the following parameters:
17457 @item r, ratio, dar (@code{setdar} only), sar (@code{setsar} only)
17458 Set the aspect ratio used by the filter.
17460 The parameter can be a floating point number string, an expression, or
17461 a string of the form @var{num}:@var{den}, where @var{num} and
17462 @var{den} are the numerator and denominator of the aspect ratio. If
17463 the parameter is not specified, it is assumed the value "0".
17464 In case the form "@var{num}:@var{den}" is used, the @code{:} character
17468 Set the maximum integer value to use for expressing numerator and
17469 denominator when reducing the expressed aspect ratio to a rational.
17470 Default value is @code{100}.
17474 The parameter @var{sar} is an expression containing
17475 the following constants:
17479 These are approximated values for the mathematical constants e
17480 (Euler's number), pi (Greek pi), and phi (the golden ratio).
17483 The input width and height.
17486 These are the same as @var{w} / @var{h}.
17489 The input sample aspect ratio.
17492 The input display aspect ratio. It is the same as
17493 (@var{w} / @var{h}) * @var{sar}.
17496 Horizontal and vertical chroma subsample values. For example, for the
17497 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
17500 @subsection Examples
17505 To change the display aspect ratio to 16:9, specify one of the following:
17512 To change the sample aspect ratio to 10:11, specify:
17518 To set a display aspect ratio of 16:9, and specify a maximum integer value of
17519 1000 in the aspect ratio reduction, use the command:
17521 setdar=ratio=16/9:max=1000
17529 Force field for the output video frame.
17531 The @code{setfield} filter marks the interlace type field for the
17532 output frames. It does not change the input frame, but only sets the
17533 corresponding property, which affects how the frame is treated by
17534 following filters (e.g. @code{fieldorder} or @code{yadif}).
17536 The filter accepts the following options:
17541 Available values are:
17545 Keep the same field property.
17548 Mark the frame as bottom-field-first.
17551 Mark the frame as top-field-first.
17554 Mark the frame as progressive.
17561 Force frame parameter for the output video frame.
17563 The @code{setparams} filter marks interlace and color range for the
17564 output frames. It does not change the input frame, but only sets the
17565 corresponding property, which affects how the frame is treated by
17570 Available values are:
17574 Keep the same field property (default).
17577 Mark the frame as bottom-field-first.
17580 Mark the frame as top-field-first.
17583 Mark the frame as progressive.
17587 Available values are:
17591 Keep the same color range property (default).
17593 @item unspecified, unknown
17594 Mark the frame as unspecified color range.
17596 @item limited, tv, mpeg
17597 Mark the frame as limited range.
17599 @item full, pc, jpeg
17600 Mark the frame as full range.
17603 @item color_primaries
17604 Set the color primaries.
17605 Available values are:
17609 Keep the same color primaries property (default).
17626 Set the color transfer.
17627 Available values are:
17631 Keep the same color trc property (default).
17653 Set the colorspace.
17654 Available values are:
17658 Keep the same colorspace property (default).
17671 @item chroma-derived-nc
17672 @item chroma-derived-c
17679 Show a line containing various information for each input video frame.
17680 The input video is not modified.
17682 This filter supports the following options:
17686 Calculate checksums of each plane. By default enabled.
17689 The shown line contains a sequence of key/value pairs of the form
17690 @var{key}:@var{value}.
17692 The following values are shown in the output:
17696 The (sequential) number of the input frame, starting from 0.
17699 The Presentation TimeStamp of the input frame, expressed as a number of
17700 time base units. The time base unit depends on the filter input pad.
17703 The Presentation TimeStamp of the input frame, expressed as a number of
17707 The position of the frame in the input stream, or -1 if this information is
17708 unavailable and/or meaningless (for example in case of synthetic video).
17711 The pixel format name.
17714 The sample aspect ratio of the input frame, expressed in the form
17715 @var{num}/@var{den}.
17718 The size of the input frame. For the syntax of this option, check the
17719 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
17722 The type of interlaced mode ("P" for "progressive", "T" for top field first, "B"
17723 for bottom field first).
17726 This is 1 if the frame is a key frame, 0 otherwise.
17729 The picture type of the input frame ("I" for an I-frame, "P" for a
17730 P-frame, "B" for a B-frame, or "?" for an unknown type).
17731 Also refer to the documentation of the @code{AVPictureType} enum and of
17732 the @code{av_get_picture_type_char} function defined in
17733 @file{libavutil/avutil.h}.
17736 The Adler-32 checksum (printed in hexadecimal) of all the planes of the input frame.
17738 @item plane_checksum
17739 The Adler-32 checksum (printed in hexadecimal) of each plane of the input frame,
17740 expressed in the form "[@var{c0} @var{c1} @var{c2} @var{c3}]".
17743 The mean value of pixels in each plane of the input frame, expressed in the form
17744 "[@var{mean0} @var{mean1} @var{mean2} @var{mean3}]".
17747 The standard deviation of pixel values in each plane of the input frame, expressed
17748 in the form "[@var{stdev0} @var{stdev1} @var{stdev2} @var{stdev3}]".
17752 @section showpalette
17754 Displays the 256 colors palette of each frame. This filter is only relevant for
17755 @var{pal8} pixel format frames.
17757 It accepts the following option:
17761 Set the size of the box used to represent one palette color entry. Default is
17762 @code{30} (for a @code{30x30} pixel box).
17765 @section shuffleframes
17767 Reorder and/or duplicate and/or drop video frames.
17769 It accepts the following parameters:
17773 Set the destination indexes of input frames.
17774 This is space or '|' separated list of indexes that maps input frames to output
17775 frames. Number of indexes also sets maximal value that each index may have.
17776 '-1' index have special meaning and that is to drop frame.
17779 The first frame has the index 0. The default is to keep the input unchanged.
17781 @subsection Examples
17785 Swap second and third frame of every three frames of the input:
17787 ffmpeg -i INPUT -vf "shuffleframes=0 2 1" OUTPUT
17791 Swap 10th and 1st frame of every ten frames of the input:
17793 ffmpeg -i INPUT -vf "shuffleframes=9 1 2 3 4 5 6 7 8 0" OUTPUT
17797 @section shuffleplanes
17799 Reorder and/or duplicate video planes.
17801 It accepts the following parameters:
17806 The index of the input plane to be used as the first output plane.
17809 The index of the input plane to be used as the second output plane.
17812 The index of the input plane to be used as the third output plane.
17815 The index of the input plane to be used as the fourth output plane.
17819 The first plane has the index 0. The default is to keep the input unchanged.
17821 @subsection Examples
17825 Swap the second and third planes of the input:
17827 ffmpeg -i INPUT -vf shuffleplanes=0:2:1:3 OUTPUT
17831 @anchor{signalstats}
17832 @section signalstats
17833 Evaluate various visual metrics that assist in determining issues associated
17834 with the digitization of analog video media.
17836 By default the filter will log these metadata values:
17840 Display the minimal Y value contained within the input frame. Expressed in
17844 Display the Y value at the 10% percentile within the input frame. Expressed in
17848 Display the average Y value within the input frame. Expressed in range of
17852 Display the Y value at the 90% percentile within the input frame. Expressed in
17856 Display the maximum Y value contained within the input frame. Expressed in
17860 Display the minimal U value contained within the input frame. Expressed in
17864 Display the U value at the 10% percentile within the input frame. Expressed in
17868 Display the average U value within the input frame. Expressed in range of
17872 Display the U value at the 90% percentile within the input frame. Expressed in
17876 Display the maximum U value contained within the input frame. Expressed in
17880 Display the minimal V value contained within the input frame. Expressed in
17884 Display the V value at the 10% percentile within the input frame. Expressed in
17888 Display the average V value within the input frame. Expressed in range of
17892 Display the V value at the 90% percentile within the input frame. Expressed in
17896 Display the maximum V value contained within the input frame. Expressed in
17900 Display the minimal saturation value contained within the input frame.
17901 Expressed in range of [0-~181.02].
17904 Display the saturation value at the 10% percentile within the input frame.
17905 Expressed in range of [0-~181.02].
17908 Display the average saturation value within the input frame. Expressed in range
17912 Display the saturation value at the 90% percentile within the input frame.
17913 Expressed in range of [0-~181.02].
17916 Display the maximum saturation value contained within the input frame.
17917 Expressed in range of [0-~181.02].
17920 Display the median value for hue within the input frame. Expressed in range of
17924 Display the average value for hue within the input frame. Expressed in range of
17928 Display the average of sample value difference between all values of the Y
17929 plane in the current frame and corresponding values of the previous input frame.
17930 Expressed in range of [0-255].
17933 Display the average of sample value difference between all values of the U
17934 plane in the current frame and corresponding values of the previous input frame.
17935 Expressed in range of [0-255].
17938 Display the average of sample value difference between all values of the V
17939 plane in the current frame and corresponding values of the previous input frame.
17940 Expressed in range of [0-255].
17943 Display bit depth of Y plane in current frame.
17944 Expressed in range of [0-16].
17947 Display bit depth of U plane in current frame.
17948 Expressed in range of [0-16].
17951 Display bit depth of V plane in current frame.
17952 Expressed in range of [0-16].
17955 The filter accepts the following options:
17961 @option{stat} specify an additional form of image analysis.
17962 @option{out} output video with the specified type of pixel highlighted.
17964 Both options accept the following values:
17968 Identify @var{temporal outliers} pixels. A @var{temporal outlier} is a pixel
17969 unlike the neighboring pixels of the same field. Examples of temporal outliers
17970 include the results of video dropouts, head clogs, or tape tracking issues.
17973 Identify @var{vertical line repetition}. Vertical line repetition includes
17974 similar rows of pixels within a frame. In born-digital video vertical line
17975 repetition is common, but this pattern is uncommon in video digitized from an
17976 analog source. When it occurs in video that results from the digitization of an
17977 analog source it can indicate concealment from a dropout compensator.
17980 Identify pixels that fall outside of legal broadcast range.
17984 Set the highlight color for the @option{out} option. The default color is
17988 @subsection Examples
17992 Output data of various video metrics:
17994 ffprobe -f lavfi movie=example.mov,signalstats="stat=tout+vrep+brng" -show_frames
17998 Output specific data about the minimum and maximum values of the Y plane per frame:
18000 ffprobe -f lavfi movie=example.mov,signalstats -show_entries frame_tags=lavfi.signalstats.YMAX,lavfi.signalstats.YMIN
18004 Playback video while highlighting pixels that are outside of broadcast range in red.
18006 ffplay example.mov -vf signalstats="out=brng:color=red"
18010 Playback video with signalstats metadata drawn over the frame.
18012 ffplay example.mov -vf signalstats=stat=brng+vrep+tout,drawtext=fontfile=FreeSerif.ttf:textfile=signalstat_drawtext.txt
18015 The contents of signalstat_drawtext.txt used in the command are:
18018 Y (%@{metadata:lavfi.signalstats.YMIN@}-%@{metadata:lavfi.signalstats.YMAX@})
18019 U (%@{metadata:lavfi.signalstats.UMIN@}-%@{metadata:lavfi.signalstats.UMAX@})
18020 V (%@{metadata:lavfi.signalstats.VMIN@}-%@{metadata:lavfi.signalstats.VMAX@})
18021 saturation maximum: %@{metadata:lavfi.signalstats.SATMAX@}
18029 Calculates the MPEG-7 Video Signature. The filter can handle more than one
18030 input. In this case the matching between the inputs can be calculated additionally.
18031 The filter always passes through the first input. The signature of each stream can
18032 be written into a file.
18034 It accepts the following options:
18038 Enable or disable the matching process.
18040 Available values are:
18044 Disable the calculation of a matching (default).
18046 Calculate the matching for the whole video and output whether the whole video
18047 matches or only parts.
18049 Calculate only until a matching is found or the video ends. Should be faster in
18054 Set the number of inputs. The option value must be a non negative integer.
18055 Default value is 1.
18058 Set the path to which the output is written. If there is more than one input,
18059 the path must be a prototype, i.e. must contain %d or %0nd (where n is a positive
18060 integer), that will be replaced with the input number. If no filename is
18061 specified, no output will be written. This is the default.
18064 Choose the output format.
18066 Available values are:
18070 Use the specified binary representation (default).
18072 Use the specified xml representation.
18076 Set threshold to detect one word as similar. The option value must be an integer
18077 greater than zero. The default value is 9000.
18080 Set threshold to detect all words as similar. The option value must be an integer
18081 greater than zero. The default value is 60000.
18084 Set threshold to detect frames as similar. The option value must be an integer
18085 greater than zero. The default value is 116.
18088 Set the minimum length of a sequence in frames to recognize it as matching
18089 sequence. The option value must be a non negative integer value.
18090 The default value is 0.
18093 Set the minimum relation, that matching frames to all frames must have.
18094 The option value must be a double value between 0 and 1. The default value is 0.5.
18097 @subsection Examples
18101 To calculate the signature of an input video and store it in signature.bin:
18103 ffmpeg -i input.mkv -vf signature=filename=signature.bin -map 0:v -f null -
18107 To detect whether two videos match and store the signatures in XML format in
18108 signature0.xml and signature1.xml:
18110 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 -
18118 Blur the input video without impacting the outlines.
18120 It accepts the following options:
18123 @item luma_radius, lr
18124 Set the luma radius. The option value must be a float number in
18125 the range [0.1,5.0] that specifies the variance of the gaussian filter
18126 used to blur the image (slower if larger). Default value is 1.0.
18128 @item luma_strength, ls
18129 Set the luma strength. The option value must be a float number
18130 in the range [-1.0,1.0] that configures the blurring. A value included
18131 in [0.0,1.0] will blur the image whereas a value included in
18132 [-1.0,0.0] will sharpen the image. Default value is 1.0.
18134 @item luma_threshold, lt
18135 Set the luma threshold used as a coefficient to determine
18136 whether a pixel should be blurred or not. The option value must be an
18137 integer in the range [-30,30]. A value of 0 will filter all the image,
18138 a value included in [0,30] will filter flat areas and a value included
18139 in [-30,0] will filter edges. Default value is 0.
18141 @item chroma_radius, cr
18142 Set the chroma radius. The option value must be a float number in
18143 the range [0.1,5.0] that specifies the variance of the gaussian filter
18144 used to blur the image (slower if larger). Default value is @option{luma_radius}.
18146 @item chroma_strength, cs
18147 Set the chroma strength. The option value must be a float number
18148 in the range [-1.0,1.0] that configures the blurring. A value included
18149 in [0.0,1.0] will blur the image whereas a value included in
18150 [-1.0,0.0] will sharpen the image. Default value is @option{luma_strength}.
18152 @item chroma_threshold, ct
18153 Set the chroma threshold used as a coefficient to determine
18154 whether a pixel should be blurred or not. The option value must be an
18155 integer in the range [-30,30]. A value of 0 will filter all the image,
18156 a value included in [0,30] will filter flat areas and a value included
18157 in [-30,0] will filter edges. Default value is @option{luma_threshold}.
18160 If a chroma option is not explicitly set, the corresponding luma value
18164 Apply sobel operator to input video stream.
18166 The filter accepts the following option:
18170 Set which planes will be processed, unprocessed planes will be copied.
18171 By default value 0xf, all planes will be processed.
18174 Set value which will be multiplied with filtered result.
18177 Set value which will be added to filtered result.
18180 @subsection Commands
18182 This filter supports the all above options as @ref{commands}.
18187 Apply a simple postprocessing filter that compresses and decompresses the image
18188 at several (or - in the case of @option{quality} level @code{6} - all) shifts
18189 and average the results.
18191 The filter accepts the following options:
18195 Set quality. This option defines the number of levels for averaging. It accepts
18196 an integer in the range 0-6. If set to @code{0}, the filter will have no
18197 effect. A value of @code{6} means the higher quality. For each increment of
18198 that value the speed drops by a factor of approximately 2. Default value is
18202 Force a constant quantization parameter. If not set, the filter will use the QP
18203 from the video stream (if available).
18206 Set thresholding mode. Available modes are:
18210 Set hard thresholding (default).
18212 Set soft thresholding (better de-ringing effect, but likely blurrier).
18215 @item use_bframe_qp
18216 Enable the use of the QP from the B-Frames if set to @code{1}. Using this
18217 option may cause flicker since the B-Frames have often larger QP. Default is
18218 @code{0} (not enabled).
18221 @subsection Commands
18223 This filter supports the following commands:
18225 @item quality, level
18226 Set quality level. The value @code{max} can be used to set the maximum level,
18227 currently @code{6}.
18233 Scale the input by applying one of the super-resolution methods based on
18234 convolutional neural networks. Supported models:
18238 Super-Resolution Convolutional Neural Network model (SRCNN).
18239 See @url{https://arxiv.org/abs/1501.00092}.
18242 Efficient Sub-Pixel Convolutional Neural Network model (ESPCN).
18243 See @url{https://arxiv.org/abs/1609.05158}.
18246 Training scripts as well as scripts for model file (.pb) saving can be found at
18247 @url{https://github.com/XueweiMeng/sr/tree/sr_dnn_native}. Original repository
18248 is at @url{https://github.com/HighVoltageRocknRoll/sr.git}.
18250 Native model files (.model) can be generated from TensorFlow model
18251 files (.pb) by using tools/python/convert.py
18253 The filter accepts the following options:
18257 Specify which DNN backend to use for model loading and execution. This option accepts
18258 the following values:
18262 Native implementation of DNN loading and execution.
18265 TensorFlow backend. To enable this backend you
18266 need to install the TensorFlow for C library (see
18267 @url{https://www.tensorflow.org/install/install_c}) and configure FFmpeg with
18268 @code{--enable-libtensorflow}
18271 Default value is @samp{native}.
18274 Set path to model file specifying network architecture and its parameters.
18275 Note that different backends use different file formats. TensorFlow backend
18276 can load files for both formats, while native backend can load files for only
18280 Set scale factor for SRCNN model. Allowed values are @code{2}, @code{3} and @code{4}.
18281 Default value is @code{2}. Scale factor is necessary for SRCNN model, because it accepts
18282 input upscaled using bicubic upscaling with proper scale factor.
18285 This feature can also be finished with @ref{dnn_processing} filter.
18289 Obtain the SSIM (Structural SImilarity Metric) between two input videos.
18291 This filter takes in input two input videos, the first input is
18292 considered the "main" source and is passed unchanged to the
18293 output. The second input is used as a "reference" video for computing
18296 Both video inputs must have the same resolution and pixel format for
18297 this filter to work correctly. Also it assumes that both inputs
18298 have the same number of frames, which are compared one by one.
18300 The filter stores the calculated SSIM of each frame.
18302 The description of the accepted parameters follows.
18305 @item stats_file, f
18306 If specified the filter will use the named file to save the SSIM of
18307 each individual frame. When filename equals "-" the data is sent to
18311 The file printed if @var{stats_file} is selected, contains a sequence of
18312 key/value pairs of the form @var{key}:@var{value} for each compared
18315 A description of each shown parameter follows:
18319 sequential number of the input frame, starting from 1
18321 @item Y, U, V, R, G, B
18322 SSIM of the compared frames for the component specified by the suffix.
18325 SSIM of the compared frames for the whole frame.
18328 Same as above but in dB representation.
18331 This filter also supports the @ref{framesync} options.
18333 @subsection Examples
18338 movie=ref_movie.mpg, setpts=PTS-STARTPTS [main];
18339 [main][ref] ssim="stats_file=stats.log" [out]
18342 On this example the input file being processed is compared with the
18343 reference file @file{ref_movie.mpg}. The SSIM of each individual frame
18344 is stored in @file{stats.log}.
18347 Another example with both psnr and ssim at same time:
18349 ffmpeg -i main.mpg -i ref.mpg -lavfi "ssim;[0:v][1:v]psnr" -f null -
18353 Another example with different containers:
18355 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 -
18361 Convert between different stereoscopic image formats.
18363 The filters accept the following options:
18367 Set stereoscopic image format of input.
18369 Available values for input image formats are:
18372 side by side parallel (left eye left, right eye right)
18375 side by side crosseye (right eye left, left eye right)
18378 side by side parallel with half width resolution
18379 (left eye left, right eye right)
18382 side by side crosseye with half width resolution
18383 (right eye left, left eye right)
18387 above-below (left eye above, right eye below)
18391 above-below (right eye above, left eye below)
18395 above-below with half height resolution
18396 (left eye above, right eye below)
18400 above-below with half height resolution
18401 (right eye above, left eye below)
18404 alternating frames (left eye first, right eye second)
18407 alternating frames (right eye first, left eye second)
18410 interleaved rows (left eye has top row, right eye starts on next row)
18413 interleaved rows (right eye has top row, left eye starts on next row)
18416 interleaved columns, left eye first
18419 interleaved columns, right eye first
18421 Default value is @samp{sbsl}.
18425 Set stereoscopic image format of output.
18429 side by side parallel (left eye left, right eye right)
18432 side by side crosseye (right eye left, left eye right)
18435 side by side parallel with half width resolution
18436 (left eye left, right eye right)
18439 side by side crosseye with half width resolution
18440 (right eye left, left eye right)
18444 above-below (left eye above, right eye below)
18448 above-below (right eye above, left eye below)
18452 above-below with half height resolution
18453 (left eye above, right eye below)
18457 above-below with half height resolution
18458 (right eye above, left eye below)
18461 alternating frames (left eye first, right eye second)
18464 alternating frames (right eye first, left eye second)
18467 interleaved rows (left eye has top row, right eye starts on next row)
18470 interleaved rows (right eye has top row, left eye starts on next row)
18473 anaglyph red/blue gray
18474 (red filter on left eye, blue filter on right eye)
18477 anaglyph red/green gray
18478 (red filter on left eye, green filter on right eye)
18481 anaglyph red/cyan gray
18482 (red filter on left eye, cyan filter on right eye)
18485 anaglyph red/cyan half colored
18486 (red filter on left eye, cyan filter on right eye)
18489 anaglyph red/cyan color
18490 (red filter on left eye, cyan filter on right eye)
18493 anaglyph red/cyan color optimized with the least squares projection of dubois
18494 (red filter on left eye, cyan filter on right eye)
18497 anaglyph green/magenta gray
18498 (green filter on left eye, magenta filter on right eye)
18501 anaglyph green/magenta half colored
18502 (green filter on left eye, magenta filter on right eye)
18505 anaglyph green/magenta colored
18506 (green filter on left eye, magenta filter on right eye)
18509 anaglyph green/magenta color optimized with the least squares projection of dubois
18510 (green filter on left eye, magenta filter on right eye)
18513 anaglyph yellow/blue gray
18514 (yellow filter on left eye, blue filter on right eye)
18517 anaglyph yellow/blue half colored
18518 (yellow filter on left eye, blue filter on right eye)
18521 anaglyph yellow/blue colored
18522 (yellow filter on left eye, blue filter on right eye)
18525 anaglyph yellow/blue color optimized with the least squares projection of dubois
18526 (yellow filter on left eye, blue filter on right eye)
18529 mono output (left eye only)
18532 mono output (right eye only)
18535 checkerboard, left eye first
18538 checkerboard, right eye first
18541 interleaved columns, left eye first
18544 interleaved columns, right eye first
18550 Default value is @samp{arcd}.
18553 @subsection Examples
18557 Convert input video from side by side parallel to anaglyph yellow/blue dubois:
18563 Convert input video from above below (left eye above, right eye below) to side by side crosseye.
18569 @section streamselect, astreamselect
18570 Select video or audio streams.
18572 The filter accepts the following options:
18576 Set number of inputs. Default is 2.
18579 Set input indexes to remap to outputs.
18582 @subsection Commands
18584 The @code{streamselect} and @code{astreamselect} filter supports the following
18589 Set input indexes to remap to outputs.
18592 @subsection Examples
18596 Select first 5 seconds 1st stream and rest of time 2nd stream:
18598 sendcmd='5.0 streamselect map 1',streamselect=inputs=2:map=0
18602 Same as above, but for audio:
18604 asendcmd='5.0 astreamselect map 1',astreamselect=inputs=2:map=0
18611 Draw subtitles on top of input video using the libass library.
18613 To enable compilation of this filter you need to configure FFmpeg with
18614 @code{--enable-libass}. This filter also requires a build with libavcodec and
18615 libavformat to convert the passed subtitles file to ASS (Advanced Substation
18616 Alpha) subtitles format.
18618 The filter accepts the following options:
18622 Set the filename of the subtitle file to read. It must be specified.
18624 @item original_size
18625 Specify the size of the original video, the video for which the ASS file
18626 was composed. For the syntax of this option, check the
18627 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18628 Due to a misdesign in ASS aspect ratio arithmetic, this is necessary to
18629 correctly scale the fonts if the aspect ratio has been changed.
