Overview

Facet is an open-source live coding system for algorithmic music and synthesis. With a code editor in the browser and a pair of NodeJS servers running locally on your machine, Facet can generate and sequence audio, MIDI, OSC, and image data.

Facet runs on MacOS, Linux, and Windows.

Installation and getting started

1. Download and install Node.js (must be v14 or greater) and npm: https://docs.npmjs.com/getting-started
2. Download or clone the Facet repository.
3. In a terminal, navigate to the root of the Facet repository, and run npm install.
4. After the previous command completes, run npm run facet. This will start the servers that run in the background for generating and patterns and keeping time. If running on Windows: Windows has a firewall by default for local connections (on the same private network), and it needs to be disabled, or you can manually allow the connection via the confirmation dialog from the Windows firewall system when starting up the servers.
5. In a browser tab (Firefox, Chrome, and Edge work best), navigate to http://localhost:1124. This is the browser-based code editor which can also handle stereo audio playback.
6. Copy this command into the code editor in the browser: test.sine(100).play(); Move your cursor so it's on the line. Hit [ctrl + enter] to run the command. The code editor application will always briefly highlights to illustrate what command(s) ran. You should hear a sine wave playing through your browser tab. Hit [ctrl + .] or [ctrl + /] (Windows) to stop.

Facet command syntax

Facet commands are based entirely around JavaScript, using a class called a FacetPattern. In order to produce audio or MIDI output, create an instance of a FacetPattern by typing any name, such as example, and running some methods:

example.sine(100).play();
// 1 second, 100Hz sine wave plays at beginning of each loop

Some FacetPatterns might contain other FacetPatterns. FacetPatterns contained inside other commands can be created with a shorthand, _:

example.sine(100).times(_.sine(100)).play();

There are lots of methods to generate, translate, and orchestrate playback on FacetPattern data:

example.sine(100).times(random()).play();
// each time you run ^, the same sine wave at a different volume

Certain operations (e.g. sometimes(), iter(), slices(), mix(), run()) allow you to supply functions as arguments:

example.iter(16,()=>{this.append(_.randsamp('808').speed(10))}).play();
// stitches together 16 random samples, each playing at 10x normal speed

In-browser text editor and user interface

Bottom controls

At the bottom of the page, there are several UI elements which control the servers running in the background. Moving from left to right:

• Server connection status indicator (green = online; red = offline)
• CPU% indicator
• Number input for setting the BPM of the global transport (note: when the .bpm() operation runs, this value updates automatically)
• Number inputs for setting the time signature numerator and denominator of the global transport. (note: when the .time() operation runs, these values update automatically)
• MIDI output selector / refresh button
• ■ = stop playback
• ⊖ = stop regenerating patterns but continue playback
• ↻ = restart system (in case it becomes unresponsive)
• 🔉 = toggle browser sound on / off

Right sidebar controls

In the top-right corner, a UI element lists every command that is currently producing audio, MIDI, or OSC. Each command will be listed by its FacetPattern name, so if your command looked like this:

example.sine(440).play();

The corresponding button would be listed with the name "example". Each voice control includes several buttons:

• Pattern name button: mutes/unmutes this pattern.
• Solo button: soloes this pattern (mutes all others). Multiple patterns can be soloed simultaneously.
• Lock button: locks this pattern so that its solo or mute status is not affected when other patterns are soloed. Prevents the .solo() method from targeting this pattern.
• Stop button: stops regeneration and playback of the pattern and removes it from the controls.
• Gain slider: adjusts the volume of the pattern (audio only).

Key-combination shortcuts

• Run command(s): [ctrl + enter] or [ctrl + r]. All commands not separated by multiple newlines will run together.
• Stop command(s): [ctrl + ']. All commands not separated by multiple newlines will be stopped, if they are currently running.
• Keep command(s): [ctrl + ;]. All commands not separated by multiple newlines will continue to play back as-is, without regenerating.
• Once command(s): [ctrl + \]. All commands not separated by multiple newlines will play back once and not regenerate.
• Stop all playback: [ctrl + .] or [ctrl + /]
• Stop regenerating all patterns: [ctrl + ,]
• Autocomplete / list methods: [ctrl + space]. This will list all available methods including their arguments in a dropdown menu, filtered by the text preceding the cursor position. If only one matching method is found, it will autocomplete that method.
• Autoformat code: [ctrl + f]

Running "npm run facet"

When you run the npm run facet command in the terminal, the following sequence of events occurs:

A server, known as the process manager, starts up on http://localhost:5831. This server is responsible for managing the startup and shutdown of the two servers listed below:

1. The transport server starts up on http://localhost:3211. This server is responsible for handling the timing and playback of audio, MIDI, and OSC events.

2. The pattern generator server starts up on http://localhost:1123. This server listens to requests from the text editor UI in the browser located at http://localhost:1124 and interprets those commands into data that can be outputted as audio, MIDI, or OSC data.

Configuration for Max

While Facet runs completely as a standalone process, it is possible to send data from Facet into a Max patch using a custom facet.maxpat object which is included in this repository. First, follow these setup steps:

1. Open Max. In the Max navbar, go to > Options > File Preferences, click "Add Path", and add the facet directory (the folder that contains this file). Make sure that the Subfolders checkbox is checked.
2. Create a new Max patcher, and add a facet object.
3. The single outlet in the facet object will pass OSC commands from Facet into your patcher. Use the /route Max object to route the OSC data in your Max patcher.
4. See the examples/osc.md example file for more details on receiving OSC in Max or any other application.

Global event resolution

By default, Facet checks every 10 milliseconds whether it needs to fire any events that produce output, such as playing audio, MIDI, or osc. You can change EVENT_RESOLUTION_MS in js/config.js to set a different integer value. Slower speeds (e.g. 20 = 20ms) will produce less tightly-timed events but can help make it possible for Facet to run on computers with less CPU resources, at the expense of slight timing accuracy. Faster speeds (e.g. 4 = 4ms) will produce tighter event scheduling but can overload computers with less CPU resources.

Command reference

The rest of this document lists all the methods and variables available in Facet. Since Facet is an extension of JavaScript, you can also incoporate core functions such as Math.pow(), Date.now(), etc.

Variables

Variables output a single floating-point number that can be used as an argument to another method.

mousex / mousey

Both mousex and mousey, as floating-point number representations of your cursor's position in the browser window, are available for use in commands, e.g.:

example.sine(100).times(mousey).play();
// cursor y position controls volume every time the code runs

notevalues

There are 128 "notevalues" variables, corresponding to the number of audio samples relative to geometric divisions of one transport loop (a "whole note"). A whole note is n1, a half note is n2, etc... up to n128.

example.noise(n1);
// ^ at 120bpm 4/4  = 2 second transport loop = 88200 samples
// ^ at 120bpm 3/4  = 1.5 second transport loop = 66150 samples

example.noise(n128);
// ^ at 120bpm 4/4 = 517 samples

bpm

The variable bpm represents the current BPM in the Facet transport when the FacetPattern is generated.

bars

The variable bars represents how many loops have occurred since the time the server was started. This is especially useful with the modulo % operator, e.g.: bars%4, which could be either 0, 1, 2, or 3, depending on how many loops have occurred.

time signature

The variables time_num and time_denom represent the current time signature of the Facet transport. These variables will update when you modify either the time signature numerator or time signature denominator number inputs in the user interface, or if the time() method is dynamically updating the time signature.

sample rate

The sample rate for audio generated and played back with Facet can be changed by modifying SAMPLE_RATE in js/config.js to any integer.

In Facet commands, the constant SAMPLE_RATE refers to the configured sample rate, which is useful for doing something for a specific number of seconds. The constant NYQUIST refers to the Nyquist frequency which is SAMPLE_RATE/2.

example.noise(SAMPLE_RATE).play();
// generate and continually play back exactly 1 second of noise

Outputs

Facet can synthesize and orchestrate the playback of multiple FacetPatterns simultaneously, producing audio, MIDI, or OSC output. By default, patterns will continually regenerate each loop, but methods such as .whenmod(), .keep(), and .once() modify this behavior.

Audio output

**play** ( _PlaybackFacetPattern_, _pitchSequenceData_ = 1 )

• plays the FacetPattern as audio, at however many positions are specified in PlaybackFacetPattern, as the global transport loops through a whole note.
• PlaybackFacetPattern should contain floating-point numbers between 0 and 1, corresponding to the relative point in the transport between 0 and 1 when the generated audio should play.
• pitchSequenceData is an optional argument that should contain an array of pitch shift values. These values are used to adjust the pitch of the sound at each step of the sequence. The pitch shift values are spread out over the sequence steps, so if there are fewer pitch shift values than sequence steps, the pitch shift values will be repeated.
• With no arguments, the command will regenerate at point 0, i.e. at the beginning of each whole note. You can supply a number, array, or FacetPattern as the argument.
• By default, the FacetPattern will continue to regenerate and play. To prevent it from regenerating, include a keep() operation. To stop playback, use the key command [ctrl + .] or [ctrl + /], or press the stop button "■".

    example
      .randsamp('808')
      .play();
    // plays once at beginning of loop
    example
      .randsamp('808')
      .play(0.5);
    // plays once at middle point
    example
      .randsamp('808')
      .play(_.noise(4)
        .scale(0, 1));
    // plays once at 4 random positions
    example
      .randsamp('808')
      .play(_.noise(4)
        .scale(0, 1), _.ramp(1, 4, 4));
    // plays once at 4 random positions, each with a higher pitch

**pan** ( _PanningFacetPattern_ = 0 )

• dynamically moves the FacetPattern between however many channels are specified in a seperate .channels() call. Without a call to .channels(), it will default to spatially positioning the FacetPattern between channels 1 and 2.
• the values in PanningFacetPattern should be between 0 and 1. Values beyond that will be clipped to the 0 - 1 range. A value of 0 will hard-pan the sound to the first active channel that is set via a .channels() call (or defaulting to stereo). A value of 1 will hard-pan the sound to the last active channel. Values between 0 and 1 will crossfade between all the specified active channels.

example
  .noise(n1)
  .times(_.ramp(1, 0, n1))
  .pan(_.sine(1, n1)
    .scale(0, 1))
  .play();
// no channels are specified; defaults to stereo panning
example
  .noise(n1)
  .times(_.ramp(1, 0, n1))
  .channels([1, 2, 4])
  .pan(_.sine(1, n1)
    .scale(0, 1))
  .play();
// pans the noise smoothly around channels 1, 2, and 4

**channel** ( _channels_ )

• The .channel() method (and equivalent channels() method) allow you to route the output of a FacetPattern onto the specified channel(s) in the channels input array.

    example
      .randsamp('808')
      .channel(1)
      .play();
    // first channel only
    example
      .randsamp('808')
      .channels([1, 3])
      .play();
    // channels 1 & 3 only
    example
      .randsamp('808')
      .channel(_.from([9, 10, 11, 12, 13, 14, 15, 16])
        .shuffle()
        .reduce(ri(1, 8)))
      .play();
    // play on a random number of channels from 9-16

**saveas** ( _filename_, _absolute_path_ = false )

• saves a monophonic, 32-bit depth wav file in the samples/ directory or a sub-directory, containing the FacetPattern. If the directory doesn't exist, it will be created.
• to save a file to a location outside of the samples/ directory, supply a truthy absolute_path argument.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

    example
      .iter(6, () => {
          this.append(_.sine(ri(1, 40)))
            .saveas('/myNoiseStuff/' + Date.now())})
            .once();
            // creates 6 wav files in the myNoiseStuff directory.
            // each filename is the UNIX timestamp to preserve order.

**stop** ( )

• stops the command from regenerating and playing back in future loops.
• any time a .stop() is found in a command, the entire command will be skipped and not executed. This helps to preserve CPU.
• you can also stop any command with the [ctrl] + ' key combination while the cursor is on that command.

example
  .noise(n16)
  .play()
  .stop();
// you only hear sound when you remove the stop()

MIDI / OSC output

Facet has various methods for generating and outputting MIDI and OSC data to control other synthesizers or DAWs. You need to connect the MIDI device you want to use before starting Facet.

**note** ( _VelocityPattern_ = 100, _DurationPattern_ = 125, _channel_ = 1, _PositionPattern_ )

• sends a MIDI note on/off pair for every value in the FacetPattern's data.
• the VelocityPattern and DurationPattern will automatically scale to match the note pattern. This allows you to modulate MIDI velocity and duration over the course of the whole note.
• the channel argument by default sends the MIDI out channel 1. It can be set to any channel between 1-16.
• the PositionPattern argument is optional. When not supplied, the positions will calculate automatically and spread pattern data across the entire whole note, playing the first value in the data at relative position 0, and playing the last value in the data at relative position 1. By supplying a PositionPattern, you can program when each note should play.

    example
      .sine(1)
      .size(32)
      .scale(36, 90)
      .round()
      .note();
    // sine wave of 32 notes going from 36 to 90 and back each loop
    example
      .noise(16)
      .scale(60, 80)
      .sort()
      .note(100, 125, 1, _.ramp(0, 0.25, 16));
    // play all notes during the first 25% of the whole note

**note2d** ( _VelocityPattern_ = 100, _DurationPattern_ = 125, _channel_ = 1, _lowNote_ = 0, _highNote_ = 127 )

• sends a MIDI note on/off pair for every value in the FacetPattern's data, arranged as if it were a square 2D grid.
• for more information on generating 2D patterns, see the Methods for image generation and processing section further below in the README document.
• the VelocityPattern and DurationPattern will automatically scale to match the note pattern. This allows you to modulate MIDI velocity and duration over the course of the whole note.
• the channel argument by default sends the MIDI out channel 1. It can be set to any channel between 1-16.
• the lowNote and highNote arguments define the range of MIDI notes that can be played. By default, this range is from 0 (inclusive) to 127 (inclusive).
• the data length must be a perfect square, as it is processed in columns to form a 2D grid of notes.

example
  .silence(2500)
  .circle2d(25, 25, 25, 1)
  .saveimg()
  .note2d(100, 125, 1, 30, 80)
  .once();
// saves 50x50 image with a circle in the middle; then plays that circle shape between the MIDI notes 30 and 80

**cc** ( _controller_number_ = 70, _channel_ = 1 )

• sends a MIDI cc event bound to controller # controller_number for every value in the FacetPattern's data.
• Note: This method is automatically scaled into the expected data range for MIDI CC data. It expects a FacetPattern of values between 0 and 1.
• The channel argument by default sends the MIDI out channel 1. It can be set to any channel between 1-16.

example.drunk(64,0.1).cc();
// sends out cc channel 70
example2.drunk(64,0.1).cc(71);
// sends out cc channel 71

**chord** ( _chordTypePattern_, _inversion_mode_ = 0 )

• creates a chord of MIDI notes for every value in the FacetPattern's data.
• chordTypePattern can be a string, or a FacetPattern, or an array of either. If chordTypePattern is a FacetPattern, the chord intervals will correspond to the data of the chordTypePattern FacetPattern.
• if chordTypePattern is an array with more than one value in it, then the chord type will change dynamically over the course of the loop. For example, a chordTypePattern of ['major','maj7','minor'] will produce major chords for the first third of the loop, then switch to maj7 chords for the middle third, then switch to minor chords for the last third.
• if chordTypePattern is a string, it must be from the below list of chord names:

'maj' / 'major'             =   [0,4,7]
'min' / 'minor'             =   [0,3,7]
'fifth' / '5th'              =   [0,5]
'seventh' / '7th'           =   [0,4,7,10]
'major seventh' / 'maj7'    =   [0,4,7,11]
'minor seventh' / 'm7'      =   [0,3,7,10]
'diminished' / 'dim'        =   [-1,2,5]
'add2'                      =   [0,2,4,7]
'add9'                      =   [0,4,7,14]

• if chord_type is a string, the inversion_mode can be 0, 1, 2, or 3. This number represents how many of the values in the chord have been inverted and are now below the root.
• Note: to force chords into a certain key, use the key() operation after the chord() operation.

    example
      .ramp(36, 72, 32)
      .chord('maj7')
      .add((bars % 4) * 12)
      .key('F#', 'major')
      .note(50, 100, 1);
    // same maj7 chord run, getting an octave higher for 4 octaves then resets
    example
      .noise(16)
      .scale(36, 90)
      .chord(_.from([3, 5, 7, 10, 11, 14, 16, 20, 25]))
      .key('c', 'major')
      .note();
    // 9-note chords mapped onto c major
    example
      .noise(8)
      .scale(30, 80)
      .chord('maj7')
      .key(['c', 'f#'], ['major', 'minor'])
      .note(100, 500);
    // maj7 chords, first in c major, then in f# minor

**key** ( _keyLetterPattern_, _keyScalePattern_ )

• translates a FacetPattern with data in the range of MIDI note numbers (0-127) so all its values now adhere to the supplied keyLetterPattern and keyScalePattern.
• keyLetterPattern values can be a string: "A", "A#", "B", "C", "C#", "D", "D#", "E", "F", "F#", "G", "G#", or an array, or strings: ["A", "D"], or a FacetPattern of strings: _.from(['A','D']).dup(3).shuffle(). When keyLetterPattern contains multiple values, the key will change dynamically over the course of the loop. So a keyLetterPattern of ["A", "D", "G"] will be in the key of A for the first third, then switch to D for the middle third, then switch to G for the last third.
• keyScalePattern values can either be a string (see list below), or a FacetPattern containing 1-12 binary numbers (see examples), or an array of strings/FacetPatterns, in which case the values change dynamically over the course of the loop.
• possible scales:

["major pentatonic", "major", "minor", "major blues", "minor blues", "melodic minor", "harmonic minor", "bebop", "diminished", "dorian", "c", "mixolydian", "phrygian", "locrian", "ionian pentatonic", "mixolydian pentatonic", "ritusen", "egyptian", "neopolitan major pentatonic", "vietnamese 1", "pelog", "kumoijoshi", "hirajoshi", "iwato", "in-sen", "lydian pentatonic", "malkos raga", "locrian pentatonic", "minor pentatonic", "minor six pentatonic", "flat three pentatonic", "flat six pentatonic", "scriabin", "whole tone pentatonic", "lydian #5P pentatonic", "lydian dominant pentatonic", "minor #7M pentatonic", "super locrian pentatonic", "minor hexatonic", "augmented", "piongio", "prometheus neopolitan", "prometheus", "mystery #1", "six tone symmetric", "whole tone", "messiaen's mode #5", "locrian major", "double harmonic lydian", "altered", "locrian #2", "mixolydian b6", "lydian dominant", "lydian augmented", "dorian b2", "ultralocrian", "locrian 6", "augmented heptatonic", "dorian #4", "lydian diminished", "leading whole tone", "lydian minor", "phrygian dominant", "balinese", "neopolitan major", "harmonic major", "double harmonic major", "hungarian minor", "hungarian major", "oriental", "flamenco", "todi raga", "persian", "enigmatic", "major augmented", "lydian #9", "messiaen's mode #4", "purvi raga", "spanish heptatonic", "bebop minor", "bebop major", "bebop locrian", "minor bebop", "ichikosucho", "minor six diminished", "half-whole diminished", "kafi raga", "messiaen's mode #6", "composite blues", "messiaen's mode #3", "messiaen's mode #7", "chromatic"]

example
  .randsamp('808')
  .reduce(32)
  .scale(36, 51)
  .key('F#', 'bebop')
  .note();
example
  .noise(16)
  .scale(30, 80)
  .key('c', _.from([1]))
  .note();
// octave scale, via custom FacetPattern
example
  .noise(16)
  .scale(30, 80)
  .key('c', _.from([1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]))
  .note();
// equivalent to the above custom octave scale; the padded zeroes are optional
example
  .noise(16)
  .scale(30, 80)
  .key('c', _.from([1, 0, 0, 0, 0, 0, 1]))
  .note();
// octave + perfect fifth scale, via custom FacetPattern
example
  .noise(16)
  .scale(30, 80)
  .key(['c', 'f#'], ['major', 'minor'])
  .note();
// the first half is c major; the second half is f# major

**osc** ( _address_ )

• sends a packet of OSC data to OSC address (or array of addresses) address, for every value in the FacetPattern's data.
• The OSC server sends output to port 5813 by default but can be modified by changing OSC_OUTPORT in js/config.js.
• The address argument must begin with a backslash: /.
• Since this is intended for control-rate data, large patterns are automatically reduced in size, if ncessary, so they will never output more than 100 values per second.
• Note: This method does not automatically scale the FacetPattern values between 0 and 1, so the user can send any range of numbers over OSC.

example.noise(128).osc('/test');

**pitchbend** ( _channel_ = 1 )

• sends a MIDI pitchbend event for every value in the FacetPattern's data.
• The channel argument by default sends the MIDI out channel 1. It can be set to any channel between 1-16.
• Note: This method is automatically scaled into the expected range for MIDI pitchbend data. It expects a FacetPattern of values between -1 and 1, with 0 meaning no pitchbend.

example.sine(1).size(128).pitchbend();

**savemidi** (_midifilename_ = Date.now(), _velocityPattern_ = 64, _durationPattern_ = 16, _wraps_ = 1, _tick_mode_ = false, _absolute_path_ = false)

• creates a MIDI file of MIDI notes named midifilename in the midi/ directory or a sub-directory, with the FacetPattern's data.
• to save a file to a location outside of the midi/ directory, supply a truthy absolute_path argument.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)
• VelocityPattern and DurationPattern will automatically scale to match the note pattern. This allows you to modulate MIDI velocity and duration over the course of the whole note.
• The velocityPattern expects values between 1 and 100. (This range is set by the midi-writer-js npm package).
• The wraps parameter controls how many times to wrap the data back around onto itself, allowing for MIDI polyphony. For example, if your FacetPattern has 8 different 128-note patterns appended together in a sequence, a wraps of 8 would superpose all of those patterns on top of each other, while a wraps value of 4 would produce four patterns on top of each other, followed by four more patterns on top of each other.
• When tick_mode is set to a truthy value, the numbers in durationPattern represent the number of ticks to last, rather than a whole-note divisions. 1 tick = 1/256th note. This allows for durations smaller and larger than the valid duration values for when tick_mode is set to false.
• When tick_mode is set to false or excluded from the command, the following values are the only valid durationPattern argument values:

    '1'   =   whole note
    '2'   =   half note
    'd2'  =   dotted half note
    'dd2' =   double dotted half note
    '4'   =   quarter note
    '4t'  =   quarter triplet note
    'd4'  =   dotted quarter note
    'dd4' =   double dotted quarter note
    '8'   =   eighth note
    '8t'  =   eighth triplet note
    'd8'  =   dotted eighth note
    'dd8' =   double dotted eighth note
    '16'  =   sixteenth note
    '16t' =   sixteenth triplet note
    '32'  =   thirty-second note
    '64'  =   sixty-fourth note

example
  .noise(64)
  .scale(20, 90)
  .key('c major')
  .savemidi(ts(), 64, 16)
  .once();
// 64 random notes in c major at 64 velocity, each lasting a 16th note
example
  .noise(64)
  .scale(20, 90)
  .key('c major')
  .savemidi(ts(), _.noise(64)
    .scale(1, 100), 4, 1, true)
  .once();
// 64 random notes in c major, each with a random velocity between 1 - 100, each lasting 4 ticks
example
  .iter(8, () => {
    this.append(_.sine(choose([1, 2, 3, 4]))
      .size(128)
      .scale(ri(30, 50), ri(60, 90))
      .key('c major'))
  })
  .savemidi(ts(), 64, 16, 8)
  .once();
// 8 sine wave patterns playing notes in c major, superposed on top of each other. try changing the wraps argument to values other than 8

**savemidi2d()** (_midifilename_ = Date.now(), _velocityPattern_ = 64, _durationPattern_ = 16, _tick_mode_ = false, _min_note_ = 0, _max_note_ = 127, _absolute_path_ = false)

• creates a MIDI file of MIDI notes named midifilename in the midi/ directory or a sub-directory, with the FacetPattern's data, assuming that the data was created using the 2d methods for image generation and processing. All nonzero "pixel" values will be translated into a MIDI note. Go to the methods for image generation and processing section for more details on these methods.
• to save a file to a location outside of the midi/ directory, supply a truthy absolute_path argument.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)
• VelocityPattern and DurationPattern will automatically scale to match the note pattern. This allows you to modulate MIDI velocity and duration over the course of the whole note.
• The velocityPattern expects values between 1 and 100. (This range is set by the midi-writer-js npm package).
• The min_note and max_note values control the range of notes that the corresponding 2d pattern will be generated between.
• When tick_mode is set to a truthy value, the numbers in durationPattern represent the number of ticks to last, rather than a whole-note divisions. 1 tick = 1/256th note. This allows for durations smaller and larger than the valid duration values for when tick_mode is set to false.
• When tick_mode is set to false or excluded from the command, the following values are the only valid durationPattern argument values:

        '1'   =   whole note
        '2'   =   half note
        'd2'  =   dotted half note
        'dd2' =   double dotted half note
        '4'   =   quarter note
        '4t'  =   quarter triplet note
        'd4'  =   dotted quarter note
        'dd4' =   double dotted quarter note
        '8'   =   eighth note
        '8t'  =   eighth triplet note
        'd8'  =   dotted eighth note
        'dd8' =   double dotted eighth note
        '16'  =   sixteenth note
        '16t' =   sixteenth triplet note
        '32'  =   thirty-second note
        '64'  =   sixty-fourth note

example
  .silence(2500)
  .iter(8, () => {
    this.tri2d(ri(0, 49), ri(0, 49), ri(0, 49), ri(0, 49), ri(0, 49), ri(0, 49), 1, 0)
  })
  .savemidi2d(ts(), 64, 16)
  .once();
  // 8 randomly sized triangles in 2d space, all at velocity 64, 16th note durations
example
  .silence(2500)
  .iter(10, () => {
    this.circle2d(ri(10, 40), ri(10, 40), 10, 1, 0)
  })
  .savemidi2d(ts(), 64, 64)
  .once();
  // 10 randomly sized circles in 2d space, all at velocity 64, 64th note durations

Methods for controlling time and user interface

**bpm** ( _bpm_pattern_ )

• stores the data of bpm_pattern in the transport as BPM values to be cycled through over each loop.
• the minimum BPM value is 1, and the maximum BPM value is 10000.
• using the over() method on the bpm_pattern will cycle through the BPM pattern over multiple loops.

example
  .bpm(_.from([20, 40, 80, 160, 320])
    .shuffle()); // each loop will be all 5 of these BPM, randomly ordered
example
  .bpm(_.ramp(20, 200, 64)
    .over(4)); // ramps from 20BPM to 200BPM over 4 loops

**randsolo** ( _control_pattern_ )

• triggers a random re-solo of all unlocked voices whenever a value in control_pattern exceeds 0.5.
• the subset of voices soloed and the number of voices in the subset are both chosen randomly each time it fires.
• locked patterns (using the lock button in the UI) are excluded from random solo targeting.
• appears in the bottom solo panel in the right sidebar alongside .solo() patterns.

example
.randsolo(_.noise(16,0,1))); // randomly re-solos voices on each step where noise > 0.5
  // note that other audio/MIDI/OSC patterns will need to be playing for this to take effect

**solo** ( _target_pattern_ = 0 )

• automatically targets other patterns for soloing based on target_pattern.
• when target_pattern is provided, those values should be normalized between 0-1 and will be mapped onto the available unlocked voices in the right sidebar.
• locked patterns (using the lock button in the UI) are excluded from automatic solo targeting.
• multiple .solo() patterns can run at once, creating potential for meta-modulation in the system output.
• appears in the bottom solo panel in the right sidebar alongside .randsolo() patterns.

example
  .solo(_.ramp(0,1,16).echo(3)); // solos different voices based on the ramp values
  // note that other audio/MIDI/OSC patterns will need to be playing for this to take effect

**time** ( _time_signature_numerator_pattern_ = 4, _time_signature_denominator_pattern_ = 4 )

• sets the time signature for the transport.
• if time_signature_numerator_pattern or time_signature_denominator_pattern has more than one value, those values will be cycled through over each loop.
• using the over() method on time_signature_numerator_pattern or time_signature_denominator_pattern will cycle through the time signature pattern over multiple loops.
• the minimum _time_signature_numerator_pattern_ value is 1, and the maximum is 32.
• the minimum _time_signature_denominator_pattern_ value is 1, and the maximum is 16.

example
  .time(4, 4); // set time to 4/4
example
  .time(_.ramp(8, 1, 8)
    .over(8), 2); // ramps from 8/2 to 1/2 time over 8 loops

Methods for controlling pattern regeneration

**whenmod** ( _modulo_operand_, _equals_value_ = 0 )

• only regenerates the pattern when the number of bars elapsed, modulo the modulo_operand, equals the equals_value argument. This can be useful when you want to conditionally skip certain commands, only rerunning every n bars.

example_1
  .sine(ri(100, 400))
  .whenmod(4, 0); // regenerates the pattern once every 4 bars
example_2
  .sine(ri(100, 400))
  .whenmod(4, 2); // regenerates the pattern once every 4 bars, but 2 bars away from the above example

**keep** ( )

• preserve the generated FacetPattern so that it plays each loop. Without including keep(), the FacetPattern will regenerate each loop by default.

example
  .sine(ri(10, 500))
  .keep()
  .play();

**once** ( )

• only play the generated FacetPattern a single time. Without including once(), the FacetPattern will regenerate and play back each loop by default.

example
  .noise(4096)
  .play()
  .once();

**over** ( _n_loops_ = 1 )

• distributes all the events that a FacetPattern would fire over n_loops so the pattern can last any number of loops before regenerating.
• works with audio playback, MIDI note/cc/pitchbend, OSC, BPM (via .bpm()), and time signature (via .time()).

example
  .randsamp('808')
  .play(_.ramp(0, 1, 16))
  .over(ri(1, 4)); // random sample played 16 times over 1,2,3 or 4 bars
example
  .drunk(2048, 0.01)
  .cc()
  .over(128); // drunk walk over 128 bars, creates a drifty process that you can map onto device paramters to slowly randomize something

Methods for setting variables

These can be useful when you want to access or modify the same pattern across commands or inside of one command.

