Fluent ffmpeg-API for node.js

Fluent-ffmpeg is looking for new maintainers More details on the wiki

This library abstracts the complex command-line usage of ffmpeg into a fluent, easy to use node.js module. In order to be able to use this module, make sure you have ffmpeg installed on your system (including all necessary encoding libraries like libmp3lame or libx264).

This is the documentation for fluent-ffmpeg 2.x. You can still access the code and documentation for fluent-ffmpeg 1.7 here.

Installation

Via npm:

1$ npm install fluent-ffmpeg

Or as a submodule:

1$ git submodule add git://github.com/schaermu/node-fluent-ffmpeg.git vendor/fluent-ffmpeg

Usage

You will find a lot of usage examples (including a real-time streaming example using flowplayer and express!) in the examples folder.

Prerequisites

ffmpeg and ffprobe

fluent-ffmpeg requires ffmpeg >= 0.9 to work. It may work with previous versions but several features won't be available (and the library is not tested with lower versions anylonger).

If the FFMPEG_PATH environment variable is set, fluent-ffmpeg will use it as the full path to the ffmpeg executable. Otherwise, it will attempt to call ffmpeg directly (so it should be in your PATH). You must also have ffprobe installed (it comes with ffmpeg in most distributions). Similarly, fluent-ffmpeg will use the FFPROBE_PATH environment variable if it is set, otherwise it will attempt to call it in the PATH.

Most features should work when using avconv and avprobe instead of ffmpeg and ffprobe, but they are not officially supported at the moment.

Windows users: most probably ffmpeg and ffprobe will not be in your %PATH, so you must set %FFMPEG_PATH and %FFPROBE_PATH.

Debian/Ubuntu users: the official repositories have the ffmpeg/ffprobe executable in the libav-tools package, and they are actually rebranded avconv/avprobe executables (avconv is a fork of ffmpeg). They should be mostly compatible, but should you encounter any issue, you may want to use the real ffmpeg instead. You can either compile it from source or find a pre-built .deb package at https://ffmpeg.org/download.html (For Ubuntu, the ppa:mc3man/trusty-media PPA provides recent builds).

flvtool2 or flvmeta

If you intend to encode FLV videos, you must have either flvtool2 or flvmeta installed and in your PATH or fluent-ffmpeg won't be able to produce streamable output files. If you set either the FLVTOOL2_PATH or FLVMETA_PATH, fluent-ffmpeg will try to use it instead of searching in the PATH.

Setting binary paths manually

Alternatively, you may set the ffmpeg, ffprobe and flvtool2/flvmeta binary paths manually by using the following API commands:

• Ffmpeg.setFfmpegPath(path) Argument path is a string with the full path to the ffmpeg binary.
• Ffmpeg.setFfprobePath(path) Argument path is a string with the full path to the ffprobe binary.
• Ffmpeg.setFlvtoolPath(path) Argument path is a string with the full path to the flvtool2 or flvmeta binary.

Creating an FFmpeg command

The fluent-ffmpeg module returns a constructor that you can use to instanciate FFmpeg commands.

1var FfmpegCommand = require('fluent-ffmpeg');
2var command = new FfmpegCommand();

You can also use the constructor without the new operator.

1var ffmpeg = require('fluent-ffmpeg');
2var command = ffmpeg();

You may pass an input file name or readable stream, a configuration object, or both to the constructor.

1var command = ffmpeg('/path/to/file.avi');
2var command = ffmpeg(fs.createReadStream('/path/to/file.avi'));
3var command = ffmpeg({ option: "value", ... });
4var command = ffmpeg('/path/to/file.avi', { option: "value", ... });

The following options are available:

• source: input file name or readable stream (ignored if an input file is passed to the constructor)
• timeout: ffmpeg timeout in seconds (defaults to no timeout)
• preset or presets: directory to load module presets from (defaults to the lib/presets directory in fluent-ffmpeg tree)
• niceness or priority: ffmpeg niceness value, between -20 and 20; ignored on Windows platforms (defaults to 0)
• logger: logger object with debug(), info(), warn() and error() methods (defaults to no logging)
• stdoutLines: maximum number of lines from ffmpeg stdout/stderr to keep in memory (defaults to 100, use 0 for unlimited storage)

Specifying inputs

You can add any number of inputs to an Ffmpeg command. An input can be:

• a file name (eg. /path/to/file.avi);
• an image pattern (eg. /path/to/frame%03d.png);
• a readable stream; only one input stream may be used for a command, but you can use both an input stream and one or several file names.

1// Note that all fluent-ffmpeg methods are chainable
2ffmpeg('/path/to/input1.avi')
3  .input('/path/to/input2.avi')
4  .input(fs.createReadStream('/path/to/input3.avi'));
5
6// Passing an input to the constructor is the same as calling .input()
7ffmpeg()
8  .input('/path/to/input1.avi')
9  .input('/path/to/input2.avi');
10
11// Most methods have several aliases, here you may use addInput or mergeAdd instead
12ffmpeg()
13  .addInput('/path/to/frame%02d.png')
14  .addInput('/path/to/soundtrack.mp3');
15
16ffmpeg()
17  .mergeAdd('/path/to/input1.avi')
18  .mergeAdd('/path/to/input2.avi');

Input options

The following methods enable passing input-related options to ffmpeg. Each of these methods apply on the last input added (including the one passed to the constructor, if any). You must add an input before calling those, or an error will be thrown.

inputFormat(format): specify input format

Aliases: fromFormat(), withInputFormat().

This is only useful for raw inputs, as ffmpeg can determine the input format automatically.

1ffmpeg()
2  .input('/dev/video0')
3  .inputFormat('mov')
4  .input('/path/to/file.avi')
5  .inputFormat('avi');

Fluent-ffmpeg checks for format availability before actually running the command, and throws an error when a specified input format is not available.

inputFPS(fps): specify input framerate

Aliases: withInputFps(), withInputFPS(), withFpsInput(), withFPSInput(), inputFps(), fpsInput(), FPSInput().

This is only valid for raw inputs, as ffmpeg can determine the input framerate automatically.

1ffmpeg('/dev/video0').inputFPS(29.7);

native(): read input at native framerate

Aliases: nativeFramerate(), withNativeFramerate().

