pixl-canvas-plus

A universal library for manipulating images, built on canvas.


Keywords
image, canvas, rgba, brightness, contrast, curves, html5, hue, javascript, quantize, resize-images, rotate
License
MIT
Install
npm install pixl-canvas-plus@2.1.9

Documentation

Overview

CanvasPlus is a universal Node.js and browser library, which adds image processing capabilities onto the HTML5 Canvas API. In Node.js we rely on the amazing canvas module, which is built on the Cairo graphics library. In the browser we simply extend (well, wrap) the native canvas.

Check out our live demo playground to try out the library in your favorite browser.

Features

  • Load images from buffers, files or URLs
  • Resize with special modes including cover, contain, and letterbox
  • Render text with extra features
    • Multi-line, paragraph text
    • Custom character spacing and line spacing
    • Word-wrap
    • Auto-scale to fit line or paragraph
  • Quantization into indexed mode for PNGs and GIFs
    • Optional dithering
    • Super-fast quantize mode
  • Curves
    • Similar to Adobe Photoshop's Curves filter
    • Optionally apply different curves to each channel
    • Macro functions including Posterize, Solarize, Invert, Gamma, Sepia, etc.
  • Adjust hue, saturation, brightness, contrast and color temperature
  • Normalize (Auto-Enhance)
  • Generate histogram data
  • Access EXIF Metadata
  • Auto-Orient
  • Convolution Kernel
    • Macro functions including Emboss, Find Edges, Blur, Gaussian Blur, Sharpen
  • Crop images
    • Manual or automatic based on corner pixel
  • Rotation and flip horizontal / vertical
  • Read/Write JPEG, PNG, GIF and WebP images
    • Full support for indexed palette PNGs and GIFs
    • Full support for alpha PNGs, even indexed alpha
    • Full support for alpha WebP images.

Why

There are already some extremely cool image processing libraries out there. I played with several, but none of them provided all the features I needed. Specifically, I needed fast JPEG load and save, fast resize, TrueType / OpenType font support, advanced text rendering features like word wrap and auto-scale, quantization to 8-bit indexed mode with dithering, and curves. I also wanted a universal library that worked in the browser as well as Node.js. That being said, I want to recommend some amazing free libraries that you should absolutely check out:

Library Notes
LWIP Very nice, fast engine with a clean API. Only missing text rendering, curves, and 8-bit quantization.
JIMP Incredible pure-JS engine that does just about everything. However, JPEG load/save and resize are too slow for me. Also, no TrueType / OpenType font support, and there are some licensing issues. Otherwise this is very, very cool.
Sharp Very fast (uses libvips), and cool SVG integration. Missing text rendering, curves, 8-bit quantization, and browser support.
GM I love this library, but it actually just shells out to the command-line GraphicsMagick/ImageMagick binary, so there are some performance concerns. Also, no advanced text features, no curves, and no browser support.

Table of Contents

Usage

Node.js

Use npm to install the module:

npm install pixl-canvas-plus

Please note that the canvas module dependency is a binary compiled library, which depends on Cairo being preinstalled on your machine. See their installation wiki for assistance. That being said, canvas v2.x comes with precompiled binaries for macOS, Windows and Linux, so it might just slip right in.

Here is a simple usage example:

const CanvasPlus = require('pixl-canvas-plus');
let canvas = new CanvasPlus();

canvas.load( 'waterfall.jpg', function(err) {
	if (err) throw err;
	
	canvas.resize({
		"width": 640,
		"height": 480,
		"mode": "fit"
	});
	
	canvas.adjust({
		"brightness": -20,
		"contrast": 20
	});
	
	canvas.write({"format":"jpeg", "quality":90}, function(err, buf) {
		if (err) throw err;
		
		// 'buf' will be a binary buffer containing final image...
		require('fs').writeFileSync('my_image.jpg', buf);
	});
	
});

Browser

For the browser we use the native built-in HTML5 Canvas API, and there are no dependencies. Simply download and host this pre-built file (compiled using browserify) on your own web server:

https://raw.githubusercontent.com/jhuckaby/canvas-plus/master/canvas-plus.js

This will expose a global CanvasPlus class in the window object. Here is how to use it:

<script src="canvas-plus.js"></script>
<script>
	let canvas = new CanvasPlus();
	
	canvas.load( 'waterfall.jpg', function(err) {
		if (err) throw err;
		
		canvas.resize({
			"width": 640,
			"height": 480,
			"mode": "fit"
		});
		
		canvas.adjust({
			"brightness": -20,
			"contrast": 20
		});
		
		canvas.write({"format":"jpeg","quality":90}, function(err, buf) {
			if (err) throw err;
			
			// 'buf' will be a binary buffer containing final image...
			let blob = new Blob( [ buf ], { type: "image/jpeg" } );
			let object_url = URL.createObjectURL( blob );
			
			// insert new image into DOM
			let img = new Image();
			img.src = object_url;
			document.body.appendChild( img );
		});
		
	});
</script>

Creating

To create a blank canvas, and optionally fill with a background color, you can pass arguments to the constructor, like this:

let canvas = new CanvasPlus( 640, 480 );

Or with an optional background color:

let canvas = new CanvasPlus( 640, 480, "#FF0000" );

Or you can use the explicit create() method:

let canvas = new CanvasPlus();
canvas.create({
	width: 640,
	height: 480,
	background: "#FF0000"
});

The background color can be any CSS-compatible color string, including RGBA, e.g. rgba(255, 0, 0, 0.5).