18632 Set a directory path containing fonts that can be used by the filter.
18633 These fonts will be used in addition to whatever the font provider uses.
18636 Process alpha channel, by default alpha channel is untouched.
18639 Set subtitles input character encoding. @code{subtitles} filter only. Only
18640 useful if not UTF-8.
18642 @item stream_index, si
18643 Set subtitles stream index. @code{subtitles} filter only.
18646 Override default style or script info parameters of the subtitles. It accepts a
18647 string containing ASS style format @code{KEY=VALUE} couples separated by ",".
18650 If the first key is not specified, it is assumed that the first value
18651 specifies the @option{filename}.
18653 For example, to render the file @file{sub.srt} on top of the input
18654 video, use the command:
18659 which is equivalent to:
18661 subtitles=filename=sub.srt
18664 To render the default subtitles stream from file @file{video.mkv}, use:
18666 subtitles=video.mkv
18669 To render the second subtitles stream from that file, use:
18671 subtitles=video.mkv:si=1
18674 To make the subtitles stream from @file{sub.srt} appear in 80% transparent blue
18675 @code{DejaVu Serif}, use:
18677 subtitles=sub.srt:force_style='Fontname=DejaVu Serif,PrimaryColour=&HCCFF0000'
18680 @section super2xsai
18682 Scale the input by 2x and smooth using the Super2xSaI (Scale and
18683 Interpolate) pixel art scaling algorithm.
18685 Useful for enlarging pixel art images without reducing sharpness.
18689 Swap two rectangular objects in video.
18691 This filter accepts the following options:
18701 Set 1st rect x coordinate.
18704 Set 1st rect y coordinate.
18707 Set 2nd rect x coordinate.
18710 Set 2nd rect y coordinate.
18712 All expressions are evaluated once for each frame.
18715 The all options are expressions containing the following constants:
18720 The input width and height.
18723 same as @var{w} / @var{h}
18726 input sample aspect ratio
18729 input display aspect ratio, it is the same as (@var{w} / @var{h}) * @var{sar}
18732 The number of the input frame, starting from 0.
18735 The timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
18738 the position in the file of the input frame, NAN if unknown
18745 Blend successive video frames.
18751 Apply telecine process to the video.
18753 This filter accepts the following options:
18762 The default value is @code{top}.
18766 A string of numbers representing the pulldown pattern you wish to apply.
18767 The default value is @code{23}.
18771 Some typical patterns:
18776 24p: 2332 (preferred)
18783 24p: 222222222223 ("Euro pulldown")
18788 @section thistogram
18790 Compute and draw a color distribution histogram for the input video across time.
18792 Unlike @ref{histogram} video filter which only shows histogram of single input frame
18793 at certain time, this filter shows also past histograms of number of frames defined
18794 by @code{width} option.
18796 The computed histogram is a representation of the color component
18797 distribution in an image.
18799 The filter accepts the following options:
18803 Set width of single color component output. Default value is @code{0}.
18804 Value of @code{0} means width will be picked from input video.
18805 This also set number of passed histograms to keep.
18806 Allowed range is [0, 8192].
18808 @item display_mode, d
18810 It accepts the following values:
18813 Per color component graphs are placed below each other.
18816 Per color component graphs are placed side by side.
18819 Presents information identical to that in the @code{parade}, except
18820 that the graphs representing color components are superimposed directly
18823 Default is @code{stack}.
18825 @item levels_mode, m
18826 Set mode. Can be either @code{linear}, or @code{logarithmic}.
18827 Default is @code{linear}.
18829 @item components, c
18830 Set what color components to display.
18831 Default is @code{7}.
18834 Set background opacity. Default is @code{0.9}.
18837 Show envelope. Default is disabled.
18840 Set envelope color. Default is @code{gold}.
18845 Available values for slide is:
18848 Draw new frame when right border is reached.
18851 Replace old columns with new ones.
18854 Scroll from right to left.
18857 Scroll from left to right.
18860 Draw single picture.
18863 Default is @code{replace}.
18868 Apply threshold effect to video stream.
18870 This filter needs four video streams to perform thresholding.
18871 First stream is stream we are filtering.
18872 Second stream is holding threshold values, third stream is holding min values,
18873 and last, fourth stream is holding max values.
18875 The filter accepts the following option:
18879 Set which planes will be processed, unprocessed planes will be copied.
18880 By default value 0xf, all planes will be processed.
18883 For example if first stream pixel's component value is less then threshold value
18884 of pixel component from 2nd threshold stream, third stream value will picked,
18885 otherwise fourth stream pixel component value will be picked.
18887 Using color source filter one can perform various types of thresholding:
18889 @subsection Examples
18893 Binary threshold, using gray color as threshold:
18895 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=black -f lavfi -i color=white -lavfi threshold output.avi
18899 Inverted binary threshold, using gray color as threshold:
18901 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -f lavfi -i color=black -lavfi threshold output.avi
18905 Truncate binary threshold, using gray color as threshold:
18907 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=gray -lavfi threshold output.avi
18911 Threshold to zero, using gray color as threshold:
18913 ffmpeg -i 320x240.avi -f lavfi -i color=gray -f lavfi -i color=white -i 320x240.avi -lavfi threshold output.avi
18917 Inverted threshold to zero, using gray color as threshold:
18919 ffmpeg -i 320x240.avi -f lavfi -i color=gray -i 320x240.avi -f lavfi -i color=white -lavfi threshold output.avi
18924 Select the most representative frame in a given sequence of consecutive frames.
18926 The filter accepts the following options:
18930 Set the frames batch size to analyze; in a set of @var{n} frames, the filter
18931 will pick one of them, and then handle the next batch of @var{n} frames until
18932 the end. Default is @code{100}.
18935 Since the filter keeps track of the whole frames sequence, a bigger @var{n}
18936 value will result in a higher memory usage, so a high value is not recommended.
18938 @subsection Examples
18942 Extract one picture each 50 frames:
18948 Complete example of a thumbnail creation with @command{ffmpeg}:
18950 ffmpeg -i in.avi -vf thumbnail,scale=300:200 -frames:v 1 out.png
18957 Tile several successive frames together.
18959 The @ref{untile} filter can do the reverse.
18961 The filter accepts the following options:
18966 Set the grid size (i.e. the number of lines and columns). For the syntax of
18967 this option, check the
18968 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
18971 Set the maximum number of frames to render in the given area. It must be less
18972 than or equal to @var{w}x@var{h}. The default value is @code{0}, meaning all
18973 the area will be used.
18976 Set the outer border margin in pixels.
18979 Set the inner border thickness (i.e. the number of pixels between frames). For
18980 more advanced padding options (such as having different values for the edges),
18981 refer to the pad video filter.
18984 Specify the color of the unused area. For the syntax of this option, check the
18985 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
18986 The default value of @var{color} is "black".
18989 Set the number of frames to overlap when tiling several successive frames together.
18990 The value must be between @code{0} and @var{nb_frames - 1}.
18993 Set the number of frames to initially be empty before displaying first output frame.
18994 This controls how soon will one get first output frame.
18995 The value must be between @code{0} and @var{nb_frames - 1}.
18998 @subsection Examples
19002 Produce 8x8 PNG tiles of all keyframes (@option{-skip_frame nokey}) in a movie:
19004 ffmpeg -skip_frame nokey -i file.avi -vf 'scale=128:72,tile=8x8' -an -vsync 0 keyframes%03d.png
19006 The @option{-vsync 0} is necessary to prevent @command{ffmpeg} from
19007 duplicating each output frame to accommodate the originally detected frame
19011 Display @code{5} pictures in an area of @code{3x2} frames,
19012 with @code{7} pixels between them, and @code{2} pixels of initial margin, using
19013 mixed flat and named options:
19015 tile=3x2:nb_frames=5:padding=7:margin=2
19019 @section tinterlace
19021 Perform various types of temporal field interlacing.
19023 Frames are counted starting from 1, so the first input frame is
19026 The filter accepts the following options:
19031 Specify the mode of the interlacing. This option can also be specified
19032 as a value alone. See below for a list of values for this option.
19034 Available values are:
19038 Move odd frames into the upper field, even into the lower field,
19039 generating a double height frame at half frame rate.
19043 Frame 1 Frame 2 Frame 3 Frame 4
19045 11111 22222 33333 44444
19046 11111 22222 33333 44444
19047 11111 22222 33333 44444
19048 11111 22222 33333 44444
19062 Only output odd frames, even frames are dropped, generating a frame with
19063 unchanged height at half frame rate.
19068 Frame 1 Frame 2 Frame 3 Frame 4
19070 11111 22222 33333 44444
19071 11111 22222 33333 44444
19072 11111 22222 33333 44444
19073 11111 22222 33333 44444
19083 Only output even frames, odd frames are dropped, generating a frame with
19084 unchanged height at half frame rate.
19089 Frame 1 Frame 2 Frame 3 Frame 4
19091 11111 22222 33333 44444
19092 11111 22222 33333 44444
19093 11111 22222 33333 44444
19094 11111 22222 33333 44444
19104 Expand each frame to full height, but pad alternate lines with black,
19105 generating a frame with double height at the same input frame rate.
19110 Frame 1 Frame 2 Frame 3 Frame 4
19112 11111 22222 33333 44444
19113 11111 22222 33333 44444
19114 11111 22222 33333 44444
19115 11111 22222 33333 44444
19118 11111 ..... 33333 .....
19119 ..... 22222 ..... 44444
19120 11111 ..... 33333 .....
19121 ..... 22222 ..... 44444
19122 11111 ..... 33333 .....
19123 ..... 22222 ..... 44444
19124 11111 ..... 33333 .....
19125 ..... 22222 ..... 44444
19129 @item interleave_top, 4
19130 Interleave the upper field from odd frames with the lower field from
19131 even frames, generating a frame with unchanged height at half frame rate.
19136 Frame 1 Frame 2 Frame 3 Frame 4
19138 11111<- 22222 33333<- 44444
19139 11111 22222<- 33333 44444<-
19140 11111<- 22222 33333<- 44444
19141 11111 22222<- 33333 44444<-
19151 @item interleave_bottom, 5
19152 Interleave the lower field from odd frames with the upper field from
19153 even frames, generating a frame with unchanged height at half frame rate.
19158 Frame 1 Frame 2 Frame 3 Frame 4
19160 11111 22222<- 33333 44444<-
19161 11111<- 22222 33333<- 44444
19162 11111 22222<- 33333 44444<-
19163 11111<- 22222 33333<- 44444
19173 @item interlacex2, 6
19174 Double frame rate with unchanged height. Frames are inserted each
19175 containing the second temporal field from the previous input frame and
19176 the first temporal field from the next input frame. This mode relies on
19177 the top_field_first flag. Useful for interlaced video displays with no
19178 field synchronisation.
19183 Frame 1 Frame 2 Frame 3 Frame 4
19185 11111 22222 33333 44444
19186 11111 22222 33333 44444
19187 11111 22222 33333 44444
19188 11111 22222 33333 44444
19191 11111 22222 22222 33333 33333 44444 44444
19192 11111 11111 22222 22222 33333 33333 44444
19193 11111 22222 22222 33333 33333 44444 44444
19194 11111 11111 22222 22222 33333 33333 44444
19199 Move odd frames into the upper field, even into the lower field,
19200 generating a double height frame at same frame rate.
19205 Frame 1 Frame 2 Frame 3 Frame 4
19207 11111 22222 33333 44444
19208 11111 22222 33333 44444
19209 11111 22222 33333 44444
19210 11111 22222 33333 44444
19213 11111 33333 33333 55555
19214 22222 22222 44444 44444
19215 11111 33333 33333 55555
19216 22222 22222 44444 44444
19217 11111 33333 33333 55555
19218 22222 22222 44444 44444
19219 11111 33333 33333 55555
19220 22222 22222 44444 44444
19225 Numeric values are deprecated but are accepted for backward
19226 compatibility reasons.
19228 Default mode is @code{merge}.
19231 Specify flags influencing the filter process.
19233 Available value for @var{flags} is:
19236 @item low_pass_filter, vlpf
19237 Enable linear vertical low-pass filtering in the filter.
19238 Vertical low-pass filtering is required when creating an interlaced
19239 destination from a progressive source which contains high-frequency
19240 vertical detail. Filtering will reduce interlace 'twitter' and Moire
19243 @item complex_filter, cvlpf
19244 Enable complex vertical low-pass filtering.
19245 This will slightly less reduce interlace 'twitter' and Moire
19246 patterning but better retain detail and subjective sharpness impression.
19249 Bypass already interlaced frames, only adjust the frame rate.
19252 Vertical low-pass filtering and bypassing already interlaced frames can only be
19253 enabled for @option{mode} @var{interleave_top} and @var{interleave_bottom}.
19258 Pick median pixels from several successive input video frames.
19260 The filter accepts the following options:
19264 Set radius of median filter.
19265 Default is 1. Allowed range is from 1 to 127.
19268 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
19271 Set median percentile. Default value is @code{0.5}.
19272 Default value of @code{0.5} will pick always median values, while @code{0} will pick
19273 minimum values, and @code{1} maximum values.
19278 Mix successive video frames.
19280 A description of the accepted options follows.
19284 The number of successive frames to mix. If unspecified, it defaults to 3.
19287 Specify weight of each input video frame.
19288 Each weight is separated by space. If number of weights is smaller than
19289 number of @var{frames} last specified weight will be used for all remaining
19293 Specify scale, if it is set it will be multiplied with sum
19294 of each weight multiplied with pixel values to give final destination
19295 pixel value. By default @var{scale} is auto scaled to sum of weights.
19298 @subsection Examples
19302 Average 7 successive frames:
19304 tmix=frames=7:weights="1 1 1 1 1 1 1"
19308 Apply simple temporal convolution:
19310 tmix=frames=3:weights="-1 3 -1"
19314 Similar as above but only showing temporal differences:
19316 tmix=frames=3:weights="-1 2 -1":scale=1
19322 Tone map colors from different dynamic ranges.
19324 This filter expects data in single precision floating point, as it needs to
19325 operate on (and can output) out-of-range values. Another filter, such as
19326 @ref{zscale}, is needed to convert the resulting frame to a usable format.
19328 The tonemapping algorithms implemented only work on linear light, so input
19329 data should be linearized beforehand (and possibly correctly tagged).
19332 ffmpeg -i INPUT -vf zscale=transfer=linear,tonemap=clip,zscale=transfer=bt709,format=yuv420p OUTPUT
19335 @subsection Options
19336 The filter accepts the following options.
19340 Set the tone map algorithm to use.
19342 Possible values are:
19345 Do not apply any tone map, only desaturate overbright pixels.
19348 Hard-clip any out-of-range values. Use it for perfect color accuracy for
19349 in-range values, while distorting out-of-range values.
19352 Stretch the entire reference gamut to a linear multiple of the display.
19355 Fit a logarithmic transfer between the tone curves.
19358 Preserve overall image brightness with a simple curve, using nonlinear
19359 contrast, which results in flattening details and degrading color accuracy.
19362 Preserve both dark and bright details better than @var{reinhard}, at the cost
19363 of slightly darkening everything. Use it when detail preservation is more
19364 important than color and brightness accuracy.
19367 Smoothly map out-of-range values, while retaining contrast and colors for
19368 in-range material as much as possible. Use it when color accuracy is more
19369 important than detail preservation.
19375 Tune the tone mapping algorithm.
19377 This affects the following algorithms:
19383 Specifies the scale factor to use while stretching.
19387 Specifies the exponent of the function.
19391 Specify an extra linear coefficient to multiply into the signal before clipping.
19395 Specify the local contrast coefficient at the display peak.
19396 Default to 0.5, which means that in-gamut values will be about half as bright
19403 Specify the transition point from linear to mobius transform. Every value
19404 below this point is guaranteed to be mapped 1:1. The higher the value, the
19405 more accurate the result will be, at the cost of losing bright details.
19406 Default to 0.3, which due to the steep initial slope still preserves in-range
19407 colors fairly accurately.
19411 Apply desaturation for highlights that exceed this level of brightness. The
19412 higher the parameter, the more color information will be preserved. This
19413 setting helps prevent unnaturally blown-out colors for super-highlights, by
19414 (smoothly) turning into white instead. This makes images feel more natural,
19415 at the cost of reducing information about out-of-range colors.
19417 The default of 2.0 is somewhat conservative and will mostly just apply to
19418 skies or directly sunlit surfaces. A setting of 0.0 disables this option.
19420 This option works only if the input frame has a supported color tag.
19423 Override signal/nominal/reference peak with this value. Useful when the
19424 embedded peak information in display metadata is not reliable or when tone
19425 mapping from a lower range to a higher range.
19430 Temporarily pad video frames.
19432 The filter accepts the following options:
19436 Specify number of delay frames before input video stream. Default is 0.
19439 Specify number of padding frames after input video stream.
19440 Set to -1 to pad indefinitely. Default is 0.
19443 Set kind of frames added to beginning of stream.
19444 Can be either @var{add} or @var{clone}.
19445 With @var{add} frames of solid-color are added.
19446 With @var{clone} frames are clones of first frame.
19447 Default is @var{add}.
19450 Set kind of frames added to end of stream.
19451 Can be either @var{add} or @var{clone}.
19452 With @var{add} frames of solid-color are added.
19453 With @var{clone} frames are clones of last frame.
19454 Default is @var{add}.
19456 @item start_duration, stop_duration
19457 Specify the duration of the start/stop delay. See
19458 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19459 for the accepted syntax.
19460 These options override @var{start} and @var{stop}. Default is 0.
19463 Specify the color of the padded area. For the syntax of this option,
19464 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
19465 manual,ffmpeg-utils}.
19467 The default value of @var{color} is "black".
19473 Transpose rows with columns in the input video and optionally flip it.
19475 It accepts the following parameters:
19480 Specify the transposition direction.
19482 Can assume the following values:
19484 @item 0, 4, cclock_flip
19485 Rotate by 90 degrees counterclockwise and vertically flip (default), that is:
19493 Rotate by 90 degrees clockwise, that is:
19501 Rotate by 90 degrees counterclockwise, that is:
19508 @item 3, 7, clock_flip
19509 Rotate by 90 degrees clockwise and vertically flip, that is:
19517 For values between 4-7, the transposition is only done if the input
19518 video geometry is portrait and not landscape. These values are
19519 deprecated, the @code{passthrough} option should be used instead.
19521 Numerical values are deprecated, and should be dropped in favor of
19522 symbolic constants.
19525 Do not apply the transposition if the input geometry matches the one
19526 specified by the specified value. It accepts the following values:
19529 Always apply transposition.
19531 Preserve portrait geometry (when @var{height} >= @var{width}).
19533 Preserve landscape geometry (when @var{width} >= @var{height}).
19536 Default value is @code{none}.
19539 For example to rotate by 90 degrees clockwise and preserve portrait
19542 transpose=dir=1:passthrough=portrait
19545 The command above can also be specified as:
19547 transpose=1:portrait
19550 @section transpose_npp
19552 Transpose rows with columns in the input video and optionally flip it.
19553 For more in depth examples see the @ref{transpose} video filter, which shares mostly the same options.
19555 It accepts the following parameters:
19560 Specify the transposition direction.
19562 Can assume the following values:
19565 Rotate by 90 degrees counterclockwise and vertically flip. (default)
19568 Rotate by 90 degrees clockwise.
19571 Rotate by 90 degrees counterclockwise.
19574 Rotate by 90 degrees clockwise and vertically flip.
19578 Do not apply the transposition if the input geometry matches the one
19579 specified by the specified value. It accepts the following values:
19582 Always apply transposition. (default)
19584 Preserve portrait geometry (when @var{height} >= @var{width}).
19586 Preserve landscape geometry (when @var{width} >= @var{height}).
19592 Trim the input so that the output contains one continuous subpart of the input.
19594 It accepts the following parameters:
19597 Specify the time of the start of the kept section, i.e. the frame with the
19598 timestamp @var{start} will be the first frame in the output.
19601 Specify the time of the first frame that will be dropped, i.e. the frame
19602 immediately preceding the one with the timestamp @var{end} will be the last
19603 frame in the output.
19606 This is the same as @var{start}, except this option sets the start timestamp
19607 in timebase units instead of seconds.
19610 This is the same as @var{end}, except this option sets the end timestamp
19611 in timebase units instead of seconds.
19614 The maximum duration of the output in seconds.
19617 The number of the first frame that should be passed to the output.
19620 The number of the first frame that should be dropped.
19623 @option{start}, @option{end}, and @option{duration} are expressed as time
19624 duration specifications; see
19625 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
19626 for the accepted syntax.
19628 Note that the first two sets of the start/end options and the @option{duration}
19629 option look at the frame timestamp, while the _frame variants simply count the
19630 frames that pass through the filter. Also note that this filter does not modify
19631 the timestamps. If you wish for the output timestamps to start at zero, insert a
19632 setpts filter after the trim filter.
19634 If multiple start or end options are set, this filter tries to be greedy and
19635 keep all the frames that match at least one of the specified constraints. To keep
19636 only the part that matches all the constraints at once, chain multiple trim
19639 The defaults are such that all the input is kept. So it is possible to set e.g.
19640 just the end values to keep everything before the specified time.
19645 Drop everything except the second minute of input:
19647 ffmpeg -i INPUT -vf trim=60:120
19651 Keep only the first second:
19653 ffmpeg -i INPUT -vf trim=duration=1
19658 @section unpremultiply
19659 Apply alpha unpremultiply effect to input video stream using first plane
19660 of second stream as alpha.
19662 Both streams must have same dimensions and same pixel format.
19664 The filter accepts the following option:
19668 Set which planes will be processed, unprocessed planes will be copied.
19669 By default value 0xf, all planes will be processed.
19671 If the format has 1 or 2 components, then luma is bit 0.
19672 If the format has 3 or 4 components:
19673 for RGB formats bit 0 is green, bit 1 is blue and bit 2 is red;
19674 for YUV formats bit 0 is luma, bit 1 is chroma-U and bit 2 is chroma-V.
19675 If present, the alpha channel is always the last bit.
19678 Do not require 2nd input for processing, instead use alpha plane from input stream.
19684 Sharpen or blur the input video.
19686 It accepts the following parameters:
19689 @item luma_msize_x, lx
19690 Set the luma matrix horizontal size. It must be an odd integer between
19691 3 and 23. The default value is 5.
19693 @item luma_msize_y, ly
19694 Set the luma matrix vertical size. It must be an odd integer between 3
19695 and 23. The default value is 5.
19697 @item luma_amount, la
19698 Set the luma effect strength. It must be a floating point number, reasonable
19699 values lay between -1.5 and 1.5.
19701 Negative values will blur the input video, while positive values will
19702 sharpen it, a value of zero will disable the effect.
19704 Default value is 1.0.
19706 @item chroma_msize_x, cx
19707 Set the chroma matrix horizontal size. It must be an odd integer
19708 between 3 and 23. The default value is 5.
19710 @item chroma_msize_y, cy
19711 Set the chroma matrix vertical size. It must be an odd integer
19712 between 3 and 23. The default value is 5.
19714 @item chroma_amount, ca
19715 Set the chroma effect strength. It must be a floating point number, reasonable
19716 values lay between -1.5 and 1.5.
19718 Negative values will blur the input video, while positive values will
19719 sharpen it, a value of zero will disable the effect.
19721 Default value is 0.0.
19725 All parameters are optional and default to the equivalent of the
19726 string '5:5:1.0:5:5:0.0'.