**set** ( _name_ )

• saves a FacetPattern's data in memory, for reference as a variable in operations. Any FacetPatterns stored via .set() will only be stored until the server is closed.
• if a pattern stored with set() has more than one piece of data in it, the corresponding variable will be an array. If the pattern has one piece of data in it, the corresponding variable will be a float.
• NOTE: when you run the .set() command for the first time after starting the system, if you're also running commands that reference that variable in the same block, for the first evaluation, the variable will have a value of 0 as it has not fully propagated into the variable storage system.

set_example
  .noise(8)
  .scale(0, 1)
  .set('my_var')
  .once();
  // first, set the variable here

example
  .sine(100)
  .times(my_var)
  .play();
  // now, you can use my_var in commands

**drift** ( _seedPattern_, _patternName_, _command_ = function )

• creates and runs a drifting process on a pattern.
• when the command is executed manually by the user, it will create and store (internally, via set()) a new pattern, named patternName, using seedPattern.
• each time the command reruns, it checks if a pattern named patternName has been stored. If so, it will modify and replace the saved pattern based on the code in command.

example
  .drift(_.noise(16)
    .scale(36, 72)
    .sort(), 'mynotes', () => {
      this.walk(0.1, 1)
    })
  .note();
  // starts as ascending melody and drifts away into randomness

**inc** ( _name_, _amount_to_add_ = 1 )

• increments a variable called name by amount_to_add. This variable can be used in operations.
• similar to the set() method, when you run the .inc() command for the first time after starting the system, if you're also running commands that reference that variable in the same block, for the first evaluation, the variable will have a value of 0 as it has not fully propagated into the variable storage system.

example
  .inc('abc')
  .iter(abc, () => {
    this.sup(_.randsamp('808'), i / iters)
  })
  .play();
  // more 808 samples each iteration

**dec** ( _name_, _amount_to_subtract_ = 1 )

• decrements a variable called name by amount_to_add. This variable can be used in operations.
• similar to the set() method, when you run the .dec() command for the first time after starting the system, if you're also running commands that reference that variable in the same block, for the first evaluation, the variable will have a value of 0 as it has not fully propagated into the variable storage system.

example
  .from(8)
  .set('abc')
  .sometimes(0.5, () => {
    this.dec('abc')
  })
  .sometimes(0.5, () => {
    this.inc('abc')
  })
  .iter(abc, () => {
    this.sup(_.randsamp('k'), i / iters)
  })
  .play();
  // start at 8, sometimes increment & sometimes decrement the total number of 808 samples

**setlocal** ( _name_ )

• saves a FacetPattern's data locally, making it immediately accessible to later operations in the same command.
• in order to acesss a pattern saved with setlocal(), use getlocal().
• NOTE: the data is available only internally, to the command where it runs and cannot be accessed in other commands.

example
  .drunk(1000, 0.2)
  .setlocal('mylocalpattern')
  .reduce(0)
  .iter(1000, () => {
    this.append(_.getlocal('mylocalpattern')
      .jam(0.1, 0.1)
      .setlocal('mylocalpattern'))
  })
  .saveimg()
  .once();
  // initial 1000-value drunk walk, each row 10% of the pixels are +/- 10% modified from previous row

**getlocal** ( _name_ )

• retrieves a FacetPattern's data that was stored locally via setlocal().

example
  .drunk(1000, 0.2)
  .setlocal('mylocalpattern')
  .reduce(0)
  .iter(1000, () => {
    this.append(_.getlocal('mylocalpattern')
      .jam(0.1, 0.1)
      .setlocal('mylocalpattern'))
  })
  .saveimg()
  .once();
  // initial 1000-value drunk walk, each row 10% of the pixels are +/- 10% modified from previous row

Utility functions

**barmod** ( _modulo_, _values_ )

• returns values that depend on the current value of bars. (bars is a global variable that starts at 0 and increments at the completion of a loop.)
• selects a value from the values array, based on bars % modulo. If the bars value currently is 9, and the modulo argument to this method is 4, since 9 % 4 = 1, this method will return the value from the values array immediately following the number 1.
• NOTE: It first checks if the values array contains an even number of elements. If not, it throws an error.
• NOTE: It also checks if every integer from 0 to (mod-1) is one of the even-numbered keys of the values array. If not, it throws an error.

example
  .sine(barmod(4, [0, 100, 1, 150, 2, 200, 3, 300]))
  .play();
  // when bars % 4 == 0, plays a 100Hz sine.
  // when bars % 4 == 1, plays a 150 Hz sine.
  // when bars % 4 == 2, plays a 200Hz sine.
  // when bars % 4 == 3, plays a 300Hz sine.

**choose** ( _pattern_ )

• returns a randomly selected value from a supplied array.

example
  .sine(choose([10, 200, 1000]))
  .play();
  // sine wave with either 10, 200, or 1000 cycles

**cof** ( _index_ )

• returns the element at position index in the circle of fifths, starting with C and ending with F.
• the set of notes: ['C', 'G', 'D', 'A', 'E', 'B', 'F#', 'C#', 'G#', 'D#', 'A#', 'F'].

example
  .noise(16)
  .scale(36, 90)
  .key(cof(2))
  .note();
  // MIDI notes in D major
example
  .noise(16)
  .scale(36, 90)
  .key(cof(ri(0, 11)))
  .note();
  // MIDI notes in random major key

**decide** ()

• returns a 1 or 0 randomly.

example
  .sine(choose([10, 200, 1000]))
  .dup(decide())
  .play();
  // duplicate half the time

**ftom** ( _hzfrequency_ )

• converts the supplied hzfrequency value to its corresponding MIDI note number.

example
  .from(440)
  .ftom()
  .note();
  // plays an A440 MIDI note

**ms** ( _milliseconds_ )

• converts the supplied milliseconds value to that many samples, at whatever sample rate the user has configured.

example
  .noise(4096)
  .size(ms(5))
  .play();
  // 5ms noise
example
  .noise(4096)
  .size(ms(50))
  .play();
  // 50ms noise

**mtof** ( _midi_note_number_ )

• converts the supplied midi_note_number value to its corresponding frequency in Hz.

example
  .sine(mtof(choose([36, 38, 40, 41, 43, 45, 47, 48])))
  .play();
  // random sine wave each loop in C major key

**mtos** ( _midi_note_number_ )

• converts the supplied midi_note_number value to its corresponding number of samples.

example
  .noise(n4)
  .delay(mtos(choose([36, 38, 40, 41, 43, 45, 47, 48])))
  .delay(mtos(choose([36, 38, 40, 41, 43, 45, 47, 48])))
  .delay(mtos(choose([36, 38, 40, 41, 43, 45, 47, 48])))
  .play();
  // noise is delayed by amounts that are harmonic with C major key

**random** ( _min_ = 0, _max_ = 1, _int_mode_ = 0, _weight = 1_ )

• returns a random number between min and max. If int_mode = 1, returns an integer. Otherwise, returns a float by default.
• you can also use these shorthands for a random float: rf(min,max) and a random integer: ri(min,max).
• The weight argument allows you to specify an exponential weight for the probability of random values. For instance, rf(0.125,8,3) will generate half of its values between 0.125 and 1; and the other half will be between 1 and 8. By default, the weighting is linear, i.e. all values between min and max have equal probability.

example
  .sine(ri(20, 1000))
  .play();
  // a sine wave with 20 - 1000 cycles

**ts** ( )

• returns the current timestamp Date.now() as a string.

example
  .sine(100, n1)
  .saveas('mytest' + ts())
  .once();
  // saves a file in the samples directory named like this: mytest1704420621454.wav

**just** ()

• returns an array of frequency ratios for the just intonation tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(just()));
  // sine waves in ascending just intonation scale

**pythagorean** ()

• returns an array of frequency ratios for the Pythagorean tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(pythagorean()));
  // sine waves in ascending pythagorean intonation scale

**equaltemp** ()

• returns an array of frequency ratios for the 12-tone equal temperament tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(equaltemp()));
  // sine waves in ascending equal temperament intonation scale

**meantone** ()

• returns an array of frequency ratios for the meantone temperament tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(meantone()));
  // sine waves in ascending meantone intonation scale

**edo19** ()

• returns an array of frequency ratios for the 19-tone equal division of the octave (EDO) tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(edo19()));
  // sine waves in ascending edo19 intonation scale

**edo31** ()

• returns an array of frequency ratios for the 31-tone equal division of the octave (EDO) tuning system.

example
  .sine(100, n16)
  .play(_.ramp(0, 1, 12), _.from(edo31()));
  // sine waves in ascending edo31 intonation scale

Methods for controlling MIDI scales

**randscale** ( )

• returns a random scale for MIDI notes out. The set of possible scales is listed in the key() method.

example
  .noise(32)
  .scale(30, 80)
  .sort()
  .key('f#', randscale())
  .note();
  // random scale in f#

FacetPattern generators that can take a FacetPattern, number, array, or object as an argument

When a generator takes a FacetPattern or an array as an argument, it uses that pattern to dynamically change its behavior over time, affecting the output in a more complex way than if a single number were supplied. For example, with the command example.sine(440).play();, the output is a static 440Hz wave. But with the command example.sine(_.sine(5).scale(20,2000))).play();, the frequency of the sine wave is being modulated by a 5 Hz sine wave which is generating values between 20 and 2000. This produces a classic frequency modulation sound, but since you can supply any FacetPattern as an argument, there are lots of sound design possibilities.

**sine** ( _frequencyPattern_, _duration_ = sample_rate, _fade_in_and_out_ = false )

• generates a sine wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from -1 - 1.
• by default, the fade_in_and_out argument is set to false, which will generate the signal without applying any fade to its start or end values, which can produce audible clicks. Use a truthy value for fade_in_and_out to apply a 30-millisecond fade-in and fade-out, or add a fadeout(0.01) after the command.

example
  .sine(440, n4).fadeout(0.01)
  .play();
  // 440 Hz sine wave for a quarter note
example
  .sine(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .sine(_.sine(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

**cosine** ( _frequencyPattern_, _duration_ = 1 second, _fade_in_and_out_ = true )

• generates a cosine wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from -1 - 1.
• by default, the fade_in_and_out argument is set to false, which will generate the signal without applying any fade to its start or end values, which can produce audible clicks. Use a truthy value for fade_in_and_out to apply a 30-millisecond fade-in and fade-out, or add a fadeout(0.01) after the command.

example
  .cosine(440, n4).fadeout(0.01)
  .play();
  // 440 Hz cosine wave for a quarter note
example
  .cosine(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .cosine(_.cosine(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

**circle** ( _frequencyPattern_, _duration_ = 1 second )

• generates a half-circle wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from 0 - 1.

example
  .circle(440, n4)
  .play();
  // 440 Hz circle wave for a quarter note
example
  .noise(n1)
  .times(_.circle(4))
  .play()
  .once();
  // amplitude modulation of noise with a quarter note circular waveform
example
  .noise(n1)
  .ffilter(_.circle(1)
    .invert()
    .size(128)
    .scale(0, NYQUIST / 2), _.circle(1)
    .size(128)
    .scale(NYQUIST / 2, NYQUIST))
  .play()
  .once();
  // circular spectral filtering of a whole note of noise

**markov** ( _states_ = [{name, value, probs}, {name, value, probs}, ...] )

- modifies the input FacetPattern according to a Markov chain. The `states` parameter is an array where each entry is an object representing a state. Each state object has a `name`, a `value` that corresponds to a value in `this.data`, and a `probs` object that defines the transition probabilities to other states.
- the method modifies `this.data` by transitioning each value to a new state based on the probabilities defined in the `states` array. The transition probabilities are normalized so that they add up to 1 for each state.
- example: 

example
.iter(choose([3,4,8]), () => {
this.prepend(_.from(['k*', 'h*', 's*', 'h*', '_', '_','_', '_'])
  .markov([{
      name: "state1",
      value: 'k*',
      probs: {
        "state1": 0.8,
        "state2": 0.5,
        "state3": 0.5
      }
    },
    {
      name: "state2",
      value: 'h*',
      probs: {
        "state3": 1
      }
    },
    {
      name: "state3",
      value: 's*',
      probs: {
        "state1": 1
      }
    },
    {
      name: "state4",
      value: '_',
      probs: {
        "state1": 0.5,
        "state4": 0.5
      }
    }
  ]))
})
.run(() => {
this.seq(this.data)
}).play();

**phasor** ( _frequencyPattern_, _duration_ = 1 second, _fade_in_and_out_ = true )

• generates a phasor wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from -1 - 1.
• by default, the fade_in_and_out argument is set to false, which will generate the signal without applying any fade to its start or end values, which can produce audible clicks. Use a truthy value for fade_in_and_out to apply a 30-millisecond fade-in and fade-out, or add a fadeout(0.01) after the command.

example
  .phasor(440, n4).fadeout(0.01)
  .play();
  // 440 Hz phasor wave for a quarter note
example
  .phasor(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .phasor(_.phasor(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

**rect** ( _frequencyPattern_, _duration_ = 1 second, _pulse_width_ = 0.5, _fade_in_and_out_ = true )

• generates a rectangle wave at frequencyPattern Hertz, with a pulse width defined by pulse_width, lasting for duration samples.
• output range is from -1 - 1.
• by default, the fade_in_and_out argument is set to false, which will generate the signal without applying any fade to its start or end values, which can produce audible clicks. Use a truthy value for fade_in_and_out to apply a 30-millisecond fade-in and fade-out, or add a fadeout(0.01) after the command.

example
  .rect(440, n4, rf()).fadeout(0.01)
  .play();
  // 440 Hz rectangle wave for a quarter note, different bandwidth each time
example
  .rect(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .rect(_.rect(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

**square** ( _frequencyPattern_, _duration_ = sample_rate )

• generates a square wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from -1 - 1.

example
  .square(440, n4).fadeout(0.01)
  .play();
  // 440 Hz square wave for a quarter note
example
  .square(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .square(_.square(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

**tri** ( _frequencyPattern_, _duration_ = sample_rate, _fade_in_and_out_ = true )

• generates a triangle wave at frequencyPattern Hertz, lasting for duration samples.
• output range is from -1 - 1.
• by default, the fade_in_and_out argument is set to false, which will generate the signal without applying any fade to its start or end values, which can produce audible clicks. Use a truthy value for fade_in_and_out to apply a 30-millisecond fade-in and fade-out, or add a fadeout(0.01) after the command.

example
  .tri(440, n4).fadeout(0.01)
  .play();
  // 440 Hz triangle wave for a quarter note
example
  .tri(_.ramp(10, 2000, 300)).fadeout(0.01)
  .play();
  // ramp from 10Hz to 2000 Hz over 300 values
example
  .tri(_.tri(5)
    .scale(20, 2000)).fadeout(0.01)
  .play();
  // 5Hz frequency modulation with output frequencies oscillating between 20Hz and 2000Hz

FacetPattern generators

**binary** ( _integer_, _length_)

• Computes the binary representation of integer. If length is not present, the output FacetPattern will be the actual length of the binary representation of integer.
• output range is from 0 - 1.