Loading

You can load images from a variety of sources, including buffers, files and URLs. JPEGs, PNGs, GIFs and WebP are supported, and possibly others depending on your environment:

Environment Sources Image Formats
Node.js Buffer, File Path, URL JPEG, PNG, GIF, WebP
Browser ArrayBuffer, File, Blob, URL JPEG, PNG, GIF, BMP, WebP

To load an image, pass the source (e.g. file path, URL) to the load() method. Please note that this is an asynchronous call, so you need to provide a callback, or you can use the promise / async / await pattern with Node 8+. Example with callback:

let canvas = new CanvasPlus();

canvas.load( 'waterfall.jpg', function(err) {
	if (err) throw err;
});

Or with Node 8+:

try {
	await canvas.load( 'waterfall.jpg' );
}
catch (err) {
	// handle error here
}

In Node.js this example would look for a waterfall.jpg file on disk in the current directory. In the browser this would be treated as a URL to waterfall.jpg relative to the current page.

Note that loading images in the browser requires that the file is hosted on the same domain as the page, or is hosted on a server that sends back proper CORS headers. However, if you need to load an image from a 3rd party URL that does not support CORS, you can use the special loadRemote() method. This works well enough, but due to browser security restrictions we will not have access to the raw binary bytes, so features like EXIF data and Auto-Orient are not available.

Saving

To save the contents of your canvas to a binary image format, use the write() method. This will compress the image into a JPEG, PNG or GIF, and provide you with a Buffer object. Please note that this is an asynchronous call, so you need to provide a callback, or you can use the promise / async / await pattern with Node 8+. Example with callback:

canvas.write({"format":"jpeg", "quality":90}, function(err, buf) {
	if (err) throw err;
	
	// 'buf' will be a binary buffer containing final image
});

Or with async/await:

try {
	let buf = await canvas.write({"format":"jpeg", "quality":90});
	// 'buf' will be a binary buffer containing final image
}
catch (err) {
	// handle error here
}

The format property should be one of jpeg, png, gif or webp. The quality property is only used for JPEGs and WebP, and should be a number from 0 to 100.

Note that in the browser the buffer is provided using the buffer module, which is is a subclass of Uint8Array. So there is no need to explicitly convert to typed array -- just use the buffer as you would a native Uint8Array.

Errors

All filter methods are synchronous, so they do not follow the callback pattern. So by default all filter methods do not throw (however you can enable this behavior if you want, see below). Instead, they set an internal error state which you can query after the fact. Example:

canvas.adjust({
	"brightness": -20,
	"contrast": 20
});
if (canvas.getLastError()) {
	// an error occurred
	let err = canvas.getLastError();
}

This works even if you chain multiple filter calls together. CanvasPlus will "abort" the chain on the first error. Example:

canvas.desaturate().normalize().solarize().sepia().rotate(45);

if (canvas.getLastError()) {
	// an error occurred
	let err = canvas.getLastError();
}

If you would prefer to use try/catch, you can enable throw mode by calling set() before running any filters:

canvas.set('throw', true);

try {
	canvas.adjust({
		"brightness": -20,
		"contrast": 20
	});
}
catch (err) {
	// an error occurred
}

Logging

To enable basic debug logging, use the set() method to enable the debug parameter. Example:

canvas.set('debug', true);

Example debug log, which is emitted using console.log() (and console.error() for errors):

[DEBUG] Setting property: debug: true
[DEBUG] Loading image data from file: 640x480.jpg 
[DEBUG] Loading image from buffer {"size":296673}
[DEBUG] Image ping: {"height":480,"width":640,"type":"jpg"}
[DEBUG] Setting property: width: 640 
[DEBUG] Setting property: height: 480 
[DEBUG] Setting property: format: jpg 
[DEBUG] Setting property: mode: image 
[DEBUG] Setting property: origWidth: 640 
[DEBUG] Setting property: origHeight: 480 
[DEBUG] Setting property: origFormat: jpg 
[DEBUG] Image load complete 
[DEBUG] Resizing image to target size: 320x240 {"mode":"Fit","gravity":"center","background":"","orig_width":640,"orig_height":480,"target_width":320,"target_height":240,"dest_width":320,"dest_height":240,"canvas_width":320,"canvas_height":240,"x":0,"y":0}
[DEBUG] Setting property: width: 320 
[DEBUG] Setting property: height: 240 
[DEBUG] Creating new canvas: 320x240 {"background":"(transparent)"}
[DEBUG] Setting property: mode: rgba 
[DEBUG] Canvas created successfully 
[DEBUG] Image resize complete 
[DEBUG] Adjusting image {"brightness":-20,"contrast":20,"hue":0,"saturation":0}
[DEBUG] Image adjustment complete 
[DEBUG] Setting property: file: out.png 
[DEBUG] Setting property: format: png 
[DEBUG] Compressing image to format: png 
[DEBUG] Compressing into 32-bit PNG {"compression":9,"filter":"PNG_ALL_FILTERS"}
[DEBUG] PNG compression complete 
[DEBUG] Image compression complete {"size":165279}
[DEBUG] Saving to file: out.png 
[DEBUG] Image write complete

In Node.js, you can attach a log agent compatible with our pixl-logger module, or write your own that implements the interface. Example of the former:

const Logger = require('pixl-logger');
let columns = ['hires_epoch', 'date', 'hostname', 'component', 'category', 'code', 'msg', 'data'];
let logger = new Logger( 'logs/debug.log', columns );

canvas.attachLogAgent( logger );

Example custom logger implementation:

let logger = {
	debug: function(level, msg, data) {
		if (data) msg += " (" + JSON.stringify(data) + ")";
		console.log('[DEBUG]['+level+'] ' + msg);
	},
	error: function(code, msg, data) {
		if (data) msg += " (" + JSON.stringify(data) + ")";
		console.log('[ERROR]['+code+'] ' + msg);
	}
};

canvas.attachLogAgent( logger );

API

General

create

The create() method creates a new blank canvas, or replaces one if it already exists. It accepts an object containing the following properties:

Property Name Type Description
width Integer (Required) Canvas width in pixels.
height Integer (Required) Canvas height in pixels.
background String Optional background color (any CSS color string).