19728 @subsection Examples
19732 Apply strong luma sharpen effect:
19734 unsharp=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5
19738 Apply a strong blur of both luma and chroma parameters:
19740 unsharp=7:7:-2:7:7:-2
19747 Decompose a video made of tiled images into the individual images.
19749 The frame rate of the output video is the frame rate of the input video
19750 multiplied by the number of tiles.
19752 This filter does the reverse of @ref{tile}.
19754 The filter accepts the following options:
19759 Set the grid size (i.e. the number of lines and columns). For the syntax of
19760 this option, check the
19761 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
19764 @subsection Examples
19768 Produce a 1-second video from a still image file made of 25 frames stacked
19769 vertically, like an analogic film reel:
19771 ffmpeg -r 1 -i image.jpg -vf untile=1x25 movie.mkv
19777 Apply ultra slow/simple postprocessing filter that compresses and decompresses
19778 the image at several (or - in the case of @option{quality} level @code{8} - all)
19779 shifts and average the results.
19781 The way this differs from the behavior of spp is that uspp actually encodes &
19782 decodes each case with libavcodec Snow, whereas spp uses a simplified intra only 8x8
19783 DCT similar to MJPEG.
19785 The filter accepts the following options:
19789 Set quality. This option defines the number of levels for averaging. It accepts
19790 an integer in the range 0-8. If set to @code{0}, the filter will have no
19791 effect. A value of @code{8} means the higher quality. For each increment of
19792 that value the speed drops by a factor of approximately 2. Default value is
19796 Force a constant quantization parameter. If not set, the filter will use the QP
19797 from the video stream (if available).
19802 Convert 360 videos between various formats.
19804 The filter accepts the following options:
19810 Set format of the input/output video.
19818 Equirectangular projection.
19823 Cubemap with 3x2/6x1/1x6 layout.
19825 Format specific options:
19830 Set padding proportion for the input/output cubemap. Values in decimals.
19837 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)
19840 Default value is @b{@samp{0}}.
19841 Maximum value is @b{@samp{0.1}}.
19845 Set fixed padding for the input/output cubemap. Values in pixels.
19847 Default value is @b{@samp{0}}. If greater than zero it overrides other padding options.
19851 Set order of faces for the input/output cubemap. Choose one direction for each position.
19853 Designation of directions:
19869 Default value is @b{@samp{rludfb}}.
19873 Set rotation of faces for the input/output cubemap. Choose one angle for each position.
19875 Designation of angles:
19878 0 degrees clockwise
19880 90 degrees clockwise
19882 180 degrees clockwise
19884 270 degrees clockwise
19887 Default value is @b{@samp{000000}}.
19891 Equi-Angular Cubemap.
19898 Format specific options:
19903 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19905 If diagonal field of view is set it overrides horizontal and vertical field of view.
19910 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19912 If diagonal field of view is set it overrides horizontal and vertical field of view.
19918 Format specific options:
19923 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19925 If diagonal field of view is set it overrides horizontal and vertical field of view.
19930 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19932 If diagonal field of view is set it overrides horizontal and vertical field of view.
19938 Facebook's 360 formats.
19941 Stereographic format.
19943 Format specific options:
19948 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19950 If diagonal field of view is set it overrides horizontal and vertical field of view.
19955 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19957 If diagonal field of view is set it overrides horizontal and vertical field of view.
19964 Ball format, gives significant distortion toward the back.
19967 Hammer-Aitoff map projection format.
19970 Sinusoidal map projection format.
19973 Fisheye projection.
19975 Format specific options:
19980 Set output horizontal/vertical/diagonal field of view. Values in degrees.
19982 If diagonal field of view is set it overrides horizontal and vertical field of view.
19987 Set input horizontal/vertical/diagonal field of view. Values in degrees.
19989 If diagonal field of view is set it overrides horizontal and vertical field of view.
19993 Pannini projection.
19995 Format specific options:
19998 Set output pannini parameter.
20001 Set input pannini parameter.
20005 Cylindrical projection.
20007 Format specific options:
20012 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20014 If diagonal field of view is set it overrides horizontal and vertical field of view.
20019 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20021 If diagonal field of view is set it overrides horizontal and vertical field of view.
20025 Perspective projection. @i{(output only)}
20027 Format specific options:
20030 Set perspective parameter.
20034 Tetrahedron projection.
20037 Truncated square pyramid projection.
20041 Half equirectangular projection.
20046 Format specific options:
20051 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20053 If diagonal field of view is set it overrides horizontal and vertical field of view.
20058 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20060 If diagonal field of view is set it overrides horizontal and vertical field of view.
20064 Orthographic format.
20066 Format specific options:
20071 Set output horizontal/vertical/diagonal field of view. Values in degrees.
20073 If diagonal field of view is set it overrides horizontal and vertical field of view.
20078 Set input horizontal/vertical/diagonal field of view. Values in degrees.
20080 If diagonal field of view is set it overrides horizontal and vertical field of view.
20084 Octahedron projection.
20088 Set interpolation method.@*
20089 @i{Note: more complex interpolation methods require much more memory to run.}
20099 Bilinear interpolation.
20101 Lagrange9 interpolation.
20104 Bicubic interpolation.
20107 Lanczos interpolation.
20110 Spline16 interpolation.
20113 Gaussian interpolation.
20115 Mitchell interpolation.
20118 Default value is @b{@samp{line}}.
20122 Set the output video resolution.
20124 Default resolution depends on formats.
20128 Set the input/output stereo format.
20139 Default value is @b{@samp{2d}} for input and output format.
20144 Set rotation for the output video. Values in degrees.
20147 Set rotation order for the output video. Choose one item for each position.
20158 Default value is @b{@samp{ypr}}.
20163 Flip the output video horizontally(swaps left-right)/vertically(swaps up-down)/in-depth(swaps back-forward). Boolean values.
20167 Set if input video is flipped horizontally/vertically. Boolean values.
20170 Set if input video is transposed. Boolean value, by default disabled.
20173 Set if output video needs to be transposed. Boolean value, by default disabled.
20176 Build mask in alpha plane for all unmapped pixels by marking them fully transparent. Boolean value, by default disabled.
20179 @subsection Examples
20183 Convert equirectangular video to cubemap with 3x2 layout and 1% padding using bicubic interpolation:
20185 ffmpeg -i input.mkv -vf v360=e:c3x2:cubic:out_pad=0.01 output.mkv
20188 Extract back view of Equi-Angular Cubemap:
20190 ffmpeg -i input.mkv -vf v360=eac:flat:yaw=180 output.mkv
20193 Convert transposed and horizontally flipped Equi-Angular Cubemap in side-by-side stereo format to equirectangular top-bottom stereo format:
20195 v360=eac:equirect:in_stereo=sbs:in_trans=1:ih_flip=1:out_stereo=tb
20199 @subsection Commands
20201 This filter supports subset of above options as @ref{commands}.
20203 @section vaguedenoiser
20205 Apply a wavelet based denoiser.
20207 It transforms each frame from the video input into the wavelet domain,
20208 using Cohen-Daubechies-Feauveau 9/7. Then it applies some filtering to
20209 the obtained coefficients. It does an inverse wavelet transform after.
20210 Due to wavelet properties, it should give a nice smoothed result, and
20211 reduced noise, without blurring picture features.
20213 This filter accepts the following options:
20217 The filtering strength. The higher, the more filtered the video will be.
20218 Hard thresholding can use a higher threshold than soft thresholding
20219 before the video looks overfiltered. Default value is 2.
20222 The filtering method the filter will use.
20224 It accepts the following values:
20227 All values under the threshold will be zeroed.
20230 All values under the threshold will be zeroed. All values above will be
20231 reduced by the threshold.
20234 Scales or nullifies coefficients - intermediary between (more) soft and
20235 (less) hard thresholding.
20238 Default is garrote.
20241 Number of times, the wavelet will decompose the picture. Picture can't
20242 be decomposed beyond a particular point (typically, 8 for a 640x480
20243 frame - as 2^9 = 512 > 480). Valid values are integers between 1 and 32. Default value is 6.
20246 Partial of full denoising (limited coefficients shrinking), from 0 to 100. Default value is 85.
20249 A list of the planes to process. By default all planes are processed.
20252 The threshold type the filter will use.
20254 It accepts the following values:
20257 Threshold used is same for all decompositions.
20260 Threshold used depends also on each decomposition coefficients.
20263 Default is universal.
20266 @section vectorscope
20268 Display 2 color component values in the two dimensional graph (which is called
20271 This filter accepts the following options:
20275 Set vectorscope mode.
20277 It accepts the following values:
20281 Gray values are displayed on graph, higher brightness means more pixels have
20282 same component color value on location in graph. This is the default mode.
20285 Gray values are displayed on graph. Surrounding pixels values which are not
20286 present in video frame are drawn in gradient of 2 color components which are
20287 set by option @code{x} and @code{y}. The 3rd color component is static.
20290 Actual color components values present in video frame are displayed on graph.
20293 Similar as color2 but higher frequency of same values @code{x} and @code{y}
20294 on graph increases value of another color component, which is luminance by
20295 default values of @code{x} and @code{y}.
20298 Actual colors present in video frame are displayed on graph. If two different
20299 colors map to same position on graph then color with higher value of component
20300 not present in graph is picked.
20303 Gray values are displayed on graph. Similar to @code{color} but with 3rd color
20304 component picked from radial gradient.
20308 Set which color component will be represented on X-axis. Default is @code{1}.
20311 Set which color component will be represented on Y-axis. Default is @code{2}.
20314 Set intensity, used by modes: gray, color, color3 and color5 for increasing brightness
20315 of color component which represents frequency of (X, Y) location in graph.
20320 No envelope, this is default.
20323 Instant envelope, even darkest single pixel will be clearly highlighted.
20326 Hold maximum and minimum values presented in graph over time. This way you
20327 can still spot out of range values without constantly looking at vectorscope.
20330 Peak and instant envelope combined together.
20334 Set what kind of graticule to draw.
20343 Set graticule opacity.
20346 Set graticule flags.
20350 Draw graticule for white point.
20353 Draw graticule for black point.
20356 Draw color points short names.
20360 Set background opacity.
20362 @item lthreshold, l
20363 Set low threshold for color component not represented on X or Y axis.
20364 Values lower than this value will be ignored. Default is 0.
20365 Note this value is multiplied with actual max possible value one pixel component
20366 can have. So for 8-bit input and low threshold value of 0.1 actual threshold
20369 @item hthreshold, h
20370 Set high threshold for color component not represented on X or Y axis.
20371 Values higher than this value will be ignored. Default is 1.
20372 Note this value is multiplied with actual max possible value one pixel component
20373 can have. So for 8-bit input and high threshold value of 0.9 actual threshold
20374 is 0.9 * 255 = 230.
20376 @item colorspace, c
20377 Set what kind of colorspace to use when drawing graticule.
20387 Set color tint for gray/tint vectorscope mode. By default both options are zero.
20388 This means no tint, and output will remain gray.
20391 @anchor{vidstabdetect}
20392 @section vidstabdetect
20394 Analyze video stabilization/deshaking. Perform pass 1 of 2, see
20395 @ref{vidstabtransform} for pass 2.
20397 This filter generates a file with relative translation and rotation
20398 transform information about subsequent frames, which is then used by
20399 the @ref{vidstabtransform} filter.
20401 To enable compilation of this filter you need to configure FFmpeg with
20402 @code{--enable-libvidstab}.
20404 This filter accepts the following options:
20408 Set the path to the file used to write the transforms information.
20409 Default value is @file{transforms.trf}.
20412 Set how shaky the video is and how quick the camera is. It accepts an
20413 integer in the range 1-10, a value of 1 means little shakiness, a
20414 value of 10 means strong shakiness. Default value is 5.
20417 Set the accuracy of the detection process. It must be a value in the
20418 range 1-15. A value of 1 means low accuracy, a value of 15 means high
20419 accuracy. Default value is 15.
20422 Set stepsize of the search process. The region around minimum is
20423 scanned with 1 pixel resolution. Default value is 6.
20426 Set minimum contrast. Below this value a local measurement field is
20427 discarded. Must be a floating point value in the range 0-1. Default
20431 Set reference frame number for tripod mode.
20433 If enabled, the motion of the frames is compared to a reference frame
20434 in the filtered stream, identified by the specified number. The idea
20435 is to compensate all movements in a more-or-less static scene and keep
20436 the camera view absolutely still.
20438 If set to 0, it is disabled. The frames are counted starting from 1.
20441 Show fields and transforms in the resulting frames. It accepts an
20442 integer in the range 0-2. Default value is 0, which disables any
20446 @subsection Examples
20450 Use default values:
20456 Analyze strongly shaky movie and put the results in file
20457 @file{mytransforms.trf}:
20459 vidstabdetect=shakiness=10:accuracy=15:result="mytransforms.trf"
20463 Visualize the result of internal transformations in the resulting
20466 vidstabdetect=show=1
20470 Analyze a video with medium shakiness using @command{ffmpeg}:
20472 ffmpeg -i input -vf vidstabdetect=shakiness=5:show=1 dummy.avi
20476 @anchor{vidstabtransform}
20477 @section vidstabtransform
20479 Video stabilization/deshaking: pass 2 of 2,
20480 see @ref{vidstabdetect} for pass 1.
20482 Read a file with transform information for each frame and
20483 apply/compensate them. Together with the @ref{vidstabdetect}
20484 filter this can be used to deshake videos. See also
20485 @url{http://public.hronopik.de/vid.stab}. It is important to also use
20486 the @ref{unsharp} filter, see below.
20488 To enable compilation of this filter you need to configure FFmpeg with
20489 @code{--enable-libvidstab}.
20491 @subsection Options
20495 Set path to the file used to read the transforms. Default value is
20496 @file{transforms.trf}.
20499 Set the number of frames (value*2 + 1) used for lowpass filtering the
20500 camera movements. Default value is 10.
20502 For example a number of 10 means that 21 frames are used (10 in the
20503 past and 10 in the future) to smoothen the motion in the video. A
20504 larger value leads to a smoother video, but limits the acceleration of
20505 the camera (pan/tilt movements). 0 is a special case where a static
20506 camera is simulated.
20509 Set the camera path optimization algorithm.
20511 Accepted values are:
20514 gaussian kernel low-pass filter on camera motion (default)
20516 averaging on transformations
20520 Set maximal number of pixels to translate frames. Default value is -1,
20524 Set maximal angle in radians (degree*PI/180) to rotate frames. Default
20525 value is -1, meaning no limit.
20528 Specify how to deal with borders that may be visible due to movement
20531 Available values are:
20534 keep image information from previous frame (default)
20536 fill the border black
20540 Invert transforms if set to 1. Default value is 0.
20543 Consider transforms as relative to previous frame if set to 1,
20544 absolute if set to 0. Default value is 0.
20547 Set percentage to zoom. A positive value will result in a zoom-in
20548 effect, a negative value in a zoom-out effect. Default value is 0 (no
20552 Set optimal zooming to avoid borders.
20554 Accepted values are:
20559 optimal static zoom value is determined (only very strong movements
20560 will lead to visible borders) (default)
20562 optimal adaptive zoom value is determined (no borders will be
20563 visible), see @option{zoomspeed}
20566 Note that the value given at zoom is added to the one calculated here.
20569 Set percent to zoom maximally each frame (enabled when
20570 @option{optzoom} is set to 2). Range is from 0 to 5, default value is
20574 Specify type of interpolation.
20576 Available values are:
20581 linear only horizontal
20583 linear in both directions (default)
20585 cubic in both directions (slow)
20589 Enable virtual tripod mode if set to 1, which is equivalent to
20590 @code{relative=0:smoothing=0}. Default value is 0.
20592 Use also @code{tripod} option of @ref{vidstabdetect}.
20595 Increase log verbosity if set to 1. Also the detected global motions
20596 are written to the temporary file @file{global_motions.trf}. Default
20600 @subsection Examples
20604 Use @command{ffmpeg} for a typical stabilization with default values:
20606 ffmpeg -i inp.mpeg -vf vidstabtransform,unsharp=5:5:0.8:3:3:0.4 inp_stabilized.mpeg
20609 Note the use of the @ref{unsharp} filter which is always recommended.
20612 Zoom in a bit more and load transform data from a given file:
20614 vidstabtransform=zoom=5:input="mytransforms.trf"
20618 Smoothen the video even more:
20620 vidstabtransform=smoothing=30
20626 Flip the input video vertically.
20628 For example, to vertically flip a video with @command{ffmpeg}:
20630 ffmpeg -i in.avi -vf "vflip" out.avi
20635 Detect variable frame rate video.
20637 This filter tries to detect if the input is variable or constant frame rate.
20639 At end it will output number of frames detected as having variable delta pts,
20640 and ones with constant delta pts.
20641 If there was frames with variable delta, than it will also show min, max and
20642 average delta encountered.
20646 Boost or alter saturation.
20648 The filter accepts the following options:
20651 Set strength of boost if positive value or strength of alter if negative value.
20652 Default is 0. Allowed range is from -2 to 2.
20655 Set the red balance. Default is 1. Allowed range is from -10 to 10.
20658 Set the green balance. Default is 1. Allowed range is from -10 to 10.
20661 Set the blue balance. Default is 1. Allowed range is from -10 to 10.
20664 Set the red luma coefficient.
20667 Set the green luma coefficient.
20670 Set the blue luma coefficient.
20673 If @code{intensity} is negative and this is set to 1, colors will change,
20674 otherwise colors will be less saturated, more towards gray.
20677 @subsection Commands
20679 This filter supports the all above options as @ref{commands}.
20684 Make or reverse a natural vignetting effect.
20686 The filter accepts the following options:
20690 Set lens angle expression as a number of radians.
20692 The value is clipped in the @code{[0,PI/2]} range.
20694 Default value: @code{"PI/5"}
20698 Set center coordinates expressions. Respectively @code{"w/2"} and @code{"h/2"}
20702 Set forward/backward mode.
20704 Available modes are:
20707 The larger the distance from the central point, the darker the image becomes.
20710 The larger the distance from the central point, the brighter the image becomes.
20711 This can be used to reverse a vignette effect, though there is no automatic
20712 detection to extract the lens @option{angle} and other settings (yet). It can
20713 also be used to create a burning effect.
20716 Default value is @samp{forward}.
20719 Set evaluation mode for the expressions (@option{angle}, @option{x0}, @option{y0}).
20721 It accepts the following values:
20724 Evaluate expressions only once during the filter initialization.
20727 Evaluate expressions for each incoming frame. This is way slower than the
20728 @samp{init} mode since it requires all the scalers to be re-computed, but it
20729 allows advanced dynamic expressions.
20732 Default value is @samp{init}.
20735 Set dithering to reduce the circular banding effects. Default is @code{1}
20739 Set vignette aspect. This setting allows one to adjust the shape of the vignette.
20740 Setting this value to the SAR of the input will make a rectangular vignetting
20741 following the dimensions of the video.
20743 Default is @code{1/1}.
20746 @subsection Expressions
20748 The @option{alpha}, @option{x0} and @option{y0} expressions can contain the
20749 following parameters.
20754 input width and height
20757 the number of input frame, starting from 0
20760 the PTS (Presentation TimeStamp) time of the filtered video frame, expressed in
20761 @var{TB} units, NAN if undefined
20764 frame rate of the input video, NAN if the input frame rate is unknown
20767 the PTS (Presentation TimeStamp) of the filtered video frame,
20768 expressed in seconds, NAN if undefined
20771 time base of the input video
20775 @subsection Examples
20779 Apply simple strong vignetting effect:
20785 Make a flickering vignetting:
20787 vignette='PI/4+random(1)*PI/50':eval=frame
20792 @section vmafmotion
20794 Obtain the average VMAF motion score of a video.
20795 It is one of the component metrics of VMAF.
20797 The obtained average motion score is printed through the logging system.
20799 The filter accepts the following options:
20803 If specified, the filter will use the named file to save the motion score of
20804 each frame with respect to the previous frame.
20805 When filename equals "-" the data is sent to standard output.
20810 ffmpeg -i ref.mpg -vf vmafmotion -f null -
20814 Stack input videos vertically.
20816 All streams must be of same pixel format and of same width.
20818 Note that this filter is faster than using @ref{overlay} and @ref{pad} filter
20819 to create same output.
20821 The filter accepts the following options:
20825 Set number of input streams. Default is 2.
20828 If set to 1, force the output to terminate when the shortest input
20829 terminates. Default value is 0.
20834 Deinterlace the input video ("w3fdif" stands for "Weston 3 Field
20835 Deinterlacing Filter").
20837 Based on the process described by Martin Weston for BBC R&D, and
20838 implemented based on the de-interlace algorithm written by Jim
20839 Easterbrook for BBC R&D, the Weston 3 field deinterlacing filter
20840 uses filter coefficients calculated by BBC R&D.
20842 This filter uses field-dominance information in frame to decide which
20843 of each pair of fields to place first in the output.
20844 If it gets it wrong use @ref{setfield} filter before @code{w3fdif} filter.
20846 There are two sets of filter coefficients, so called "simple"
20847 and "complex". Which set of filter coefficients is used can
20848 be set by passing an optional parameter:
20852 Set the interlacing filter coefficients. Accepts one of the following values:
20856 Simple filter coefficient set.
20858 More-complex filter coefficient set.
20860 Default value is @samp{complex}.
20863 Specify which frames to deinterlace. Accepts one of the following values:
20867 Deinterlace all frames,
20869 Only deinterlace frames marked as interlaced.
20872 Default value is @samp{all}.
20876 Video waveform monitor.
20878 The waveform monitor plots color component intensity. By default luminance
20879 only. Each column of the waveform corresponds to a column of pixels in the
20882 It accepts the following options:
20886 Can be either @code{row}, or @code{column}. Default is @code{column}.
20887 In row mode, the graph on the left side represents color component value 0 and
20888 the right side represents value = 255. In column mode, the top side represents
20889 color component value = 0 and bottom side represents value = 255.
20892 Set intensity. Smaller values are useful to find out how many values of the same
20893 luminance are distributed across input rows/columns.
20894 Default value is @code{0.04}. Allowed range is [0, 1].
20897 Set mirroring mode. @code{0} means unmirrored, @code{1} means mirrored.
20898 In mirrored mode, higher values will be represented on the left
20899 side for @code{row} mode and at the top for @code{column} mode. Default is
20900 @code{1} (mirrored).
20904 It accepts the following values:
20907 Presents information identical to that in the @code{parade}, except
20908 that the graphs representing color components are superimposed directly
20911 This display mode makes it easier to spot relative differences or similarities
20912 in overlapping areas of the color components that are supposed to be identical,
20913 such as neutral whites, grays, or blacks.
20916 Display separate graph for the color components side by side in
20917 @code{row} mode or one below the other in @code{column} mode.
20920 Display separate graph for the color components side by side in
20921 @code{column} mode or one below the other in @code{row} mode.
20923 Using this display mode makes it easy to spot color casts in the highlights
20924 and shadows of an image, by comparing the contours of the top and the bottom
20925 graphs of each waveform. Since whites, grays, and blacks are characterized
20926 by exactly equal amounts of red, green, and blue, neutral areas of the picture
20927 should display three waveforms of roughly equal width/height. If not, the
20928 correction is easy to perform by making level adjustments the three waveforms.
20930 Default is @code{stack}.
20932 @item components, c
20933 Set which color components to display. Default is 1, which means only luminance
20934 or red color component if input is in RGB colorspace. If is set for example to
20935 7 it will display all 3 (if) available color components.
20940 No envelope, this is default.
20943 Instant envelope, minimum and maximum values presented in graph will be easily
20944 visible even with small @code{step} value.
20947 Hold minimum and maximum values presented in graph across time. This way you
20948 can still spot out of range values without constantly looking at waveforms.