1ffmpeg('/path/to/file.avi').native();

seekInput(time): set input start time

Alias: setStartTime().

Seeks an input and only start decoding at given time offset. The time argument may be a number (in seconds) or a timestamp string (with format [[hh:]mm:]ss[.xxx]).

1ffmpeg('/path/to/file.avi').seekInput(134.5);
2ffmpeg('/path/to/file.avi').seekInput('2:14.500');

loop([duration]): loop over input

1ffmpeg('/path/to/file.avi').loop();
2ffmpeg('/path/to/file.avi').loop(134.5);
3ffmpeg('/path/to/file.avi').loop('2:14.500');

inputOptions(option...): add custom input options

Aliases: inputOption(), addInputOption(), addInputOptions(), withInputOption(), withInputOptions().

This method allows passing any input-related option to ffmpeg. You can call it with a single argument to pass a single option, optionally with a space-separated parameter:

1/* Single option */
2ffmpeg('/path/to/file.avi').inputOptions('-someOption');
3
4/* Single option with parameter */
5ffmpeg('/dev/video0').inputOptions('-r 24');

You may also pass multiple options at once by passing an array to the method:

1ffmpeg('/path/to/file.avi').inputOptions([
2  '-option1',
3  '-option2 param2',
4  '-option3',
5  '-option4 param4'
6]);

Finally, you may also directly pass command line tokens as separate arguments to the method:

1ffmpeg('/path/to/file.avi').inputOptions(
2  '-option1',
3  '-option2', 'param2',
4  '-option3',
5  '-option4', 'param4'
6);

Audio options

The following methods change the audio stream(s) in the produced output.

noAudio(): disable audio altogether

Aliases: withNoAudio().

Disables audio in the output and remove any previously set audio option.

1ffmpeg('/path/to/file.avi').noAudio();

audioCodec(codec): set audio codec

Aliases: withAudioCodec().

1ffmpeg('/path/to/file.avi').audioCodec('libmp3lame');

Fluent-ffmpeg checks for codec availability before actually running the command, and throws an error when a specified audio codec is not available.

audioBitrate(bitrate): set audio bitrate

Aliases: withAudioBitrate().

Sets the audio bitrate in kbps. The bitrate parameter may be a number or a string with an optional k suffix. This method is used to enforce a constant bitrate; use audioQuality() to encode using a variable bitrate.

1ffmpeg('/path/to/file.avi').audioBitrate(128);
2ffmpeg('/path/to/file.avi').audioBitrate('128');
3ffmpeg('/path/to/file.avi').audioBitrate('128k');

audioChannels(count): set audio channel count

Aliases: withAudioChannels().

1ffmpeg('/path/to/file.avi').audioChannels(2);

audioFrequency(freq): set audio frequency

Aliases: withAudioFrequency().

The freq parameter specifies the audio frequency in Hz.

1ffmpeg('/path/to/file.avi').audioFrequency(22050);

audioQuality(quality): set audio quality

Aliases: withAudioQuality().

This method fixes a quality factor for the audio codec (VBR encoding). The quality scale depends on the actual codec used.

1ffmpeg('/path/to/file.avi')
2  .audioCodec('libmp3lame')
3  .audioQuality(0);

audioFilters(filter...): add custom audio filters

Aliases: audioFilter(), withAudioFilter(), withAudioFilters().

This method enables adding custom audio filters. You may add multiple filters at once by passing either several arguments or an array. See the Ffmpeg documentation for available filters and their syntax.

Each filter pased to this method can be either a filter string (eg. volume=0.5) or a filter specification object with the following keys:

• filter: filter name
• options: optional; either an option string for the filter (eg. n=-50dB:d=5), an options array for unnamed options (eg. ['-50dB', 5]) or an object mapping option names to values (eg. { n: '-50dB', d: 5 }). When options is not specified, the filter will be added without any options.

1ffmpeg('/path/to/file.avi')
2  .audioFilters('volume=0.5')
3  .audioFilters('silencedetect=n=-50dB:d=5');
4
5ffmpeg('/path/to/file.avi')
6  .audioFilters('volume=0.5', 'silencedetect=n=-50dB:d=5');
7
8ffmpeg('/path/to/file.avi')
9  .audioFilters(['volume=0.5', 'silencedetect=n=-50dB:d=5']);
10
11ffmpeg('/path/to/file.avi')
12  .audioFilters([
13    {
14      filter: 'volume',
15      options: '0.5'
16    },
17    {
18      filter: 'silencedetect',
19      options: 'n=-50dB:d=5'
20    }
21  ]);
22
23ffmpeg('/path/to/file.avi')
24  .audioFilters(
25    {
26      filter: 'volume',
27      options: ['0.5']
28    },
29    {
30      filter: 'silencedetect',
31      options: { n: '-50dB', d: 5 }
32    }
33  ]);

Video options

The following methods change the video stream(s) in the produced output.

noVideo(): disable video altogether

Aliases: withNoVideo().

This method disables video output and removes any previously set video option.

1ffmpeg('/path/to/file.avi').noVideo();

videoCodec(codec): set video codec

Aliases: withVideoCodec().

1ffmpeg('/path/to/file.avi').videoCodec('libx264');

Fluent-ffmpeg checks for codec availability before actually running the command, and throws an error when a specified video codec is not available.

videoBitrate(bitrate[, constant=false]): set video bitrate

Aliases: withVideoBitrate().

Sets the target video bitrate in kbps. The bitrate argument may be a number or a string with an optional k suffix. The constant argument specifies whether a constant bitrate should be enforced (defaults to false).

Keep in mind that, depending on the codec used, enforcing a constant bitrate often comes at the cost of quality. The best way to have a constant video bitrate without losing too much quality is to use 2-pass encoding (see Fffmpeg documentation).

1ffmpeg('/path/to/file.avi').videoBitrate(1000);
2ffmpeg('/path/to/file.avi').videoBitrate('1000');
3ffmpeg('/path/to/file.avi').videoBitrate('1000k');
4ffmpeg('/path/to/file.avi').videoBitrate('1000k', true);

videoFilters(filter...): add custom video filters

Aliases: videoFilter(), withVideoFilter(), withVideoFilters().

This method enables adding custom video filters. You may add multiple filters at once by passing either several arguments or an array. See the Ffmpeg documentation for available filters and their syntax.