Example use:

canvas.create({
	width: 640,
	height: 480,
	background: "#FF0000"
});

The background color can be any CSS-compatible color string, including RGBA, e.g. rgba(255, 0, 0, 0.5). If omitted, the canvas is initialized to transparent black.

Canvases are always created in 32-bit RGBA mode.

load

The load() method loads an image into CanvasPlus. It accepts buffers, files and URLs. For image formats, JPEGs, PNGs, GIFs and WebP are supported, and possibly others depending on your environment. Please note that this is an asynchronous call, so you need to provide a callback as the 2nd argument. Example use:

canvas.load( 'waterfall.jpg', function(err) {
	if (err) throw err;
});

In Node.js this example would look for a waterfall.jpg file on disk in the current directory. In the browser this would be treated as a URL to waterfall.jpg relative to the current page.

Alternatively you can import an existing canvas using importCanvas(), or an image object using importImage().

Auto-Orient

By default, CanvasPlus will automatically orient (rotate) your image on load, so it is always right-side up. This is accomplished by detecting the image's EXIF data, if present, and reading the special orientation flag. Many digital cameras do not fix image rotation when shots are taken, and instead simply include an EXIF flag in the files, indicating the camera's orientation.

If you do not want this behavior for some reason, you can disable the feature by calling set() before loading your image, and disabling the autoOrient parameter. Example:

let canvas = new CanvasPlus();
canvas.set('autoOrient', false);

canvas.load( 'waterfall.jpg', function(err) {
	if (err) throw err;
});

loadRemote

The loadRemote() method is a special, browser-only API, which allows you to load 3rd party image URLs. That is, images that are hosted on 3rd party domains, and do not provide proper CORS headers. Example use:

let url = "https://upload.wikimedia.org/wikipedia/commons/thumb/6/61/HTML5_logo_and_wordmark.svg/1024px-HTML5_logo_and_wordmark.svg.png";

canvas.loadRemote( url, function(err) {
	if (err) throw err;
});

Note that due to browser security restrictions we have no access to the raw binary bytes, so features like EXIF data and Auto-Orient are not available when using loadRemote().

write

The write() method will compress your canvas into a JPEG, PNG, GIF or WebP, and provide you with a Buffer object. Please note that this is an asynchronous call, so you need to provide a callback.

canvas.write({"format":"jpeg", "quality":90}, function(err, buf) {
	if (err) throw err;
	
	// 'buf' will be a binary buffer containing final image
});

The format property should be one of jpeg, png, gif or webp. The quality property is only used for JPEGs and WebP, and should be a number from 0 to 100.

Note that in the browser the buffer is provided using the buffer module, which is is a subclass of Uint8Array. So there is no need to explicitly convert to typed array. Just use the buffer as you would a native Uint8Array.

clone

The clone() method makes a copy of your canvas, including the raw pixel data, settings and all. This is a synchronous call, and the new cloned object is returned. Example use:

let copy = canvas.clone();

Filters

All filter methods are synchronous, and do not require a callback. You can also chain them together, like this:

canvas.desaturate().brightness(-15).contrast(30).invert();

adjust

Live Demo: Adjust Brightness/Contrast

The adjust() method allows you to adjust the image hue, saturation, brightness, and/or contrast. It accepts an object containing the following properties:

Property Name Type Description
hue Integer Adjust hue (-360 to 360).
saturation Integer Adjust saturation (-255 to 255).
brightness Integer Adjust brightness (-255 to 255).
contrast Integer Adjust contrast (-255 to 255).
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.adjust({
	"brightness": -15,
	"contrast": 30
});

The following method shortcuts are provided for convenience:

Method Name Example
hue canvas.hue( 180 );
saturation canvas.saturation( -128 );
brightness canvas.brightness( 64 );
contrast canvas.contrast( 32 );

Note: If you want to completely desaturate an image, check out the desaturate() filter below, as it is considerably faster than using adjust().

blur

Live Demo: Blur

The blur() method applies a simple box blur to your image, using a custom convolution kernel. This is a fast blur filter, but isn't particularly high quality. It accepts an object containing the following properties:

Property Name Type Description
amount Integer (Required) Blur amount (2 to 100).
edges String How to calculate out-of-bounds pixels, can be one of repeat, wrap, mirror or discard (case-insensitive). Defaults to repeat.
channels String Which channels to apply the filter to, defaults to rgba. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.blur({
	"amount": 5
});

See Also: gaussianBlur()

border

Live Demo: Border

The border() method draws a border around your canvas, with a custom size and color. It accepts an object containing the following properties:

Property Name Type Description
size Integer (Required) Size of the border in pixels.
color String (Required) Color of the border (any CSS color string).
mode String (Required) Border render mode (inside, center or outside).

The mode can be one of inside (draw on top of edge pixels), outside (expand canvas by border size), or center (draw border half inside, half outside). The latter two modes will actually expand the canvas to accommodate the border.

Example use:

canvas.border({
	"size": 5,
	"color": "#FF0000",
	"mode": "inside"
});

Note that transparent colors aren't supported for this filter.

composite

Live Demo: Composite

The composite() method superimposes a separate image or canvas atop the current one. The source can be another CanvasPlus object, a native HTML5 Canvas object, or a native Image object. Several parameters are provided for positioning, scaling and setting the compositing mode. The method accepts an object containing the following properties:

Property Name Type Description
image Object (Required) Source image or canvas to superimpose atop current canvas.
width Integer Optional desired width, will trigger a resize during composite.
height Integer Optional desired height, will trigger a resize during composite.
gravity String Image starting position (alignment), which can be further adjusted with margins and/or offsets. See Gravity.
marginX Integer Horizontal margin in pixels, defaults to 0.
marginY Integer Vertical margin in pixels, defaults to 0.
offsetX Integer Horizontal offset in pixels, defaults to 0.
offsetY Integer Vertical offset in pixels, defaults to 0.
opacity Float Image opacity (0.0 to 1.0), defaults to 1.0.
mode String Composite rendering mode, see globalCompositeOperation, defaults to source-over.
antialias String Image scaling quality, one of best, good, fast or nearest. Defaults to best. See Anti-Aliasing.