20951 Peak and instant envelope combined together.
20957 No filtering, this is default.
20960 Luma and chroma combined together.
20963 Similar as above, but shows difference between blue and red chroma.
20966 Similar as above, but use different colors.
20969 Similar as above, but again with different colors.
20972 Displays only chroma.
20975 Displays actual color value on waveform.
20978 Similar as above, but with luma showing frequency of chroma values.
20982 Set which graticule to display.
20986 Do not display graticule.
20989 Display green graticule showing legal broadcast ranges.
20992 Display orange graticule showing legal broadcast ranges.
20995 Display invert graticule showing legal broadcast ranges.
20999 Set graticule opacity.
21002 Set graticule flags.
21006 Draw numbers above lines. By default enabled.
21009 Draw dots instead of lines.
21013 Set scale used for displaying graticule.
21020 Default is digital.
21023 Set background opacity.
21027 Set tint for output.
21028 Only used with lowpass filter and when display is not overlay and input
21029 pixel formats are not RGB.
21032 @section weave, doubleweave
21034 The @code{weave} takes a field-based video input and join
21035 each two sequential fields into single frame, producing a new double
21036 height clip with half the frame rate and half the frame count.
21038 The @code{doubleweave} works same as @code{weave} but without
21039 halving frame rate and frame count.
21041 It accepts the following option:
21045 Set first field. Available values are:
21049 Set the frame as top-field-first.
21052 Set the frame as bottom-field-first.
21056 @subsection Examples
21060 Interlace video using @ref{select} and @ref{separatefields} filter:
21062 separatefields,select=eq(mod(n,4),0)+eq(mod(n,4),3),weave
21067 Apply the xBR high-quality magnification filter which is designed for pixel
21068 art. It follows a set of edge-detection rules, see
21069 @url{https://forums.libretro.com/t/xbr-algorithm-tutorial/123}.
21071 It accepts the following option:
21075 Set the scaling dimension: @code{2} for @code{2xBR}, @code{3} for
21076 @code{3xBR} and @code{4} for @code{4xBR}.
21077 Default is @code{3}.
21082 Apply cross fade from one input video stream to another input video stream.
21083 The cross fade is applied for specified duration.
21085 The filter accepts the following options:
21089 Set one of available transition effects:
21137 Default transition effect is fade.
21140 Set cross fade duration in seconds.
21141 Default duration is 1 second.
21144 Set cross fade start relative to first input stream in seconds.
21145 Default offset is 0.
21148 Set expression for custom transition effect.
21150 The expressions can use the following variables and functions:
21155 The coordinates of the current sample.
21159 The width and height of the image.
21162 Progress of transition effect.
21165 Currently processed plane.
21168 Return value of first input at current location and plane.
21171 Return value of second input at current location and plane.
21177 Return the value of the pixel at location (@var{x},@var{y}) of the
21178 first/second/third/fourth component of first input.
21184 Return the value of the pixel at location (@var{x},@var{y}) of the
21185 first/second/third/fourth component of second input.
21189 @subsection Examples
21193 Cross fade from one input video to another input video, with fade transition and duration of transition
21194 of 2 seconds starting at offset of 5 seconds:
21196 ffmpeg -i first.mp4 -i second.mp4 -filter_complex xfade=transition=fade:duration=2:offset=5 output.mp4
21201 Pick median pixels from several input videos.
21203 The filter accepts the following options:
21207 Set number of inputs.
21208 Default is 3. Allowed range is from 3 to 255.
21209 If number of inputs is even number, than result will be mean value between two median values.
21212 Set which planes to filter. Default value is @code{15}, by which all planes are processed.
21215 Set median percentile. Default value is @code{0.5}.
21216 Default value of @code{0.5} will pick always median values, while @code{0} will pick
21217 minimum values, and @code{1} maximum values.
21221 Stack video inputs into custom layout.
21223 All streams must be of same pixel format.
21225 The filter accepts the following options:
21229 Set number of input streams. Default is 2.
21232 Specify layout of inputs.
21233 This option requires the desired layout configuration to be explicitly set by the user.
21234 This sets position of each video input in output. Each input
21235 is separated by '|'.
21236 The first number represents the column, and the second number represents the row.
21237 Numbers start at 0 and are separated by '_'. Optionally one can use wX and hX,
21238 where X is video input from which to take width or height.
21239 Multiple values can be used when separated by '+'. In such
21240 case values are summed together.
21242 Note that if inputs are of different sizes gaps may appear, as not all of
21243 the output video frame will be filled. Similarly, videos can overlap each
21244 other if their position doesn't leave enough space for the full frame of
21247 For 2 inputs, a default layout of @code{0_0|w0_0} is set. In all other cases,
21248 a layout must be set by the user.
21251 If set to 1, force the output to terminate when the shortest input
21252 terminates. Default value is 0.
21255 If set to valid color, all unused pixels will be filled with that color.
21256 By default fill is set to none, so it is disabled.
21259 @subsection Examples
21263 Display 4 inputs into 2x2 grid.
21267 input1(0, 0) | input3(w0, 0)
21268 input2(0, h0) | input4(w0, h0)
21272 xstack=inputs=4:layout=0_0|0_h0|w0_0|w0_h0
21275 Note that if inputs are of different sizes, gaps or overlaps may occur.
21278 Display 4 inputs into 1x4 grid.
21285 input4(0, h0+h1+h2)
21289 xstack=inputs=4:layout=0_0|0_h0|0_h0+h1|0_h0+h1+h2
21292 Note that if inputs are of different widths, unused space will appear.
21295 Display 9 inputs into 3x3 grid.
21299 input1(0, 0) | input4(w0, 0) | input7(w0+w3, 0)
21300 input2(0, h0) | input5(w0, h0) | input8(w0+w3, h0)
21301 input3(0, h0+h1) | input6(w0, h0+h1) | input9(w0+w3, h0+h1)
21305 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
21308 Note that if inputs are of different sizes, gaps or overlaps may occur.
21311 Display 16 inputs into 4x4 grid.
21315 input1(0, 0) | input5(w0, 0) | input9 (w0+w4, 0) | input13(w0+w4+w8, 0)
21316 input2(0, h0) | input6(w0, h0) | input10(w0+w4, h0) | input14(w0+w4+w8, h0)
21317 input3(0, h0+h1) | input7(w0, h0+h1) | input11(w0+w4, h0+h1) | input15(w0+w4+w8, h0+h1)
21318 input4(0, h0+h1+h2)| input8(w0, h0+h1+h2)| input12(w0+w4, h0+h1+h2)| input16(w0+w4+w8, h0+h1+h2)
21322 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|
21323 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
21326 Note that if inputs are of different sizes, gaps or overlaps may occur.
21333 Deinterlace the input video ("yadif" means "yet another deinterlacing
21336 It accepts the following parameters:
21342 The interlacing mode to adopt. It accepts one of the following values:
21345 @item 0, send_frame
21346 Output one frame for each frame.
21347 @item 1, send_field
21348 Output one frame for each field.
21349 @item 2, send_frame_nospatial
21350 Like @code{send_frame}, but it skips the spatial interlacing check.
21351 @item 3, send_field_nospatial
21352 Like @code{send_field}, but it skips the spatial interlacing check.
21355 The default value is @code{send_frame}.
21358 The picture field parity assumed for the input interlaced video. It accepts one
21359 of the following values:
21363 Assume the top field is first.
21365 Assume the bottom field is first.
21367 Enable automatic detection of field parity.
21370 The default value is @code{auto}.
21371 If the interlacing is unknown or the decoder does not export this information,
21372 top field first will be assumed.
21375 Specify which frames to deinterlace. Accepts one of the following
21380 Deinterlace all frames.
21381 @item 1, interlaced
21382 Only deinterlace frames marked as interlaced.
21385 The default value is @code{all}.
21388 @section yadif_cuda
21390 Deinterlace the input video using the @ref{yadif} algorithm, but implemented
21391 in CUDA so that it can work as part of a GPU accelerated pipeline with nvdec
21394 It accepts the following parameters:
21400 The interlacing mode to adopt. It accepts one of the following values:
21403 @item 0, send_frame
21404 Output one frame for each frame.
21405 @item 1, send_field
21406 Output one frame for each field.
21407 @item 2, send_frame_nospatial
21408 Like @code{send_frame}, but it skips the spatial interlacing check.
21409 @item 3, send_field_nospatial
21410 Like @code{send_field}, but it skips the spatial interlacing check.
21413 The default value is @code{send_frame}.
21416 The picture field parity assumed for the input interlaced video. It accepts one
21417 of the following values:
21421 Assume the top field is first.
21423 Assume the bottom field is first.
21425 Enable automatic detection of field parity.
21428 The default value is @code{auto}.
21429 If the interlacing is unknown or the decoder does not export this information,
21430 top field first will be assumed.
21433 Specify which frames to deinterlace. Accepts one of the following
21438 Deinterlace all frames.
21439 @item 1, interlaced
21440 Only deinterlace frames marked as interlaced.
21443 The default value is @code{all}.
21448 Apply blur filter while preserving edges ("yaepblur" means "yet another edge preserving blur filter").
21449 The algorithm is described in
21450 "J. S. Lee, Digital image enhancement and noise filtering by use of local statistics, IEEE Trans. Pattern Anal. Mach. Intell. PAMI-2, 1980."
21452 It accepts the following parameters:
21456 Set the window radius. Default value is 3.
21459 Set which planes to filter. Default is only the first plane.
21462 Set blur strength. Default value is 128.
21465 @subsection Commands
21466 This filter supports same @ref{commands} as options.
21470 Apply Zoom & Pan effect.
21472 This filter accepts the following options:
21476 Set the zoom expression. Range is 1-10. Default is 1.
21480 Set the x and y expression. Default is 0.
21483 Set the duration expression in number of frames.
21484 This sets for how many number of frames effect will last for
21485 single input image.
21488 Set the output image size, default is 'hd720'.
21491 Set the output frame rate, default is '25'.
21494 Each expression can contain the following constants:
21513 Output frame count.
21516 The input timestamp expressed in seconds. It's NAN if the input timestamp is unknown.
21518 @item out_time, time, ot
21519 The output timestamp expressed in seconds.
21523 Last calculated 'x' and 'y' position from 'x' and 'y' expression
21524 for current input frame.
21528 'x' and 'y' of last output frame of previous input frame or 0 when there was
21529 not yet such frame (first input frame).
21532 Last calculated zoom from 'z' expression for current input frame.
21535 Last calculated zoom of last output frame of previous input frame.
21538 Number of output frames for current input frame. Calculated from 'd' expression
21539 for each input frame.
21542 number of output frames created for previous input frame
21545 Rational number: input width / input height
21548 sample aspect ratio
21551 display aspect ratio
21555 @subsection Examples
21559 Zoom in up to 1.5x and pan at same time to some spot near center of picture:
21561 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
21565 Zoom in up to 1.5x and pan always at center of picture:
21567 zoompan=z='min(zoom+0.0015,1.5)':d=700:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21571 Same as above but without pausing:
21573 zoompan=z='min(max(zoom,pzoom)+0.0015,1.5)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21577 Zoom in 2x into center of picture only for the first second of the input video:
21579 zoompan=z='if(between(in_time,0,1),2,1)':d=1:x='iw/2-(iw/zoom/2)':y='ih/2-(ih/zoom/2)'
21586 Scale (resize) the input video, using the z.lib library:
21587 @url{https://github.com/sekrit-twc/zimg}. To enable compilation of this
21588 filter, you need to configure FFmpeg with @code{--enable-libzimg}.
21590 The zscale filter forces the output display aspect ratio to be the same
21591 as the input, by changing the output sample aspect ratio.
21593 If the input image format is different from the format requested by
21594 the next filter, the zscale filter will convert the input to the
21597 @subsection Options
21598 The filter accepts the following options.
21603 Set the output video dimension expression. Default value is the input
21606 If the @var{width} or @var{w} value is 0, the input width is used for
21607 the output. If the @var{height} or @var{h} value is 0, the input height
21608 is used for the output.
21610 If one and only one of the values is -n with n >= 1, the zscale filter
21611 will use a value that maintains the aspect ratio of the input image,
21612 calculated from the other specified dimension. After that it will,
21613 however, make sure that the calculated dimension is divisible by n and
21614 adjust the value if necessary.
21616 If both values are -n with n >= 1, the behavior will be identical to
21617 both values being set to 0 as previously detailed.
21619 See below for the list of accepted constants for use in the dimension
21623 Set the video size. For the syntax of this option, check the
21624 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
21627 Set the dither type.
21629 Possible values are:
21634 @item error_diffusion
21640 Set the resize filter type.
21642 Possible values are:
21652 Default is bilinear.
21655 Set the color range.
21657 Possible values are:
21664 Default is same as input.
21667 Set the color primaries.
21669 Possible values are:
21679 Default is same as input.
21682 Set the transfer characteristics.
21684 Possible values are:
21698 Default is same as input.
21701 Set the colorspace matrix.
21703 Possible value are:
21714 Default is same as input.
21717 Set the input color range.
21719 Possible values are:
21726 Default is same as input.
21728 @item primariesin, pin
21729 Set the input color primaries.
21731 Possible values are:
21741 Default is same as input.
21743 @item transferin, tin
21744 Set the input transfer characteristics.
21746 Possible values are:
21757 Default is same as input.
21759 @item matrixin, min
21760 Set the input colorspace matrix.
21762 Possible value are:
21774 Set the output chroma location.
21776 Possible values are:
21787 @item chromalin, cin
21788 Set the input chroma location.
21790 Possible values are:
21802 Set the nominal peak luminance.
21805 The values of the @option{w} and @option{h} options are expressions
21806 containing the following constants:
21811 The input width and height
21815 These are the same as @var{in_w} and @var{in_h}.
21819 The output (scaled) width and height
21823 These are the same as @var{out_w} and @var{out_h}
21826 The same as @var{iw} / @var{ih}
21829 input sample aspect ratio
21832 The input display aspect ratio. Calculated from @code{(iw / ih) * sar}.
21836 horizontal and vertical input chroma subsample values. For example for the
21837 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21841 horizontal and vertical output chroma subsample values. For example for the
21842 pixel format "yuv422p" @var{hsub} is 2 and @var{vsub} is 1.
21845 @subsection Commands
21847 This filter supports the following commands:
21851 Set the output video dimension expression.
21852 The command accepts the same syntax of the corresponding option.
21854 If the specified expression is not valid, it is kept at its current
21858 @c man end VIDEO FILTERS
21860 @chapter OpenCL Video Filters
21861 @c man begin OPENCL VIDEO FILTERS
21863 Below is a description of the currently available OpenCL video filters.
21865 To enable compilation of these filters you need to configure FFmpeg with
21866 @code{--enable-opencl}.
21868 Running OpenCL filters requires you to initialize a hardware device and to pass that device to all filters in any filter graph.
21871 @item -init_hw_device opencl[=@var{name}][:@var{device}[,@var{key=value}...]]
21872 Initialise a new hardware device of type @var{opencl} called @var{name}, using the
21873 given device parameters.
21875 @item -filter_hw_device @var{name}
21876 Pass the hardware device called @var{name} to all filters in any filter graph.
21880 For more detailed information see @url{https://www.ffmpeg.org/ffmpeg.html#Advanced-Video-options}
21884 Example of choosing the first device on the second platform and running avgblur_opencl filter with default parameters on it.
21886 -init_hw_device opencl=gpu:1.0 -filter_hw_device gpu -i INPUT -vf "hwupload, avgblur_opencl, hwdownload" OUTPUT
21890 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.
21892 @section avgblur_opencl
21894 Apply average blur filter.
21896 The filter accepts the following options:
21900 Set horizontal radius size.
21901 Range is @code{[1, 1024]} and default value is @code{1}.
21904 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
21907 Set vertical radius size. Range is @code{[1, 1024]} and default value is @code{0}. If zero, @code{sizeX} value will be used.
21910 @subsection Example
21914 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.
21916 -i INPUT -vf "hwupload, avgblur_opencl=3, hwdownload" OUTPUT
21920 @section boxblur_opencl
21922 Apply a boxblur algorithm to the input video.
21924 It accepts the following parameters:
21928 @item luma_radius, lr
21929 @item luma_power, lp
21930 @item chroma_radius, cr
21931 @item chroma_power, cp
21932 @item alpha_radius, ar
21933 @item alpha_power, ap
21937 A description of the accepted options follows.
21940 @item luma_radius, lr
21941 @item chroma_radius, cr
21942 @item alpha_radius, ar
21943 Set an expression for the box radius in pixels used for blurring the
21944 corresponding input plane.
21946 The radius value must be a non-negative number, and must not be
21947 greater than the value of the expression @code{min(w,h)/2} for the
21948 luma and alpha planes, and of @code{min(cw,ch)/2} for the chroma
21951 Default value for @option{luma_radius} is "2". If not specified,
21952 @option{chroma_radius} and @option{alpha_radius} default to the
21953 corresponding value set for @option{luma_radius}.
21955 The expressions can contain the following constants:
21959 The input width and height in pixels.
21963 The input chroma image width and height in pixels.
21967 The horizontal and vertical chroma subsample values. For example, for the
21968 pixel format "yuv422p", @var{hsub} is 2 and @var{vsub} is 1.
21971 @item luma_power, lp
21972 @item chroma_power, cp
21973 @item alpha_power, ap
21974 Specify how many times the boxblur filter is applied to the
21975 corresponding plane.
21977 Default value for @option{luma_power} is 2. If not specified,
21978 @option{chroma_power} and @option{alpha_power} default to the
21979 corresponding value set for @option{luma_power}.
21981 A value of 0 will disable the effect.
21984 @subsection Examples
21986 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.
21990 Apply a boxblur filter with the luma, chroma, and alpha radius
21991 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.
21993 -i INPUT -vf "hwupload, boxblur_opencl=luma_radius=2:luma_power=3, hwdownload" OUTPUT
21994 -i INPUT -vf "hwupload, boxblur_opencl=2:3, hwdownload" OUTPUT
21998 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.
22000 For the luma plane, a 2x2 box radius will be run once.
22002 For the chroma plane, a 4x4 box radius will be run 5 times.
22004 For the alpha plane, a 3x3 box radius will be run 7 times.
22006 -i INPUT -vf "hwupload, boxblur_opencl=2:1:4:5:3:7, hwdownload" OUTPUT
22010 @section colorkey_opencl
22011 RGB colorspace color keying.
22013 The filter accepts the following options:
22017 The color which will be replaced with transparency.
22020 Similarity percentage with the key color.
22022 0.01 matches only the exact key color, while 1.0 matches everything.
22027 0.0 makes pixels either fully transparent, or not transparent at all.
22029 Higher values result in semi-transparent pixels, with a higher transparency
22030 the more similar the pixels color is to the key color.
22033 @subsection Examples
22037 Make every semi-green pixel in the input transparent with some slight blending:
22039 -i INPUT -vf "hwupload, colorkey_opencl=green:0.3:0.1, hwdownload" OUTPUT
22043 @section convolution_opencl
22045 Apply convolution of 3x3, 5x5, 7x7 matrix.
22047 The filter accepts the following options:
22054 Set matrix for each plane.
22055 Matrix is sequence of 9, 25 or 49 signed numbers.
22056 Default value for each plane is @code{0 0 0 0 1 0 0 0 0}.
22062 Set multiplier for calculated value for each plane.
22063 If unset or 0, it will be sum of all matrix elements.
22064 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{1.0}.
22070 Set bias for each plane. This value is added to the result of the multiplication.
22071 Useful for making the overall image brighter or darker.
22072 The option value must be a float number greater or equal to @code{0.0}. Default value is @code{0.0}.
22076 @subsection Examples
22082 -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
22088 -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
22092 Apply edge enhance:
22094 -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
22100 -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
22104 Apply laplacian edge detector which includes diagonals:
22106 -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
22112 -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
22116 @section erosion_opencl
22118 Apply erosion effect to the video.
22120 This filter replaces the pixel by the local(3x3) minimum.
22122 It accepts the following options:
22129 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22130 If @code{0}, plane will remain unchanged.
22133 Flag which specifies the pixel to refer to.
22134 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22136 Flags to local 3x3 coordinates region centered on @code{x}:
22145 @subsection Example
22149 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.
22151 -i INPUT -vf "hwupload, erosion_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22155 @section deshake_opencl
22156 Feature-point based video stabilization filter.
22158 The filter accepts the following options:
22162 Simulates a tripod by preventing any camera movement whatsoever from the original frame. Defaults to @code{0}.
22165 Whether or not additional debug info should be displayed, both in the processed output and in the console.
22167 Note that in order to see console debug output you will also need to pass @code{-v verbose} to ffmpeg.
22169 Viewing point matches in the output video is only supported for RGB input.
22171 Defaults to @code{0}.
22173 @item adaptive_crop
22174 Whether or not to do a tiny bit of cropping at the borders to cut down on the amount of mirrored pixels.
22176 Defaults to @code{1}.
22178 @item refine_features
22179 Whether or not feature points should be refined at a sub-pixel level.
22181 This can be turned off for a slight performance gain at the cost of precision.
22183 Defaults to @code{1}.
22185 @item smooth_strength
22186 The strength of the smoothing applied to the camera path from @code{0.0} to @code{1.0}.
22188 @code{1.0} is the maximum smoothing strength while values less than that result in less smoothing.
22190 @code{0.0} causes the filter to adaptively choose a smoothing strength on a per-frame basis.
22192 Defaults to @code{0.0}.
22194 @item smooth_window_multiplier
22195 Controls the size of the smoothing window (the number of frames buffered to determine motion information from).
22197 The size of the smoothing window is determined by multiplying the framerate of the video by this number.
22199 Acceptable values range from @code{0.1} to @code{10.0}.
22201 Larger values increase the amount of motion data available for determining how to smooth the camera path,
22202 potentially improving smoothness, but also increase latency and memory usage.
22204 Defaults to @code{2.0}.
22208 @subsection Examples
22212 Stabilize a video with a fixed, medium smoothing strength:
22214 -i INPUT -vf "hwupload, deshake_opencl=smooth_strength=0.5, hwdownload" OUTPUT
22218 Stabilize a video with debugging (both in console and in rendered video):
22220 -i INPUT -filter_complex "[0:v]format=rgba, hwupload, deshake_opencl=debug=1, hwdownload, format=rgba, format=yuv420p" -v verbose OUTPUT
22224 @section dilation_opencl
22226 Apply dilation effect to the video.
22228 This filter replaces the pixel by the local(3x3) maximum.
22230 It accepts the following options:
22237 Limit the maximum change for each plane. Range is @code{[0, 65535]} and default value is @code{65535}.
22238 If @code{0}, plane will remain unchanged.
22241 Flag which specifies the pixel to refer to.
22242 Range is @code{[0, 255]} and default value is @code{255}, i.e. all eight pixels are used.
22244 Flags to local 3x3 coordinates region centered on @code{x}:
22253 @subsection Example
22257 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.
22259 -i INPUT -vf "hwupload, dilation_opencl=30:40:50:coordinates=231, hwdownload" OUTPUT
22263 @section nlmeans_opencl
22265 Non-local Means denoise filter through OpenCL, this filter accepts same options as @ref{nlmeans}.
22267 @section overlay_opencl
22269 Overlay one video on top of another.
22271 It takes two inputs and has one output. The first input is the "main" video on which the second input is overlaid.
22272 This filter requires same memory layout for all the inputs. So, format conversion may be needed.
22274 The filter accepts the following options:
22279 Set the x coordinate of the overlaid video on the main video.
22280 Default value is @code{0}.
22283 Set the y coordinate of the overlaid video on the main video.
22284 Default value is @code{0}.
22288 @subsection Examples
22292 Overlay an image LOGO at the top-left corner of the INPUT video. Both inputs are yuv420p format.
22294 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuv420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22297 The inputs have same memory layout for color channels , the overlay has additional alpha plane, like INPUT is yuv420p, and the LOGO is yuva420p.
22299 -i INPUT -i LOGO -filter_complex "[0:v]hwupload[a], [1:v]format=yuva420p, hwupload[b], [a][b]overlay_opencl, hwdownload" OUTPUT
22304 @section pad_opencl
22306 Add paddings to the input image, and place the original input at the
22307 provided @var{x}, @var{y} coordinates.
22309 It accepts the following options:
22314 Specify an expression for the size of the output image with the
22315 paddings added. If the value for @var{width} or @var{height} is 0, the
22316 corresponding input size is used for the output.
22318 The @var{width} expression can reference the value set by the
22319 @var{height} expression, and vice versa.