example
  .binary(8);
  // 1000
example
  .binary(490321, 13);
  // 1110111101101: truncated at 13 values
example
  .binary(8, 12);
  // 000000001000: padded with 0s

**dirsamp** ( _dir_ = `../samples/`, _dirPosPattern_, _command_ = function(), _channel_index_ = 0 )

• loads a wav file from the dir directory into memory, based on its alphabetical position. The position is determined by the dir_pos argument, which can be a FacetPattern, array, or float between 0 and 1. A dir_pos value of 0 would load the first wav file in alphabetical order, a dir_pos value of 0.5 would load the middle file, and so on.
• The patterns loaded by dirPosPatern will be spread out across the duration of the loop. So if dirPosPattern has 16 values, those will be spread out into 16 evenly spaced positions.
• The default directory is ../samples/, but you can supply any directory as an argument.
• command allows to run code on each sample as it is loaded.
• By default, it loads the first channel (channel_index = 0) but you can specify any channel to load.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .iter(8, () => {
    this.sup(_.dirsamp('808', i / iters), i / iters)
  })
  .play();
  // 808 sequence

**drunk** ( _length_, _intensity_, _starting_value_ = Math.random() )

• generates a random walk of values between 0 and 1 for length values, starting at starting_value which is a random value between 0 and 1 by default. intensity controls how much to add.
• output range is from 0 - 1.

example
  .drunk(16, 0.1);
  // slight random movement

**envelope** ( _values_ )

• Generates an envelope using the supplied array values, which must have a total number of entries equal to a multiple of 3. The numbers inside the values array should be continually ordered in groups of three: from, to, size, just like the ramp() method.

example
  .noise(ms(500))
  .times(_.envelope([0, 1, ms(10), 1, 0.1, ms(200), 0.1, 0, ms(290)]))
  .play();
  // transient noise burst

**euclid** ( _pulses_, _steps_ )

• generates a Euclidean sequence with pulses pulses over steps steps.
• outputs the relative positions from 0 - 1 where the pulses occur.

example
  .randsamp('808')
  .play(_.euclid(5,8)) // plays at positions [ 0, 0.125, 0.375, 0.5, 0.75 ]

**file** ( _filepath_ )

• loads the raw data of any file into memory. You can supply any file type.
• output range is from -1 - 1.
• By default, it checks for a file in the files subdirectory. If no file exists there, it will try to load the file as an absolute path on your hard drive.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .file('my_image.png')
  .play();
  // if my_image.png is in the files directory, this will play the file's raw data. NOTE: this could be very noisy!
example
  .file('/Users/my_username/Desktop/myfile.zip')
  .play();
  // example with a supplied absolute file path

**from** ( _pattern_, _size_ )

• allows the user to specify their own pattern as an array or number.
• size is optional. if supplied, it will re-size the pattern to have a total length of size samples.

example
  .from([1, 2, 3, 4]);
  // pattern has 4 values: [1,2,3,4]
example
  .from(0.1234,1000);
  // pattern has 1000 identical values: all 0.1234

**image** ( _filepath_, _columnsPerSecond_ = 512, _minimumFrequency_ = 20, _maximumFrequency_ = sample_rate / 2, _frequencyPattern_ )

• transposes an image into audio by superposing sine waves across the audio spectrum, with one sine wave for each row of pixels in the image. The amplitudes of each sine wave are modulated by the corresponding brightness of each pixel in the image, producing an analog of the image in the audio spectrum.
• the lowest pixels in the image correspond to the lowest frequencies in the output, and the highest pixels in the image correspond to the highest frequencies in the output.
• the default columnsPerSecond value of 512 means that each second of audio will contain 512 columns of pixels. This value can be larger or smaller, but keep in mind that as this value decreases, the file will take more time to generate. This method can be CPU intensive and works best with smaller image files or larger columnsPerSecond values.
• since pixel brightness corresponds with loudness, images with dark backgrounds and high contrast will produce clearer tones.
• This method currently only works with PNG files.
• the minimumFrequency and maximumFrequency values control the range of frequencies that the pixels will map onto.
• the frequencyPattern argument allows you to remap the rows of pixels with a FacetPattern. It should be scaled between 0 and 1. It will automatically be resized so its data length matches the height of the image in pixels. Lower values in frequencyPattern will map onto lower frequencies inside the range of minimumFrequency and maximumFrequency. Higher values in frequencyPattern will map onto higher frequencies inside the range of minimumFrequency and maximumFrequency.
• output range is from -1 - 1.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .image('/path/to/file/goes/here.png', 1024)
  .play();
  // each column lasts 1024 samples

**noise** ( _length_, _min_ = -1, _max_ = 1 )

• generates a random series of values between min and max for length.

example
  .noise(1024)
  .play();

**pluck** ( _frequency_, _damping_ = 0, _feedback_ = 0.5 )

• generates a Karplus-Strong type string pluck emulation at frequency Hertz. damping and feedback values should be between 0 and 1.
• output range is from -1 - 1.

example
  .pluck(440, rf(), rf())
  .play(); // different 440 Hz quarter note pluck each time

**primes** ( _n_, _offset_from_first_prime_ = 2, _skip_ = 1 )

• generates the first n prime numbers starting at offset, skipping skip prime numbers before including the next one in the list.
• n specifies the number of prime numbers to generate.
• offset specifies the first number to be included in the list of prime numbers. The default value is 2.
• skip specifies the number of prime numbers to skip before including the next one in the list. The default value is 1.

example
  .noise(n4)
  .times(_.ramp(1, 0, n4))
  .iter(12, () => {
    this.allpass()
      .delay(_.primes(60, 1000, ri(20, 2000))
        .data[i])
      .full()
  })
  .full()
  .play();
  // generates a quarter note transient burst of noise, then iteratively sends it through delays that are all primes

**ramp** ( _from_, _to_, _size_ = 128 )

• moves from from to to over size values.

example
  .ramp(250, 100, 1000);
  // go from 250 to 100 over 1000 values

**randfile** ( _dir_ = `../files/` )

• loads a random file from the files directory into memory. The default directory is ../files/, but you can supply any directory as an argument.
• output range is from -1 - 1.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .randfile()
  .play();
  // random new file converted to audio every time

**randsamp** ( _dir_ = `../samples/` _channel_index_ = 0 )

• loads a random wav file from the dir directory into memory. The default directory is ../samples/, but you can supply any directory as an argument.
• By default, it loads the first channel (channel_index = 0) but you can specify any channel to load.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .randsamp('808')
  .reverse()
  .play();
  // random backwards sample

**sample** ( _filepath_, _channel_index_ = 0)

• loads a wav file from the samples/ directory into memory. You can also specify any file with an absolute file path. The .wav can be omitted from filename; in this case .wav it will be automatically appended to filename. By default, it loads the first channel (channel_index = 0) but you can specify any channel to load.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)

example
  .sample('1234')
  .play();
  // if 1234.wav is in the samples directory, you're good to go
example
  .sample('./myfolder/myfile.wav');
  // or point to the file with a relative path

**silence** ( _length_ )

• generates silence (many 0s in a row) for length samples.

example
  .silence(n2)
  .append(_.noise(n2))
  .play();
  // first half of loop is silence; second half is noise

**spiral** ( _length_, _degrees_ = 360/length, _angle_phase_offset_ = 0 )

• generates a spiral of length length of continually ascending values in a circular loop between 0 and 1, where each value is degrees away from the previous value. degrees can be any number or FacetPattern with values expected between 0 and 360. By default degrees is set to 360/length which produces an output pattern similar to branching leaves, where each value is as far away as possible from the previous value.
• The angle_phase_offset argument changes where the sequence starts. At its default value of 0, the first value will be 0. You can supply any float between 0 and 1, and the sequence will begin at that value instead.
• output range is from 0 - 1.

example
  .sine(1)
  .times(_.spiral(1000, ri(1, 360)))
  .play();
  // a 1Hz sine wave with amplitude modulated by a spiral

**steps** ( _semitones_ )

• converts semitone intervals to frequency ratios using the formula 2^(semitones/12).
• example

example
  .randsamp('808')
  .pitch(_.steps([0, 12, -12, 7]))
  .play();
  // random 808 sample at 1x pitch, 0.5x pitch, 2x pitch, 1.5x pitch

**turing** ( _length_ )

• generates a pattern of length length with random 1s and 0s.

example
  .turing(64); // instant rhythmic triggers

FacetPattern modulators

**abs** ( )

• returns the absolute value of all numbers in the FacetPattern.

example
  .sine(100)
  .add(-0.3)
  .abs()
  .play();
  // a wonky sine

**allpass** ( _frequency_ = default_sample_rate/2 )

• runs the FacetPattern through an allpass filter.
• frequency changes the amount of phase shift introduced by the filter at different frequencies. It will change the phase response of the filter while leaving the magnitude response unchanged.

 example
  .randsamp('808')
  .iter(12, () => {
    this.allpass()
      .delay(ri(1, 6000))
  })
  .scale(-1, 1)
  .play(); // reverb

**at** ( _position_, _value_ )

• replaces the value of a FacetPattern at the relative position position with value.

example
  .turing(16)
  .at(0, 1);
  // the 1st value of the 16-step Turing sequence (i.e. 0% position) is always 1
example
  .turing(16)
  .at(0.5, 2);
  // the 9th value of the 16-step Turing sequence (i.e. 50% position) is always 2

**audio** ( )

• removes any DC offset on the FacetPattern by running it through a high-pass biquadratic filter at 5Hz.

example
  .sine(440)
  .add(0.9)
  .audio()
  .play();
  // .audio() removes the DC offset that is added by the .add() command

**avg** ( )

• computes the average of the pattern.

example
  .noise(4)
  .avg()
  // [0.2349882227034955, 0.2349882227034955, 0.2349882227034955, 0.2349882227034955]

**bitshift** ( _shift_ = 16 )

• performs a bitwise rotation on the elements of the FacetPattern object’s data array by shift bits.
• shift is an optional parameter that specifies the number of bits to rotate. It defaults to 16 if not provided. The value of shift is converted to a non-negative integer and taken modulo 32 before being used.
• The method first scales the values in the data array to a range of 0 to 1000000 and rounds them to integers. It then performs a bitwise rotation on each element using a combination of the left shift (<<) and right shift (>>>) operators. Finally, it restores the original scale of the data.

example
  .sine(1000, n2)
  .bitshift(16)
  .play();
  // rotates the bits of a 1000Hz sine wave by 16 positions

**bounce** ( _num_ = 15, _sizeChange_ = 0.95, _amplitudeChange_ = 0.95 )

• creates a "bouncing ball" effect by appending multiple copies of the FacetPattern with progressively smaller sizes and amplitudes.
• num is the number of bounces to create.
• sizeChange is the factor by which the size decreases with each bounce.
• amplitudeChange is the factor by which the amplitude decreases with each bounce.

example
  .sine(440, 1000)
  .bounce(10, 0.9, 0.8)
  .play();
  // water droplet effect from a 440Hz sine wave that bounces 10 times, with each bounce 90% the size and 80% the amplitude of the previous

**changed** ( )

• returns a 1 or 0 for each value in the FacetPattern. If the value is different than the previous value, returns a 1. Otherwise returns a 0. (The first value is compared against the last value in the FacetPattern.)

example
  .from([1, 1, 3, 4])
  .changed(); // 1 0 1 1

**clip** ( _min_, _max_ )

• clips any numbers in the FacetPattern to a min and max range.

example
  .from([1, 2, 3, 4])
  .clip(2, 3);
  // 2 2 3 3

**compress** ( _ratio_, _threshold_, _attackTime_, _releaseTime_ )

• compresses the FacetPattern into a smaller dynamic range. ratio is a float between 0 and 1 corresponding to n:1 so 0.5 would be 2:1, 0.2 would be 5:1, etc. threshold is the sample amplitude at which compression kicks in. attackTime and releaseTime are expressed as relations to a second, so 0.1 would be 1/10th of a second.

example
  .randsamp('808')
  .compress(0.1, 0.001, 0.01, 0.01)
  .play();

**crab** ( )

• superposes a reversed copy of the FacetPattern on top of iself, so it plays backwards and forwards at the same time..

example
  .sine(_.ramp(20, 2000, 1000))
  .crab()
  .full()
  .play();
  // sine wave ramps from 20Hz to 2000Hz both backwards and forwards at the same time

**curve** ( _tension_ = 0.5, _segments_ = 25 )

• returns a curved version of the FacetPattern. Tension and number of segments in the curve can be included but default to 0.5 and 25, respectively.

example
  .noise(16)
  .curve(); // not so noisy
example
  .noise(16)
  .curve(0.5, 10); // fewer segments per curve
example
  .noise(16)
  .curve(0.9); // different curve type

**distavg** ( )

• computes the distance from the average of the FacetPattern, for each element in the FacetPattern.

example
  .from([0.1, 4, 3.14])
  .distavg();
  // -2.3133 1.5867 0.7267

**dup** ( _num_ )

• duplicates the FacetPattern num times.

example
  .noise(n16)
  .dup(ri(2, 12))
  .play();
  // 16th note of noise repeats between 2 and 12 times each loop 

**echo** ( _num_, _feedback_ = 0.666 )

• repeats the FacetPattern num times, with amplitude multiplied by feedback each repeat.

example
  .from([1])
  .echo(5);
  // 1 0.666 0.4435 0.29540 0.19674 0.13103
example
  .phasor(50)
  .size(n8)
  .echo(7)
  .play();
  // echoing out over a whole note 

**expo** ( _exponent_ )

• applies exponential scaling to the FacetPattern based on its minimum and maximum. exponent values larger than 1 will emphasize lower numbers more and more. Values less than 1 will emphasize higher numbers more and more.

example
  .sine(_.ramp(100, 2000, 512)
    .expo(6), n1)
  .play()
  .once();
  // experiment with different expo() values to hear the difference

**fade** ( _fade_percent_ = 0.1 )

• applies a crossfade window to the FacetPattern, where fade_percent of the beginning and end are faded in/out.

example
  .noise(1024)
  .fade()
  .play();

**fadein** ( _fade_percent_ = 0.5 )

• applies a fade to the beginning of the FacetPattern, where fade_percent of the beginning is faded in.

example
  .noise(20000)
  .fadein()
  .play();

**fadeout** ( _fade_percent_ = 0.5 )

• applies a fade to the ending 50% of the FacetPattern, where fade_percent of the beginning is faded out.

example
  .noise(20000)
  .fadeout()
  .play();

**fkey** ( _MIDI_note_scale_, _binThreshold_ = 0.005, _maxHarmonic_ = 10 )

• applies a spectral gate to the FacetPattern, muting any frequency bins that do not closely map onto a MIDI note frequency included in MIDI_note_scale.
• binThreshold controls how close a bin frequency must be to a MIDI note frequency or its harmonic in order to be kept. For example, if binThreshold is set to 0.1, then a bin frequency must be within 10% of a MIDI note frequency or its harmonic in order to be kept.
• maxHarmonic controls how many integer harmonics of MIDI notes in MIDI_note_scale to include in the output.

example
  .noise(n1)
  .times(_.ramp(1, 0, n1))
  .fkey(_.from([48, 50, 52, 53, 55, 57, 59, 60]), 0.005, 6)
  .play();
  // noise spectrally filtered to bins matching C major notes 48,50,52,53,55,57,59,60 and their 6 next harmonics

**flange** ( _delaySamples_ = 220, _depth_ = 110 )

• applies a flanger effect to the FacetPattern.
• delaySamples is the base delay in samples. Controls the delay of the flanging effect.
• depth is the maximum amount by which the delay is modulated. Controls the depth of the flanging effect.

example
  .sine(100, n1)
  .flange(220, 110)
  .play();
  // flanged whole note sine wave at 100Hz

**flipabove** ( _maximum_ )

• for all values above maximum, it returns maximum minus how far above the value was.

example
  .sine(100)
  .flipabove(0.2)
  .play();
  // wonky sine

**flipbelow** ( _min_ )

• for all values below minimum, it returns minimum plus how far below the value was.

example
  .sine(100)
  .flipbelow(0.2)
  .play();
  // inverse wonky sine

**follow** ( _attackTime_ = default_sample_rate / 10, _releaseTime_ = default_sample_rate / 4 )

• performs envelope following on a FacetPattern.
• attackTime is the attack time in samples. It controls the speed at which the envelope rises. Its default value is 100ms.
• releaseTime is the release time in samples. It controls the speed at which the envelope falls. Its default value is 250ms.

example
  .noise(n1)
  .times(_.noise(32)
    .scale(0, 1)
    .size(n1)
    .follow(n16, n16))
  .play();
  // controlling the amplitude of a whole note of noise, with 32 samples of noise sent through the envelope follower

**fracture** ( _pieces_ )

• divides and scrambles the FacetPattern into pieces pieces.

example
  .sine(100)
  .fracture(10)
  .play();
  // the sine has shattered into 10 pieces!