Example use:

canvas.composite({
	"image": my_other_canvas,
	"gravity": "northwest",
	"marginX": 10,
	"marginY": 10
});

Note that when width and/or height are specified here, the image is resized in exact (scale) mode.

convolve

Live Demo: 3x3 Sharpen Kernel

The convolve() method applies a custom convolution kernel to the canvas. This is used to apply effects such as blur, sharpen, emboss, find edges, and more. The method accepts an object containing the following properties:

Property Name Type Description
matrix Array (Required) An array of numbers for the convolution matrix. The length should be a perfect square (e.g. 9, 16, 25).
offset Integer Optionally offset the destination channel values, causing a brightening effect (used by emboss).
edges String How to calculate out-of-bounds pixels, can be one of repeat, wrap, mirror or discard. Defaults to repeat.
channels String Which channels to apply the filter to, defaults to rgba. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.convolve({
	"matrix": [ // 3x3 sharpen
		0, -1, 0, 
		-1, 5, -1, 
		0, -1, 0
	]
});

crop

Live Demo: Square Earth

The crop() method crops the canvas down to the specified size, at the specified coordinates (top-left corner). All values are specified in pixels. The method accepts an object containing the following properties:

Property Name Type Description
x Integer (Required) The coordinate of the left side of the crop rectangle.
y Integer (Required) The coordinate of the top side of the crop rectangle.
width Integer (Required) The width of the crop rectangle.
height Integer (Required) The width of the crop rectangle.

Example use:

canvas.crop({
	"x": 50,
	"y": 50,
	"width": 200,
	"height": 200
});

Note that cropping regenerates the underlying canvas object. It effectively creates a new canvas at the final cropped size, then copies the source into the destination, then discards the source.

See Also: trim()

curves

Live Demo: Increase mid-tones (see below for more)

The curves() method applies a tonality curve to all the pixels in your canvas. You need only provide key points, and all the curve values are interpolated using monotone cubic interpolation (similar to how Photoshop does it). Curves are used to implement all kinds of common image filters, including gamma adjustment, inversion, posterize, solarize, normalize, sepia, and color temperature. But here you can supply your own custom curve by specifying key points, including different curves per channel if desired. The method accepts an object containing the following properties:

Property Name Type Description
rgb Array An array of curve points for the red, green and blue channels together (same curve applied to all).
red Array An array of curve points for the red channel specifically.
green Array An array of curve points for the green channel specifically.
blue Array An array of curve points for the blue channel specifically.
alpha Array An array of curve points for the alpha channel specifically.
clip Object Optional clipping rectangle (see Clipping below).

You can specify the curve points in two ways. First, a simple one-dimensional array of "Y" axis values (each 0 - 255), which are evenly distributed over the "X" axis, from darkest to lightest. And second, an array of X/Y pairs specifying a more precise curve. Example of the former:

canvas.curves({
	"rgb": [0, 191, 255] // increase mid-tones
});

To better explain how this works, here is a table with some common curves, a visualization, and a live demo of each:

Curve Points Visualization Description Demo Link
[0, 255] Baseline (no change) Live Demo
[63, 255] Increase brightness Live Demo
[0, 127] Decrease brightness Live Demo
[0, 0, 255, 255] Increase contrast Live Demo
[63, 191] Decrease contrast Live Demo
[0, 191, 255] Increase mid-tones Live Demo
[0, 63, 255] Decrease mid-tones Live Demo
[0, 96, 160, 216, 255] Increase shadow detail Live Demo
[255, 0] Invert image Live Demo
[0, 127, 0] Solarize effect Live Demo
[0, 255, 0, 255] Alice in Wonderland Live Demo

You can also specify different curves to each channel. For example, here is one way to apply a sepia tone effect to an image:

canvas.desaturate().curves({
	"green": [0, 108, 255],
	"blue": [0, 64, 255]
});

As mentioned above, you can alternatively specify an array of X/Y pairs for the key points, to describe a more precise curve. Here is an example of that:

canvas.curves({
	"rgb": [
		[0, 0],
		[32, 192],
		[255, 255]
	]
});

Note that if your point array contains exactly 256 elements, and is a simple one-dimensional array of integers, then no interpolation is performed, and the points are mapped exactly to each value as specified.

desaturate

Live Demo: Desaturate

The desaturate() method removes all color from your canvas, resulting in a pure grayscale image. Instead of averaging the channels together, they are weighted in such a way that the grayscale output is more natural-looking (see relative luminance).

This is similar to calling adjust() and setting the saturation to -255, however desaturate() is much faster. The method accepts an object containing the following properties:

Property Name Type Description
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.desaturate();

draw

Live Demo: Draw Primitives

The draw() method allows you to draw primitive geometric objects onto the canvas, such as rectangles, lines, or custom shapes. This is basically just a convenience wrapper around the HTML5 Canvas drawing API. The method accepts an object containing the following properties:

Property Name Type Description
params Object Optional parameters to set on the canvas context, e.g. fillStyle, strokeStyle.
commands Object An array of commands to run on the canvas context, e.g. ['fill'].
antialias String Draw rendering quality, one of best, good, fast or nearest. Defaults to best. See Anti-Aliasing.

