brennan-node-vibrant v2.1.6

node-vibrant

Extract prominent colors from an image.

Update Notes

• Some major refactor/rewriting comes with node-vibrant@2.0.0:

  • Closely matches latest Android M Palette API with new features such as Builder and Filter
  • Optimization: ~4x speed-up over original implementation
  • Implement image downsampling for both node.js and browsers
  • More stream-lined design
  • Better test coverage for both node.js and browser environment

• node-vibrant@1.x is a node.js port of Vibrant.js, which is a javascript port of the awesome Palette class in the Android support library.

Features

• Identical (asynchronous) API for both node.js and browser environment
• Support browserify
• Consistent results (*See Result Consistency)

Install

$ npm install node-vibrant

Usage

node.js / browserify

# Use in node.js or bundle with browserify
Vibrant = require('node-vibrant')

# Using builder
Vibrant.from('path/to/image').getPalette (err, palette) ->
    console.log palette

# Using constructor
v = new Vibrant('path/to/image', opts)
v.getPalette (err, palette) ->
  console.log(swatches)

Browser

<!-- Debug version -->
<script src="/path/to/dist/vibrant.js"></script>
<!-- Uglified version -->
<script src="/path/to/dist/vibrant.min.js"></script>

<script>
  // Use `Vibrant` in script
  // Vibrant is exported to global. window.Vibrant === Vibrant
  Vibrant.from('path/to/image').getPalette(function(err, palette) {});
  // Or
  var v = new Vibrant('/path/to/image', opts);
  // ... same as in node.js
</script>

Contribution Guidelines

1. Make changes
2. Write test specs if necessary
3. Pass tests
4. Commit source files only (without compiled files)

References

Vibrant

Main class of node-vibrant.

Vibrant.from(image)

Make a Builder for an image. Returns a Builder instance.

constructor(image, opts)

opts

getPalette(cb)

cb(err, palette)

getSwatches(cb)

Alias of getPalette.

Vibrant.Builder

Helper class for change configurations and create a Vibrant instance. Methods of a Builder instance can be chained like:

Vibrant.from(src)
  .quality(1)
  .clearFilters()
  # ...
  getPalette (err, palette) ->

constructor(src, opts)

Arguments are the same as Vibrant.constructor.

quality(q)

Sets opts.quality to q. Returns this Builder instance.

maxColorCount(n)

Sets opts.colorCount to n. Returns this Builder instance.

maxDimension(d)

Sets opts.maxDimension to d. Returns this Builder instance.

addFilter(f)

Adds a filter function. Returns this Builder instance.

removeFilter(f)

Removes a filter function. Returns this Builder instance.

clearFilters()

Clear all filters. Returns this Builder instance.

useImage(image)

Specifies which Image implementation class to use. Returns this Builder instance.

useQuantizer(quantizer)

Specifies which Quantizer implementation class to use. Returns this Builder instance.

useGenerator(generator)

Sets opts.generator to generator. Returns this Builder instance.

build()

Builds and returns a Vibrant instance as configured.

getPalette(cb)

Builds a Vibrant instance as configured and calls its getPalette method.

getSwatches(cb)

Alias of getPalette.

Vibrant.Swatch

Represents a color swatch generated from an image's palette.

constructor(rgb, population)

Internal use.

getHsl()

getPopulation()

getRgb()

getHex()

getTitleTextColor()

Returns an appropriate color to use for any 'body' text which is displayed over this Swatch's color.

getBodyTextColor()

Returns an appropriate color to use for any 'title' text which is displayed over this Swatch's color.

Vibrant.Util

Utility methods. Internal usage.

clone(o)

Make a deep copy of o.

defaults()

Same as underscore.js's defaults. Re-implemented to reduce browserify package size.

hexToRgb(hex)

rgbToHex(r, g, b)

hslToRgb(h, s, l)

rgbToHsl(r, g, b)

xyzToRgb(x, y, z)

rgbToXyz(r, g, b)

xyzToCIELab(x, y, z)

rgbToCIELab(l, a, b)

deltaE94(lab1, lab2)

Computes CIE delta E 1994 diff between lab1 and lab2. The 2 colors are in CIE-Lab color space. Used in tests to compare 2 colors' perceptual similarity.

rgbDiff(rgb1, rgb2)

Compute CIE delta E 1994 diff between rgb1 and rgb2.

hexDiff(hex1, hex2)

Compute CIE delta E 1994 diff between hex1 and hex2.

getColorDiffStatus(d)

Gets a string to describe the meaning of the color diff. Used in tests.

Vibrant.Filter

A collection of built-in filters.

Filter(r, g, b, a): function<boolean>

A Filter provides a mechanism for exercising fine-grained control over which colors are valid within a resulting. It takes color's rgba value and returns true if the color is allowed.

Filter.Default

Keeps the original vibrant.js's filtering behavior as reference.

Vibrant.Quantizer

Base class of a Quantizer.

Quantizer.MMCQ

Default quantizer. ~4x faster than baseline quantizer.

Vibrant.Generator

Base class for Generator.

Generator.Default

Default Generator implementation.

constructor(opts)

opts

Gulp Tasks

Notes

Intentional Deviation From vibrant.js

• node-vibrant takes image path, not the image object as parameter for the obvious reason that node.js environment has no access to HTML DOM object.
• node-vibrant provides asynchronous API since most node.js image processing library is asynchronous. And the original vibrant.js workflow is asynchronous any way (though you will have to handle the image loading yourself, while node-vibrant does it for you).
• node-vibrant uses one single opts object to hold all options for future expansions. And it feels more node.js-like.
• node-vibrant uses method call to initiate image processing instead of constructor so that developers can use it with Promise.

Result Consistency

The results is consistent within each user's browser instance regardelss of visible region or display size of the image, unlike the original vibrant.js implementation.

However, due to the very nature of HTML5 canvas element, image rendering is platform/machine-dependent. Thus the resulting swatches in browser environment varies and may not be the same as in node.js nor in another machine. See Canvas Fingerprinting.

The test specs use CIE delta E 1994 color difference to measure inconsistencies across platforms. It compares the generated color on node.js, Chrome, Firefox and IE11. At quality == 1 (no downsampling) and no filters, the results are rather consistent. Color diffs between browsers are mostly not perceptible by human eyes. Downsampling will cause perceptible inconsistent results across browsers due to differences in canvas implementations.