**ftom** ( )

• converts all values in the FacetPattern from frequency values (Hz) to MIDI note values.

example
  .ramp(1000, 250, 16)
  .ftom();
  // 83, 82, 82, 81, 80, 79, 77, 76, 75, 74, 72, 71, 69, 67, 65, 62

**full** ( )

• rescales the FacetPattern to a full dynamic range between -1 and 1, without any dynamic range compression, in a more efficient way than scale(-1,1).

example
  .noise(n2)
  .times(0.1)
  .loud()
  .play();
  // remove loud() to hear the difference

**gate** ( _threshold_, _attackSamples_, _releaseSamples_ )

• gates the incoming FacetPattern so that any values below threshold, after attackSamples have occurred, will be set to 0, until the values go back above threshold for releaseSamples.

example
  .sine(50)
  .gate(0.1, 20, 20)
  .play();

**gt** ( _amt_ )

• returns 1 for every value in the FacetPattern greater than amt and 0 for all other values.

example
  .from([0.1, 0.3, 0.5, 0.7])
  .gt(0.6);
  // 0 0 0 1

**gte** ( _amt_ )

• returns 1 for every value in the FacetPattern greater than or equal to amt and 0 for all other values.

example
  .from([0.1, 0.3, 0.5, 0.7])
  .gte(0.5);
  // 0 0 1 1

**interp** ( _weight_ = 0.5, _name_ )

• interpolates the FacetPattern with a FacetPattern. A weight of 0.5 gives equal weight to both patterns.

example
  .sine(100)
  .interp(0.5, _.randsamp('808'))
  .play();
  // 50% sine wave; 50% random sample

**invert** ( )

• computes the minimum and maximum values in the FacetPattern, then scales every number to the opposite position, relative to minimum and maximum.

example
  .from([0, 0.1, 0.5, 0.667, 1])
  .invert();
  // 1 0.9 0.5 0.333 0

**jam** ( _prob_, _amt_ )

• changes values in the FacetPattern. prob (float 0-1) sets the likelihood of each value changing. amt is how much bigger or smaller the changed values can be. If amt is set to 2, and prob is set to 0.5 half the values could have any number between 2 and -2 added to them.

example
  .drunk(128, 0.05)
  .jam(0.1, 0.7);
  // small 128 step random walk with larger deviations from the jam

**lt** ( _amt_ )

• returns 1 for every value in the FacetPattern less than amt and 0 for all other values.

example
  .from([0.1, 0.3, 0.5, 0.7])
  .lt(0.6);
  // 1 1 0 0

**lte** ( _amt_ )

• returns 1 for every value in the FacetPattern less than or equal to amt and 0 for all other values.

example
  .from([0.1, 0.3, 0.5, 0.7])
  .lte(0.5);
  // 1 1 1 0

**log** ( _intensity_ , _direction_ )

• stretches a FacetPattern according to a logarithmic curve, where the values at the end can be stretched for a significant portion of the FacetPattern, and the values at the beginning can be squished together. The intensity of the curve is controlled by intensity, which accepts a float between 0 and 1. If direction is negative, it returns the FacetPattern in reverse.

example
  .noise(n8)
  .log(rf())
  .play();
  // each time a different logarithmic curve on the 8th note of noise

**mtof** ( )

• converts all values in the FacetPattern from MIDI note values to frequency values (Hz).

example
  .from([60, 55, 76, 100])
  .mtof();
  // 261.63, 220, 659.26, 2637.02

**mtos** ( )

• converts all values in the FacetPattern from MIDI note values to samples.

example
  .noise(n4)
  .comb(_.noise(128)
    .scale(0, 127)
    .key('c', 'major')
    .mtos()
    .sort())
  .play();
  // comb filter delayed by sample values in c major on a quarter note of noise

**modulo** ( _amt_ )

• returns the modulo i.e. % amt calculation for each value in the FacetPattern.

example
  .from([1, 2, 3, 4])
  .modulo(3);
  // 1 2 0 1

**mutechunks** ( _chunks_, _prob_, _yes_fade_ = true )

• slices the input FacetPattern into chunks chunks and mutes prob percent of them. Note: this is intended for use with FacetPatterns with a large enough amount of data to be played back at audio rate. For a similar effect on smaller FacetPatterns, use prob().
• The default behavior is for each chunk to be slightly faded in and out to prevent audible clicks, but if you want to run this on smaller chunks of information for control-signal purposes, you can supply a falsy yes_fade argument.

example
  .randsamp('808')
  .mutechunks(16, 0.33)
  .play();
  // 33% of 16 audio slices muted

**nonzero** ( )

• replaces all instances of 0 with the previous nonzero value. Useful after with probability controls, which by default will set some values to 0. Chaining a nonzero() after that would replace the 0s with the other values the pattern. Particularly in a MIDI context with .prob(), you probably don't want to send MIDI note values of 0, so this will effectively sample and hold each nonzero value, keeping the MIDI note values in the expected range.

example
  .from([1, 2, 3, 4])
  .prob(0.5)
  .nonzero();
  // if 2 and 4 are set to 0 by prob(0.5), the output of .nonzero() would be 1 1 3 3

**norepeat** ( )

• rearranges the values in the FacetPattern to minimize consecutive repeating values.

example
  .from([1, 1, 1, 2, 2, 3])
  .norepeat();
  // 1 2 1 2 1 3

**normalize** ( )

• scales the FacetPattern to the 0 - 1 range.

example
  .sine(1)
  .times(4000)
  .normalize();
  // the *4000 gain is undone!
example
  .sine(1)
  .scale(-10, 10)
  .normalize();
  // works with negative values

**palindrome** ( )

• returns the original FacetPattern plus the reversed FacetPattern.

example
  .from([0, 1, 2, 3])
  .palindrome();
  // 0 1 2 3 3 2 1 0

**pow** ( _expo_, _direction_ = 1 )

• stretches a FacetPattern according to an exponential power expo, where the values at the beginning can be stretched for a significant portion of the FacetPattern, and the values at the end can be squished together. If direction is negative, returns the FacetPattern in reverse.

example
  .sine(100)
  .pow(6.5)
  .play();
  // squished into the end
example
  .sine(100)
  .pow(6.5, -1)
  .play();
  // squished at the beginning

**prob** ( _amtPattern_ )

• sets some values in the FacetPattern to 0. amtPattern (clipped to range 0-1) sets the likelihood of each value changing.

example
  .from([1, 2, 3, 4])
  .prob(0.5);
  // 1 0 3 0 first time it runs
example
  .from([1, 2, 3, 4])
  .prob(0.5);
  // 0 0 3 4 second time it runs
example
  .from([1, 2, 3, 4])
  .prob(0.5);
  // 0 2 3 4 third time it runs

**quantize** ( _resolution_ )

• returns 0 for every step in the FacetPattern whose position is not a multiple of resolution.

example
  .drunk(16, 0.5)
  .quantize(4);
  // 0.5241 0 0 0 0.7420 0 0 0 1.0 0 0 0 0.4268 0 0 0

**range** ( _new_start_, _new_end_ )

• returns the subset of the FacetPattern from the relative positions of new_start (float 0-1) and new_end (float 0-1).

example
  .from([0.1, 0.2, 0.3, 0.4])
  .range(0.5, 1);
  // 0.3 0.4

**rangesamps** ( _start_, _length_ )

• returns a subset of the FacetPattern, using a relative start position (between 0 - 1) and a total length in samples.

example
  .sine(n1)
  .log(0.9)
  .rangesamps(rf(0, 0.875), n8)
  .play();
  // plays a different 8th note from the same de-pitched sine wave every time
example
  .silence(n1)
  .iter(128, () => {
    this.sup(_.noise(n64)
      .lpf(_.ramp(250, 40, 20), 50)
      .times(_.ramp(1, 0, n64))
      .rangesamps(rf(), n64)
      .fade(0.1), rf())
  })
  .play();
  // granular synthesis of 128 synthesized kick drums

**rechunk** ( _chunks_, _probability_ )

• slices the input FacetPattern into chunks chunks and shuffles the chunks around. The probability argument controls the percentage of chunks to reorder, and it expects a float between 0 and 1.
• The default behavior is for each chunk to be slightly faded in and out to prevent audible clicks, but if you want to run this on smaller chunks of information for control-signal purposes, you can supply a falsy yes_fade argument.

example
  .randsamp('808')
  .rechunk(16)
  .play();
  // 16 slices from the sample in random order

**reduce** ( _new_size_ )

• reduces the FacetPattern length to new_size. If new_size is larger than the FacetPattern length, no change.

example
  .from([1, 2, 3, 4])
  .reduce(2);
  // 1 3

**replace** ( _original_value_, _new_value_ )

• replaces all instances of original_value with new_value in the FacetPattern.

example
  .from([42, 0, 0, 36])
  .replace(0, -1);
  // 42,-1,-1,36

**resonate** ( _baseFrequency_, _coefficients_, _q_ = 80, _wet_ = 1 )

• resonates the FacetPattern running it through parallel bandpass filters. Each number in the coefficients FacetPattern is multiplied by the baseFrequency to determine the frequency for that bandpass filter.

example
  .noise(n16)
  .times(_.ramp(1, 0, n16))
  .resonate(mtof(36), _.ramp(1, 20, 20), 80)
  .play();
  // 16th note transient noise burst, resonating at its first 20 harmonics starting at 65.41 Hz (MIDI note C2, mtof(36))

**reverb** ( _size_ = 1, _feedback_ = 0.85 )

• applies the Schroeder reverb algorithm to the FacetPattern. The size argument should be between 0 and 2 for most use cases but can go up to 10.
• the feedback argument controls feedback in the reverb algorithm. It should be between 0 and 0.98.

example
  .randsamp('808')
  .reverb(rf())
  .play();
  // different reverb size for random sample each loop

**reverse** ( )

• returns the reversed FacetPattern.

example
  .ramp(0, 1, 128)
  .reverse();
  // goes from 1 to 0 over 128 values

**round** ( )

• rounds all values in the FacetPattern to an integer.

example
  .from([0.1, 0.5, 0.9, 1.1])
  .round();
  // 0 1 1 1

**saheach** ( _n_ )

• samples and holds every nth value in the FacetPattern.

example
  .noise(6)
  .saheach(2);
  // 0.33173470944031735, 0.33173470944031735, 0.17466890792169742, 0.17466890792169742, 0.5601080880419886,  0.5601080880419886  

**size** ( _new_size_ )

• upscales or downscales the FacetPattern prior to playback, so its length is new_size samples.

example
  .noise(1000)
  .size(n1)
  .play();
  // upscaling 1000 samples of noise to be 1 whole note long

**scale** ( _new_min_, _new_max_, _exponent_ = 1 )

• moves the FacetPattern to a new range, from new_min to new_max, with exponent allowing for nonlinear transformations.
• exponent values larger than 1 will emphasize lower numbers more and more. Values less than 1 will emphasize higher numbers more and more.
• NOTE: this method will return the average of new_min and new_max if the FacetPattern is only 1 value long. since you cannot interpolate where the value would fall in the new range, without a larger FacetPattern to provide initial context of the value's relative position. This operation works better with sequences larger than 3 or 4.

example
  .sine(10, 100)
  .scale(0, 1);
  // unipolar signal

**shift** ( _amt_ )

• moves the FacetPattern to the left or the right. amt gets wrapped to values between -1 and 1, since you can't shift more than 100% left or 100% right.

example
  .from([1, 2, 3, 4])
  .shift(-0.5);
  // 3 4 2 1

**shuffle** ( _prob_ = 1 )

• randomizes the order of the elements in the FacetPattern.
• The prob argument controls the percentage of data to shuffle. It should be a float between 0 and 1. A prob of 1 means 100% of the elements will shuffle; a prob of 0.5 means 50% of the elements will shuffle, etc.

example
  .from([1, 2, 3, 4])
  .shuffle();
  // first time: 3 2 1 4
example
  .from([1, 2, 3, 4])
  .shuffle();
  // second time: 1 3 4 2

**skip** ( _prob_ )

• Sometimes, skip executing the command, as if it had never been attempted. Useful if you only want to update the FacetPattern some of the time, but otherwise want to preserve the previous data.

example
  .spiral(16, random(1, 360))
  .skip(0.95);
  // new pattern 5% of the time when this command runs

**slew** ( _depth_ = 25, _up_speed_ = 1, _down_speed_ = 1 )

• adds upwards and/or downwards slew to the FacetPattern. depth controls how many slew values exist between each value. up_speed and down_speed control how long the slew lasts: at 0, the slew has no effect, whereas at 1, the slew occurs over the entire depth between each FacetPattern value.

example
  .from([0, 0.5, 0.9, 0.1])
  .slew(25, 0, 1)
  // the first three numbers will jump immediately because upwards slew is 0. then it will slew from 0.9 to 0.1 over the course of the entire depth range

**smooth** ( )

• interpolates each value so it falls exactly between the values that precede and follow it.

example
  .noise(64)
  .smooth();
  // less noisy

**sort** ( )

• returns the FacetPattern ordered lowest to highest.

example
  .noise(128)
  .sort();
  // ascending values originally from noise

**speed** ( _amt_ )

• increases or decreases the playback speed of the FacetPattern, similar to transposing audio samples up or down. An amt value of 0.5 will play at half speed. An amt value of 2 will play at double speed.