Here is an example:

canvas.draw({
	params: {
		fillStyle: '#888888',
		strokeStyle: '#000000',
		lineWidth: 2
	},
	commands: [
		['rect', 50, 50, 100, 100],
		['fill'],
		['stroke']
	]
});

This example would draw a filled + stroked rectangle onto the canvas. The params object can contain any properties you want to set on the canvas context before running the draw commands (i.e. colors, stroke width, line caps, etc.), and the commands array contains each draw call to make. Each element in the commands array should be a nested array containing the draw call name (e.g. rect), followed by any additional arguments required by the call, such as coordinates for a rectangle, or points for a moveTo / lineTo.

Using this syntax, you can draw complex shapes by calling any of the HTML5 Canvas API draw methods. Here is an example which draws a filled triangle:

canvas.draw({
	params: {
		fillStyle: '#000000'
	},
	commands: [
		['beginPath'],
		['moveTo', 75, 70],
		['lineTo', 100, 75],
		['lineTo', 100, 25],
		['fill']
	]
});

Convenience properties are provided for drawing simple rectangles and lines, to make the syntax a little easier for common use cases. Here is the list of optional properties:

Property Name Type Description
line Array An array of 4 coordinates, the first two being the X/Y of the line start point, and the latter two being the X/Y of the line end point.
rect Array An array of 4 coordinates, the first two being the X/Y of the rectangle's top-left corner, and the latter two being the width and height.
fill String Setting this to a color will set the canvas context fillStyle property, and append a ['fill'] command to the end of the list.
stroke String Setting this to a color will set the canvas context strokeStyle property, and append a ['stroke'] command to the end of the list.

Examples:

canvas.draw({
	"rect": [ 75, 75, 200, 200 ],
	"fill": "#888888",
	"stroke": "#000000"
});

canvas.draw({
	"line": [ 150, 100, 350, 300 ],
	"stroke": "#ff0000"
});

emboss

Live Demo: Emboss

The emboss() method applies an emboss filter to the canvas, which is implemented using a 3x3 convolution kernel. The method accepts an object containing the following properties:

Property Name Type Description
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.emboss();

This is equivalent to calling desaturate(), followed by convolve() with the following parameters:

canvas.desaturate().convolve({ 
	matrix: [2, 0, 0, 0, -1, 0, 0, 0, -1], 
	offset: 127, 
	edges: 'repeat',
	channels: 'rgb'
});

expand

Live Demo: Expand

The expand() method increases the canvas size, while not resizing the image itself. This is essentially a "reverse crop". You can specify an alignment (gravity) for positioning the existing image on the expanded canvas, and an optional background color. The method accepts an object containing the following properties:

Property Name Type Description
width Integer Number of pixels to increase the width by (this is a delta).
height Integer Number of pixels to increase the height by (this is a delta).
gravity String Image position (alignment) of pre-existing pixels on new expanded canvas. See Gravity.
background String Background color for canvas expansion (any CSS color string). Fully transparent if omitted.

Example use:

canvas.expand({
	"width": 100,
	"height": 100,
	"gravity": "center"
});

This would increase the canvas size by 100px horizontally and vertically, essentially adding a 50px transparent border on all sides of the image.

Note that expanding regenerates the underlying canvas object. It effectively creates a new canvas at the expanded size, then copies the source into the destination, then discards the source.

exposure

Live Demo: Exposure

The exposure() method simulates increasing or decreasing the camera film exposure. Internally, this is accomplished by a special curve. To use, specify an amount between -100 (darken) and 100 (lighten). The method accepts an object containing the following properties:

Property Name Type Description
amount Integer Exposure adjustment, from -100 (darken) to 100 (lighten).
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.exposure({
	"amount": 25
});

Note that if amount is the sole argument, you can simply pass the number itself:

canvas.exposure( 25 );

findEdges

Live Demo: Find Edges

The findEdges() method applies a Sobel operator to the canvas, resulting in a strong highlight of all the apparent edges in the image. This uses a 3x3 convolution kernel. The method accepts an object containing the following properties:

Property Name Type Description
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.findEdges();

This is equivalent to calling desaturate(), followed by convolve() with the following parameters:

canvas.desaturate().convolve({ 
	matrix: [-1, -1, -1, -1, 8, -1, -1, -1, -1], 
	offset: 0, 
	edges: 'repeat',
	channels: 'rgb'
});

flatten

Live Demo: Flatten

The flatten() method effectively removes the alpha channel from your canvas, resulting in a fully opaque image. Of course, the alpha channel still technically exists, as all canvases are RGBA, but this allows you to remove all transparency. You can choose to place a background color matte behind the image, or simply set the direct alpha values to 255. The method accepts an object containing the following properties:

Property Name Type Description
background String Optional background color (any CSS color string).
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.flatten({
	"background": "#000000"
});

When a background is specified, the underlying canvas object is regenerated. The library effectively creates a new canvas with the specified background color, superimposes the source onto the destination, then discards the source.

If you omit the background property (or it evaluates to false) then the alpha pixel values are all set to 255 in a direct sense. This may expose otherwise "hidden" visual elements in your image, depending on how it was encoded.

Note that flattening the canvas sets an internal alpha flag to false, which indicates that PNG and GIF output should not contain any alpha information. You can manually get/set this flag using get() and set(), although you should rarely need to.

gamma

Live Demo: Gamma Adjust

The gamma() method applies a gamma correction curve to your canvas. Values less than 1.0 will lighten the image, whereas values greater than 1.0 will darken it. The method accepts an object containing the following properties:

Property Name Type Description
amount Float Gamma correction value, typically in the range 0.25 to 4.0. The default value is 1.0 (no change).
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.gamma({
	"amount": 0.5
});

gaussianBlur

Live Demo: Gaussian Blur

The gaussianBlur() method applies a Gaussian blur to your image, using a custom convolution kernel. This is a rather computationally expensive filter, but is much higher quality than the standard blur. It accepts an object containing the following properties:

Property Name Type Description
amount Integer (Required) Blur amount (3 to 100).
sigma Integer Gaussian sigma operator (defaults to amount / 3).
edges String How to calculate out-of-bounds pixels, can be one of repeat, wrap, mirror or discard (case-insensitive). Defaults to repeat.
channels String Which channels to apply the filter to, defaults to rgba. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

The amount is really just the size of one dimension of the Gaussian convolution matrix (it is squared to compute the full matrix size), so the larger the amount the slower the operation is. The sigma actually controls the amount of blurring that occurs. For best results, set sigma to 1/3 of amount (this is the default behavior if sigma is omitted).