22321 The default value of @var{width} and @var{height} is 0.
22325 Specify the offsets to place the input image at within the padded area,
22326 with respect to the top/left border of the output image.
22328 The @var{x} expression can reference the value set by the @var{y}
22329 expression, and vice versa.
22331 The default value of @var{x} and @var{y} is 0.
22333 If @var{x} or @var{y} evaluate to a negative number, they'll be changed
22334 so the input image is centered on the padded area.
22337 Specify the color of the padded area. For the syntax of this option,
22338 check the @ref{color syntax,,"Color" section in the ffmpeg-utils
22339 manual,ffmpeg-utils}.
22342 Pad to an aspect instead to a resolution.
22345 The value for the @var{width}, @var{height}, @var{x}, and @var{y}
22346 options are expressions containing the following constants:
22351 The input video width and height.
22355 These are the same as @var{in_w} and @var{in_h}.
22359 The output width and height (the size of the padded area), as
22360 specified by the @var{width} and @var{height} expressions.
22364 These are the same as @var{out_w} and @var{out_h}.
22368 The x and y offsets as specified by the @var{x} and @var{y}
22369 expressions, or NAN if not yet specified.
22372 same as @var{iw} / @var{ih}
22375 input sample aspect ratio
22378 input display aspect ratio, it is the same as (@var{iw} / @var{ih}) * @var{sar}
22381 @section prewitt_opencl
22383 Apply the Prewitt operator (@url{https://en.wikipedia.org/wiki/Prewitt_operator}) to input video stream.
22385 The filter accepts the following option:
22389 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22392 Set value which will be multiplied with filtered result.
22393 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22396 Set value which will be added to filtered result.
22397 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22400 @subsection Example
22404 Apply the Prewitt operator with scale set to 2 and delta set to 10.
22406 -i INPUT -vf "hwupload, prewitt_opencl=scale=2:delta=10, hwdownload" OUTPUT
22410 @anchor{program_opencl}
22411 @section program_opencl
22413 Filter video using an OpenCL program.
22418 OpenCL program source file.
22421 Kernel name in program.
22424 Number of inputs to the filter. Defaults to 1.
22427 Size of output frames. Defaults to the same as the first input.
22431 The @code{program_opencl} filter also supports the @ref{framesync} options.
22433 The program source file must contain a kernel function with the given name,
22434 which will be run once for each plane of the output. Each run on a plane
22435 gets enqueued as a separate 2D global NDRange with one work-item for each
22436 pixel to be generated. The global ID offset for each work-item is therefore
22437 the coordinates of a pixel in the destination image.
22439 The kernel function needs to take the following arguments:
22442 Destination image, @var{__write_only image2d_t}.
22444 This image will become the output; the kernel should write all of it.
22446 Frame index, @var{unsigned int}.
22448 This is a counter starting from zero and increasing by one for each frame.
22450 Source images, @var{__read_only image2d_t}.
22452 These are the most recent images on each input. The kernel may read from
22453 them to generate the output, but they can't be written to.
22460 Copy the input to the output (output must be the same size as the input).
22462 __kernel void copy(__write_only image2d_t destination,
22463 unsigned int index,
22464 __read_only image2d_t source)
22466 const sampler_t sampler = CLK_NORMALIZED_COORDS_FALSE;
22468 int2 location = (int2)(get_global_id(0), get_global_id(1));
22470 float4 value = read_imagef(source, sampler, location);
22472 write_imagef(destination, location, value);
22477 Apply a simple transformation, rotating the input by an amount increasing
22478 with the index counter. Pixel values are linearly interpolated by the
22479 sampler, and the output need not have the same dimensions as the input.
22481 __kernel void rotate_image(__write_only image2d_t dst,
22482 unsigned int index,
22483 __read_only image2d_t src)
22485 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22486 CLK_FILTER_LINEAR);
22488 float angle = (float)index / 100.0f;
22490 float2 dst_dim = convert_float2(get_image_dim(dst));
22491 float2 src_dim = convert_float2(get_image_dim(src));
22493 float2 dst_cen = dst_dim / 2.0f;
22494 float2 src_cen = src_dim / 2.0f;
22496 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22498 float2 dst_pos = convert_float2(dst_loc) - dst_cen;
22500 cos(angle) * dst_pos.x - sin(angle) * dst_pos.y,
22501 sin(angle) * dst_pos.x + cos(angle) * dst_pos.y
22503 src_pos = src_pos * src_dim / dst_dim;
22505 float2 src_loc = src_pos + src_cen;
22507 if (src_loc.x < 0.0f || src_loc.y < 0.0f ||
22508 src_loc.x > src_dim.x || src_loc.y > src_dim.y)
22509 write_imagef(dst, dst_loc, 0.5f);
22511 write_imagef(dst, dst_loc, read_imagef(src, sampler, src_loc));
22516 Blend two inputs together, with the amount of each input used varying
22517 with the index counter.
22519 __kernel void blend_images(__write_only image2d_t dst,
22520 unsigned int index,
22521 __read_only image2d_t src1,
22522 __read_only image2d_t src2)
22524 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22525 CLK_FILTER_LINEAR);
22527 float blend = (cos((float)index / 50.0f) + 1.0f) / 2.0f;
22529 int2 dst_loc = (int2)(get_global_id(0), get_global_id(1));
22530 int2 src1_loc = dst_loc * get_image_dim(src1) / get_image_dim(dst);
22531 int2 src2_loc = dst_loc * get_image_dim(src2) / get_image_dim(dst);
22533 float4 val1 = read_imagef(src1, sampler, src1_loc);
22534 float4 val2 = read_imagef(src2, sampler, src2_loc);
22536 write_imagef(dst, dst_loc, val1 * blend + val2 * (1.0f - blend));
22542 @section roberts_opencl
22543 Apply the Roberts cross operator (@url{https://en.wikipedia.org/wiki/Roberts_cross}) to input video stream.
22545 The filter accepts the following option:
22549 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22552 Set value which will be multiplied with filtered result.
22553 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22556 Set value which will be added to filtered result.
22557 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22560 @subsection Example
22564 Apply the Roberts cross operator with scale set to 2 and delta set to 10
22566 -i INPUT -vf "hwupload, roberts_opencl=scale=2:delta=10, hwdownload" OUTPUT
22570 @section sobel_opencl
22572 Apply the Sobel operator (@url{https://en.wikipedia.org/wiki/Sobel_operator}) to input video stream.
22574 The filter accepts the following option:
22578 Set which planes to filter. Default value is @code{0xf}, by which all planes are processed.
22581 Set value which will be multiplied with filtered result.
22582 Range is @code{[0.0, 65535]} and default value is @code{1.0}.
22585 Set value which will be added to filtered result.
22586 Range is @code{[-65535, 65535]} and default value is @code{0.0}.
22589 @subsection Example
22593 Apply sobel operator with scale set to 2 and delta set to 10
22595 -i INPUT -vf "hwupload, sobel_opencl=scale=2:delta=10, hwdownload" OUTPUT
22599 @section tonemap_opencl
22601 Perform HDR(PQ/HLG) to SDR conversion with tone-mapping.
22603 It accepts the following parameters:
22607 Specify the tone-mapping operator to be used. Same as tonemap option in @ref{tonemap}.
22610 Tune the tone mapping algorithm. same as param option in @ref{tonemap}.
22613 Apply desaturation for highlights that exceed this level of brightness. The
22614 higher the parameter, the more color information will be preserved. This
22615 setting helps prevent unnaturally blown-out colors for super-highlights, by
22616 (smoothly) turning into white instead. This makes images feel more natural,
22617 at the cost of reducing information about out-of-range colors.
22619 The default value is 0.5, and the algorithm here is a little different from
22620 the cpu version tonemap currently. A setting of 0.0 disables this option.
22623 The tonemapping algorithm parameters is fine-tuned per each scene. And a threshold
22624 is used to detect whether the scene has changed or not. If the distance between
22625 the current frame average brightness and the current running average exceeds
22626 a threshold value, we would re-calculate scene average and peak brightness.
22627 The default value is 0.2.
22630 Specify the output pixel format.
22632 Currently supported formats are:
22639 Set the output color range.
22641 Possible values are:
22647 Default is same as input.
22650 Set the output color primaries.
22652 Possible values are:
22658 Default is same as input.
22661 Set the output transfer characteristics.
22663 Possible values are:
22672 Set the output colorspace matrix.
22674 Possible value are:
22680 Default is same as input.
22684 @subsection Example
22688 Convert HDR(PQ/HLG) video to bt2020-transfer-characteristic p010 format using linear operator.
22690 -i INPUT -vf "format=p010,hwupload,tonemap_opencl=t=bt2020:tonemap=linear:format=p010,hwdownload,format=p010" OUTPUT
22694 @section unsharp_opencl
22696 Sharpen or blur the input video.
22698 It accepts the following parameters:
22701 @item luma_msize_x, lx
22702 Set the luma matrix horizontal size.
22703 Range is @code{[1, 23]} and default value is @code{5}.
22705 @item luma_msize_y, ly
22706 Set the luma matrix vertical size.
22707 Range is @code{[1, 23]} and default value is @code{5}.
22709 @item luma_amount, la
22710 Set the luma effect strength.
22711 Range is @code{[-10, 10]} and default value is @code{1.0}.
22713 Negative values will blur the input video, while positive values will
22714 sharpen it, a value of zero will disable the effect.
22716 @item chroma_msize_x, cx
22717 Set the chroma matrix horizontal size.
22718 Range is @code{[1, 23]} and default value is @code{5}.
22720 @item chroma_msize_y, cy
22721 Set the chroma matrix vertical size.
22722 Range is @code{[1, 23]} and default value is @code{5}.
22724 @item chroma_amount, ca
22725 Set the chroma effect strength.
22726 Range is @code{[-10, 10]} and default value is @code{0.0}.
22728 Negative values will blur the input video, while positive values will
22729 sharpen it, a value of zero will disable the effect.
22733 All parameters are optional and default to the equivalent of the
22734 string '5:5:1.0:5:5:0.0'.
22736 @subsection Examples
22740 Apply strong luma sharpen effect:
22742 -i INPUT -vf "hwupload, unsharp_opencl=luma_msize_x=7:luma_msize_y=7:luma_amount=2.5, hwdownload" OUTPUT
22746 Apply a strong blur of both luma and chroma parameters:
22748 -i INPUT -vf "hwupload, unsharp_opencl=7:7:-2:7:7:-2, hwdownload" OUTPUT
22752 @section xfade_opencl
22754 Cross fade two videos with custom transition effect by using OpenCL.
22756 It accepts the following options:
22760 Set one of possible transition effects.
22764 Select custom transition effect, the actual transition description
22765 will be picked from source and kernel options.
22777 Default transition is fade.
22781 OpenCL program source file for custom transition.
22784 Set name of kernel to use for custom transition from program source file.
22787 Set duration of video transition.
22790 Set time of start of transition relative to first video.
22793 The program source file must contain a kernel function with the given name,
22794 which will be run once for each plane of the output. Each run on a plane
22795 gets enqueued as a separate 2D global NDRange with one work-item for each
22796 pixel to be generated. The global ID offset for each work-item is therefore
22797 the coordinates of a pixel in the destination image.
22799 The kernel function needs to take the following arguments:
22802 Destination image, @var{__write_only image2d_t}.
22804 This image will become the output; the kernel should write all of it.
22807 First Source image, @var{__read_only image2d_t}.
22808 Second Source image, @var{__read_only image2d_t}.
22810 These are the most recent images on each input. The kernel may read from
22811 them to generate the output, but they can't be written to.
22814 Transition progress, @var{float}. This value is always between 0 and 1 inclusive.
22821 Apply dots curtain transition effect:
22823 __kernel void blend_images(__write_only image2d_t dst,
22824 __read_only image2d_t src1,
22825 __read_only image2d_t src2,
22828 const sampler_t sampler = (CLK_NORMALIZED_COORDS_FALSE |
22829 CLK_FILTER_LINEAR);
22830 int2 p = (int2)(get_global_id(0), get_global_id(1));
22831 float2 rp = (float2)(get_global_id(0), get_global_id(1));
22832 float2 dim = (float2)(get_image_dim(src1).x, get_image_dim(src1).y);
22835 float2 dots = (float2)(20.0, 20.0);
22836 float2 center = (float2)(0,0);
22839 float4 val1 = read_imagef(src1, sampler, p);
22840 float4 val2 = read_imagef(src2, sampler, p);
22841 bool next = distance(fract(rp * dots, &unused), (float2)(0.5, 0.5)) < (progress / distance(rp, center));
22843 write_imagef(dst, p, next ? val1 : val2);
22849 @c man end OPENCL VIDEO FILTERS
22851 @chapter VAAPI Video Filters
22852 @c man begin VAAPI VIDEO FILTERS
22854 VAAPI Video filters are usually used with VAAPI decoder and VAAPI encoder. Below is a description of VAAPI video filters.
22856 To enable compilation of these filters you need to configure FFmpeg with
22857 @code{--enable-vaapi}.
22859 To use vaapi filters, you need to setup the vaapi device correctly. For more information, please read @url{https://trac.ffmpeg.org/wiki/Hardware/VAAPI}
22861 @section tonemap_vaapi
22863 Perform HDR(High Dynamic Range) to SDR(Standard Dynamic Range) conversion with tone-mapping.
22864 It maps the dynamic range of HDR10 content to the SDR content.
22865 It currently only accepts HDR10 as input.
22867 It accepts the following parameters:
22871 Specify the output pixel format.
22873 Currently supported formats are:
22882 Set the output color primaries.
22884 Default is same as input.
22887 Set the output transfer characteristics.
22892 Set the output colorspace matrix.
22894 Default is same as input.
22898 @subsection Example
22902 Convert HDR(HDR10) video to bt2020-transfer-characteristic p010 format
22904 tonemap_vaapi=format=p010:t=bt2020-10
22908 @c man end VAAPI VIDEO FILTERS
22910 @chapter Video Sources
22911 @c man begin VIDEO SOURCES
22913 Below is a description of the currently available video sources.
22917 Buffer video frames, and make them available to the filter chain.
22919 This source is mainly intended for a programmatic use, in particular
22920 through the interface defined in @file{libavfilter/buffersrc.h}.
22922 It accepts the following parameters:
22927 Specify the size (width and height) of the buffered video frames. For the
22928 syntax of this option, check the
22929 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
22932 The input video width.
22935 The input video height.
22938 A string representing the pixel format of the buffered video frames.
22939 It may be a number corresponding to a pixel format, or a pixel format
22943 Specify the timebase assumed by the timestamps of the buffered frames.
22946 Specify the frame rate expected for the video stream.
22948 @item pixel_aspect, sar
22949 The sample (pixel) aspect ratio of the input video.
22952 This option is deprecated and ignored. Prepend @code{sws_flags=@var{flags};}
22953 to the filtergraph description to specify swscale flags for automatically
22954 inserted scalers. See @ref{Filtergraph syntax}.
22956 @item hw_frames_ctx
22957 When using a hardware pixel format, this should be a reference to an
22958 AVHWFramesContext describing input frames.
22963 buffer=width=320:height=240:pix_fmt=yuv410p:time_base=1/24:sar=1
22966 will instruct the source to accept video frames with size 320x240 and
22967 with format "yuv410p", assuming 1/24 as the timestamps timebase and
22968 square pixels (1:1 sample aspect ratio).
22969 Since the pixel format with name "yuv410p" corresponds to the number 6
22970 (check the enum AVPixelFormat definition in @file{libavutil/pixfmt.h}),
22971 this example corresponds to:
22973 buffer=size=320x240:pixfmt=6:time_base=1/24:pixel_aspect=1/1
22976 Alternatively, the options can be specified as a flat string, but this
22977 syntax is deprecated:
22979 @var{width}:@var{height}:@var{pix_fmt}:@var{time_base.num}:@var{time_base.den}:@var{pixel_aspect.num}:@var{pixel_aspect.den}
22983 Create a pattern generated by an elementary cellular automaton.
22985 The initial state of the cellular automaton can be defined through the
22986 @option{filename} and @option{pattern} options. If such options are
22987 not specified an initial state is created randomly.
22989 At each new frame a new row in the video is filled with the result of
22990 the cellular automaton next generation. The behavior when the whole
22991 frame is filled is defined by the @option{scroll} option.
22993 This source accepts the following options:
22997 Read the initial cellular automaton state, i.e. the starting row, from
22998 the specified file.
22999 In the file, each non-whitespace character is considered an alive
23000 cell, a newline will terminate the row, and further characters in the
23001 file will be ignored.
23004 Read the initial cellular automaton state, i.e. the starting row, from
23005 the specified string.
23007 Each non-whitespace character in the string is considered an alive
23008 cell, a newline will terminate the row, and further characters in the
23009 string will be ignored.
23012 Set the video rate, that is the number of frames generated per second.
23015 @item random_fill_ratio, ratio
23016 Set the random fill ratio for the initial cellular automaton row. It
23017 is a floating point number value ranging from 0 to 1, defaults to
23020 This option is ignored when a file or a pattern is specified.
23022 @item random_seed, seed
23023 Set the seed for filling randomly the initial row, must be an integer
23024 included between 0 and UINT32_MAX. If not specified, or if explicitly
23025 set to -1, the filter will try to use a good random seed on a best
23029 Set the cellular automaton rule, it is a number ranging from 0 to 255.
23030 Default value is 110.
23033 Set the size of the output video. For the syntax of this option, check the
23034 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23036 If @option{filename} or @option{pattern} is specified, the size is set
23037 by default to the width of the specified initial state row, and the
23038 height is set to @var{width} * PHI.
23040 If @option{size} is set, it must contain the width of the specified
23041 pattern string, and the specified pattern will be centered in the
23044 If a filename or a pattern string is not specified, the size value
23045 defaults to "320x518" (used for a randomly generated initial state).
23048 If set to 1, scroll the output upward when all the rows in the output
23049 have been already filled. If set to 0, the new generated row will be
23050 written over the top row just after the bottom row is filled.
23053 @item start_full, full
23054 If set to 1, completely fill the output with generated rows before
23055 outputting the first frame.
23056 This is the default behavior, for disabling set the value to 0.
23059 If set to 1, stitch the left and right row edges together.
23060 This is the default behavior, for disabling set the value to 0.
23063 @subsection Examples
23067 Read the initial state from @file{pattern}, and specify an output of
23070 cellauto=f=pattern:s=200x400
23074 Generate a random initial row with a width of 200 cells, with a fill
23077 cellauto=ratio=2/3:s=200x200
23081 Create a pattern generated by rule 18 starting by a single alive cell
23082 centered on an initial row with width 100:
23084 cellauto=p=@@:s=100x400:full=0:rule=18
23088 Specify a more elaborated initial pattern:
23090 cellauto=p='@@@@ @@ @@@@':s=100x400:full=0:rule=18
23095 @anchor{coreimagesrc}
23096 @section coreimagesrc
23097 Video source generated on GPU using Apple's CoreImage API on OSX.
23099 This video source is a specialized version of the @ref{coreimage} video filter.
23100 Use a core image generator at the beginning of the applied filterchain to
23101 generate the content.
23103 The coreimagesrc video source accepts the following options:
23105 @item list_generators
23106 List all available generators along with all their respective options as well as
23107 possible minimum and maximum values along with the default values.
23109 list_generators=true
23113 Specify the size of the sourced video. For the syntax of this option, check the
23114 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23115 The default value is @code{320x240}.
23118 Specify the frame rate of the sourced video, as the number of frames
23119 generated per second. It has to be a string in the format
23120 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23121 number or a valid video frame rate abbreviation. The default value is
23125 Set the sample aspect ratio of the sourced video.
23128 Set the duration of the sourced video. See
23129 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23130 for the accepted syntax.
23132 If not specified, or the expressed duration is negative, the video is
23133 supposed to be generated forever.
23136 Additionally, all options of the @ref{coreimage} video filter are accepted.
23137 A complete filterchain can be used for further processing of the
23138 generated input without CPU-HOST transfer. See @ref{coreimage} documentation
23139 and examples for details.
23141 @subsection Examples
23146 Use CIQRCodeGenerator to create a QR code for the FFmpeg homepage,
23147 given as complete and escaped command-line for Apple's standard bash shell:
23149 ffmpeg -f lavfi -i coreimagesrc=s=100x100:filter=CIQRCodeGenerator@@inputMessage=https\\\\\://FFmpeg.org/@@inputCorrectionLevel=H -frames:v 1 QRCode.png
23151 This example is equivalent to the QRCode example of @ref{coreimage} without the
23152 need for a nullsrc video source.
23157 Generate several gradients.
23161 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23162 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23165 Set frame rate, expressed as number of frames per second. Default
23168 @item c0, c1, c2, c3, c4, c5, c6, c7
23169 Set 8 colors. Default values for colors is to pick random one.
23171 @item x0, y0, y0, y1
23172 Set gradient line source and destination points. If negative or out of range, random ones
23176 Set number of colors to use at once. Allowed range is from 2 to 8. Default value is 2.
23179 Set seed for picking gradient line points.
23182 Set the duration of the sourced video. See
23183 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23184 for the accepted syntax.
23186 If not specified, or the expressed duration is negative, the video is
23187 supposed to be generated forever.
23190 Set speed of gradients rotation.
23194 @section mandelbrot
23196 Generate a Mandelbrot set fractal, and progressively zoom towards the
23197 point specified with @var{start_x} and @var{start_y}.
23199 This source accepts the following options:
23204 Set the terminal pts value. Default value is 400.
23207 Set the terminal scale value.
23208 Must be a floating point value. Default value is 0.3.
23211 Set the inner coloring mode, that is the algorithm used to draw the
23212 Mandelbrot fractal internal region.
23214 It shall assume one of the following values:
23219 Show time until convergence.
23221 Set color based on point closest to the origin of the iterations.
23226 Default value is @var{mincol}.
23229 Set the bailout value. Default value is 10.0.
23232 Set the maximum of iterations performed by the rendering
23233 algorithm. Default value is 7189.
23236 Set outer coloring mode.
23237 It shall assume one of following values:
23239 @item iteration_count
23240 Set iteration count mode.
23241 @item normalized_iteration_count
23242 set normalized iteration count mode.
23244 Default value is @var{normalized_iteration_count}.
23247 Set frame rate, expressed as number of frames per second. Default
23251 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23252 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23255 Set the initial scale value. Default value is 3.0.
23258 Set the initial x position. Must be a floating point value between
23259 -100 and 100. Default value is -0.743643887037158704752191506114774.
23262 Set the initial y position. Must be a floating point value between
23263 -100 and 100. Default value is -0.131825904205311970493132056385139.
23268 Generate various test patterns, as generated by the MPlayer test filter.
23270 The size of the generated video is fixed, and is 256x256.
23271 This source is useful in particular for testing encoding features.
23273 This source accepts the following options:
23278 Specify the frame rate of the sourced video, as the number of frames
23279 generated per second. It has to be a string in the format
23280 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23281 number or a valid video frame rate abbreviation. The default value is
23285 Set the duration of the sourced video. See
23286 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23287 for the accepted syntax.
23289 If not specified, or the expressed duration is negative, the video is
23290 supposed to be generated forever.
23294 Set the number or the name of the test to perform. Supported tests are:
23308 @item max_frames, m
23309 Set the maximum number of frames generated for each test, default value is 30.
23313 Default value is "all", which will cycle through the list of all tests.
23318 mptestsrc=t=dc_luma
23321 will generate a "dc_luma" test pattern.
23323 @section frei0r_src
23325 Provide a frei0r source.
23327 To enable compilation of this filter you need to install the frei0r
23328 header and configure FFmpeg with @code{--enable-frei0r}.
23330 This source accepts the following parameters:
23335 The size of the video to generate. For the syntax of this option, check the
23336 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23339 The framerate of the generated video. It may be a string of the form
23340 @var{num}/@var{den} or a frame rate abbreviation.