Each filter pased to this method can be either a filter string (eg. fade=in:0:30) or a filter specification object with the following keys:

• filter: filter name
• options: optional; either an option string for the filter (eg. in:0:30), an options array for unnamed options (eg. ['in', 0, 30]) or an object mapping option names to values (eg. { t: 'in', s: 0, n: 30 }). When options is not specified, the filter will be added without any options.

1ffmpeg('/path/to/file.avi')
2  .videoFilters('fade=in:0:30')
3  .videoFilters('pad=640:480:0:40:violet');
4
5ffmpeg('/path/to/file.avi')
6  .videoFilters('fade=in:0:30', 'pad=640:480:0:40:violet');
7
8ffmpeg('/path/to/file.avi')
9  .videoFilters(['fade=in:0:30', 'pad=640:480:0:40:violet']);
10
11ffmpeg('/path/to/file.avi')
12  .videoFilters([
13    {
14      filter: 'fade',
15      options: 'in:0:30'
16    },
17    {
18      filter: 'pad',
19      options: '640:480:0:40:violet'
20    }
21  ]);
22
23ffmpeg('/path/to/file.avi')
24    .videoFilters(
25    {
26      filter: 'fade',
27      options: ['in', 0, 30]
28    },
29    {
30      filter: 'filter2',
31      options: { w: 640, h: 480, x: 0, y: 40, color: 'violet' }
32    }
33  );

fps(fps): set output framerate

Aliases: withOutputFps(), withOutputFPS(), withFpsOutput(), withFPSOutput(), withFps(), withFPS(), outputFPS(), outputFps(), fpsOutput(), FPSOutput(), FPS().

1ffmpeg('/path/to/file.avi').fps(29.7);

frames(count): specify frame count

Aliases: takeFrames(), withFrames().

Set ffmpeg to only encode a certain number of frames.

1ffmpeg('/path/to/file.avi').frames(240);

Video frame size options

The following methods enable resizing the output video frame size. They all work together to generate the appropriate video filters.

size(size): set output frame size

Aliases: videoSize(), withSize().

This method sets the output frame size. The size argument may have one of the following formats:

• 640x480: set a fixed output frame size. Unless autopad() is called, this may result in the video being stretched or squeezed to fit the requested size.
• 640x?: set a fixed width and compute height automatically. If aspect() is also called, it is used to compute video height; otherwise it is computed so that the input aspect ratio is preserved.
• ?x480: set a fixed height and compute width automatically. If aspect() is also called, it is used to compute video width; otherwise it is computed so that the input aspect ratio is preserved.
• 50%: rescale both width and height to the given percentage. Aspect ratio is always preserved.

Note that for compatibility with some codecs, computed dimensions are always rounded down to multiples of 2.

1ffmpeg('/path/to/file.avi').size('640x480');
2ffmpeg('/path/to/file.avi').size('640x?');
3ffmpeg('/path/to/file.avi').size('640x?').aspect('4:3');
4ffmpeg('/path/to/file.avi').size('50%');

aspect(aspect): set output frame aspect ratio

Aliases: withAspect(), withAspectRatio(), setAspect(), setAspectRatio(), aspectRatio().

This method enforces a specific output aspect ratio. The aspect argument may either be a number or a X:Y string.

Note that calls to aspect() are ignored when size() has been called with a fixed width and height or a percentage, and also when size() has not been called at all.

1ffmpeg('/path/to/file.avi').size('640x?').aspect('4:3');
2ffmpeg('/path/to/file.avi').size('640x?').aspect(1.33333);

autopad([color='black']): enable auto-padding the output video

Aliases: applyAutopadding(), applyAutoPadding(), applyAutopad(), applyAutoPad(), withAutopadding(), withAutoPadding(), withAutopad(), withAutoPad(), autoPad().

This method enables applying auto-padding to the output video. The color parameter specifies which color to use for padding, and must be a color code or name supported by ffmpeg (defaults to 'black').

The behaviour of this method depends on calls made to other video size methods:

• when size() has been called with a percentage or has not been called, it is ignored;
• when size() has been called with WxH, it adds padding so that the input aspect ratio is kept;
• when size() has been called with either Wx? or ?xH, padding is only added if aspect() was called (otherwise the output dimensions are computed from the input aspect ratio and padding is not needed).

1// No size specified, autopad() is ignored
2ffmpeg('/path/to/file.avi').autopad();
3
4// Adds padding to keep original aspect ratio.
5// - with a 640x400 input, 40 pixels of padding are added on both sides
6// - with a 600x480 input, 20 pixels of padding are added on top and bottom
7// - with a 320x200 input, video is scaled up to 640x400 and 40px of padding
8//   is added on both sides
9// - with a 320x240 input, video is scaled up to 640x480 and and no padding
10//   is needed
11ffmpeg('/path/to/file.avi').size('640x480').autopad();
12ffmpeg('/path/to/file.avi').size('640x480').autopad('white');
13ffmpeg('/path/to/file.avi').size('640x480').autopad('#35A5FF');
14
15// Size computed from input, autopad() is ignored
16ffmpeg('/path/to/file.avi').size('50%').autopad();
17ffmpeg('/path/to/file.avi').size('640x?').autopad();
18ffmpeg('/path/to/file.avi').size('?x480').autopad();
19
20// Calling .size('640x?').aspect('4:3') is similar to calling .size('640x480')
21// - with a 640x400 input, 40 pixels of padding are added on both sides
22// - with a 600x480 input, 20 pixels of padding are added on top and bottom
23// - with a 320x200 input, video is scaled up to 640x400 and 40px of padding
24//   is added on both sides
25// - with a 320x240 input, video is scaled up to 640x480 and and no padding
26//   is needed
27ffmpeg('/path/to/file.avi').size('640x?').aspect('4:3').autopad();
28ffmpeg('/path/to/file.avi').size('640x?').aspect('4:3').autopad('white');
29ffmpeg('/path/to/file.avi').size('640x?').aspect('4:3').autopad('#35A5FF');
30
31// Calling .size('?x480').aspect('4:3') is similar to calling .size('640x480')
32ffmpeg('/path/to/file.avi').size('?x480').aspect('4:3').autopad();
33ffmpeg('/path/to/file.avi').size('?x480').aspect('4:3').autopad('white');
34ffmpeg('/path/to/file.avi').size('?x480').aspect('4:3').autopad('#35A5FF');

For compatibility with previous fluent-ffmpeg versions, this method also accepts an additional boolean first argument, which specifies whether to apply auto-padding.