Example use:

canvas.gaussianBlur({
	"amount": 9
});

invert

Live Demo: Invert

The invert() method inverts all colors in the image (i.e. creates a photographic negative), or can be restricted to certain specified channels. The method accepts an object containing the following properties:

Property Name Type Description
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.invert();

Inversion is implemented using a curve, and is functionally equivalent to calling:

canvas.curves({
	"rgb": [255, 0]
});

lighting

Live Demo: Shadow Detail

The lighting() method can adjust both shadows and highlights, to bring out hidden details. Internally, this is accomplished by a special multi-point curve. The method accepts an object containing the following properties:

Property Name Type Description
shadows Integer Shadow detail adjustment, from -100 (darken) to 100 (lighten).
highlights Integer Highlight detail adjustment, from -100 (darken) to 100 (lighten).
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.lighting({
	"shadows": 25
});

mask

Live Demo: Apply Mask

The mask() method applies a separate image or canvas as a mask onto the current one. Meaning, the alpha channel of one is applied to the other. The source can be another CanvasPlus object, a native HTML5 Canvas object, or a native Image object. Several parameters are provided for positioning and scaling the mask. The method accepts an object containing the following properties:

Property Name Type Description
image Object (Required) Source image or canvas to apply as a mask.
width Integer Optional desired width, will trigger a resize during composite.
height Integer Optional desired height, will trigger a resize during composite.
gravity String Image starting position (alignment), which can be further adjusted with margins and/or offsets. See Gravity.
marginX Integer Horizontal margin in pixels, defaults to 0.
marginY Integer Vertical margin in pixels, defaults to 0.
offsetX Integer Horizontal offset in pixels, defaults to 0.
offsetY Integer Vertical offset in pixels, defaults to 0.
opacity Float Mask opacity (0.0 to 1.0), defaults to 1.0.
antialias String Mask scaling quality, one of best, good, fast or nearest. Defaults to best. See Anti-Aliasing.

Example use:

canvas.mask({
	"image": my_mask_image,
	"gravity": "center"
});

Masking is effectively the same as calling composite() with the mode set to destination-in.

Note that when width and/or height are specified, the mask is resized in exact (scale) mode.

normalize

Live Demo: Normalize

The normalize() method stretches the image contrast to cover the entire range of possible values. So for example, if the image never reaches full back and/or full white, this will increase the contrast so both sides are maxed. The method accepts an object containing the following properties:

Property Name Type Description
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.normalize();

Normalize is implemented by requesting a histogram to locate the white and black points, and then applying a custom curve.

opacity

Live Demo: Half Transparent

The opacity() method fades the canvas out towards pure transparent, or a custom background color. The method accepts an opacity float, or an object containing the following properties:

Property Name Type Description
opacity Float The target image opacity from 0.0 (transparent) to 1.0 (opaque).
background String Optional background color, defaults to transparent.

Example use:

canvas.opacity({
	opacity: 0.5,
	background: 'red'
});

If you are only specifying opacity (no background), you can just pass the number as the sole argument:

canvas.opacity( 0.5 );

posterize

Live Demo: Posterize

The posterize() method reduces the image to a limited number of color levels per channel. Smaller level amounts have the most obvious effect. The method accepts an object containing the following properties:

Property Name Type Description
levels Integer (Required) Number of levels to allow per channel.
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.posterize({
	"levels": 4
});

Posterize is implemented using a custom stair-step curve.

quantize

Live Demo: Quantize

The quantize() method reduces your image to a fixed number of unique colors, then generates a palette and converts the canvas into indexed mode. This is used for saving GIFs and indexed PNGs. We rely on the awesome image-q library here, which is slow but produces beautiful results. The method accepts an object containing the following properties:

Property Name Type Description
colors Integer (Required) Number of unique colors in the palette (2 - 256).
dither Boolean Dither option for quantization. Defaults to false (disabled).
ditherType String Dithering algorithm for quantization. Defaults to FloydSteinberg.

Example use:

canvas.quantize({
	"colors": 32,
	"dither": true
});

Quantize should always be the last filter applied before calling write(). Calling any other filters after quantization will convert the canvas back to RGBA mode, which is likely not what you want.

The supported dithering algorithms are FloydSteinberg, Stucki, Atkinson, Jarvis, Burkes, and Sierra. They each have different trade-offs, and produce slightly different results. See Dithering Algorithms for more information.

Note that quantization includes and preserves alpha transparency in the palette, unless you first flatten your canvas.

quantizeFast

Live Demo: Quantize Fast

The quantizeFast() method works similarly to quantize(), in that it reduces your image to a fixed number of unique colors, then generates a palette and converts the canvas into indexed mode. However, this algorithm is approximately 10X faster. It does this by "crushing" (a.k.a posterizing) the image, optionally dithering it at the same time, then simply building a palette of all unique colors. This generally produces lower quality results than quantize() (depending on the image), but at a fraction of the CPU cost. The method accepts an object containing the following properties:

Property Name Type Description
colors Integer Target number of colors in the palette, defaults to 256.
crushRGB Integer Optional starting number of posterization levels for the RGB channels, defaults to 16.
crushAlpha Integer Optional starting number of posterization levels for the Alpha channel, defaults to 16.
dither Boolean Optional pattern dither, defaults to false (disabled).