example
  .randsamp('808')
  .speed(0.2); // slow sample
example
  .randsamp('808')
  .speed(1.5); // fast sample

**sticky** ( _amt_ )

• samples and holds values in the FacetPattern based on probability. amt (float 0-1) sets the likelihood of each value being sampled and held.
• example

example
  .noise(n4)
  .sticky(0.98);
  // quarter note of "sticky" noise

**stretchto** ( _num_samples_ )

• time-stretches the FacetPattern while preserving pitch so it now lasts num_samples samples.

example
  .sine(1000, n2)
  .stretchto(n1)
  .play();
  // 1000Hz sine wave originally a half note long, stretched to a whole note

**stutter** ( _number_of_repeats_ )

• creates _number_of_repeats_ identical chunks of data.
• example

example
  .iter(8, () => {
    this.sup(_.randsamp('808')
      .stutter(8), i / iters)
  })
  .play();
  // 8 random 808 samples, each stuttered 8 times

**subset** ( _percentage_ )

• returns a subset of the FacetPattern with percentage% values in it.

example
  .phasor(1)
  .size(50)
  .subset(0.3);
  // originally 50 values long, now 0.02 0.08 0.50 0.58 0.62 0.700 0.76 0.78 0.92

**truncate** ( _length_ )

• truncates the FacetPattern so it's now length values long. If length is longer than the FacetPattern, return the whole FacetPattern.

example
  .from([0, 1, 2, 3])
  .truncate(2);
  // now 2 values long
example
  .from([0, 1, 2, 3])
  .truncate(6);
  // still 4 values long

**tune** ( _note_ = "c", _binThreshold_ = 0.005 )

• applies a spectral gate to the FacetPattern, muting any frequency bins that do not closely map onto the supplied note.
• note values: "A", "A#", "B", "C", "C#", "D", "D#", "E", "F", "F#", "G", "G#"
• binThreshold controls how close a bin frequency must be to the supplied note's harmonics in order to be kept. For example, if binThreshold is set to 0.1, then a bin frequency must be within 10% of a harmonic in the supplied note in order to be kept.

example
  .noise(n1)
  .tune('c', 0.0001)
  .play();
  // tuning a whole note of noise to c

**unique** ( )

• returns the set of unique values in the FacetPattern.

example
  .from([1, 2, 3, 0, 0.4, 2])
  .unique();
  // 1 2 3 0 0.4

**walk** ( _prob_, _amt_ )

• changes positions in the FacetPattern. prob (float 0-1) sets the likelihood of each position changing. amt controls how many steps the values can move. If amt is set to 10, and prob is set to 0.5 half the values could move 10 positions to the left or the right.

example
  .from([0, 1, 2, 0, 1, 0.5, 2, 0])
  .walk(0.25, 3);

**wrap** ( _min_, _max_ )

• folds FacetPattern values greater than max so their output continues at min. If the values are twice greater than max, their output continues at min again. Similar for values less than min, such that they wrap around the min/max thresholds.
• if no value is entered for max, then the first argument will be used to create the min and max, centered around 0. For instance, wrap(0.3) == wrap(-0.3,0.3)

example
  .sine(100)
  .add(-0.1)
  .wrap(0.2, 0.5)
  .play();

Pattern modulators that can take a FacetPattern, number, or array as an argument

When a modulator takes a FacetPattern or an array as an argument, it uses that pattern to dynamically change its behavior over time, affecting the output in a more complex way than if a single number were supplied. For example, with the command example.noise(16).add(4), all 16 output values will be between 3 and 5, because 4 is added to every noise value, and noise values are between -1 and 1 by default. But with the command example.noise(16).add(_.ramp(0,4,16)), the output values will ramp from between [-1, 1] at the beginning to between [4, 5] at the end, since the FacetPattern that is being added is a ramp of values starting at 0 and ending at 4.

**add** ( _FacetPattern_, _match_sizes_ = true )

• adds the first FacetPattern and the second FacetPattern. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .randsamp('808')
  .add(_.randsamp('808'))
  .play();
  // two random samples each loop

**bpf** ( _cutoffPattern_ = 1000, _qPattern_ = 2.5 )

• applies a bandpass filter with configurable cutoffPattern and q to the FacetPattern.

example
  .noise(n1)
  .bpf(1000, 6)
  .times(0.1)
  .play();
  // band-passed noise
example
  .noise(n1)
  .bpf(_.sine(4)
    .scale(10, 1000))
  .play();
  // 4-cycle LFO modulating the bandpass cutoff between 10 and 1000 Hz

**comb** ( _delaySamplesPattern_ = sample_rate / 100, _feedforward_ = 0.5, _feedback_ = 0.5 )

• applies a comb filter to the input data. The _delaySamplesPattern_ parameter is equal to 10ms by default and specifies the number of samples to delay the input signal. The feedforward parameter controls the amount of the input signal that is fed directly to the output. The feedback parameter controls the amount of feedback applied to the delay, allowing the delayed signal to be mixed back into the input.
• The feedback and feedforward values are clamped between 0 and 0.98.

example
  .noise(n4)
  .comb(ms(10), 0.5, 0.5)
  .play();

**crush** ( _numberOfBitsPattern_, _downsamplingPattern_ )

• applies bit crushing and / or downsampling to the incoming FacetPattern.
• numberOfBitsPattern controls the bit depth for the output pattern. To hear the effect, the values need to be integers between 1 and 8. Lower values produce more drastic results.
• downsamplingPattern controls the fator by which to reduce the sample rate. Values need to be integers greater than 1. Higher values produce more drastic results.

example
  .sine(100)
  .crush(2)
  .play();
  // redux on the sine wave
example
  .sine(100, n1)
  .crush(_.ramp(8, 1, 8))
  .play();
  // ramping bit depth on 100Hz sine wave from 8 bits to 1
example
  .sine(100, n1)
  .crush(_.ramp(8, 1, 8), _.noise(16)
    .scale(1, 40))
  .play();
  // ramping bit depth on 100Hz sine wave from 8 bits to 1, and dynamically changing the downsampling amount between 1 and 40 samples

**delay** ( _delaySamplesPattern_, _feedback_ = 0.5, _byPassSafety_ = false )

• delays the input FacetPattern by delaySamplesPattern samples. The feedback parameter controls the amount of feedback applied to the delay, allowing the delayed signal to be mixed back into the input.
• the maximum feedback value is 0.975, unless byPassSafety is set to a truthy value.

example
  .randsamp('808')
  .delay(random(1700, 10000))
  .play();

**divide** ( _FacetPattern_, _match_sizes_ = true )

• divides the first FacetPattern by the second. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .sine(1)
  .divide(_.from([0.5, 0.25, 0.1, 1]));

**ffilter** ( _minFreqPattern_, _maxFreqPattern_, _invertMode_ = false)

• applies a spectral filter to the FacetPattern, passing only the frequency bins between minFreqPattern and maxFreqPattern.
• you can invert the filter mode so it cuts all the bins between minFreqPattern and maxFreqPattern by sending a truthy value for invertMode.

example
  .noise(n16)
  .ffilter(200, 2000)
  .play();
  // noise between 200Hz - 2000Hz

**fgate** ( _gateThresholdPattern_ = 0.1, _invertMode_ = false )

• applies a spectral gate to the FacetPattern, muting any frequency bins lower than gateThresholdPattern. The magnitudes of each FFT bin are normalized from 0 - 1. A gateThresholdPattern of 0 will pass every bin, and a gateThresholdPattern of 1 will mute every bin.
• you can invert the gate mode so it keeps only bins with magnitudes lower than gateThresholdPattern.

example
  .noise(n16)
  .fgate(0.7)
  .play();
  // try experimenting with different threshold values

**flookup** ( _lookupPattern_ )

• applies a spectral bin rearrangement to the FacetPattern based on the lookupPattern. Similar to ichunk() but using FFT. The lookupPattern maps the current spectral bins to new positions. The length of the lookupPattern determines the number of frames in the resynthesized signal.

example
  .randsamp('808')
  .flookup(_.ramp(1, 0, SAMPLE_RATE))
  .play()
  .once();
  // backwards 808 sample, ramping from relative position 1 to 0

**fshift** ( _shiftAmountPattern_ )

• applies a spectral bin shift to the FacetPattern. shiftAmountPattern values lower than 0 will cause the bottom to wrap the top, and the rest of the spectrum moves downwards. shiftAmountPattern values higher than 0 will cause the top of the spectrum to wrap to the bottom, and the rest of the spectrum moves upwards.

example
  .sine(100)
  .fshift(0.04)
  .play();
  // try experimenting with different shift values

**ftilt** ( _tiltAmountPattern_ )

- applies spectral harmonic tilting to the FacetPattern. This method allows for the rotation of the harmonic bands of the signal, enabling each band to be repositioned independently in time. The `tiltAmountPattern` should be normalized to values between -1 and 1, where -1 represents a full counter-clockwise rotation and 1 represents a full clockwise rotation.

example
  .sample('808/808-Clap03')
  .ftilt(_.ramp(rf(), rf(), 100))
  .play();
  // split the clap sample into 100 frequency bands and disperse them randomly in time

**harmonics** ( _numHarmonicsPattern_ )

• adds numHarmonicsPattern harmonics to the input signal.

example
  .sine(10)
  .harmonics(200)
  .play();
  // 10Hz sine wave with 200 harmonics added on top
example
  .sine(10, n1)
  .harmonics(_.ramp(0, 200, 200))
  .play();
  // ramping up from 0 harmonics on the 10Hz wave to 200 harmonics

**hpf** ( _cutoffPattern_ = 100, _qPattern_ = 2.5 )

• applies a high pass filter with configurable cutoffPattern and q to the FacetPattern.

example
  .noise(n1)
  .hpf(2000, 6)
  .times(0.1)
  .play();
  // high-passed noise
example
  .noise(n1)
  .hpf(_.sine(4)
    .scale(10000, 20000))
  .play();
  // 4-cycle LFO modulating the high pass cutoff between 10000 and 20000 Hz

**lpf** ( _cutoffPattern_, _qPattern_ = 2.5 )

• applies a low pass filter with configurable cutoffPattern and q to the FacetPattern.

example
  .noise(n1)
  .lpf(1000, 6)
  .times(0.1)
  .play();
  // low-passed noise
example
  .noise(n1)
  .lpf(_.sine(4)
    .scale(10, 2000))
  .play();
  // 4-cycle LFO modulating the high pass cutoff between 10 and 2000 Hz

**onepole** ( _cutoffPattern_ )

• applies a one-pole low pass filter with 6dB/octave rolloff to the FacetPattern.

example
  .noise(n1)
  .onepole(100)
  .full()
  .play();
  // gently low-passed noise with 6dB/octave rolloff
example
  .noise(n1)
  .lpf1(_.sine(4)
    .scale(20, 1000))
  .play();
  // 4-cycle LFO modulating the one-pole filter cutoff

**pitch** ( _pitchShiftPattern_ )

• pitch-shifts the FacetPattern. pitchShiftPattern values between 0 and 1 will lower the pitch; e.g. a value of 0.5 will shift it down an octave. Values higher than 1 will increase the pitch; e.g. a value of 2 will be an octave higher.

example
  .sine(100)
  .shift(rf(0.5, 2));
  // sometimes lower pitch, sometimes higher pitch
example
  .sine(100)
  .pitch(_.noise(16)
    .scale(0.5, 2))
  .play();
  // pitch shifts a 100Hz wave at 16 places, sometimes lower and sometimes higher

**stretch** ( _shiftAmountPattern_, _chunksPerSecondPattern_ = 128 )

• time-stretches the FacetPattern while preserving pitch. shiftAmountPattern values less than 1 will shorten its overall length; values greater than 1 will increase its length. chunksPerSecondPattern is the number of chunks that the timestretching algorithm will generate per second. Smaller values will produce more discrete repetitions; larger values will produce more of a bitcrushing, harmonic distortion effect. The largest chunksPerSecondPattern value is SAMPLE_RATE / (SAMPLE_RATE * 0.002), which is 500 a sample rate of 44100.

example
  .sine(100, n4)
  .stretch(4)
  .play();
  // stretching a quarter note sine wave to last a whole note
example
  .noise(n1)
  .stretch(_.ramp(0.125, 4, 16))
  .play()
  .once();
  // stretching a whole note of noise over 16 ramped values, starting at 8x faster and ending at 4x slower

**subtract** ( _FacetPattern_, _match_sizes_ = true )

• subtracts the second FacetPattern from the first. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .sine(100)
  .subtract(_.cosine(50))
  .play();

**tanh** ( _gainPattern_ = 20 )

• outputs the hyperbolic tangent method for the input FacetPattern, always returning values between -1 and 1. Higher gainPattern values will create more intense distortion.

example
  .phasor(1, 20)
  .times(10)
  .tanh(6);
  // 0 0.995 0.9999 0.99999996 0.9999999999 0.999999999999 0.9999999999999996 1 1 1 1 1 1 1 1 1 1 1 1 1
example
  .sine(100)
  .tanh(_.ramp(0, 100, 100))
  .play();
  // ramping tanh distortion up on a 100Hz sine wave

**times** ( _FacetPattern_, _match_sizes_ = true)

• multiplies the first FacetPattern by the second. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .sine(50)
  .times(_.sine(50))
  .play();

Pattern modulators that must take a second FacetPattern as an argument

**and** ( _FacetPattern_, _match_sizes_ = true )

• computes the logical AND of both FacetPattern, returning a 0 if one of the values is 0 and returning a 1 if both of the values are nonzero. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .from([1, 0, 1, 0])
  .and(_.from([0, 1]));
  // 0 0 1 0

**append** ( _FacetPattern_ )

• concatenates the second FacetPattern onto the first.

example
  .sine(1)
  .append(_.phasor(1))
  .append(_.from([1, 2, 3, 4]));

**chaos** ( _FacetPattern_, _iterations_ = 100, _cx_ = 0, _cy_ = 0)

• each piece of data in the FacetPattern is paired with the corresponding value in the second FacetPattern. The resulting complex number x,y coordinate is run through a function: f(x) = x2 + c, over iterations iterations. The output is a value between 0 and 1, which corresponds to how stable or unstable that particular point is in the complex number plane.
• By default, both cx and cy are set to 0 (Mandelbrot set). But you can set them to other values from -1 to 1, which can produce all sorts of Julia set variations.

example
  .sine(n1)
  .chaos(_.drunk(n1, 0.01))
  .play();

// generate a 1000px * 1000px mandelbrot set between [-2,2]
example
  .ramp(-2, 2, 1000)
  .setlocal('x')
  .ramp(-2, 2, 1000)
  .setlocal('y')
  .reduce(0)
  .iter(1000, () => {
    this.append(_.getlocal('x')
      .chaos(_.from(_.getlocal('y')
        .data[i]))
      .size(1000))
  })
  .saveimg()
  .once();

**convolve** ( _FacetPattern_ )

• computes the convolution between the two FacetPatterns.

example
  .randsamp('808')
  .convolve(_.randsamp('808'))
  .play();
  // convolving random samples