1ffmpeg('/path/to/file.avi').size('640x480').autopad(true);
2ffmpeg('/path/to/file.avi').size('640x480').autopad(true, 'pink');

keepDAR(): force keeping display aspect ratio

Aliases: keepPixelAspect(), keepDisplayAspect(), keepDisplayAspectRatio().

This method is useful when converting an input with non-square pixels to an output format that does not support non-square pixels (eg. most image formats). It rescales the input so that the display aspect ratio is the same.

1ffmpeg('/path/to/file.avi').keepDAR();

Specifying multiple outputs

output(target[, options]): add an output to the command

Aliases: addOutput().

Adds an output to the command. The target argument may be an output filename or a writable stream (but at most one output stream may be used with a single command).

When target is a stream, an additional options object may be passed. If it is present, it will be passed ffmpeg output stream pipe() method.

Adding an output switches the "current output" of the command, so that any fluent-ffmpeg method that applies to an output is indeed applied to the last output added. For backwards compatibility reasons, you may as well call those methods before adding the first output (in which case they will apply to the first output when it is added). Methods that apply to an output are all non-input-related methods, except for complexFilter(), which is global.

Also note that when calling output(), you should not use the save() or stream() (formerly saveToFile() and writeToStream()) methods, as they already add an output. Use the run() method to start processing.

1var stream  = fs.createWriteStream('outputfile.divx');
2
3ffmpeg('/path/to/file.avi')
4  .output('outputfile.mp4')
5  .output(stream);
6
7ffmpeg('/path/to/file.avi')
8  // You may pass a pipe() options object when using a stream
9  .output(stream, { end:true });
10
11// Output-related methods apply to the last output added
12ffmpeg('/path/to/file.avi')
13
14  .output('outputfile.mp4')
15  .audioCodec('libfaac')
16  .videoCodec('libx264')
17  .size('320x200')
18
19  .output(stream)
20  .preset('divx')
21  .size('640x480');
22
23// Use the run() method to run commands with multiple outputs
24ffmpeg('/path/to/file.avi')
25  .output('outputfile.mp4')
26  .output(stream)
27  .on('end', function() {
28    console.log('Finished processing');
29  })
30  .run();

Output options

duration(time): set output duration

Aliases: withDuration(), setDuration().

Forces ffmpeg to stop transcoding after a specific output duration. The time parameter may be a number (in seconds) or a timestamp string (with format [[hh:]mm:]ss[.xxx]).

1ffmpeg('/path/to/file.avi').duration(134.5);
2ffmpeg('/path/to/file.avi').duration('2:14.500');

seek(time): seek output

Aliases: seekOutput().

Seeks streams before encoding them into the output. This is different from calling seekInput() in that the offset will only apply to one output. This is also slower, as skipped frames will still be decoded (but dropped).

The time argument may be a number (in seconds) or a timestamp string (with format [[hh:]mm:]ss[.xxx]).

1ffmpeg('/path/to/file.avi')
2  .seekInput('1:00')
3
4  .output('from-1m30s.avi')
5  .seek(30)
6
7  .output('from-1m40s.avi')
8  .seek('0:40');

format(format): set output format

Aliases: withOutputFormat(), toFormat(), outputFormat().

1ffmpeg('/path/to/file.avi').format('flv');

flvmeta(): update FLV metadata after transcoding

Aliases: updateFlvMetadata().

Calling this method makes fluent-ffmpeg run flvmeta or flvtool2 on the output file to add FLV metadata and make files streamable. It does not work when outputting to a stream, and is only useful when outputting to FLV format.

1ffmpeg('/path/to/file.avi').flvmeta().format('flv');

outputOptions(option...): add custom output options

Aliases: outputOption(), addOutputOption(), addOutputOptions(), withOutputOption(), withOutputOptions(), addOption(), addOptions().

This method allows passing any output-related option to ffmpeg. You can call it with a single argument to pass a single option, optionally with a space-separated parameter:

1/* Single option */
2ffmpeg('/path/to/file.avi').outputOptions('-someOption');
3
4/* Single option with parameter */
5ffmpeg('/dev/video0').outputOptions('-r 24');

You may also pass multiple options at once by passing an array to the method:

1ffmpeg('/path/to/file.avi').outputOptions([
2  '-option1',
3  '-option2 param2',
4  '-option3',
5  '-option4 param4'
6]);

Finally, you may also directly pass command line tokens as separate arguments to the method:

1ffmpeg('/path/to/file.avi').outputOptions(
2  '-option1',
3  '-option2', 'param2',
4  '-option3',
5  '-option4', 'param4'
6);

Miscellaneous options

preset(preset): use fluent-ffmpeg preset

Aliases: usingPreset().

There are two kinds of presets supported by fluent-ffmpeg. The first one is preset modules; to use those, pass the preset name as the preset argument. Preset modules are loaded from the directory specified by the presets constructor option (defaults to the lib/presets fluent-ffmpeg subdirectory).

1// Uses <path-to-fluent-ffmpeg>/lib/presets/divx.js
2ffmpeg('/path/to/file.avi').preset('divx');
3
4// Uses /my/presets/foo.js
5ffmpeg('/path/to/file.avi', { presets: '/my/presets' }).preset('foo');

Preset modules must export a load() function that takes an FfmpegCommand as an argument. fluent-ffmpeg comes with the following preset modules preinstalled:

• divx
• flashvideo
• podcast

Here is the code from the included divx preset as an example:

1exports.load = function(ffmpeg) {
2  ffmpeg
3    .format('avi')
4    .videoBitrate('1024k')
5    .videoCodec('mpeg4')
6    .size('720x?')
7    .audioBitrate('128k')
8    .audioChannels(2)
9    .audioCodec('libmp3lame')
10    .outputOptions(['-vtag DIVX']);
11};

The second kind of preset is preset functions. To use those, pass a function which takes an FfmpegCommand as a parameter.