Example use:

canvas.quantizeFast({
	"colors": 256,
	"dither": true
});

The Quantize Fast algorithm will begin by crushing the image at the specified (or default) posterize levels, and see if the target number of colors was achieved. If not, it will crush further by reducing the posterization levels, and keep trying until the palette size is equal to or less than the colors parameter. To optimize for speed, depending on the image and with possible quality loss, you can change the starting crushRGB and/or crushAlpha to values below 16.

It is important to note that Quantize Fast may not achieve the exact number of colors you want in your palette, but it tries to get as close as possible without going over. The algorithm builds its own palette based on the total unique colors after "crushing" (posterizing) the image. You can control the level of posterization of course, and adjust it differently for the RGB and Alpha channels. The higher the crushRGB and crushAlpha values are, the more unique colors are produced. Also, while dithering is provided here as an option, only a pattern dither is supported.

Note that quantization includes and preserves alpha transparency in the palette, unless you first flatten your canvas.

resize

Live Demo: Resize

The resize() method scales the canvas to the desired pixel width and/or height. Several modes are available, to control how your image is scaled to fit the target dimensions. You can also control the gravity (alignment), direction (e.g. shrink only, enlarge only), and anti-alias settings. The method accepts an object containing the following properties:

Property Name Type Description
width Integer Desired canvas width in pixels. Can be omitted if height is specified.
height Integer Desired canvas height in pixels. Can be omitted if width is specified.
mode String Resize mode, one of fit, fitPad, fitOver or scale (case-insensitive). Defaults to fit. See below.
background String Background padding color, for fitPad mode only.
gravity String Image alignment, which can be further adjusted with offsets. See Gravity.
direction String Resize direction restriction, one of shrink, enlarge, or both. Defaults to both.
antialias String Resize scaling quality, one of best, good, fast or nearest. Defaults to best. See Anti-Aliasing.
offsetX Integer Horizontal offset in pixels, defaults to 0.
offsetY Integer Vertical offset in pixels, defaults to 0.
delta Boolean Interpret width and height as delta offsets, not absolute values. Defaults to false.

Example use:

canvas.resize({
	"width": 640,
	"height": 480,
	"mode": "fit"
});

If either width or height are omitted, they will be extrapolated from the other, maintaining the original aspect ratio. Here is an example which scales the canvas down to 50% width, maintaining its aspect ratio:

canvas.resize({
	"width": canvas.get('width') / 2
});

Actually, you can just specify percentage as a string with a % suffix, which has the same effect:

canvas.resize({
	"width": "50%"
});

Here is a description of all the available resize modes:

Resize Mode Alias Description
fit contain The default mode. Will resize so the image "fits" into the destination width/height without exceeding either value, and maintaining the original aspect ratio. Similar to the CSS background-size: contain rule.
fitPad letterbox This mode scales the image similar to fit, but then adds background padding to fill in the extra area (if any), so the final image size exactly matches your desired dimensions, but the original image aspect ratio is also preserved. This is similar to film letterboxing. The background property controls the padding color, and gravity controls the image alignment.
fitOver cover This mode scales the image in such a way that it fits "over" the destination width/height, covering the entire area. The aspect ratio is maintained, and the extra content is cropped off. This is similar to the CSS background-size: cover rule.
scale exact This mode scales the image to your exact specifications, ignoring the aspect ratio, and distorting as needed.

Note that resizing regenerates the underlying canvas object. It effectively creates a new canvas at the target size, then copies the source into the destination, applying the resize at the same time, then finally discards the source.

sepia

Live Demo: Sepia Tone

The sepia() method applies a sepia tone filter to your image, using a custom curve. This is a commonly used filter which gives a photograph an "old western" look. The method accepts an object containing the following properties:

Property Name Type Description
clip Object Optional clipping rectangle (see Clipping below).

Example:

canvas.sepia();

This is equivalent to calling desaturate(), followed by curves() using the following properties:

canvas.desaturate().curves({
	green: [0, 108, 255],
	blue: [0, 64, 255]
});

sharpen

Live Demo: Sharpen

The sharpen() method applies a simple sharpen effect to your image, using a custom 3x3 convolution kernel. It accepts an object containing the following properties:

Property Name Type Description
edges String How to calculate out-of-bounds pixels, can be one of repeat, wrap, mirror or discard (case-insensitive). Defaults to repeat.
channels String Which channels to apply the filter to, defaults to rgba. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.sharpen({
	"edges": "repeat"
});

This is equivalent to calling convolve() with the following properties:

canvas.convolve({ 
	"matrix": [0, -1, 0, -1, 5, -1, 0, -1, 0],
	"edges": "repeat"
});

solarize

Live Demo: Solarize

The solarize() method applies a solarisation filter to your canvas, which basically inverts all colors above half brightness. This can be restricted to certain specified channels if desired. The method accepts an object containing the following properties:

Property Name Type Description
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.solarize();

Solarisation is implemented using a custom sawtooth inverted-V curve, and is similar to calling:

canvas.curves({
	"rgb": [0, 127, 0]
});

However, the solarize() method uses a 256-element curve array, with an exact sawtooth inverted-V shape.

temperature

Live Demo: Color Temperature

The temperature() method simulates increasing or decreasing the image's color temperature. Increasing the temperature produces a "warmer" (redder) image, and decreasing it produces a "cooler" (bluer) image. The method accepts an object containing the following properties:

Property Name Type Description
amount Integer Amount to increase or decrease temperature. Range is -255 to 255.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.temperature({
	"amount": -128
});

This is a very simple implementation of color temperature, which applies a curve to increase the red or blue channels in a linear fashion, based on the supplied amount. This filter is only for visual effect, and doesn't pertain to any photographic or color standard. I'm sure there are much better and more accurate algorithms out there.