**equals** ( _FacetPattern_, _match_sizes_ = true )

• computes the logical EQUALS of both FacetPattern, returning a 0 if the values don't equal each other and returning a 1 if they do. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .sine(1)
  .equals(_.sine(2));

**ichunk** ( _FacetPattern_ )

• slices the input into FacetPattern.length windowed chunks (to avoid audible clicks). Loops through every value of FacetPattern as a lookup table, determining which ordered chunk of audio from the input sequence it corresponds to, and appends that window to the output buffer.

example
  .randsamp('808')
  .ichunk(_.ramp(rf(), rf(), 256))
  .play();
  // play 256 slices between two random points of a random sample... timestretching :)

**interlace** ( _FacetPattern_ )

• interlaces two FacetPatterns. If one FacetPattern is smaller, it will be interspersed evenly throughout the other FacetPattern.

example
  .sine(1)
  .interlace(_.phasor(1, 20));

**map** ( _FacetPattern_ )

• forces all values of the input FacetPattern to be mapped onto a new set of values from a second FacetPattern.**

example
  .from([1, 2, 3, 4])
  .map([11, 12, 13, 14]);
  // 11 11 11 11
example
  .from([1, 2, 3, 4])
  .scale(30, 34)
  .map(_.from([31, 31.5, 32, 32.5]));
  // 31 31.5 32.5 32.5

**or** ( _FacetPattern_, _match_sizes_ = false )

• computes the logical OR of both FacetPattern, returning a 0 if both of the values are 0 and returning a 1 if either of the values are nonzero. If match_sizes is false, the output FacetPattern will be the longer pattern's length, and the "missing" values from the shorter pattern will be set to 0. If match_sizes is true, both FacetPatterns will be made the same size before the calculations occur.

example
  .from([1, 0, 1, 0])
  .or(_.from([0, 1]));
  // 1 0 1 1

**sieve** ( _FacetPattern_ )

• uses the second FacetPattern as a lookup table, with each value's relative value determining which value from the input sequence to select.

example
  .noise(1024)
  .sieve(_.sine(10));
  // sieving noise with a sine wave into the audio rate :D

**splice** ( _FacetPattern_, _position_ )

• inserts the second FacetPattern into the input FacetPattern at relative position between 0 and 1.

example
  .randsamp('808')
  .splice(_.noise(n16), 0.5)
  .play();
  // inserts a 16th note of noise halfway through the random sample

**sup** ( _FacetPattern_, _startPositionPattern_, _maxFrameSize_ = whole_note_samples )

• superposes a second FacetPattern onto the first. The startPositionPattern value can be any value between 0 and 1, or an array, or a FacetPattern. It controls the relative position(s) in the input FacetPattern to begin superposing FacetPattern. The maxFrameSize value specifies the farthest sample value from the first FacetPattern, which would be equal to a startPosition of 1.

example
  .silence(n1)
  .sup(_.randsamp('808'), 0, n1)
  .sup(_.randsamp('808'), 0.5, n1)
  .play();
  // superpose two samples at the 0% and 50% points through each loop

**vocode** ( _carrierPattern_, _bands_ = [1000,2000,3000,4000,5000,6000,7000,8000,9000,1000], _q_ = 2.5 )

• creates a vocoder effect where the amplitude envelope of each frequency bin in the input FacetPattern controls the amplitude of each frequency bin in carrierPattern.
• for a "classic" vocoding effect, use a rhythmic sample as the input FacetPattern and a melodic pattern for carrierPattern.
• the Q argument controls the Q factor of the bandpass filters in the bands pattern.

example
  .seq('808/* 808/* 808/* 808/* 808/* 808/* 808/* 808/*')
  .vocode(_.square([220, 440, 110, 110]))
  .play();
  // vocode sequence of random 808 sample with simple square wave pattern

Pattern modulators with a function as one of the arguments

For more examples, refer to the examples/this.md file.

**mix** ( _wet_, _command_ = function )

• Mixes the input FacetPattern with a second FacetPattern generated by command.
• The command that will be mixed must start with the reserved word: this (see example).

example
  .randsamp('808')
  .mix(0.5, () => {
    this.reverse()
      .speed(2)
      .echo(8)
      .speed(10)
  })
  .play();

**iter** ( _num_times_, _command_ = function, _prob_ = 1 )

• A shorthand for rerunning a certain command over and over, with prob as a float between 0 and 1 controlling the likelihood that the code actually runs.
• You can refer to the current iteration of the algorithm via the reserved word: this (see example).
• The variable i, referring to the current iteration number starting at 0, is also available for use in commands.
• The variable iters, referring to the total number of iterations, is also available for use in commands.
• The variable this.original_data, referring to the original data before any iterations are proessed, is also available for use in commands.

example
  .randsamp('808')
  .iter(8, () => {
    this.delay(ri(1, 2000))
  })
  .play();
  // 8 delay lines between 1 and 2000 samples

**parallel** ( _commands_ = [function, function] )

• applies multiple commands in parallel to the input FacetPattern. The commands parameter is an array where each entry is a function. Each command is applied to a copy of the original input data, and the results are combined back together afterwards. The final output is normalized to have the same maximum value as the original input data.

example
  .noise(n4)
  .scale(-1, 1)
  .allpass(347)
  .allpass(113)
  .allpass(37)
  .parallel([() => {
    this.delay(1687, 0.975)
  }, () => {
    this.delay(1601, 0.975)
  }, () => {
    this.delay(2053, 0.975)
  }, () => {
    this.delay(2251, 0.975)
  }])
  .play()
  .full();
  // schroeder reverb on a quarter note of noise

**run** ( _commands_ = function )

• equivalent to a sometimes(1,()=>{}) method. See sometimes() below for more details.
• allows you to run arbitrary JS code on the FacetPattern every loop.

example
  .noise(n1)
  .run(() => {
    if (bars % 2 == 0) {
      this.tune('c')
    } else {
      this.tune('g')
    }
  })
  .play();
  // alternating c and g whole notes made from tuned noise

**seq** ( _sequencePattern_, _commands_ = function, _maxFrameSize_ = getWholeNoteNumSamples() )

• superposes the samples specified in sequencePattern across the loop. sequencePattern can either be a string or a FacetPattern composed of strings.
• the character * at the end of a member of the sequencePattern string will select a random sample from that directory (see examples).
• the character _ in a sequencePattern specifies to insert silence instead of a sample.
• the commands will run on each sample as it is superposed onto the output pattern.
• maxFrameSize controls the length across which the sequence is spread. Defaults to the whole note duration in samples.
• example:

example
  .seq('kicks/* hats/* snares/003 hats/003')
  .play();
  // random kick, random hat, snares/003, hats/003
example
  .seq(_.from(['kicks/003', 'hats*', 'snares/003', 'hats/*'])
    .dup(choose([1, 3, 5, 7]))
    .palindrome()
    .rechunk(8, 0.5), () => {
      this.log(rf())
        .delay(ri(n128, n16))
    })
  .full()
  .play()
  // example using commands to proess each sample, and using a FacetPattern as the sequencePattern

**slices** ( _num_slices_, _command_ = function, _prob_ = 1, _fade_mode_ = true )

• slices the FacetPattern into num_slices slices, and for prob percent of those slices, runs command, appending all slices back together. You can refer to the current slice of the algorithm via the reserved word: this (see example).
• The variable s, referring to the current slice number starting at 0, is also available for use in commands.
• The variable num_slices, referring to the number of slices, is also available for use in commands.
• If the FacetPattern's data is >= 1024 samples, the last 1% of each slice will be faded out to prevent clicks in audio slices. If the FacetPattern's data is < 1024 samples, no fading is applied, and each slice is processed exactly as-is.
• By default, the slices() method applies a very short fadeout on each slice to prevent any clicks that might occur from any code running on each slice. This behavior can be turned off by including a falsy fade_mode value.

example.noise(n1)
  .slices(32, () => {
    this.times(rf()).log(rf())
  })
  .play();
  // whole note of noise into 32 slices, each randomized amplitude and randomly warped with log()

**sometimes** ( _prob_, _command_ = function() )

• runs command only some of the time, at a probability set by prob.
• command must start with the reserved word: this (see example).

example
  .tri(100)
  .times(_.ramp(1, 0, 1000)
    .expo(8))
  .truncate(n4)
  .dup(3)
  .sometimes(0.5, () => {
    this.reverse()
  })
  .play();
  // sometimes backwards 100Hz triangle wave, sometimes forwards

**spread** ( _iterations_, _command_ = function(), _startRelativePosition_ = 0, _endRelativePosition_ = 1, _skipIterationsPattern_ = [])

• generates iterations number of FacetPatterns using command and spreads them evenly across a range of the loop, specified by startRelativePosition and endRelativePosition.
• command must start with the reserved word: this (see example).
• the variable i, referring to the current iteration number starting at 0, is also available for use in commands.
• the variable iters, referring to the total number of iterations, is also available for use in commands.
• the FacetPattern skipIterationsPattern specifies any iterations that should be skipped. By default, it is empty, e.g. no iterations skip. For example, a skipIterationsPattern of _.from([0,4,7,10]) would skip the 0th, 4th, 7th, and 10th iterations.

example
  .spread(8, () => {
    this.randsamp('808')
  }).play();
  // 8 random 808 samples across the whole loop
  example
  .spread(8, () => {
    this.randsamp('808')
      .truncate(n32)
      .delay(n128, 0.1)
  }, 0, 0.25)
  .sup(_.spread(16, () => {
    this.randsamp('808')
  }, 0, 1))
  .sup(_.spread(32, () => {
    this.randsamp('808')
  }, 0.75, 1))
  .play();
  // 8 random 808 samples across the first 25% of the loop
  // 16 random 808 samples across the whole loop
  // 32 random 808 samples across the last 25% of the loop

**subrange** ( _min_, _max_, _command_ = function() )

• runs command on a subrange of the FacetPattern, specified by min and max.
• min and max should be floats between 0 and 1. They are relative values, so a value of 0 means the beginning of the pattern, and a value of 1 means the end of the pattern.
• command must start with the reserved word: this (see example).

example
  .tri(200)
  .subrange(0.33, 0.66, () => {
    this.times(_.from([1,0,0,0,0,0,0,1]).curve())
  })
  .play();
  // quieter in the middle third

Methods for image generation and processing

Because Facet generates and modifies a 1-dimensional array of data, it is also possible to generate images from data. NOTE: by default, imaegs will be saved with square dimensions, with the FacetPattern's data going across the image, row by row. You can set a custom width and height for the image the via .dim() method.

**blur2d** ( )

• applies a 3x3 box blur filter to the 2D image data.
• averages each pixel with its 8 surrounding neighbors to create a smoothing effect.

example
  .noise(1000000)
  .iter(7,()=>{this.blur2d()})
  .saveimg('blur2d')
  .once();
  // blurred noise

**circle2d** ( _centerX_, _centerY_, _radius_, _value_, _fillMode_ = 0 )

• adds a circle on top of the existing data in a FacetPattern.
• centerX and centerY are the x,y coordinates of the center of the circle.
• radius controls the radius of the circle.
• value is brightness value for the circle normalized between 0 - 1.
• fillMode (0 or 1) controls whether the inside of the shape is filled in with value or ignored. Default is no fill.

example
  .silence(1000000)
  .circle2d(100, 100, 100, 1)
  .saveimg('example_circle');
  // white circle in a 1000x1000 image
example
  .silence(1000000)
  .circle2d(100, 100, 100, 1, 1250, 800)
  .saveimg('circle2d', [1, 1, 1], 1250, 800)
  .once();
  // white circle in a 1250x800 image

**columns2d** ( _num_columns_, _command_ )

• slices the FacetPattern into num_columns columns in a 2D grid, and for each slice, runs command, appending all columns back together. You can refer to the current column of the algorithm via the reserved word: this (see example).
• the variable c, referring to the current column number starting at 0, is also available for use in commands.
• the variable num_columns, referring to the number of columns, is also available for use in commands.
• the columns2d() method does not apply any fading on each slice. Each slice is processed exactly as-is.
• the command is a function that is applied to each slice. The function is converted to a string and any reference to this is replaced with current_slice to ensure the function operates on the correct data.
• the method divides the FacetPattern into a grid of slices, each slice being a square segment of the original pattern.

example
  .sine(0.3, 1000*1000)
  .scale(0, 1)
  .columns2d(1000,()=>{this.times(c/999)})
  .saveimg('columns2d')
  .once();

**delay2d** (_delayX_, _delayY_, _intensityDecay_ = 0.5 )

• applies a delay effect to the data in a FacetPattern in 2 dimensions.
• delayX and delayY are the delay amounts in the x and y directions. Positive values move right/down; negative values move left/up.
• intensityDecay is a value between 0 and 1 that controls how much the intensity of the data decays with each delay step. A value of 0.5 means the intensity is halved with each delay step.

example
  .silence(1000000)
  .circle2d(500, 500, 250, 1)
  .delay2d(20, 20, 0.85)
  .saveimg('delay2d')
  .once();
  // circle echoing down-left

**dim** (_imageWidth_, _imageHeight_)

• sets image dimensions so that when an image is saved via .saveimg(), its size will be imageWidth pixels wide by imageHeight pixels.

example
  .noise(100 * 2000)
  .dim(100, 2000)
  .saveimg()
  .once();
  // 100px * 2000px image of noise

**draw2d** ( _coordinates_, _fillValue_ )

• draws a polygon on the FacetPattern using the provided coordinates and fill value. The polygon is drawn by connecting each pair of consecutive points in the coordinates array with a line filled with the fill value.
• the coordinates parameter is an array of x and y coordinates for the vertices of the polygon. The array length must be divisible by 2.
• the fillValue parameter is the value used to fill the lines of the polygon.

example
  .silence(1000000)
  .iter(32, () => {
    this.draw2d([ri(0, 999), ri(0, 999), ri(0, 999), ri(0, 999)], 1)
  })
  .saveimg()
  .once();
  // draw 32 lines from 32 random points inside a 1000px * 1000px square, with a fill value of 1

**gradient2d** ( _centerX_, _centerY_, _radius_, _innerValue_, _outerValue_, _mode_ = 1 )

• adds a gradient circle on top of the existing data in a FacetPattern, with smooth interpolation between inner and outer values.
• centerX and centerY are the x,y coordinates of the center of the gradient circle.
• radius controls the radius of the circle.
• innerValue is the brightness value(s) at the center of the circle (can be number, array, or FacetPattern).
• outerValue is the brightness value(s) at the edge of the circle (can be number, array, or FacetPattern).
• mode controls the gradient style: 0 = linear, 1 = radial (default), 2 = exponential (steeper falloff), 3 = logarithmic (gentler falloff).

example
  .silence(1000000)
  .gradient2d(500, 500, 200, 1, 0)
  .saveimg('gradient2d_basic')
  .once();
  // radial gradient from white at center to black at edge
example
  .silence(1000000)
  .gradient2d(250, 250, 150, 0.8, 0.2, 2)
  .gradient2d(750, 750, 100, 0.6, 0, 3)
  .saveimg('gradient2d_multiple')
  .once();
  // two overlapping gradients with different modes