1function myPreset(command) {
2  command.format('avi').size('720x?');
3}
4
5ffmpeg('/path/to/file.avi').preset(myPreset);

complexFilter(filters[, map]): set complex filtergraph

Aliases: filterGraph()

The complexFilter() method enables setting a complex filtergraph for a command. It expects a filter specification (or a filter specification array) and an optional output mapping parameter as arguments.

Filter specifications may be either plain ffmpeg filter strings (eg. split=3[a][b][c]) or objects with the following keys:

• filter: filter name
• options: optional; either an option string for the filter (eg. in:0:30), an options array for unnamed options (eg. ['in', 0, 30]) or an object mapping option names to values (eg. { t: 'in', s: 0, n: 30 }). When options is not specified, the filter will be added without any options.
• inputs: optional; input stream specifier(s) for the filter. The value may be either a single stream specifier string or an array of stream specifiers. Each specifier can be optionally enclosed in square brackets. When input streams are not specified, ffmpeg will use the first unused streams of the correct type.
• outputs: optional; output stream specifier(s) for the filter. The value may be either a single stream specifier string or an array of stream specifiers. Each specifier can be optionally enclosed in square brackets.

The output mapping parameter specifies which stream(s) to include in the output from the filtergraph. It may be either a single stream specifier string or an array of stream specifiers. Each specifier can be optionally enclosed in square brackets. When this parameter is not present, ffmpeg will default to saving all unused outputs to the output file.

Note that only one complex filtergraph may be set on a given command. Calling complexFilter() again will override any previously set filtergraph, but you can set as many filters as needed in a single call.

1ffmpeg('/path/to/file.avi')
2  .complexFilter([
3    // Rescale input stream into stream 'rescaled'
4    'scale=640:480[rescaled]',
5
6    // Duplicate rescaled stream 3 times into streams a, b, and c
7    {
8      filter: 'split', options: '3',
9      inputs: 'rescaled', outputs: ['a', 'b', 'c']
10    },
11
12    // Create stream 'red' by removing green and blue channels from stream 'a'
13    {
14      filter: 'lutrgb', options: { g: 0, b: 0 },
15      inputs: 'a', outputs: 'red'
16    },
17
18    // Create stream 'green' by removing red and blue channels from stream 'b'
19    {
20      filter: 'lutrgb', options: { r: 0, b: 0 },
21      inputs: 'b', outputs: 'green'
22    },
23
24    // Create stream 'blue' by removing red and green channels from stream 'c'
25    {
26      filter: 'lutrgb', options: { r: 0, g: 0 },
27      inputs: 'c', outputs: 'blue'
28    },
29
30    // Pad stream 'red' to 3x width, keeping the video on the left,
31    // and name output 'padded'
32    {
33      filter: 'pad', options: { w: 'iw*3', h: 'ih' },
34      inputs: 'red', outputs: 'padded'
35    },
36
37    // Overlay 'green' onto 'padded', moving it to the center,
38    // and name output 'redgreen'
39    {
40      filter: 'overlay', options: { x: 'w', y: 0 },
41      inputs: ['padded', 'green'], outputs: 'redgreen'
42    },
43
44    // Overlay 'blue' onto 'redgreen', moving it to the right
45    {
46      filter: 'overlay', options: { x: '2*w', y: 0 },
47      inputs: ['redgreen', 'blue'], outputs: 'output'
48    },
49  ], 'output');

Setting event handlers

Before actually running a command, you may want to set event listeners on it to be notified when it's done. The following events are available:

'start': ffmpeg process started

The start event is emitted just after ffmpeg has been spawned. It is emitted with the full command line used as an argument.

1ffmpeg('/path/to/file.avi')
2  .on('start', function(commandLine) {
3    console.log('Spawned Ffmpeg with command: ' + commandLine);
4  });

'codecData': input codec data available

The codecData event is emitted when ffmpeg outputs codec information about its input streams. It is emitted with an object argument with the following keys:

• format: input format
• duration: input duration
• audio: audio codec
• audio_details: audio encoding details
• video: video codec
• video_details: video encoding details

1ffmpeg('/path/to/file.avi')
2  .on('codecData', function(data) {
3    console.log('Input is ' + data.audio + ' audio ' +
4      'with ' + data.video + ' video');
5  });

'progress': transcoding progress information

The progress event is emitted every time ffmpeg reports progress information. It is emitted with an object argument with the following keys:

• frames: total processed frame count
• currentFps: framerate at which FFmpeg is currently processing
• currentKbps: throughput at which FFmpeg is currently processing
• targetSize: current size of the target file in kilobytes
• timemark: the timestamp of the current frame in seconds
• percent: an estimation of the progress percentage

Note that percent can be (very) inaccurate, as the only progress information fluent-ffmpeg gets from ffmpeg is the total number of frames written (and the corresponding duration). To estimate percentage, fluent-ffmpeg has to guess what the total output duration will be, and uses the first input added to the command to do so. In particular:

• percentage is not available when using an input stream
• percentage may be wrong when using multiple inputs with different durations and the first one is not the longest

1ffmpeg('/path/to/file.avi')
2  .on('progress', function(progress) {
3    console.log('Processing: ' + progress.percent + '% done');
4  });

'stderr': FFmpeg output

The stderr event is emitted every time FFmpeg outputs a line to stderr. It is emitted with a string containing the line of stderr (minus trailing new line characters).

1ffmpeg('/path/to/file.avi')
2  .on('stderr', function(stderrLine) {
3    console.log('Stderr output: ' + stderrLine);
4  });

'error': transcoding error

The error event is emitted when an error occurs when running ffmpeg or when preparing its execution. It is emitted with an error object as an argument. If the error happened during ffmpeg execution, listeners will also receive two additional arguments containing ffmpegs stdout and stderr.

If streams are used for input or output, any errors emitted from these streams will be passed through to this event, attached to the error as inputStreamError and outputStreamError for input and output streams respectively.

Warning: you should always set a handler for the error event, as node's default behaviour when an error event without any listeners is emitted is to output the error to the console and terminate the program.

1ffmpeg('/path/to/file.avi')
2  .on('error', function(err, stdout, stderr) {
3    console.log('Cannot process video: ' + err.message);
4  });

'end': processing finished

The end event is emitted when processing has finished. Listeners receive ffmpeg standard output and standard error as arguments, except when generating thumbnails (see below), in which case they receive an array of the generated filenames.