23343 The name to the frei0r source to load. For more information regarding frei0r and
23344 how to set the parameters, read the @ref{frei0r} section in the video filters
23347 @item filter_params
23348 A '|'-separated list of parameters to pass to the frei0r source.
23352 For example, to generate a frei0r partik0l source with size 200x200
23353 and frame rate 10 which is overlaid on the overlay filter main input:
23355 frei0r_src=size=200x200:framerate=10:filter_name=partik0l:filter_params=1234 [overlay]; [in][overlay] overlay
23360 Generate a life pattern.
23362 This source is based on a generalization of John Conway's life game.
23364 The sourced input represents a life grid, each pixel represents a cell
23365 which can be in one of two possible states, alive or dead. Every cell
23366 interacts with its eight neighbours, which are the cells that are
23367 horizontally, vertically, or diagonally adjacent.
23369 At each interaction the grid evolves according to the adopted rule,
23370 which specifies the number of neighbor alive cells which will make a
23371 cell stay alive or born. The @option{rule} option allows one to specify
23374 This source accepts the following options:
23378 Set the file from which to read the initial grid state. In the file,
23379 each non-whitespace character is considered an alive cell, and newline
23380 is used to delimit the end of each row.
23382 If this option is not specified, the initial grid is generated
23386 Set the video rate, that is the number of frames generated per second.
23389 @item random_fill_ratio, ratio
23390 Set the random fill ratio for the initial random grid. It is a
23391 floating point number value ranging from 0 to 1, defaults to 1/PHI.
23392 It is ignored when a file is specified.
23394 @item random_seed, seed
23395 Set the seed for filling the initial random grid, must be an integer
23396 included between 0 and UINT32_MAX. If not specified, or if explicitly
23397 set to -1, the filter will try to use a good random seed on a best
23403 A rule can be specified with a code of the kind "S@var{NS}/B@var{NB}",
23404 where @var{NS} and @var{NB} are sequences of numbers in the range 0-8,
23405 @var{NS} specifies the number of alive neighbor cells which make a
23406 live cell stay alive, and @var{NB} the number of alive neighbor cells
23407 which make a dead cell to become alive (i.e. to "born").
23408 "s" and "b" can be used in place of "S" and "B", respectively.
23410 Alternatively a rule can be specified by an 18-bits integer. The 9
23411 high order bits are used to encode the next cell state if it is alive
23412 for each number of neighbor alive cells, the low order bits specify
23413 the rule for "borning" new cells. Higher order bits encode for an
23414 higher number of neighbor cells.
23415 For example the number 6153 = @code{(12<<9)+9} specifies a stay alive
23416 rule of 12 and a born rule of 9, which corresponds to "S23/B03".
23418 Default value is "S23/B3", which is the original Conway's game of life
23419 rule, and will keep a cell alive if it has 2 or 3 neighbor alive
23420 cells, and will born a new cell if there are three alive cells around
23424 Set the size of the output video. For the syntax of this option, check the
23425 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23427 If @option{filename} is specified, the size is set by default to the
23428 same size of the input file. If @option{size} is set, it must contain
23429 the size specified in the input file, and the initial grid defined in
23430 that file is centered in the larger resulting area.
23432 If a filename is not specified, the size value defaults to "320x240"
23433 (used for a randomly generated initial grid).
23436 If set to 1, stitch the left and right grid edges together, and the
23437 top and bottom edges also. Defaults to 1.
23440 Set cell mold speed. If set, a dead cell will go from @option{death_color} to
23441 @option{mold_color} with a step of @option{mold}. @option{mold} can have a
23442 value from 0 to 255.
23445 Set the color of living (or new born) cells.
23448 Set the color of dead cells. If @option{mold} is set, this is the first color
23449 used to represent a dead cell.
23452 Set mold color, for definitely dead and moldy cells.
23454 For the syntax of these 3 color options, check the @ref{color syntax,,"Color" section in the
23455 ffmpeg-utils manual,ffmpeg-utils}.
23458 @subsection Examples
23462 Read a grid from @file{pattern}, and center it on a grid of size
23465 life=f=pattern:s=300x300
23469 Generate a random grid of size 200x200, with a fill ratio of 2/3:
23471 life=ratio=2/3:s=200x200
23475 Specify a custom rule for evolving a randomly generated grid:
23481 Full example with slow death effect (mold) using @command{ffplay}:
23483 ffplay -f lavfi life=s=300x200:mold=10:r=60:ratio=0.1:death_color=#C83232:life_color=#00ff00,scale=1200:800:flags=16
23490 @anchor{haldclutsrc}
23493 @anchor{pal100bars}
23494 @anchor{rgbtestsrc}
23496 @anchor{smptehdbars}
23499 @anchor{yuvtestsrc}
23500 @section allrgb, allyuv, color, haldclutsrc, nullsrc, pal75bars, pal100bars, rgbtestsrc, smptebars, smptehdbars, testsrc, testsrc2, yuvtestsrc
23502 The @code{allrgb} source returns frames of size 4096x4096 of all rgb colors.
23504 The @code{allyuv} source returns frames of size 4096x4096 of all yuv colors.
23506 The @code{color} source provides an uniformly colored input.
23508 The @code{haldclutsrc} source provides an identity Hald CLUT. See also
23509 @ref{haldclut} filter.
23511 The @code{nullsrc} source returns unprocessed video frames. It is
23512 mainly useful to be employed in analysis / debugging tools, or as the
23513 source for filters which ignore the input data.
23515 The @code{pal75bars} source generates a color bars pattern, based on
23516 EBU PAL recommendations with 75% color levels.
23518 The @code{pal100bars} source generates a color bars pattern, based on
23519 EBU PAL recommendations with 100% color levels.
23521 The @code{rgbtestsrc} source generates an RGB test pattern useful for
23522 detecting RGB vs BGR issues. You should see a red, green and blue
23523 stripe from top to bottom.
23525 The @code{smptebars} source generates a color bars pattern, based on
23526 the SMPTE Engineering Guideline EG 1-1990.
23528 The @code{smptehdbars} source generates a color bars pattern, based on
23529 the SMPTE RP 219-2002.
23531 The @code{testsrc} source generates a test video pattern, showing a
23532 color pattern, a scrolling gradient and a timestamp. This is mainly
23533 intended for testing purposes.
23535 The @code{testsrc2} source is similar to testsrc, but supports more
23536 pixel formats instead of just @code{rgb24}. This allows using it as an
23537 input for other tests without requiring a format conversion.
23539 The @code{yuvtestsrc} source generates an YUV test pattern. You should
23540 see a y, cb and cr stripe from top to bottom.
23542 The sources accept the following parameters:
23547 Specify the level of the Hald CLUT, only available in the @code{haldclutsrc}
23548 source. A level of @code{N} generates a picture of @code{N*N*N} by @code{N*N*N}
23549 pixels to be used as identity matrix for 3D lookup tables. Each component is
23550 coded on a @code{1/(N*N)} scale.
23553 Specify the color of the source, only available in the @code{color}
23554 source. For the syntax of this option, check the
23555 @ref{color syntax,,"Color" section in the ffmpeg-utils manual,ffmpeg-utils}.
23558 Specify the size of the sourced video. For the syntax of this option, check the
23559 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23560 The default value is @code{320x240}.
23562 This option is not available with the @code{allrgb}, @code{allyuv}, and
23563 @code{haldclutsrc} filters.
23566 Specify the frame rate of the sourced video, as the number of frames
23567 generated per second. It has to be a string in the format
23568 @var{frame_rate_num}/@var{frame_rate_den}, an integer number, a floating point
23569 number or a valid video frame rate abbreviation. The default value is
23573 Set the duration of the sourced video. See
23574 @ref{time duration syntax,,the Time duration section in the ffmpeg-utils(1) manual,ffmpeg-utils}
23575 for the accepted syntax.
23577 If not specified, or the expressed duration is negative, the video is
23578 supposed to be generated forever.
23580 Since the frame rate is used as time base, all frames including the last one
23581 will have their full duration. If the specified duration is not a multiple
23582 of the frame duration, it will be rounded up.
23585 Set the sample aspect ratio of the sourced video.
23588 Specify the alpha (opacity) of the background, only available in the
23589 @code{testsrc2} source. The value must be between 0 (fully transparent) and
23590 255 (fully opaque, the default).
23593 Set the number of decimals to show in the timestamp, only available in the
23594 @code{testsrc} source.
23596 The displayed timestamp value will correspond to the original
23597 timestamp value multiplied by the power of 10 of the specified
23598 value. Default value is 0.
23601 @subsection Examples
23605 Generate a video with a duration of 5.3 seconds, with size
23606 176x144 and a frame rate of 10 frames per second:
23608 testsrc=duration=5.3:size=qcif:rate=10
23612 The following graph description will generate a red source
23613 with an opacity of 0.2, with size "qcif" and a frame rate of 10
23616 color=c=red@@0.2:s=qcif:r=10
23620 If the input content is to be ignored, @code{nullsrc} can be used. The
23621 following command generates noise in the luminance plane by employing
23622 the @code{geq} filter:
23624 nullsrc=s=256x256, geq=random(1)*255:128:128
23628 @subsection Commands
23630 The @code{color} source supports the following commands:
23634 Set the color of the created image. Accepts the same syntax of the
23635 corresponding @option{color} option.
23640 Generate video using an OpenCL program.
23645 OpenCL program source file.
23648 Kernel name in program.
23651 Size of frames to generate. This must be set.
23654 Pixel format to use for the generated frames. This must be set.
23657 Number of frames generated every second. Default value is '25'.
23661 For details of how the program loading works, see the @ref{program_opencl}
23668 Generate a colour ramp by setting pixel values from the position of the pixel
23669 in the output image. (Note that this will work with all pixel formats, but
23670 the generated output will not be the same.)
23672 __kernel void ramp(__write_only image2d_t dst,
23673 unsigned int index)
23675 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23678 val.xy = val.zw = convert_float2(loc) / convert_float2(get_image_dim(dst));
23680 write_imagef(dst, loc, val);
23685 Generate a Sierpinski carpet pattern, panning by a single pixel each frame.
23687 __kernel void sierpinski_carpet(__write_only image2d_t dst,
23688 unsigned int index)
23690 int2 loc = (int2)(get_global_id(0), get_global_id(1));
23692 float4 value = 0.0f;
23693 int x = loc.x + index;
23694 int y = loc.y + index;
23695 while (x > 0 || y > 0) {
23696 if (x % 3 == 1 && y % 3 == 1) {
23704 write_imagef(dst, loc, value);
23710 @section sierpinski
23712 Generate a Sierpinski carpet/triangle fractal, and randomly pan around.
23714 This source accepts the following options:
23718 Set frame size. For the syntax of this option, check the @ref{video size syntax,,"Video
23719 size" section in the ffmpeg-utils manual,ffmpeg-utils}. Default value is "640x480".
23722 Set frame rate, expressed as number of frames per second. Default
23726 Set seed which is used for random panning.
23729 Set max jump for single pan destination. Allowed range is from 1 to 10000.
23732 Set fractal type, can be default @code{carpet} or @code{triangle}.
23735 @c man end VIDEO SOURCES
23737 @chapter Video Sinks
23738 @c man begin VIDEO SINKS
23740 Below is a description of the currently available video sinks.
23742 @section buffersink
23744 Buffer video frames, and make them available to the end of the filter
23747 This sink is mainly intended for programmatic use, in particular
23748 through the interface defined in @file{libavfilter/buffersink.h}
23749 or the options system.
23751 It accepts a pointer to an AVBufferSinkContext structure, which
23752 defines the incoming buffers' formats, to be passed as the opaque
23753 parameter to @code{avfilter_init_filter} for initialization.
23757 Null video sink: do absolutely nothing with the input video. It is
23758 mainly useful as a template and for use in analysis / debugging
23761 @c man end VIDEO SINKS
23763 @chapter Multimedia Filters
23764 @c man begin MULTIMEDIA FILTERS
23766 Below is a description of the currently available multimedia filters.
23770 Convert input audio to a video output, displaying the audio bit scope.
23772 The filter accepts the following options:
23776 Set frame rate, expressed as number of frames per second. Default
23780 Specify the video size for the output. For the syntax of this option, check the
23781 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23782 Default value is @code{1024x256}.
23785 Specify list of colors separated by space or by '|' which will be used to
23786 draw channels. Unrecognized or missing colors will be replaced
23790 @section adrawgraph
23791 Draw a graph using input audio metadata.
23793 See @ref{drawgraph}
23795 @section agraphmonitor
23797 See @ref{graphmonitor}.
23799 @section ahistogram
23801 Convert input audio to a video output, displaying the volume histogram.
23803 The filter accepts the following options:
23807 Specify how histogram is calculated.
23809 It accepts the following values:
23812 Use single histogram for all channels.
23814 Use separate histogram for each channel.
23816 Default is @code{single}.
23819 Set frame rate, expressed as number of frames per second. Default
23823 Specify the video size for the output. For the syntax of this option, check the
23824 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23825 Default value is @code{hd720}.
23830 It accepts the following values:
23841 reverse logarithmic
23843 Default is @code{log}.
23846 Set amplitude scale.
23848 It accepts the following values:
23855 Default is @code{log}.
23858 Set how much frames to accumulate in histogram.
23859 Default is 1. Setting this to -1 accumulates all frames.
23862 Set histogram ratio of window height.
23865 Set sonogram sliding.
23867 It accepts the following values:
23870 replace old rows with new ones.
23872 scroll from top to bottom.
23874 Default is @code{replace}.
23877 @section aphasemeter
23879 Measures phase of input audio, which is exported as metadata @code{lavfi.aphasemeter.phase},
23880 representing mean phase of current audio frame. A video output can also be produced and is
23881 enabled by default. The audio is passed through as first output.
23883 Audio will be rematrixed to stereo if it has a different channel layout. Phase value is in
23884 range @code{[-1, 1]} where @code{-1} means left and right channels are completely out of phase
23885 and @code{1} means channels are in phase.
23887 The filter accepts the following options, all related to its video output:
23891 Set the output frame rate. Default value is @code{25}.
23894 Set the video size for the output. For the syntax of this option, check the
23895 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23896 Default value is @code{800x400}.
23901 Specify the red, green, blue contrast. Default values are @code{2},
23902 @code{7} and @code{1}.
23903 Allowed range is @code{[0, 255]}.
23906 Set color which will be used for drawing median phase. If color is
23907 @code{none} which is default, no median phase value will be drawn.
23910 Enable video output. Default is enabled.
23913 @subsection phasing detection
23915 The filter also detects out of phase and mono sequences in stereo streams.
23916 It logs the sequence start, end and duration when it lasts longer or as long as the minimum set.
23918 The filter accepts the following options for this detection:
23922 Enable mono and out of phase detection. Default is disabled.
23925 Set phase tolerance for mono detection, in amplitude ratio. Default is @code{0}.
23926 Allowed range is @code{[0, 1]}.
23929 Set angle threshold for out of phase detection, in degree. Default is @code{170}.
23930 Allowed range is @code{[90, 180]}.
23933 Set mono or out of phase duration until notification, expressed in seconds. Default is @code{2}.
23936 @subsection Examples
23940 Complete example with @command{ffmpeg} to detect 1 second of mono with 0.001 phase tolerance:
23942 ffmpeg -i stereo.wav -af aphasemeter=video=0:phasing=1:duration=1:tolerance=0.001 -f null -
23946 @section avectorscope
23948 Convert input audio to a video output, representing the audio vector
23951 The filter is used to measure the difference between channels of stereo
23952 audio stream. A monaural signal, consisting of identical left and right
23953 signal, results in straight vertical line. Any stereo separation is visible
23954 as a deviation from this line, creating a Lissajous figure.
23955 If the straight (or deviation from it) but horizontal line appears this
23956 indicates that the left and right channels are out of phase.
23958 The filter accepts the following options:
23962 Set the vectorscope mode.
23964 Available values are:
23967 Lissajous rotated by 45 degrees.
23970 Same as above but not rotated.
23973 Shape resembling half of circle.
23976 Default value is @samp{lissajous}.
23979 Set the video size for the output. For the syntax of this option, check the
23980 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
23981 Default value is @code{400x400}.
23984 Set the output frame rate. Default value is @code{25}.
23990 Specify the red, green, blue and alpha contrast. Default values are @code{40},
23991 @code{160}, @code{80} and @code{255}.
23992 Allowed range is @code{[0, 255]}.
23998 Specify the red, green, blue and alpha fade. Default values are @code{15},
23999 @code{10}, @code{5} and @code{5}.
24000 Allowed range is @code{[0, 255]}.
24003 Set the zoom factor. Default value is @code{1}. Allowed range is @code{[0, 10]}.
24004 Values lower than @var{1} will auto adjust zoom factor to maximal possible value.
24007 Set the vectorscope drawing mode.
24009 Available values are:
24012 Draw dot for each sample.
24015 Draw line between previous and current sample.
24018 Default value is @samp{dot}.
24021 Specify amplitude scale of audio samples.
24023 Available values are:
24039 Swap left channel axis with right channel axis.
24049 Mirror only x axis.
24052 Mirror only y axis.
24060 @subsection Examples
24064 Complete example using @command{ffplay}:
24066 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
24067 [a] avectorscope=zoom=1.3:rc=2:gc=200:bc=10:rf=1:gf=8:bf=7 [out0]'
24071 @section bench, abench
24073 Benchmark part of a filtergraph.
24075 The filter accepts the following options:
24079 Start or stop a timer.
24081 Available values are:
24084 Get the current time, set it as frame metadata (using the key
24085 @code{lavfi.bench.start_time}), and forward the frame to the next filter.
24088 Get the current time and fetch the @code{lavfi.bench.start_time} metadata from
24089 the input frame metadata to get the time difference. Time difference, average,
24090 maximum and minimum time (respectively @code{t}, @code{avg}, @code{max} and
24091 @code{min}) are then printed. The timestamps are expressed in seconds.
24095 @subsection Examples
24099 Benchmark @ref{selectivecolor} filter:
24101 bench=start,selectivecolor=reds=-.2 .12 -.49,bench=stop
24107 Concatenate audio and video streams, joining them together one after the
24110 The filter works on segments of synchronized video and audio streams. All
24111 segments must have the same number of streams of each type, and that will
24112 also be the number of streams at output.
24114 The filter accepts the following options:
24119 Set the number of segments. Default is 2.
24122 Set the number of output video streams, that is also the number of video
24123 streams in each segment. Default is 1.
24126 Set the number of output audio streams, that is also the number of audio
24127 streams in each segment. Default is 0.
24130 Activate unsafe mode: do not fail if segments have a different format.
24134 The filter has @var{v}+@var{a} outputs: first @var{v} video outputs, then
24135 @var{a} audio outputs.
24137 There are @var{n}x(@var{v}+@var{a}) inputs: first the inputs for the first
24138 segment, in the same order as the outputs, then the inputs for the second
24141 Related streams do not always have exactly the same duration, for various
24142 reasons including codec frame size or sloppy authoring. For that reason,
24143 related synchronized streams (e.g. a video and its audio track) should be
24144 concatenated at once. The concat filter will use the duration of the longest
24145 stream in each segment (except the last one), and if necessary pad shorter
24146 audio streams with silence.
24148 For this filter to work correctly, all segments must start at timestamp 0.
24150 All corresponding streams must have the same parameters in all segments; the
24151 filtering system will automatically select a common pixel format for video
24152 streams, and a common sample format, sample rate and channel layout for
24153 audio streams, but other settings, such as resolution, must be converted
24154 explicitly by the user.
24156 Different frame rates are acceptable but will result in variable frame rate
24157 at output; be sure to configure the output file to handle it.
24159 @subsection Examples
24163 Concatenate an opening, an episode and an ending, all in bilingual version
24164 (video in stream 0, audio in streams 1 and 2):
24166 ffmpeg -i opening.mkv -i episode.mkv -i ending.mkv -filter_complex \
24167 '[0:0] [0:1] [0:2] [1:0] [1:1] [1:2] [2:0] [2:1] [2:2]
24168 concat=n=3:v=1:a=2 [v] [a1] [a2]' \
24169 -map '[v]' -map '[a1]' -map '[a2]' output.mkv
24173 Concatenate two parts, handling audio and video separately, using the
24174 (a)movie sources, and adjusting the resolution:
24176 movie=part1.mp4, scale=512:288 [v1] ; amovie=part1.mp4 [a1] ;
24177 movie=part2.mp4, scale=512:288 [v2] ; amovie=part2.mp4 [a2] ;
24178 [v1] [v2] concat [outv] ; [a1] [a2] concat=v=0:a=1 [outa]
24180 Note that a desync will happen at the stitch if the audio and video streams
24181 do not have exactly the same duration in the first file.
24185 @subsection Commands
24187 This filter supports the following commands:
24190 Close the current segment and step to the next one
24196 EBU R128 scanner filter. This filter takes an audio stream and analyzes its loudness
24197 level. By default, it logs a message at a frequency of 10Hz with the
24198 Momentary loudness (identified by @code{M}), Short-term loudness (@code{S}),
24199 Integrated loudness (@code{I}) and Loudness Range (@code{LRA}).
24201 The filter can only analyze streams which have a sampling rate of 48000 Hz and whose
24202 sample format is double-precision floating point. The input stream will be converted to
24203 this specification, if needed. Users may need to insert aformat and/or aresample filters
24204 after this filter to obtain the original parameters.
24206 The filter also has a video output (see the @var{video} option) with a real
24207 time graph to observe the loudness evolution. The graphic contains the logged
24208 message mentioned above, so it is not printed anymore when this option is set,
24209 unless the verbose logging is set. The main graphing area contains the
24210 short-term loudness (3 seconds of analysis), and the gauge on the right is for
24211 the momentary loudness (400 milliseconds), but can optionally be configured
24212 to instead display short-term loudness (see @var{gauge}).
24214 The green area marks a +/- 1LU target range around the target loudness
24215 (-23LUFS by default, unless modified through @var{target}).
24217 More information about the Loudness Recommendation EBU R128 on
24218 @url{http://tech.ebu.ch/loudness}.
24220 The filter accepts the following options:
24225 Activate the video output. The audio stream is passed unchanged whether this
24226 option is set or no. The video stream will be the first output stream if
24227 activated. Default is @code{0}.
24230 Set the video size. This option is for video only. For the syntax of this
24232 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
24233 Default and minimum resolution is @code{640x480}.
24236 Set the EBU scale meter. Default is @code{9}. Common values are @code{9} and
24237 @code{18}, respectively for EBU scale meter +9 and EBU scale meter +18. Any
24238 other integer value between this range is allowed.
24241 Set metadata injection. If set to @code{1}, the audio input will be segmented
24242 into 100ms output frames, each of them containing various loudness information
24243 in metadata. All the metadata keys are prefixed with @code{lavfi.r128.}.
24245 Default is @code{0}.
24248 Force the frame logging level.
24250 Available values are:
24253 information logging level
24255 verbose logging level
24258 By default, the logging level is set to @var{info}. If the @option{video} or
24259 the @option{metadata} options are set, it switches to @var{verbose}.
24264 Available modes can be cumulated (the option is a @code{flag} type). Possible
24268 Disable any peak mode (default).
24270 Enable sample-peak mode.
24272 Simple peak mode looking for the higher sample value. It logs a message
24273 for sample-peak (identified by @code{SPK}).
24275 Enable true-peak mode.
24277 If enabled, the peak lookup is done on an over-sampled version of the input
24278 stream for better peak accuracy. It logs a message for true-peak.
24279 (identified by @code{TPK}) and true-peak per frame (identified by @code{FTPK}).
24280 This mode requires a build with @code{libswresample}.
24284 Treat mono input files as "dual mono". If a mono file is intended for playback
24285 on a stereo system, its EBU R128 measurement will be perceptually incorrect.
24286 If set to @code{true}, this option will compensate for this effect.
24287 Multi-channel input files are not affected by this option.
24290 Set a specific pan law to be used for the measurement of dual mono files.
24291 This parameter is optional, and has a default value of -3.01dB.
24294 Set a specific target level (in LUFS) used as relative zero in the visualization.
24295 This parameter is optional and has a default value of -23LUFS as specified
24296 by EBU R128. However, material published online may prefer a level of -16LUFS
24297 (e.g. for use with podcasts or video platforms).