**grow2d** ( _iterations_, _prob_, _threshold_ = 0, _mode_ = 0 )

• applies a growth algorithm to the FacetPattern in 2D space. The algorithm iterates over each "pixel" in the pattern and, based on a probability, spreads its value to adjacent pixels.
• the iterations parameter determines how many times the algorithm is applied to the pattern.
• the prob parameter is the probability that a pixel's value will spread to its neighbors.
• the threshold parameter is a value that a pixel's value is compared to in order to determine whether it should spread. The default value is 0.
• the mode parameter determines how the threshold is applied. If mode is 0 (the default), a pixel's value will spread if it is less than the threshold. If mode is 1, a pixel's value will spread if it is greater than the threshold.

example
  .noise(1000000)
  .grow2d(5, 0.5, 0.2, 1)
  .saveimg('grow2d')
  .once();
  // apply the growth algorithm to a noise pattern

**layer2d** ( _brightness_data_, _xCoords_, _yCoords_ )

• superposes a FacetPattern in 2 dimensions on top of the existing data in a FacetPattern.
• brightness_data is a FacetPattern that should be normalized between 0 and 1. It controls how bright the corresponding pixels will be.
• xCoords and yCoords are FacetPatterns that allow the user to control the x,y position of the pixels in brightness_data.

example
  .sine(1)
  .size(10000)
  .scale(0, 1)
  .layer2d(_.noise(10000), _.ramp(0, 100, 128), _.ramp(0, 100, 128))
  .saveimg('layer2d')
  .once();
  // layers a ramp from 0,0 to 100,100 over a sine wave background

**lookup2d** ( _lookupPattern_ )

• remaps the columns of a 2D FacetPattern using a lookup pattern, reordering or repeating columns based on normalized position values.
• lookupPattern is a FacetPattern whose values should be normalized between 0 and 1. Each value maps to a column in the source data: 0 selects the leftmost column, 1 selects the rightmost column.
• the output width equals the length of lookupPattern; the height is preserved from the source data.
• when used on spectral image data (e.g. after .loadimage()), lookupPattern is automatically resized to match the image width.

example
  .silence(1000000)
  .iter(16, () => {
    this.rect2d(ri(0, 900), ri(0, 900), ri(10, 100), ri(10, 100), rf())
  })
  .lookup2d(_.ramp(0, 1, 1000).shuffle())
  .saveimg('lookup2d')
  .once();

**loadimage** ( _imagePath_, _targetHeight_ = 4096 )

• loads an image file and converts it to spectral data for audio resynthesis.
• imagePath is the file path to the image to load (PNG format supported).
• targetHeight controls the frequency resolution - higher values provide more detailed frequency mapping but use more memory.
• the image is converted to grayscale and each row represents a frequency band, with pixels mapped to spectral magnitude values.
• use with .resynthesize() to convert the spectral data back to audio.

example
  .loadimage('img/myspectrogram.png', 2048) // load a PNG file
  .resynthesize(1, ms(6000)) // make it 6 seconds long, linear frequency scale
  .play()
  .once();

**mutechunks2d** ( _num_chunks_, _probabilty_ )

• slices the input FacetPattern into chunks chunks in 2D space and mutes prob percent of them.
• num_chunks must have an integer square root, e.g. 9, 16, 25, 36.

example
  .sine(0.3, 1000)
  .scale(0, 1)
  .mutechunks2d(36, 0.5)
  .saveimg('mutechunks2d')
  .once();

**palindrome2d** ( )

• generates a 2D palindrome of the input FacetPattern, mirrored in both x and y axes.

example
  .silence(1000000)
  .iter(128, () => {
    this.rect2d(ri(0, 1000), ri(0, 1000), ri(10, 100), ri(10, 100), rf())
  })
  .palindrome2d()
  .invert()
  .saveimg('palindrome2d', [_.ramp(rf(), rf(), 1000000), _.ramp(rf(), rf(), 1000000), _.ramp(rf(), rf(), 1000000)])
  .once();
  // 128 rectangles in a 2d palindrome

**perspective3d** ( _copies_ = 5, _scaleFactor_ = 0.8, _vanishX_ = 0, _vanishY_ = 0, _intensityDecay_ = 0.5 )

• creates a 3D perspective effect by layering multiple scaled copies of the image.
• copies is the number of perspective layers to create.
• scaleFactor controls how much smaller each successive layer becomes (values less than 1 create depth).
• vanishX and vanishY set the vanishing point coordinates where the perspective converges.
• intensityDecay controls how much dimmer each successive layer becomes (0 = no decay, 1 = maximum decay).

example
  .silence(1000000)
  .rect2d(400, 400, 200, 200, 1)
  .perspective3d(8, 0.85, 500, 300, 0.3)
  .saveimg('perspective3d')
  .once();
  // white square with 3D perspective effect vanishing toward point (500, 300)

**rechunk2d** ( _num_chunks_ )

• slices the input FacetPattern into chunks chunks in 2D space and shuffles the chunks around.
• num_chunks must have an integer square root, e.g. 9, 16, 25, 36.

example
  .sine(0.3, 1000)
  .scale(0, 1)
  .rechunk2d(36)
  .saveimg('rechunk2d')
  .once();

**react2d** ( _generations_ = 100, _activatorSize_ = 1, _inhibitorSize_ = 1, _feedRate_ = 0.08, _killRate_ = 0.001, _cutoffFrequency_ = 5000, _sampleRate_ = 44100 )

• applies a reaction-diffusion simulation to create organic 2-dimensional patterns.
• generations controls how many simulation steps to run (can be a FacetPattern for variable evolution across the image).
• activatorSize and inhibitorSize control the diffusion rates of the activator and inhibitor agents.
• feedRate controls how fast the activator is fed into the system.
• killRate controls how fast the inhibitor kills the activator.
• cutoffFrequency and sampleRate are used for post-processing smoothing with a biquad low-pass filter, so that this method can also be run on audio patterns.

example
  .silence(500*500)
  .circle2d(250, 250, 250, 1, 1)
  .react2d(100, 1.5, 0.8, 0.055, 0.062)
  .saveimg('react2d')
  .once();
  // organic spreading pattern starting from a circle in the middle of the image

**resynthesize** ( _scale_ = 1, _targetSamples_ = null, _speed_ = 1 )

• converts 2D spectral image data back into audio using inverse FFT.
• scale controls frequency mapping: 1 = linear, 1.5 = moderate logarithmic, 2 = full logarithmic (like Photosounder). Can be a FacetPattern for variable scaling.
• targetSamples sets the exact output length in samples. If null, length is calculated automatically.
• speed controls time stretching: 1 = normal, 0.5 = half speed, 2 = double speed. Can be a FacetPattern for variable speed.
• each row of the image represents a frequency band, with brightness controlling amplitude.
• use after loading an image with .loadimage() or after generating 2D patterns in Facet.

example
  .loadimage('img/spectrogram.png') // load image
  .resynthesize(1.5, n1) // resynthesisze to last a whole note, linear frequency scale
  .play()
  .once();
  
  example
  .silence(100 * 100)
  .circle2d(49, 49, 20, 1, 0) // generate a circle in a 100x100 space
  .delay2d(4, 20, 0.99) // create copies of it around the 2d space
  .resynthesize(1, ms(1000)) // resynthesize to 1 second of audio
  .full()
  .play()
  .once();

**rect2d** ( _topLeftX_, _topLeftY_, _rectWidth_, _rectHeight_, _value_, _fillMode_ = 0 )

• adds a rectangle on top of the existing data in a FacetPattern.
• topLeftX and topLeftY are the x,y coordinates of the top-left corner of the rectangle.
• rectWidth and rectHeight control the size of the rectangle.
• value is brightness value for the rectamgle normalized between 0 - 1.
• fillMode (0 or 1) controls whether the inside of the shape is filled in with value or ignored. Default is no fill.

example
  .silence(1000000)
  .rect2d(0, 0, 100, 100, 1)
  .saveimg('rect2d')
  .once();
  // 100x100 white square in top-left corner of 1000x1000 image

**rotate** ( _angle_ )

• rotates the FacetPattern angle degrees around a center point, as if it were suspended in 2D space.
• the width and height arguments are optional. They default to the square root of the FacetPattern's length. Other values will rotate the data in a different way, around a different center point.

example
  .sine(1)
  .scale(0, 1)
  .size(512 * 512)
  .rotate(35)
  .saveimg('rotate')
  .once();
  // rotates a sine wave background 35 degrees

**saveimg** ( _filepath_ = Date.now(), _rgbData_, _absolute_path_ = false )

• saves the FacetPattern data as a PNG file in the img/ directory or a sub-directory. If a sub-directory is specified in the filepath argument and it doesn't exist, it will be created.
• to save a file to a location outside of the img/ directory, supply a truthy absolute_path argument.
• Note: this example uses MacOS / Linux file paths with forward slashes (e.g. my/path/here). For Windows, you will need to use back slashes (e.g my\path\here)
• the rgbData argument is optional. Without it, the image will be greyscaled. If rgbData is included, it should be an array containing three FacetPatterns normalized to between 0 and 1, representing the R, G, and B amounts. The FacetPattern data will be multipled by the three rgbData patterns to create colored pixels in the image. Values between 0 and 1 will be mapped onto RGB values 0-255.

example
  // create black background
  .silence(512 * 512)
  // add the 512 brightest-possible pixels (1s) that will be used to create a circle
  .layer2d(_.from(1)
    .size(512),
    // the circle x coordinates move from left edge (0) to right edge (512) and back
    _.ramp(0, 511, 512)
    .palindrome(),
    // the circle y coordinates, pt. 1: create a half-circle out of 512 values, defaulting to between 0 and 1
    _.circle(1)
    .size(512)
    // the circle y coordinates, pt. 2: append another half-circle out of 512 values, scaled between -1 and 0 and inverted
    .append(_.circle(1)
      .size(512)
      .scale(-1, 0)
      .invert())
    // scale the y coordinates so they move between 0 and 511
    .scale(0, 511))
  .saveimg('circle',
    // use 3 random ramps, 1 for each RGB channel, to create a gradient in the circle's pixels
    [_.ramp(rf(), rf(), 512), _.ramp(rf(), rf(), 512), _.ramp(rf(), rf(), 512)]
  )
  .once();

**savespectrogram** ( _filePath_, _windowSize_ = 2048 )

• saves a PNG file in the img/ directory named fileName.png, with the FacetPattern's spectrogram.

example
  .noise(n1)
  .ffilter(_.ramp(0, NYQUIST / 2), _.ramp(NYQUIST, NYQUIST / 2))
  .savespectrogram('mytri' + Date.now())
  .once();

**shift2d** ( _xAmt_, _yAmt_ )

• shifts the FacetPattern in 2D space, by xAmt pixels to the left/right, and by yAmt pixels up/down.

example
  .noise(100 * 100)
  .prob(0.001)
  .iter(4, () => {
    this.mix(0.5, () => {
      this.shift2d(0, 1)
    })
  })
  .saveimg('shift2d')
  .once();
  // slides all the pixels up 4

**size2d** ( _size_ )

• creates a smaller image of the FacetPattern in 2D Space, according to the relative amount size.
• size must be between 0 and 1. The new pattern will be a smaller 2D image of the input, surrounded by padding of black pixels (0s).

example
  .noise(10000)
  .size2d(0.5)
  .saveimg('size2d')
  .once();
  // 100 x 100 image with a square of noise in the center

**slices2d** ( _num_slices_, _command_ )

• slices the FacetPattern into num_slices slices in a 2D grid, and for each slice, runs command, appending all slices back together. You can refer to the current slice of the algorithm via the reserved word: this (see example).
• the variable s, referring to the current slice number starting at 0, is also available for use in commands.
• the variable num_slices, referring to the number of slices, is also available for use in commands.
• the slices2d() method does not apply any fading on each slice. Each slice is processed exactly as-is.
• the command is a function that is applied to each slice. The function is converted to a string and any reference to this is replaced with current_slice to ensure the function operates on the correct data.
• the method divides the FacetPattern into a grid of slices, each slice being a square segment of the original pattern. The number of slices is determined by the square of the nearest integer square root of num_slices. This ensures that the slices form a square grid.

example
  .noise(1000000)
  .slices2d(36, () => {
    this.times(rf())
  })
  .saveimg('slices2d')
  .once();

**spectral** ( _stretchFactor_ = 1 )

• applies a spectral transformation to the FacetPattern, treating it as if it were a 2d spectrogram and applying an Inverse Fourier Fast Transform (IFFT).
• this allows you to generate, interpret, and resynthesize a frequency-domain representation into sound.
• the stretchFactor parameter determines how much the pattern is stretched horizontally before the IFFT is applied. The default value is 1, which means no stretching. Each row of the pattern is then stretched by the stretchFactor. This is done by linearly interpolating between each pair of consecutive values in the row.

example
  .silence(1000000)
  .iter(16, () => {
    this.circle2d(ri(0, 999), ri(0, 999), ri(0, 100), rf())
  })
  .spectral()
  .play()
  .full()
  .once();
  // 16 circles randomly dispersed and superposed around the audio spectrum

**tri2d** ( _x1_, _y1_, _x2_, _y2_, _x3_, _y3_, _value_, _fillMode_ = 0 )

• adds a triangle on top of the existing data in a FacetPattern.
• x1, y1, x2, y2, x3, and y3 define the triangle's position in the 2d space.
• value is brightness value for the triangle normalized between 0 - 1.
• fillMode (0 or 1) controls whether the inside of the shape is filled in with value or ignored. Default is no fill.

example
  .silence(1000000)
  .tri2d(ri(0, 1000), ri(0, 1000), ri(0, 1000), ri(0, 1000), ri(0, 1000), ri(0, 1000), 1)
  .saveimg('tri2d')
  .once();
  // one random white triangle in a 1000x1000 image

**warp2d** ( _warpX_, _warpY_, _warpIntensity_ )

• applies a warp effect to the FacetPattern, pulling it towards a given point in 2D space.
• warpX and warpY are the x and y coordinates of the warp point.
• warpIntensity should be a float between 0 and 1 and controls the intensity of the warp effect.
• example

example
  .silence(1000000)
  .circle2d(100, 100, 100, 0.2)
  .delay2d(20, 20, 0.98)
  .iter(4, () => {
    this.warp2d(ri(0, 999), ri(0, 999), rf())
  })
  .saveimg('warp2d')
  .once();
  // echoing circles warped to a random position in the 2d space

**walk2d** ( _percentage_, _x_, _y_, _mode_ = 0 )

• generates a 2D random walk for the FacetPattern.
• percentage should be a float between 0 and 1 and controls the percentage of pixels to move.
• x and y control the maximum distance that a pixel can move. They should be non-negative integers (but random walks occur in both left/right AND up/down),
• mode controls the behavior of pixels at the boundary. A value of 0 is "wrap" mode; values move to the other side. A value of 1 is "fold" mode; values get folded back by however many they exeeded the boundary. A value of 2 is "clip" mode; values get stuck at the boundary.

example
  .silence(1000000)
  .rect2d(950, 950, 50, 50, 1)
  .walk2d(0.5, 10, 10, 0)
  .saveimg('walk2d')
  .once();
  // white square in bottom corner, 50% random walked by 10px in all 4 directions