1ffmpeg('/path/to/file.avi')
2  .on('end', function(stdout, stderr) {
3    console.log('Transcoding succeeded !');
4  });

stdout is empty when the command outputs to a stream. Both stdout and stderr are limited by the stdoutLines option (defaults to 100 lines).

Starting FFmpeg processing

save(filename): save the output to a file

Aliases: saveToFile()

Starts ffmpeg processing and saves the output to a file.

1ffmpeg('/path/to/file.avi')
2  .videoCodec('libx264')
3  .audioCodec('libmp3lame')
4  .size('320x240')
5  .on('error', function(err) {
6    console.log('An error occurred: ' + err.message);
7  })
8  .on('end', function() {
9    console.log('Processing finished !');
10  })
11  .save('/path/to/output.mp4');

Note: the save() method is actually syntactic sugar for calling both output() and run().

pipe([stream], [options]): pipe the output to a writable stream

Aliases: stream(), writeToStream().

Starts processing and pipes ffmpeg output to a writable stream. The options argument, if present, is passed to ffmpeg output stream's pipe() method (see nodejs documentation).

1var outStream = fs.createWriteStream('/path/to/output.mp4');
2
3ffmpeg('/path/to/file.avi')
4  .videoCodec('libx264')
5  .audioCodec('libmp3lame')
6  .size('320x240')
7  .on('error', function(err) {
8    console.log('An error occurred: ' + err.message);
9  })
10  .on('end', function() {
11    console.log('Processing finished !');
12  })
13  .pipe(outStream, { end: true });

When no stream argument is present, the pipe() method returns a PassThrough stream, which you can pipe to somewhere else (or just listen to events on).

Note: this is only available with node >= 0.10.

1var command = ffmpeg('/path/to/file.avi')
2  .videoCodec('libx264')
3  .audioCodec('libmp3lame')
4  .size('320x240')
5  .on('error', function(err) {
6    console.log('An error occurred: ' + err.message);
7  })
8  .on('end', function() {
9    console.log('Processing finished !');
10  });
11
12var ffstream = command.pipe();
13ffstream.on('data', function(chunk) {
14  console.log('ffmpeg just wrote ' + chunk.length + ' bytes');
15});

Note: the stream() method is actually syntactic sugar for calling both output() and run().

run(): start processing

Aliases: exec(), execute().

This method is mainly useful when producing multiple outputs (otherwise the save() or stream() methods are more straightforward). It starts processing with the specified outputs.

Warning: do not use run() when calling other processing methods (eg. save(), pipe() or screenshots()).

1ffmpeg('/path/to/file.avi')
2  .output('screenshot.png')
3  .noAudio()
4  .seek('3:00')
5
6  .output('small.avi')
7  .audioCodec('copy')
8  .size('320x200')
9
10  .output('big.avi')
11  .audioCodec('copy')
12  .size('640x480')
13
14  .on('error', function(err) {
15    console.log('An error occurred: ' + err.message);
16  })
17  .on('end', function() {
18    console.log('Processing finished !');
19  })
20  .run();

mergeToFile(filename, tmpdir): concatenate multiple inputs

Use the input and mergeToFile methods on a command to concatenate multiple inputs to a single output file. The mergeToFile needs a temporary folder as its second argument.

1ffmpeg('/path/to/part1.avi')
2  .input('/path/to/part2.avi')
3  .input('/path/to/part2.avi')
4  .on('error', function(err) {
5    console.log('An error occurred: ' + err.message);
6  })
7  .on('end', function() {
8    console.log('Merging finished !');
9  })
10  .mergeToFile('/path/to/merged.avi', '/path/to/tempDir');

screenshots(options[, dirname]): generate thumbnails

Aliases: thumbnail(), thumbnails(), screenshot(), takeScreenshots().

Use the screenshots method to extract one or several thumbnails and save them as PNG files. There are a few caveats with this implementation, though:

• It will not work on input streams.
• Progress information reported by the progress event is not accurate.
• It doesn't interract well with filters. In particular, don't use the size() method to resize thumbnails, use the size option instead.

The options argument is an object with the following keys:

• folder: output folder for generated image files. Defaults to the current folder.
• filename: output filename pattern (see below). Defaults to "tn.png".
• count: specifies how many thumbnails to generate. When using this option, thumbnails are generated at regular intervals in the video (for example, when requesting 3 thumbnails, at 25%, 50% and 75% of the video length). count is ignored when timemarks or timestamps is specified.
• timemarks or timestamps: specifies an array of timestamps in the video where thumbnails should be taken. Each timestamp may be a number (in seconds), a percentage string (eg. "50%") or a timestamp string with format "hh:mm:ss.xxx" (where hours, minutes and milliseconds are both optional).
• size: specifies a target size for thumbnails (with the same format as the .size() method). Note: you should not use the .size() method when generating thumbnails.

The filename option specifies a filename pattern for generated files. It may contain the following format tokens:

• '%s': offset in seconds
• '%w': screenshot width
• '%h': screenshot height
• '%r': screenshot resolution (same as '%wx%h')
• '%f': input filename
• '%b': input basename (filename w/o extension)
• '%i': index of screenshot in timemark array (can be zero-padded by using it like %000i)

If multiple timemarks are passed and no variable format token ('%s' or '%i') is specified in the filename pattern, _%i will be added automatically.

When generating thumbnails, an additional filenames event is dispatched with an array of generated filenames as an argument.

1ffmpeg('/path/to/video.avi')
2  .on('filenames', function(filenames) {
3    console.log('Will generate ' + filenames.join(', '))
4  })
5  .on('end', function() {
6    console.log('Screenshots taken');
7  })
8  .screenshots({
9    // Will take screens at 20%, 40%, 60% and 80% of the video
10    count: 4,
11    folder: '/path/to/output'
12  });
13
14ffmpeg('/path/to/video.avi')
15  .screenshots({
16    timestamps: [30.5, '50%', '01:10.123'],
17    filename: 'thumbnail-at-%s-seconds.png',
18    folder: '/path/to/output',
19    size: '320x240'
20  });

Controlling the FFmpeg process

kill([signal='SIGKILL']): kill any running ffmpeg process

This method sends signal (defaults to 'SIGKILL') to the ffmpeg process. It only has sense when processing has started. Sending a signal that terminates the process will result in the error event being emitted.