24300 Set the value displayed by the gauge. Valid values are @code{momentary} and s
24301 @code{shortterm}. By default the momentary value will be used, but in certain
24302 scenarios it may be more useful to observe the short term value instead (e.g.
24306 Sets the display scale for the loudness. Valid parameters are @code{absolute}
24307 (in LUFS) or @code{relative} (LU) relative to the target. This only affects the
24308 video output, not the summary or continuous log output.
24311 @subsection Examples
24315 Real-time graph using @command{ffplay}, with a EBU scale meter +18:
24317 ffplay -f lavfi -i "amovie=input.mp3,ebur128=video=1:meter=18 [out0][out1]"
24321 Run an analysis with @command{ffmpeg}:
24323 ffmpeg -nostats -i input.mp3 -filter_complex ebur128 -f null -
24327 @section interleave, ainterleave
24329 Temporally interleave frames from several inputs.
24331 @code{interleave} works with video inputs, @code{ainterleave} with audio.
24333 These filters read frames from several inputs and send the oldest
24334 queued frame to the output.
24336 Input streams must have well defined, monotonically increasing frame
24339 In order to submit one frame to output, these filters need to enqueue
24340 at least one frame for each input, so they cannot work in case one
24341 input is not yet terminated and will not receive incoming frames.
24343 For example consider the case when one input is a @code{select} filter
24344 which always drops input frames. The @code{interleave} filter will keep
24345 reading from that input, but it will never be able to send new frames
24346 to output until the input sends an end-of-stream signal.
24348 Also, depending on inputs synchronization, the filters will drop
24349 frames in case one input receives more frames than the other ones, and
24350 the queue is already filled.
24352 These filters accept the following options:
24356 Set the number of different inputs, it is 2 by default.
24359 How to determine the end-of-stream.
24363 The duration of the longest input. (default)
24366 The duration of the shortest input.
24369 The duration of the first input.
24374 @subsection Examples
24378 Interleave frames belonging to different streams using @command{ffmpeg}:
24380 ffmpeg -i bambi.avi -i pr0n.mkv -filter_complex "[0:v][1:v] interleave" out.avi
24384 Add flickering blur effect:
24386 select='if(gt(random(0), 0.2), 1, 2)':n=2 [tmp], boxblur=2:2, [tmp] interleave
24390 @section metadata, ametadata
24392 Manipulate frame metadata.
24394 This filter accepts the following options:
24398 Set mode of operation of the filter.
24400 Can be one of the following:
24404 If both @code{value} and @code{key} is set, select frames
24405 which have such metadata. If only @code{key} is set, select
24406 every frame that has such key in metadata.
24409 Add new metadata @code{key} and @code{value}. If key is already available
24413 Modify value of already present key.
24416 If @code{value} is set, delete only keys that have such value.
24417 Otherwise, delete key. If @code{key} is not set, delete all metadata values in
24421 Print key and its value if metadata was found. If @code{key} is not set print all
24422 metadata values available in frame.
24426 Set key used with all modes. Must be set for all modes except @code{print} and @code{delete}.
24429 Set metadata value which will be used. This option is mandatory for
24430 @code{modify} and @code{add} mode.
24433 Which function to use when comparing metadata value and @code{value}.
24435 Can be one of following:
24439 Values are interpreted as strings, returns true if metadata value is same as @code{value}.
24442 Values are interpreted as strings, returns true if metadata value starts with
24443 the @code{value} option string.
24446 Values are interpreted as floats, returns true if metadata value is less than @code{value}.
24449 Values are interpreted as floats, returns true if @code{value} is equal with metadata value.
24452 Values are interpreted as floats, returns true if metadata value is greater than @code{value}.
24455 Values are interpreted as floats, returns true if expression from option @code{expr}
24459 Values are interpreted as strings, returns true if metadata value ends with
24460 the @code{value} option string.
24464 Set expression which is used when @code{function} is set to @code{expr}.
24465 The expression is evaluated through the eval API and can contain the following
24470 Float representation of @code{value} from metadata key.
24473 Float representation of @code{value} as supplied by user in @code{value} option.
24477 If specified in @code{print} mode, output is written to the named file. Instead of
24478 plain filename any writable url can be specified. Filename ``-'' is a shorthand
24479 for standard output. If @code{file} option is not set, output is written to the log
24480 with AV_LOG_INFO loglevel.
24483 Reduces buffering in print mode when output is written to a URL set using @var{file}.
24487 @subsection Examples
24491 Print all metadata values for frames with key @code{lavfi.signalstats.YDIF} with values
24494 signalstats,metadata=print:key=lavfi.signalstats.YDIF:value=0:function=expr:expr='between(VALUE1,0,1)'
24497 Print silencedetect output to file @file{metadata.txt}.
24499 silencedetect,ametadata=mode=print:file=metadata.txt
24502 Direct all metadata to a pipe with file descriptor 4.
24504 metadata=mode=print:file='pipe\:4'
24508 @section perms, aperms
24510 Set read/write permissions for the output frames.
24512 These filters are mainly aimed at developers to test direct path in the
24513 following filter in the filtergraph.
24515 The filters accept the following options:
24519 Select the permissions mode.
24521 It accepts the following values:
24524 Do nothing. This is the default.
24526 Set all the output frames read-only.
24528 Set all the output frames directly writable.
24530 Make the frame read-only if writable, and writable if read-only.
24532 Set each output frame read-only or writable randomly.
24536 Set the seed for the @var{random} mode, must be an integer included between
24537 @code{0} and @code{UINT32_MAX}. If not specified, or if explicitly set to
24538 @code{-1}, the filter will try to use a good random seed on a best effort
24542 Note: in case of auto-inserted filter between the permission filter and the
24543 following one, the permission might not be received as expected in that
24544 following filter. Inserting a @ref{format} or @ref{aformat} filter before the
24545 perms/aperms filter can avoid this problem.
24547 @section realtime, arealtime
24549 Slow down filtering to match real time approximately.
24551 These filters will pause the filtering for a variable amount of time to
24552 match the output rate with the input timestamps.
24553 They are similar to the @option{re} option to @code{ffmpeg}.
24555 They accept the following options:
24559 Time limit for the pauses. Any pause longer than that will be considered
24560 a timestamp discontinuity and reset the timer. Default is 2 seconds.
24562 Speed factor for processing. The value must be a float larger than zero.
24563 Values larger than 1.0 will result in faster than realtime processing,
24564 smaller will slow processing down. The @var{limit} is automatically adapted
24565 accordingly. Default is 1.0.
24567 A processing speed faster than what is possible without these filters cannot
24572 @section select, aselect
24574 Select frames to pass in output.
24576 This filter accepts the following options:
24581 Set expression, which is evaluated for each input frame.
24583 If the expression is evaluated to zero, the frame is discarded.
24585 If the evaluation result is negative or NaN, the frame is sent to the
24586 first output; otherwise it is sent to the output with index
24587 @code{ceil(val)-1}, assuming that the input index starts from 0.
24589 For example a value of @code{1.2} corresponds to the output with index
24590 @code{ceil(1.2)-1 = 2-1 = 1}, that is the second output.
24593 Set the number of outputs. The output to which to send the selected
24594 frame is based on the result of the evaluation. Default value is 1.
24597 The expression can contain the following constants:
24601 The (sequential) number of the filtered frame, starting from 0.
24604 The (sequential) number of the selected frame, starting from 0.
24606 @item prev_selected_n
24607 The sequential number of the last selected frame. It's NAN if undefined.
24610 The timebase of the input timestamps.
24613 The PTS (Presentation TimeStamp) of the filtered video frame,
24614 expressed in @var{TB} units. It's NAN if undefined.
24617 The PTS of the filtered video frame,
24618 expressed in seconds. It's NAN if undefined.
24621 The PTS of the previously filtered video frame. It's NAN if undefined.
24623 @item prev_selected_pts
24624 The PTS of the last previously filtered video frame. It's NAN if undefined.
24626 @item prev_selected_t
24627 The PTS of the last previously selected video frame, expressed in seconds. It's NAN if undefined.
24630 The PTS of the first video frame in the video. It's NAN if undefined.
24633 The time of the first video frame in the video. It's NAN if undefined.
24635 @item pict_type @emph{(video only)}
24636 The type of the filtered frame. It can assume one of the following
24648 @item interlace_type @emph{(video only)}
24649 The frame interlace type. It can assume one of the following values:
24652 The frame is progressive (not interlaced).
24654 The frame is top-field-first.
24656 The frame is bottom-field-first.
24659 @item consumed_sample_n @emph{(audio only)}
24660 the number of selected samples before the current frame
24662 @item samples_n @emph{(audio only)}
24663 the number of samples in the current frame
24665 @item sample_rate @emph{(audio only)}
24666 the input sample rate
24669 This is 1 if the filtered frame is a key-frame, 0 otherwise.
24672 the position in the file of the filtered frame, -1 if the information
24673 is not available (e.g. for synthetic video)
24675 @item scene @emph{(video only)}
24676 value between 0 and 1 to indicate a new scene; a low value reflects a low
24677 probability for the current frame to introduce a new scene, while a higher
24678 value means the current frame is more likely to be one (see the example below)
24680 @item concatdec_select
24681 The concat demuxer can select only part of a concat input file by setting an
24682 inpoint and an outpoint, but the output packets may not be entirely contained
24683 in the selected interval. By using this variable, it is possible to skip frames
24684 generated by the concat demuxer which are not exactly contained in the selected
24687 This works by comparing the frame pts against the @var{lavf.concat.start_time}
24688 and the @var{lavf.concat.duration} packet metadata values which are also
24689 present in the decoded frames.
24691 The @var{concatdec_select} variable is -1 if the frame pts is at least
24692 start_time and either the duration metadata is missing or the frame pts is less
24693 than start_time + duration, 0 otherwise, and NaN if the start_time metadata is
24696 That basically means that an input frame is selected if its pts is within the
24697 interval set by the concat demuxer.
24701 The default value of the select expression is "1".
24703 @subsection Examples
24707 Select all frames in input:
24712 The example above is the same as:
24724 Select only I-frames:
24726 select='eq(pict_type\,I)'
24730 Select one frame every 100:
24732 select='not(mod(n\,100))'
24736 Select only frames contained in the 10-20 time interval:
24738 select=between(t\,10\,20)
24742 Select only I-frames contained in the 10-20 time interval:
24744 select=between(t\,10\,20)*eq(pict_type\,I)
24748 Select frames with a minimum distance of 10 seconds:
24750 select='isnan(prev_selected_t)+gte(t-prev_selected_t\,10)'
24754 Use aselect to select only audio frames with samples number > 100:
24756 aselect='gt(samples_n\,100)'
24760 Create a mosaic of the first scenes:
24762 ffmpeg -i video.avi -vf select='gt(scene\,0.4)',scale=160:120,tile -frames:v 1 preview.png
24765 Comparing @var{scene} against a value between 0.3 and 0.5 is generally a sane
24769 Send even and odd frames to separate outputs, and compose them:
24771 select=n=2:e='mod(n, 2)+1' [odd][even]; [odd] pad=h=2*ih [tmp]; [tmp][even] overlay=y=h
24775 Select useful frames from an ffconcat file which is using inpoints and
24776 outpoints but where the source files are not intra frame only.
24778 ffmpeg -copyts -vsync 0 -segment_time_metadata 1 -i input.ffconcat -vf select=concatdec_select -af aselect=concatdec_select output.avi
24782 @section sendcmd, asendcmd
24784 Send commands to filters in the filtergraph.
24786 These filters read commands to be sent to other filters in the
24789 @code{sendcmd} must be inserted between two video filters,
24790 @code{asendcmd} must be inserted between two audio filters, but apart
24791 from that they act the same way.
24793 The specification of commands can be provided in the filter arguments
24794 with the @var{commands} option, or in a file specified by the
24795 @var{filename} option.
24797 These filters accept the following options:
24800 Set the commands to be read and sent to the other filters.
24802 Set the filename of the commands to be read and sent to the other
24806 @subsection Commands syntax
24808 A commands description consists of a sequence of interval
24809 specifications, comprising a list of commands to be executed when a
24810 particular event related to that interval occurs. The occurring event
24811 is typically the current frame time entering or leaving a given time
24814 An interval is specified by the following syntax:
24816 @var{START}[-@var{END}] @var{COMMANDS};
24819 The time interval is specified by the @var{START} and @var{END} times.
24820 @var{END} is optional and defaults to the maximum time.
24822 The current frame time is considered within the specified interval if
24823 it is included in the interval [@var{START}, @var{END}), that is when
24824 the time is greater or equal to @var{START} and is lesser than
24827 @var{COMMANDS} consists of a sequence of one or more command
24828 specifications, separated by ",", relating to that interval. The
24829 syntax of a command specification is given by:
24831 [@var{FLAGS}] @var{TARGET} @var{COMMAND} @var{ARG}
24834 @var{FLAGS} is optional and specifies the type of events relating to
24835 the time interval which enable sending the specified command, and must
24836 be a non-null sequence of identifier flags separated by "+" or "|" and
24837 enclosed between "[" and "]".
24839 The following flags are recognized:
24842 The command is sent when the current frame timestamp enters the
24843 specified interval. In other words, the command is sent when the
24844 previous frame timestamp was not in the given interval, and the
24848 The command is sent when the current frame timestamp leaves the
24849 specified interval. In other words, the command is sent when the
24850 previous frame timestamp was in the given interval, and the
24854 The command @var{ARG} is interpreted as expression and result of
24855 expression is passed as @var{ARG}.
24857 The expression is evaluated through the eval API and can contain the following
24862 Original position in the file of the frame, or undefined if undefined
24863 for the current frame.
24866 The presentation timestamp in input.
24869 The count of the input frame for video or audio, starting from 0.
24872 The time in seconds of the current frame.
24875 The start time in seconds of the current command interval.
24878 The end time in seconds of the current command interval.
24881 The interpolated time of the current command interval, TI = (T - TS) / (TE - TS).
24886 If @var{FLAGS} is not specified, a default value of @code{[enter]} is
24889 @var{TARGET} specifies the target of the command, usually the name of
24890 the filter class or a specific filter instance name.
24892 @var{COMMAND} specifies the name of the command for the target filter.
24894 @var{ARG} is optional and specifies the optional list of argument for
24895 the given @var{COMMAND}.
24897 Between one interval specification and another, whitespaces, or
24898 sequences of characters starting with @code{#} until the end of line,
24899 are ignored and can be used to annotate comments.
24901 A simplified BNF description of the commands specification syntax
24904 @var{COMMAND_FLAG} ::= "enter" | "leave"
24905 @var{COMMAND_FLAGS} ::= @var{COMMAND_FLAG} [(+|"|")@var{COMMAND_FLAG}]
24906 @var{COMMAND} ::= ["[" @var{COMMAND_FLAGS} "]"] @var{TARGET} @var{COMMAND} [@var{ARG}]
24907 @var{COMMANDS} ::= @var{COMMAND} [,@var{COMMANDS}]
24908 @var{INTERVAL} ::= @var{START}[-@var{END}] @var{COMMANDS}
24909 @var{INTERVALS} ::= @var{INTERVAL}[;@var{INTERVALS}]
24912 @subsection Examples
24916 Specify audio tempo change at second 4:
24918 asendcmd=c='4.0 atempo tempo 1.5',atempo
24922 Target a specific filter instance:
24924 asendcmd=c='4.0 atempo@@my tempo 1.5',atempo@@my
24928 Specify a list of drawtext and hue commands in a file.
24930 # show text in the interval 5-10
24931 5.0-10.0 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=hello world',
24932 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=';
24934 # desaturate the image in the interval 15-20
24935 15.0-20.0 [enter] hue s 0,
24936 [enter] drawtext reinit 'fontfile=FreeSerif.ttf:text=nocolor',
24938 [leave] drawtext reinit 'fontfile=FreeSerif.ttf:text=color';
24940 # apply an exponential saturation fade-out effect, starting from time 25
24941 25 [enter] hue s exp(25-t)
24944 A filtergraph allowing to read and process the above command list
24945 stored in a file @file{test.cmd}, can be specified with:
24947 sendcmd=f=test.cmd,drawtext=fontfile=FreeSerif.ttf:text='',hue
24952 @section setpts, asetpts
24954 Change the PTS (presentation timestamp) of the input frames.
24956 @code{setpts} works on video frames, @code{asetpts} on audio frames.
24958 This filter accepts the following options:
24963 The expression which is evaluated for each frame to construct its timestamp.
24967 The expression is evaluated through the eval API and can contain the following
24971 @item FRAME_RATE, FR
24972 frame rate, only defined for constant frame-rate video
24975 The presentation timestamp in input
24978 The count of the input frame for video or the number of consumed samples,
24979 not including the current frame for audio, starting from 0.
24981 @item NB_CONSUMED_SAMPLES
24982 The number of consumed samples, not including the current frame (only
24985 @item NB_SAMPLES, S
24986 The number of samples in the current frame (only audio)
24988 @item SAMPLE_RATE, SR
24989 The audio sample rate.
24992 The PTS of the first frame.
24995 the time in seconds of the first frame
24998 State whether the current frame is interlaced.
25001 the time in seconds of the current frame
25004 original position in the file of the frame, or undefined if undefined
25005 for the current frame
25008 The previous input PTS.
25011 previous input time in seconds
25014 The previous output PTS.
25017 previous output time in seconds
25020 The wallclock (RTC) time in microseconds. This is deprecated, use time(0)
25024 The wallclock (RTC) time at the start of the movie in microseconds.
25027 The timebase of the input timestamps.
25031 @subsection Examples
25035 Start counting PTS from zero
25037 setpts=PTS-STARTPTS
25041 Apply fast motion effect:
25047 Apply slow motion effect:
25053 Set fixed rate of 25 frames per second:
25059 Set fixed rate 25 fps with some jitter:
25061 setpts='1/(25*TB) * (N + 0.05 * sin(N*2*PI/25))'
25065 Apply an offset of 10 seconds to the input PTS:
25071 Generate timestamps from a "live source" and rebase onto the current timebase:
25073 setpts='(RTCTIME - RTCSTART) / (TB * 1000000)'
25077 Generate timestamps by counting samples:
25086 Force color range for the output video frame.
25088 The @code{setrange} filter marks the color range property for the
25089 output frames. It does not change the input frame, but only sets the
25090 corresponding property, which affects how the frame is treated by
25093 The filter accepts the following options:
25098 Available values are:
25102 Keep the same color range property.
25104 @item unspecified, unknown
25105 Set the color range as unspecified.
25107 @item limited, tv, mpeg
25108 Set the color range as limited.
25110 @item full, pc, jpeg
25111 Set the color range as full.
25115 @section settb, asettb
25117 Set the timebase to use for the output frames timestamps.
25118 It is mainly useful for testing timebase configuration.
25120 It accepts the following parameters:
25125 The expression which is evaluated into the output timebase.
25129 The value for @option{tb} is an arithmetic expression representing a
25130 rational. The expression can contain the constants "AVTB" (the default
25131 timebase), "intb" (the input timebase) and "sr" (the sample rate,
25132 audio only). Default value is "intb".
25134 @subsection Examples
25138 Set the timebase to 1/25:
25144 Set the timebase to 1/10:
25150 Set the timebase to 1001/1000:
25156 Set the timebase to 2*intb:
25162 Set the default timebase value:
25169 Convert input audio to a video output representing frequency spectrum
25170 logarithmically using Brown-Puckette constant Q transform algorithm with
25171 direct frequency domain coefficient calculation (but the transform itself
25172 is not really constant Q, instead the Q factor is actually variable/clamped),
25173 with musical tone scale, from E0 to D#10.
25175 The filter accepts the following options:
25179 Specify the video size for the output. It must be even. For the syntax of this option,
25180 check the @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25181 Default value is @code{1920x1080}.
25184 Set the output frame rate. Default value is @code{25}.
25187 Set the bargraph height. It must be even. Default value is @code{-1} which
25188 computes the bargraph height automatically.
25191 Set the axis height. It must be even. Default value is @code{-1} which computes
25192 the axis height automatically.
25195 Set the sonogram height. It must be even. Default value is @code{-1} which
25196 computes the sonogram height automatically.
25199 Set the fullhd resolution. This option is deprecated, use @var{size}, @var{s}
25200 instead. Default value is @code{1}.
25202 @item sono_v, volume
25203 Specify the sonogram volume expression. It can contain variables:
25206 the @var{bar_v} evaluated expression
25207 @item frequency, freq, f
25208 the frequency where it is evaluated
25209 @item timeclamp, tc
25210 the value of @var{timeclamp} option
25214 @item a_weighting(f)
25215 A-weighting of equal loudness
25216 @item b_weighting(f)
25217 B-weighting of equal loudness
25218 @item c_weighting(f)
25219 C-weighting of equal loudness.
25221 Default value is @code{16}.
25223 @item bar_v, volume2
25224 Specify the bargraph volume expression. It can contain variables:
25227 the @var{sono_v} evaluated expression
25228 @item frequency, freq, f
25229 the frequency where it is evaluated
25230 @item timeclamp, tc
25231 the value of @var{timeclamp} option
25235 @item a_weighting(f)
25236 A-weighting of equal loudness
25237 @item b_weighting(f)
25238 B-weighting of equal loudness
25239 @item c_weighting(f)
25240 C-weighting of equal loudness.
25242 Default value is @code{sono_v}.
25244 @item sono_g, gamma
25245 Specify the sonogram gamma. Lower gamma makes the spectrum more contrast,
25246 higher gamma makes the spectrum having more range. Default value is @code{3}.
25247 Acceptable range is @code{[1, 7]}.
25249 @item bar_g, gamma2
25250 Specify the bargraph gamma. Default value is @code{1}. Acceptable range is
25254 Specify the bargraph transparency level. Lower value makes the bargraph sharper.
25255 Default value is @code{1}. Acceptable range is @code{[0, 1]}.
25257 @item timeclamp, tc
25258 Specify the transform timeclamp. At low frequency, there is trade-off between
25259 accuracy in time domain and frequency domain. If timeclamp is lower,
25260 event in time domain is represented more accurately (such as fast bass drum),
25261 otherwise event in frequency domain is represented more accurately
25262 (such as bass guitar). Acceptable range is @code{[0.002, 1]}. Default value is @code{0.17}.
25265 Set attack time in seconds. The default is @code{0} (disabled). Otherwise, it
25266 limits future samples by applying asymmetric windowing in time domain, useful
25267 when low latency is required. Accepted range is @code{[0, 1]}.
25270 Specify the transform base frequency. Default value is @code{20.01523126408007475},
25271 which is frequency 50 cents below E0. Acceptable range is @code{[10, 100000]}.
25274 Specify the transform end frequency. Default value is @code{20495.59681441799654},
25275 which is frequency 50 cents above D#10. Acceptable range is @code{[10, 100000]}.
25278 This option is deprecated and ignored.
25281 Specify the transform length in time domain. Use this option to control accuracy
25282 trade-off between time domain and frequency domain at every frequency sample.
25283 It can contain variables:
25285 @item frequency, freq, f
25286 the frequency where it is evaluated
25287 @item timeclamp, tc
25288 the value of @var{timeclamp} option.
25290 Default value is @code{384*tc/(384+tc*f)}.
25293 Specify the transform count for every video frame. Default value is @code{6}.
25294 Acceptable range is @code{[1, 30]}.
25297 Specify the transform count for every single pixel. Default value is @code{0},
25298 which makes it computed automatically. Acceptable range is @code{[0, 10]}.
25301 Specify font file for use with freetype to draw the axis. If not specified,
25302 use embedded font. Note that drawing with font file or embedded font is not
25303 implemented with custom @var{basefreq} and @var{endfreq}, use @var{axisfile}
25307 Specify fontconfig pattern. This has lower priority than @var{fontfile}. The
25308 @code{:} in the pattern may be replaced by @code{|} to avoid unnecessary
25312 Specify font color expression. This is arithmetic expression that should return
25313 integer value 0xRRGGBB. It can contain variables:
25315 @item frequency, freq, f
25316 the frequency where it is evaluated
25317 @item timeclamp, tc
25318 the value of @var{timeclamp} option
25323 midi number of frequency f, some midi numbers: E0(16), C1(24), C2(36), A4(69)
25324 @item r(x), g(x), b(x)
25325 red, green, and blue value of intensity x.
