Skip to content

Latest commit

 

History

History
100 lines (71 loc) · 8.2 KB

Transform.md

File metadata and controls

100 lines (71 loc) · 8.2 KB
Raw

Notes about prebuilt LUT Size

The prebuilt LUTs are built with a grid size of 33x33x33 for RGB/Lab Inputs and 17x17x17x17 for CMYK inputs. Lut size increases exponentially so 4D luts tend to be smaller

A 3D LUT 17x17x17 is 4913 entries A 4D LUT 17x17x17x17 is 83521 entries but at 33x33x33x33 its 11881376 entries.

Transform Methods

constructor(options)

Initializes a new Transform instance, See above for options

Option Type Default Description
BPC Boolean false Enables black point compensation - Note that BPC will be auto enabled for some profiles
dataFormat String 'object' 'object', 'objectFloat', 'int8', 'int16', 'device'
roundOutput Boolean true Rounds the output values
precession Number 0 Number of decimals to round to
interpolation Boolean 'tetrahedral' 'trilinear' or 'tetrahedral'
labAdaptation Boolean false If true object based Lab is adapted to D50 white point before conversion, i.e LabD65 will be converted to LabD50 before transforms
displayChromaticAdaptation Boolean false Perform chromatic adaptation across the PCS if the profiles have different whitepoints, False by default as according to specs profiles should already be adapted to PCS
optimise Boolean true Optimise the pipeline to remove un-necessary conversions
pipelineDebug Boolean false Saves a history of the values at each stage in the pipeline
buildLUT Boolean false If true, a lookup table is built for the transform. This is faster but less accurate
verbose Boolean false Logs into to console
lutGridPoints3D Number 33 The number of grid points in each dimension for a 3D LUT. typically 17, 33 or 65, anything more gets diminishing returns
lutGridPoints4D Number 17 The number of grid points in each dimension for a 4D LUT. typically 11, 17, 33, anything more gets diminishing returns
LUTinterpolation string 'tetrahedral' The interpolation method used for lookup tables, trilinear' or 'tetrahedral'

create(inputProfile, outputProfile, eIntent)

  • inputProfile: Instance of Profile or name of virtual profile, i.e '*lab'
  • outputProfile: Instance of Profile or name of virtual profile, i.e '*srgb'
  • eIntent: Rendering intent to use for the transform, i.e eIntent.perceptual
  • Returns : none, Throws error if pipeline cannot be created

createMultiStage(profileChain, customStages)

  • profileChain: Array of profiles and intents in order to use in the transform, must be [profile, eIntent, profile, eIntent, ... profile]
  • customStages: Array of custom stages to use in the transform

transform(inputColor)

  • inputColor: Color to transform, can be any of the supported color types
  • Returns : Color in the output profile

transformArray(inputArray, inputHasAlpha, outputHasAlpha, preserveAlpha, pixelCount, outputFormat)

  • inputArray: Array of input colors, can be any of the supported color types
  • inputHasAlpha: Boolean indicating if the input array has an alpha channel, expected to be the last channel, ignored if dataFormat is 'object' or 'objectFloat'
  • outputHasAlpha: Boolean indicating if the output array should have an alpha channel, expected to be the last channel, will be set to 255 unless preserveAlpha=true ignored if dataFormat is 'object' or 'objectFloat'
  • preserveAlpha: Boolean indicating if the alpha channel should be preserved, ignored if dataFormat is 'object' or 'objectFloat'
  • pixelCount: Number of pixels or items to convert, if not specified the length of the input array is used
  • outputFormat: String indicating the output array type, Standard array created if blank, options are 'int8' = Uint8ClampedArray(), 'int16' = Uint16Array(), 'float32' = Float32Array(), 'float64' = Float64Array(), 'same'
  • Returns : Array of colors in the output profile

transformArrayViaLUT(inputArray, inputHasAlpha, outputHasAlpha, preserveAlpha, pixelCount)

A High speed transform using a prebuilt LUT on Uint8ClampedArrays specifically for image data, See above for LUT options

  • inputArray: Array of input colors Uint8ClampedArray
  • inputHasAlpha: Boolean indicating if the input array has an alpha channel, expected to be the last channel
  • outputHasAlpha: Boolean indicating if the output array should have an alpha channel, expected to be the last channel, will be set to 255 unless preserveAlpha=true
  • preserveAlpha: Boolean indicating if the alpha channel should be preserved
  • pixelCount: Number of pixels or items to convert, if not specified the length of the input array is used

getStageNames()

  • Returns : Array of string with names of the stages the transform pipeline

debugInfo()

  • Returns : Object string with information about the transform including Chain, History and Optimise info

chainInfo()

  • Returns : Object string with information about the transform chain

historyInfo()

  • Returns : Object string with information about the transform history

optimiseInfo()

  • Returns : Object string with information about the optimisation

Transform Properties

Property Description
inputProfile The FIRST input profile for the transformation.
outputProfile The LAST output profile for the transformation.
inputChannels The number of channels in the input profile.
outputChannels The number of channels in the output profile.
usesBPC A boolean indicating if Black Point Compensation is applied.
usesAdaptation A boolean indicating if chromatic adaptation is applied across the PCS if the profiles have different whitepoints.
chain The profile chain, showing how this pipeline was created.