1var command = ffmpeg('/path/to/video.avi')
2  .videoCodec('libx264')
3  .audioCodec('libmp3lame')
4  .on('start', function() {
5    // Send SIGSTOP to suspend ffmpeg
6    command.kill('SIGSTOP');
7
8    doSomething(function() {
9      // Send SIGCONT to resume ffmpeg
10      command.kill('SIGCONT');
11    });
12  })
13  .save('/path/to/output.mp4');
14
15// Kill ffmpeg after 60 seconds anyway
16setTimeout(function() {
17  command.on('error', function() {
18    console.log('Ffmpeg has been killed');
19  });
20
21  command.kill();
22}, 60000);

renice([niceness=0]): change ffmpeg process priority

This method alters the niceness (priority) value of any running ffmpeg process (if any) and any process spawned in the future. The niceness parameter may range from -20 (highest priority) to 20 (lowest priority) and defaults to 0 (which is the default process niceness on most *nix systems).

Note: this method is ineffective on Windows platforms.

1// Set startup niceness
2var command = ffmpeg('/path/to/file.avi')
3  .renice(5)
4  .save('/path/to/output.mp4');
5
6// Command takes too long, raise its priority
7setTimeout(function() {
8  command.renice(-5);
9}, 60000);

Reading video metadata

You can read metadata from any valid ffmpeg input file with the modules ffprobe method.

1ffmpeg.ffprobe('/path/to/file.avi', function(err, metadata) {
2    console.dir(metadata);
3});

You may also call the ffprobe method on an FfmpegCommand to probe one of its input. You may pass a 0-based input number as a first argument to specify which input to read metadata from, otherwise the method will probe the last added input.

1ffmpeg('/path/to/file1.avi')
2  .input('/path/to/file2.avi')
3  .ffprobe(function(err, data) {
4    console.log('file2 metadata:');
5    console.dir(data);
6  });
7
8ffmpeg('/path/to/file1.avi')
9  .input('/path/to/file2.avi')
10  .ffprobe(0, function(err, data) {
11    console.log('file1 metadata:');
12    console.dir(data);
13  });

Warning: ffprobe may be called with an input stream, but in this case it will consume data from the stream, and this data will no longer be available for ffmpeg. Using both ffprobe and a transcoding command on the same input stream will most likely fail unless the stream is a live stream. Only do this if you know what you're doing.

The returned object is the same that is returned by running the following command from your shell (depending on your ffmpeg version you may have to replace -of with -print_format) :

1$ ffprobe -of json -show_streams -show_format /path/to/file.avi

It will contain information about the container (as a format key) and an array of streams (as a stream key). The format object and each stream object also contains metadata tags, depending on the format:

1{
2  "streams": [
3    {
4      "index": 0,
5      "codec_name": "h264",
6      "codec_long_name": "H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10",
7      "profile": "Constrained Baseline",
8      "codec_type": "video",
9      "codec_time_base": "1/48",
10      "codec_tag_string": "avc1",
11      "codec_tag": "0x31637661",
12      "width": 320,
13      "height": 180,
14      "has_b_frames": 0,
15      "sample_aspect_ratio": "1:1",
16      "display_aspect_ratio": "16:9",
17      "pix_fmt": "yuv420p",
18      "level": 13,
19      "r_frame_rate": "24/1",
20      "avg_frame_rate": "24/1",
21      "time_base": "1/24",
22      "start_pts": 0,
23      "start_time": "0.000000",
24      "duration_ts": 14315,
25      "duration": "596.458333",
26      "bit_rate": "702655",
27      "nb_frames": "14315",
28      "disposition": {
29        "default": 0,
30        "dub": 0,
31        "original": 0,
32        "comment": 0,
33        "lyrics": 0,
34        "karaoke": 0,
35        "forced": 0,
36        "hearing_impaired": 0,
37        "visual_impaired": 0,
38        "clean_effects": 0,
39        "attached_pic": 0
40      },
41      "tags": {
42        "creation_time": "1970-01-01 00:00:00",
43        "language": "und",
44        "handler_name": "\fVideoHandler"
45      }
46    },
47    {
48      "index": 1,
49      "codec_name": "aac",
50      "codec_long_name": "AAC (Advanced Audio Coding)",
51      "codec_type": "audio",
52      "codec_time_base": "1/48000",
53      "codec_tag_string": "mp4a",
54      "codec_tag": "0x6134706d",
55      "sample_fmt": "fltp",
56      "sample_rate": "48000",
57      "channels": 2,
58      "bits_per_sample": 0,
59      "r_frame_rate": "0/0",
60      "avg_frame_rate": "0/0",
61      "time_base": "1/48000",
62      "start_pts": 0,
63      "start_time": "0.000000",
64      "duration_ts": 28619776,
65      "duration": "596.245333",
66      "bit_rate": "159997",
67      "nb_frames": "27949",
68      "disposition": {
69        "default": 0,
70        "dub": 0,
71        "original": 0,
72        "comment": 0,
73        "lyrics": 0,
74        "karaoke": 0,
75        "forced": 0,
76        "hearing_impaired": 0,
77        "visual_impaired": 0,
78        "clean_effects": 0,
79        "attached_pic": 0
80      },
81      "tags": {
82        "creation_time": "1970-01-01 00:00:00",
83        "language": "und",
84        "handler_name": "\fSoundHandler"
85      }
86    }
87  ],
88  "format": {
89    "filename": "http://download.blender.org/peach/bigbuckbunny_movies/BigBuckBunny_320x180.mp4",
90    "nb_streams": 2,
91    "format_name": "mov,mp4,m4a,3gp,3g2,mj2",
92    "format_long_name": "QuickTime / MOV",
93    "start_time": "0.000000",
94    "duration": "596.459000",
95    "size": "64657027",
96    "bit_rate": "867211",
97    "tags": {
98      "major_brand": "isom",
99      "minor_version": "512",
100      "compatible_brands": "mp41",
101      "creation_time": "1970-01-01 00:00:00",
102      "title": "Big Buck Bunny",
103      "artist": "Blender Foundation",
104      "composer": "Blender Foundation",
105      "date": "2008",
106      "encoder": "Lavf52.14.0"
107    }
108  }
109}

Querying ffmpeg capabilities

fluent-ffmpeg enables you to query your installed ffmpeg version for supported formats, codecs, encoders and filters.