25327 Default value is @code{st(0, (midi(f)-59.5)/12);
25328 st(1, if(between(ld(0),0,1), 0.5-0.5*cos(2*PI*ld(0)), 0));
25329 r(1-ld(1)) + b(ld(1))}.
25332 Specify image file to draw the axis. This option override @var{fontfile} and
25333 @var{fontcolor} option.
25336 Enable/disable drawing text to the axis. If it is set to @code{0}, drawing to
25337 the axis is disabled, ignoring @var{fontfile} and @var{axisfile} option.
25338 Default value is @code{1}.
25341 Set colorspace. The accepted values are:
25344 Unspecified (default)
25353 BT.470BG or BT.601-6 625
25356 SMPTE-170M or BT.601-6 525
25362 BT.2020 with non-constant luminance
25367 Set spectrogram color scheme. This is list of floating point values with format
25368 @code{left_r|left_g|left_b|right_r|right_g|right_b}.
25369 The default is @code{1|0.5|0|0|0.5|1}.
25373 @subsection Examples
25377 Playing audio while showing the spectrum:
25379 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt [out0]'
25383 Same as above, but with frame rate 30 fps:
25385 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=fps=30:count=5 [out0]'
25389 Playing at 1280x720:
25391 ffplay -f lavfi 'amovie=a.mp3, asplit [a][out1]; [a] showcqt=s=1280x720:count=4 [out0]'
25395 Disable sonogram display:
25401 A1 and its harmonics: A1, A2, (near)E3, A3:
25403 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),
25404 asplit[a][out1]; [a] showcqt [out0]'
25408 Same as above, but with more accuracy in frequency domain:
25410 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),
25411 asplit[a][out1]; [a] showcqt=timeclamp=0.5 [out0]'
25417 bar_v=10:sono_v=bar_v*a_weighting(f)
25421 Custom gamma, now spectrum is linear to the amplitude.
25427 Custom tlength equation:
25429 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)))'
25433 Custom fontcolor and fontfile, C-note is colored green, others are colored blue:
25435 fontcolor='if(mod(floor(midi(f)+0.5),12), 0x0000FF, g(1))':fontfile=myfont.ttf
25439 Custom font using fontconfig:
25441 font='Courier New,Monospace,mono|bold'
25445 Custom frequency range with custom axis using image file:
25447 axisfile=myaxis.png:basefreq=40:endfreq=10000
25453 Convert input audio to video output representing the audio power spectrum.
25454 Audio amplitude is on Y-axis while frequency is on X-axis.
25456 The filter accepts the following options:
25460 Specify size of video. For the syntax of this option, check the
25461 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25462 Default is @code{1024x512}.
25466 This set how each frequency bin will be represented.
25468 It accepts the following values:
25474 Default is @code{bar}.
25477 Set amplitude scale.
25479 It accepts the following values:
25493 Default is @code{log}.
25496 Set frequency scale.
25498 It accepts the following values:
25507 Reverse logarithmic scale.
25509 Default is @code{lin}.
25512 Set window size. Allowed range is from 16 to 65536.
25514 Default is @code{2048}
25517 Set windowing function.
25519 It accepts the following values:
25542 Default is @code{hanning}.
25545 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
25546 which means optimal overlap for selected window function will be picked.
25549 Set time averaging. Setting this to 0 will display current maximal peaks.
25550 Default is @code{1}, which means time averaging is disabled.
25553 Specify list of colors separated by space or by '|' which will be used to
25554 draw channel frequencies. Unrecognized or missing colors will be replaced
25558 Set channel display mode.
25560 It accepts the following values:
25565 Default is @code{combined}.
25568 Set minimum amplitude used in @code{log} amplitude scaler.
25571 Set data display mode.
25573 It accepts the following values:
25579 Default is @code{magnitude}.
25582 @section showspatial
25584 Convert stereo input audio to a video output, representing the spatial relationship
25585 between two channels.
25587 The filter accepts the following options:
25591 Specify the video size for the output. For the syntax of this option, check the
25592 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25593 Default value is @code{512x512}.
25596 Set window size. Allowed range is from @var{1024} to @var{65536}. Default size is @var{4096}.
25599 Set window function.
25601 It accepts the following values:
25626 Default value is @code{hann}.
25629 Set ratio of overlap window. Default value is @code{0.5}.
25630 When value is @code{1} overlap is set to recommended size for specific
25631 window function currently used.
25634 @anchor{showspectrum}
25635 @section showspectrum
25637 Convert input audio to a video output, representing the audio frequency
25640 The filter accepts the following options:
25644 Specify the video size for the output. For the syntax of this option, check the
25645 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25646 Default value is @code{640x512}.
25649 Specify how the spectrum should slide along the window.
25651 It accepts the following values:
25654 the samples start again on the left when they reach the right
25656 the samples scroll from right to left
25658 frames are only produced when the samples reach the right
25660 the samples scroll from left to right
25663 Default value is @code{replace}.
25666 Specify display mode.
25668 It accepts the following values:
25671 all channels are displayed in the same row
25673 all channels are displayed in separate rows
25676 Default value is @samp{combined}.
25679 Specify display color mode.
25681 It accepts the following values:
25684 each channel is displayed in a separate color
25686 each channel is displayed using the same color scheme
25688 each channel is displayed using the rainbow color scheme
25690 each channel is displayed using the moreland color scheme
25692 each channel is displayed using the nebulae color scheme
25694 each channel is displayed using the fire color scheme
25696 each channel is displayed using the fiery color scheme
25698 each channel is displayed using the fruit color scheme
25700 each channel is displayed using the cool color scheme
25702 each channel is displayed using the magma color scheme
25704 each channel is displayed using the green color scheme
25706 each channel is displayed using the viridis color scheme
25708 each channel is displayed using the plasma color scheme
25710 each channel is displayed using the cividis color scheme
25712 each channel is displayed using the terrain color scheme
25715 Default value is @samp{channel}.
25718 Specify scale used for calculating intensity color values.
25720 It accepts the following values:
25725 square root, default
25736 Default value is @samp{sqrt}.
25739 Specify frequency scale.
25741 It accepts the following values:
25749 Default value is @samp{lin}.
25752 Set saturation modifier for displayed colors. Negative values provide
25753 alternative color scheme. @code{0} is no saturation at all.
25754 Saturation must be in [-10.0, 10.0] range.
25755 Default value is @code{1}.
25758 Set window function.
25760 It accepts the following values:
25785 Default value is @code{hann}.
25788 Set orientation of time vs frequency axis. Can be @code{vertical} or
25789 @code{horizontal}. Default is @code{vertical}.
25792 Set ratio of overlap window. Default value is @code{0}.
25793 When value is @code{1} overlap is set to recommended size for specific
25794 window function currently used.
25797 Set scale gain for calculating intensity color values.
25798 Default value is @code{1}.
25801 Set which data to display. Can be @code{magnitude}, default or @code{phase}.
25804 Set color rotation, must be in [-1.0, 1.0] range.
25805 Default value is @code{0}.
25808 Set start frequency from which to display spectrogram. Default is @code{0}.
25811 Set stop frequency to which to display spectrogram. Default is @code{0}.
25814 Set upper frame rate limit. Default is @code{auto}, unlimited.
25817 Draw time and frequency axes and legends. Default is disabled.
25820 The usage is very similar to the showwaves filter; see the examples in that
25823 @subsection Examples
25827 Large window with logarithmic color scaling:
25829 showspectrum=s=1280x480:scale=log
25833 Complete example for a colored and sliding spectrum per channel using @command{ffplay}:
25835 ffplay -f lavfi 'amovie=input.mp3, asplit [a][out1];
25836 [a] showspectrum=mode=separate:color=intensity:slide=1:scale=cbrt [out0]'
25840 @section showspectrumpic
25842 Convert input audio to a single video frame, representing the audio frequency
25845 The filter accepts the following options:
25849 Specify the video size for the output. For the syntax of this option, check the
25850 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
25851 Default value is @code{4096x2048}.
25854 Specify display mode.
25856 It accepts the following values:
25859 all channels are displayed in the same row
25861 all channels are displayed in separate rows
25863 Default value is @samp{combined}.
25866 Specify display color mode.
25868 It accepts the following values:
25871 each channel is displayed in a separate color
25873 each channel is displayed using the same color scheme
25875 each channel is displayed using the rainbow color scheme
25877 each channel is displayed using the moreland color scheme
25879 each channel is displayed using the nebulae color scheme
25881 each channel is displayed using the fire color scheme
25883 each channel is displayed using the fiery color scheme
25885 each channel is displayed using the fruit color scheme
25887 each channel is displayed using the cool color scheme
25889 each channel is displayed using the magma color scheme
25891 each channel is displayed using the green color scheme
25893 each channel is displayed using the viridis color scheme
25895 each channel is displayed using the plasma color scheme
25897 each channel is displayed using the cividis color scheme
25899 each channel is displayed using the terrain color scheme
25901 Default value is @samp{intensity}.
25904 Specify scale used for calculating intensity color values.
25906 It accepts the following values:
25911 square root, default
25921 Default value is @samp{log}.
25924 Specify frequency scale.
25926 It accepts the following values:
25934 Default value is @samp{lin}.
25937 Set saturation modifier for displayed colors. Negative values provide
25938 alternative color scheme. @code{0} is no saturation at all.
25939 Saturation must be in [-10.0, 10.0] range.
25940 Default value is @code{1}.
25943 Set window function.
25945 It accepts the following values:
25969 Default value is @code{hann}.
25972 Set orientation of time vs frequency axis. Can be @code{vertical} or
25973 @code{horizontal}. Default is @code{vertical}.
25976 Set scale gain for calculating intensity color values.
25977 Default value is @code{1}.
25980 Draw time and frequency axes and legends. Default is enabled.
25983 Set color rotation, must be in [-1.0, 1.0] range.
25984 Default value is @code{0}.
25987 Set start frequency from which to display spectrogram. Default is @code{0}.
25990 Set stop frequency to which to display spectrogram. Default is @code{0}.
25993 @subsection Examples
25997 Extract an audio spectrogram of a whole audio track
25998 in a 1024x1024 picture using @command{ffmpeg}:
26000 ffmpeg -i audio.flac -lavfi showspectrumpic=s=1024x1024 spectrogram.png
26004 @section showvolume
26006 Convert input audio volume to a video output.
26008 The filter accepts the following options:
26015 Set border width, allowed range is [0, 5]. Default is 1.
26018 Set channel width, allowed range is [80, 8192]. Default is 400.
26021 Set channel height, allowed range is [1, 900]. Default is 20.
26024 Set fade, allowed range is [0, 1]. Default is 0.95.
26027 Set volume color expression.
26029 The expression can use the following variables:
26033 Current max volume of channel in dB.
26039 Current channel number, starting from 0.
26043 If set, displays channel names. Default is enabled.
26046 If set, displays volume values. Default is enabled.
26049 Set orientation, can be horizontal: @code{h} or vertical: @code{v},
26050 default is @code{h}.
26053 Set step size, allowed range is [0, 5]. Default is 0, which means
26057 Set background opacity, allowed range is [0, 1]. Default is 0.
26060 Set metering mode, can be peak: @code{p} or rms: @code{r},
26061 default is @code{p}.
26064 Set display scale, can be linear: @code{lin} or log: @code{log},
26065 default is @code{lin}.
26069 If set to > 0., display a line for the max level
26070 in the previous seconds.
26071 default is disabled: @code{0.}
26074 The color of the max line. Use when @code{dm} option is set to > 0.
26075 default is: @code{orange}
26080 Convert input audio to a video output, representing the samples waves.
26082 The filter accepts the following options:
26086 Specify the video size for the output. For the syntax of this option, check the
26087 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26088 Default value is @code{600x240}.
26093 Available values are:
26096 Draw a point for each sample.
26099 Draw a vertical line for each sample.
26102 Draw a point for each sample and a line between them.
26105 Draw a centered vertical line for each sample.
26108 Default value is @code{point}.
26111 Set the number of samples which are printed on the same column. A
26112 larger value will decrease the frame rate. Must be a positive
26113 integer. This option can be set only if the value for @var{rate}
26114 is not explicitly specified.
26117 Set the (approximate) output frame rate. This is done by setting the
26118 option @var{n}. Default value is "25".
26120 @item split_channels
26121 Set if channels should be drawn separately or overlap. Default value is 0.
26124 Set colors separated by '|' which are going to be used for drawing of each channel.
26127 Set amplitude scale.
26129 Available values are:
26147 Set the draw mode. This is mostly useful to set for high @var{n}.
26149 Available values are:
26152 Scale pixel values for each drawn sample.
26155 Draw every sample directly.
26158 Default value is @code{scale}.
26161 @subsection Examples
26165 Output the input file audio and the corresponding video representation
26168 amovie=a.mp3,asplit[out0],showwaves[out1]
26172 Create a synthetic signal and show it with showwaves, forcing a
26173 frame rate of 30 frames per second:
26175 aevalsrc=sin(1*2*PI*t)*sin(880*2*PI*t):cos(2*PI*200*t),asplit[out0],showwaves=r=30[out1]
26179 @section showwavespic
26181 Convert input audio to a single video frame, representing the samples waves.
26183 The filter accepts the following options:
26187 Specify the video size for the output. For the syntax of this option, check the
26188 @ref{video size syntax,,"Video size" section in the ffmpeg-utils manual,ffmpeg-utils}.
26189 Default value is @code{600x240}.
26191 @item split_channels
26192 Set if channels should be drawn separately or overlap. Default value is 0.
26195 Set colors separated by '|' which are going to be used for drawing of each channel.
26198 Set amplitude scale.
26200 Available values are:
26220 Available values are:
26223 Scale pixel values for each drawn sample.
26226 Draw every sample directly.
26229 Default value is @code{scale}.
26232 Set the filter mode.
26234 Available values are:
26237 Use average samples values for each drawn sample.
26240 Use peak samples values for each drawn sample.
26243 Default value is @code{average}.
26246 @subsection Examples
26250 Extract a channel split representation of the wave form of a whole audio track
26251 in a 1024x800 picture using @command{ffmpeg}:
26253 ffmpeg -i audio.flac -lavfi showwavespic=split_channels=1:s=1024x800 waveform.png
26257 @section sidedata, asidedata
26259 Delete frame side data, or select frames based on it.
26261 This filter accepts the following options:
26265 Set mode of operation of the filter.
26267 Can be one of the following:
26271 Select every frame with side data of @code{type}.
26274 Delete side data of @code{type}. If @code{type} is not set, delete all side
26280 Set side data type used with all modes. Must be set for @code{select} mode. For
26281 the list of frame side data types, refer to the @code{AVFrameSideDataType} enum
26282 in @file{libavutil/frame.h}. For example, to choose
26283 @code{AV_FRAME_DATA_PANSCAN} side data, you must specify @code{PANSCAN}.
26287 @section spectrumsynth
26289 Synthesize audio from 2 input video spectrums, first input stream represents
26290 magnitude across time and second represents phase across time.
26291 The filter will transform from frequency domain as displayed in videos back
26292 to time domain as presented in audio output.
26294 This filter is primarily created for reversing processed @ref{showspectrum}
26295 filter outputs, but can synthesize sound from other spectrograms too.
26296 But in such case results are going to be poor if the phase data is not
26297 available, because in such cases phase data need to be recreated, usually
26298 it's just recreated from random noise.
26299 For best results use gray only output (@code{channel} color mode in
26300 @ref{showspectrum} filter) and @code{log} scale for magnitude video and
26301 @code{lin} scale for phase video. To produce phase, for 2nd video, use
26302 @code{data} option. Inputs videos should generally use @code{fullframe}
26303 slide mode as that saves resources needed for decoding video.
26305 The filter accepts the following options:
26309 Specify sample rate of output audio, the sample rate of audio from which
26310 spectrum was generated may differ.
26313 Set number of channels represented in input video spectrums.
26316 Set scale which was used when generating magnitude input spectrum.
26317 Can be @code{lin} or @code{log}. Default is @code{log}.
26320 Set slide which was used when generating inputs spectrums.
26321 Can be @code{replace}, @code{scroll}, @code{fullframe} or @code{rscroll}.
26322 Default is @code{fullframe}.
26325 Set window function used for resynthesis.
26328 Set window overlap. In range @code{[0, 1]}. Default is @code{1},
26329 which means optimal overlap for selected window function will be picked.
26332 Set orientation of input videos. Can be @code{vertical} or @code{horizontal}.
26333 Default is @code{vertical}.
26336 @subsection Examples
26340 First create magnitude and phase videos from audio, assuming audio is stereo with 44100 sample rate,
26341 then resynthesize videos back to audio with spectrumsynth:
26343 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
26344 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
26345 ffmpeg -i magnitude.nut -i phase.nut -lavfi spectrumsynth=channels=2:sample_rate=44100:win_func=hann:overlap=0.875:slide=fullframe output.flac
26349 @section split, asplit
26351 Split input into several identical outputs.
26353 @code{asplit} works with audio input, @code{split} with video.
26355 The filter accepts a single parameter which specifies the number of outputs. If
26356 unspecified, it defaults to 2.
26358 @subsection Examples
26362 Create two separate outputs from the same input:
26364 [in] split [out0][out1]
26368 To create 3 or more outputs, you need to specify the number of
26371 [in] asplit=3 [out0][out1][out2]
26375 Create two separate outputs from the same input, one cropped and
26378 [in] split [splitout1][splitout2];
26379 [splitout1] crop=100:100:0:0 [cropout];
26380 [splitout2] pad=200:200:100:100 [padout];
26384 Create 5 copies of the input audio with @command{ffmpeg}:
26386 ffmpeg -i INPUT -filter_complex asplit=5 OUTPUT
26392 Receive commands sent through a libzmq client, and forward them to
26393 filters in the filtergraph.
26395 @code{zmq} and @code{azmq} work as a pass-through filters. @code{zmq}
26396 must be inserted between two video filters, @code{azmq} between two
26397 audio filters. Both are capable to send messages to any filter type.
26399 To enable these filters you need to install the libzmq library and
26400 headers and configure FFmpeg with @code{--enable-libzmq}.
26402 For more information about libzmq see:
26403 @url{http://www.zeromq.org/}
26405 The @code{zmq} and @code{azmq} filters work as a libzmq server, which
26406 receives messages sent through a network interface defined by the
26407 @option{bind_address} (or the abbreviation "@option{b}") option.
26408 Default value of this option is @file{tcp://localhost:5555}. You may
26409 want to alter this value to your needs, but do not forget to escape any
26410 ':' signs (see @ref{filtergraph escaping}).
26412 The received message must be in the form:
26414 @var{TARGET} @var{COMMAND} [@var{ARG}]
26417 @var{TARGET} specifies the target of the command, usually the name of
26418 the filter class or a specific filter instance name. The default
26419 filter instance name uses the pattern @samp{Parsed_<filter_name>_<index>},
26420 but you can override this by using the @samp{filter_name@@id} syntax
26421 (see @ref{Filtergraph syntax}).
26423 @var{COMMAND} specifies the name of the command for the target filter.
26425 @var{ARG} is optional and specifies the optional argument list for the
26426 given @var{COMMAND}.
26428 Upon reception, the message is processed and the corresponding command
26429 is injected into the filtergraph. Depending on the result, the filter
26430 will send a reply to the client, adopting the format:
26432 @var{ERROR_CODE} @var{ERROR_REASON}
26436 @var{MESSAGE} is optional.
26438 @subsection Examples
26440 Look at @file{tools/zmqsend} for an example of a zmq client which can
26441 be used to send commands processed by these filters.
26443 Consider the following filtergraph generated by @command{ffplay}.
26444 In this example the last overlay filter has an instance name. All other
26445 filters will have default instance names.
26448 ffplay -dumpgraph 1 -f lavfi "
26449 color=s=100x100:c=red [l];
26450 color=s=100x100:c=blue [r];
26451 nullsrc=s=200x100, zmq [bg];
26452 [bg][l] overlay [bg+l];
26453 [bg+l][r] overlay@@my=x=100 "
26456 To change the color of the left side of the video, the following
26457 command can be used:
26459 echo Parsed_color_0 c yellow | tools/zmqsend
26462 To change the right side:
26464 echo Parsed_color_1 c pink | tools/zmqsend
26467 To change the position of the right side:
26469 echo overlay@@my x 150 | tools/zmqsend
26473 @c man end MULTIMEDIA FILTERS
26475 @chapter Multimedia Sources
26476 @c man begin MULTIMEDIA SOURCES
26478 Below is a description of the currently available multimedia sources.
26482 This is the same as @ref{movie} source, except it selects an audio
26488 Read audio and/or video stream(s) from a movie container.
26490 It accepts the following parameters:
26494 The name of the resource to read (not necessarily a file; it can also be a
26495 device or a stream accessed through some protocol).
26497 @item format_name, f
26498 Specifies the format assumed for the movie to read, and can be either
26499 the name of a container or an input device. If not specified, the
26500 format is guessed from @var{movie_name} or by probing.
26502 @item seek_point, sp
26503 Specifies the seek point in seconds. The frames will be output
26504 starting from this seek point. The parameter is evaluated with
26505 @code{av_strtod}, so the numerical value may be suffixed by an IS
26506 postfix. The default value is "0".
26509 Specifies the streams to read. Several streams can be specified,
26510 separated by "+". The source will then have as many outputs, in the
26511 same order. The syntax is explained in the @ref{Stream specifiers,,"Stream specifiers"
26512 section in the ffmpeg manual,ffmpeg}. Two special names, "dv" and "da" specify
26513 respectively the default (best suited) video and audio stream. Default
26514 is "dv", or "da" if the filter is called as "amovie".
26516 @item stream_index, si
26517 Specifies the index of the video stream to read. If the value is -1,
26518 the most suitable video stream will be automatically selected. The default
26519 value is "-1". Deprecated. If the filter is called "amovie", it will select
26520 audio instead of video.
26523 Specifies how many times to read the stream in sequence.
26524 If the value is 0, the stream will be looped infinitely.
26525 Default value is "1".
26527 Note that when the movie is looped the source timestamps are not
26528 changed, so it will generate non monotonically increasing timestamps.
26530 @item discontinuity
26531 Specifies the time difference between frames above which the point is
26532 considered a timestamp discontinuity which is removed by adjusting the later
26536 It allows overlaying a second video on top of the main input of
26537 a filtergraph, as shown in this graph:
26539 input -----------> deltapts0 --> overlay --> output
26542 movie --> scale--> deltapts1 -------+
26544 @subsection Examples
26548 Skip 3.2 seconds from the start of the AVI file in.avi, and overlay it
26549 on top of the input labelled "in":
26551 movie=in.avi:seek_point=3.2, scale=180:-1, setpts=PTS-STARTPTS [over];
26552 [in] setpts=PTS-STARTPTS [main];
26553 [main][over] overlay=16:16 [out]
26557 Read from a video4linux2 device, and overlay it on top of the input
26560 movie=/dev/video0:f=video4linux2, scale=180:-1, setpts=PTS-STARTPTS [over];
26561 [in] setpts=PTS-STARTPTS [main];
26562 [main][over] overlay=16:16 [out]
26566 Read the first video stream and the audio stream with id 0x81 from
26567 dvd.vob; the video is connected to the pad named "video" and the audio is
26568 connected to the pad named "audio":
26570 movie=dvd.vob:s=v:0+#0x81 [video] [audio]
26574 @subsection Commands
26576 Both movie and amovie support the following commands:
26579 Perform seek using "av_seek_frame".
26580 The syntax is: seek @var{stream_index}|@var{timestamp}|@var{flags}
26583 @var{stream_index}: If stream_index is -1, a default
26584 stream is selected, and @var{timestamp} is automatically converted
26585 from AV_TIME_BASE units to the stream specific time_base.
26587 @var{timestamp}: Timestamp in AVStream.time_base units
26588 or, if no stream is specified, in AV_TIME_BASE units.
26590 @var{flags}: Flags which select direction and seeking mode.
26594 Get movie duration in AV_TIME_BASE units.
26598 @c man end MULTIMEDIA SOURCES