1
2var Ffmpeg = require('fluent-ffmpeg');
3
4Ffmpeg.getAvailableFormats(function(err, formats) {
5  console.log('Available formats:');
6  console.dir(formats);
7});
8
9Ffmpeg.getAvailableCodecs(function(err, codecs) {
10  console.log('Available codecs:');
11  console.dir(codecs);
12});
13
14Ffmpeg.getAvailableEncoders(function(err, encoders) {
15  console.log('Available encoders:');
16  console.dir(encoders);
17});
18
19Ffmpeg.getAvailableFilters(function(err, filters) {
20  console.log("Available filters:");
21  console.dir(filters);
22});
23
24// Those methods can also be called on commands
25new Ffmpeg({ source: '/path/to/file.avi' })
26  .getAvailableCodecs(...);

These methods pass an object to their callback with keys for each available format, codec or filter.

The returned object for formats looks like:

1{
2  ...
3  mp4: {
4    description: 'MP4 (MPEG-4 Part 14)',
5    canDemux: false,
6    canMux: true
7  },
8  ...
9}

• canDemux indicates whether ffmpeg is able to extract streams from (demux) this format
• canMux indicates whether ffmpeg is able to write streams into (mux) this format

The returned object for codecs looks like:

1{
2  ...
3  mp3: {
4    type: 'audio',
5    description: 'MP3 (MPEG audio layer 3)',
6    canDecode: true,
7    canEncode: true,
8    intraFrameOnly: false,
9    isLossy: true,
10    isLossless: false
11  },
12  ...
13}

• type indicates the codec type, either "audio", "video" or "subtitle"
• canDecode tells whether ffmpeg is able to decode streams using this codec
• canEncode tells whether ffmpeg is able to encode streams using this codec

Depending on your ffmpeg version (or if you use avconv instead) other keys may be present, for example:

• directRendering tells if codec can render directly in GPU RAM; useless for transcoding purposes
• intraFrameOnly tells if codec can only work with I-frames
• isLossy tells if codec can do lossy encoding/decoding
• isLossless tells if codec can do lossless encoding/decoding

With some ffmpeg/avcodec versions, the description includes encoder/decoder mentions in the form "Foo codec (decoders: libdecodefoo) (encoders: libencodefoo)". In this case you will want to use those encoders/decoders instead (the codecs object returned by getAvailableCodecs will also include them).

The returned object for encoders looks like:

1{
2  ...
3  libmp3lame: {
4    type: 'audio',
5    description: 'MP3 (MPEG audio layer 3) (codec mp3)',
6    frameMT: false,
7    sliceMT: false,
8    experimental: false,
9    drawHorizBand: false,
10    directRendering: false
11  },
12  ...
13}

• type indicates the encoder type, either "audio", "video" or "subtitle"
• experimental indicates whether the encoder is experimental. When using such a codec, fluent-ffmpeg automatically adds the '-strict experimental' flag.

The returned object for filters looks like:

1{
2  ...
3  scale: {
4    description: 'Scale the input video to width:height size and/or convert the image format.',
5    input: 'video',
6    multipleInputs: false,
7    output: 'video',
8    multipleOutputs: false
9  },
10  ...
11}

• input tells the input type this filter operates on, one of "audio", "video" or "none". When "none", the filter likely generates output from nothing
• multipleInputs tells whether the filter can accept multiple inputs
• output tells the output type this filter generates, one of "audio", "video" or "none". When "none", the filter has no output (sink only)
• multipleInputs tells whether the filter can generate multiple outputs

Cloning an FfmpegCommand

You can create clones of an FfmpegCommand instance by calling the clone() method. The clone will be an exact copy of the original at the time it has been called (same inputs, same options, same event handlers, etc.). This is mainly useful when you want to apply different processing options on the same input.

Setting options, adding inputs or event handlers on a clone will not affect the original command.

1// Create a command to convert source.avi to MP4
2var command = ffmpeg('/path/to/source.avi')
3  .audioCodec('libfaac')
4  .videoCodec('libx264')
5  .format('mp4');
6
7// Create a clone to save a small resized version
8command.clone()
9  .size('320x200')
10  .save('/path/to/output-small.mp4');
11
12// Create a clone to save a medium resized version
13command.clone()
14  .size('640x400')
15  .save('/path/to/output-medium.mp4');
16
17// Save a converted version with the original size
18command.save('/path/to/output-original-size.mp4');

Contributing

Contributions in any form are highly encouraged and welcome! Be it new or improved presets, optimized streaming code or just some cleanup. So start forking!

Code contributions

If you want to add new features or change the API, please submit an issue first to make sure no one else is already working on the same thing and discuss the implementation and API details with maintainers and users by creating an issue. When everything is settled down, you can submit a pull request.

When fixing bugs, you can directly submit a pull request.

Make sure to add tests for your features and bugfixes and update the documentation (see below) before submitting your code!

Documentation contributions

You can directly submit pull requests for documentation changes. Make sure to regenerate the documentation before submitting (see below).

Updating the documentation

When contributing API changes (new methods for example), be sure to update the README file and JSDoc comments in the code. fluent-ffmpeg comes with a plugin that enables two additional JSDoc tags:

• @aliases: document method aliases

1/**
2 * ...
3 * @method FfmpegCommand#myMethod
4 * @aliases myMethodAlias,myOtherMethodAlias
5 */

• @category: set method category

1/**
2 * ...
3 * @category Audio
4 */

You can regenerate the JSDoc documentation by running the following command:

1$ make doc

To avoid polluting the commit history, make sure to only commit the regenerated JSDoc once and in a specific commit.

Running tests

To run unit tests, first make sure you installed npm dependencies (run npm install).

1$ make test

Make sure your ffmpeg installation is up-to-date to prevent strange assertion errors because of missing codecs/bugfixes.

Main contributors

• enobrev
• njoyard
• sadikzzz
• smremde
• spruce
• tagedieb
• tommadema
• Weltschmerz

License

(The MIT License)

Copyright (c) 2011 Stefan Schaermeli <schaermu@gmail.com>

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the 'Software'), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED 'AS IS', WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.