text

Live Demos: Basic Meme, Auto-Shrink, Word-Wrap, Shrink-Wrap

The text() method renders text onto your canvas. Beyond the built-in features of the HTML5 Canvas Text API, we provide both character and line spacing, automatic scaling to fit, word-wrap, and the almighty paragraph word-wrap + auto-shrink combo. The method accepts an object containing the following properties:

Property Name Type Description
text String (Required) Actual text to render (can be multi-line).
font String (Required) Font file path (Node.js) or font family name (browser). See Fonts.
size Integer (Required) Text size, measured in points (often equivalent to height in pixels, depending on font family).
color String (Required) Text color (any CSS color string).
background String Optional background color (any CSS color string), defaults to transparent.
gravity String Text alignment, which can be further adjusted with margins and/or offsets. See Gravity.
overflow String How to handle text overflow. See Text Overflow.
marginX Integer Horizontal margin in pixels, defaults to 0.
marginY Integer Vertical margin in pixels, defaults to 0.
offsetX Integer Horizontal offset in pixels, defaults to 0.
offsetY Integer Vertical offset in pixels, defaults to 0.
maxWidth Integer Maximum text render width in pixels, defaults to canvas width (minus margins).
maxHeight Integer Maximum text render height in pixels, defaults to canvas height (minus margins).
characterSpacing Float Adjust horizontal space between each character, with sub-pixel precision. Specify values less than 0 to condense, and greater than 0 to expand.
lineSpacing Float Adjust vertical space between each line, with sub-pixel precision. Specify values less than 0 to condense, and greater than 0 to expand.
shadowColor String Optional shadow color (any CSS color string), defaults to empty string (disabled).
shadowOffsetX Integer When shadowColor is specified, this controls the horizontal offset of the shadow (default to 0).
shadowOffsetY Integer When shadowColor is specified, this controls the vertical offset of the shadow (defaults to 0).
shadowBlur Integer When shadowColor is specified, this controls the shadow blur amount (defaults to 0).
outlineColor String Optional text outline color (any CSS color string), defaults to empty string (disabled).
outlineThickness Integer When outlineColor is specified, this controls the thickness of the outline.
outlineStyle String When outlineColor is specified, this controls the style of the outline (see lineJoin.
autoCrop String Optionally crop canvas down to rendered text size, can be horiz, vert or both. Defaults to disabled.
opacity Float Text opacity (0.0 to 1.0), defaults to 1.0.
mode String Text rendering mode, see globalCompositeOperation, defaults to source-over.

Fonts

When specifying fonts via the font property, please note that things are handled differently if you are running in a browser, vs. using Node.js on the server-side. For Node.js, you need to specify a filesystem path to the font file, in either OpenType (OTF) or TrueType (TTF) format, and preload the font using loadFont() before creating the canvas (see below). The font weight and style are expected to be embedded into the file itself, e.g. Arial-Narrow-bold.otf or Helvetica-bold-italic.ttf. Example:

// Node.js text example
let canvas = new CanvasPlus();
canvas.loadFont( "/path/to/fonts/Arial-Narrow-bold.otf" ); // do this first!
canvas.create({
	width: 640,
	height: 480,
	background: "#FFFFFF"
});

canvas.text({
	"text": "Hello there, Node.js!",
	"font": "/path/to/fonts/Arial-Narrow-bold.otf",
	"size": 48,
	"color": "#ffffff",
	"gravity": "south",
	"marginX": 10,
	"marginY": 10,
	"outlineColor": "#000000",
	"outlineThickness": 4
});

Note that you must load all fonts before creating the canvas in which they will be used. This is a limitation of node-canvas v2. See registerFont for details.

When running CanvasPlus in the browser, things are a bit different. You must specify a CSS font family instead, e.g. Arial Narrow. This can be any built-in system font, or a custom web font that you loaded previously (make sure it is completely loaded before using in CanvasPlus). In addition, if you want to specify a font weight or style, please use the special browser-only fontWeight and/or fontStyle properties, respectively. Example:

// Browser-specific text example
canvas.text({
	"text": "Hello there, browser!",
	"font": "Arial Narrow", // CSS font-family
	"fontWeight": "bold", // CSS font-weight
	"fontStyle": "normal", // CSS font-style
	"size": 48,
	"color": "#ffffff",
	"gravity": "south",
	"marginX": 10,
	"marginY": 10,
	"outlineColor": "#000000",
	"outlineThickness": 4
});

As a convenience, a special version of loadFont() is made available for use in the browser. It accepts a font family name, a URL, and an optional callback. The callback will be fired when the font is completely loaded and ready to use. Example:

// browser font loading example
canvas.loadFont( "Open Sans", "fonts/open-sans-regular.woff2", function(err) {
	if (err) throw err;
	// font is ready to use!
} );

Text Overflow

When your text does not fit into the available width/height (this is the canvas size minus margins and adjusted by maxWidth and/or maxHeight), you have several options on how to handle the overflow. The behavior is controlled by the overflow property, which accepts one of the following values:

Overflow Value Description
"" The default behavior is to take no action on overflow, and simply render the text outside the bounds.
"shrink" This mode will automatically shrink the text until it fits, both horizontally and vertically.
"wrap" This mode will apply automatic word-wrapping to prevent horizontal overflow (it disregards vertical).
"shrinkWrap" This mode will apply both word-wrapping and auto-scale to provide the best fit into the available width/height.

Example use:

canvas.text({
	"text": "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.",
	"font": "/path/to/fonts/Arial-italic.otf",
	"size": 12,
	"color": "rgba(255, 0, 0, 1.0)",
	"gravity": "northwest",
	"overflow": "shrinkwrap",
	"marginX": 10,
	"marginY": 10
});

Note that while shrink is very fast, both wrap and shrinkWrap are rather computationally expensive operations, because they need to flow your text into a smaller space that it was designed for. shrinkWrap in particular is extremely expensive, because it has to constantly re-flow and re-wrap the text as it shrinks, and also it cannot always provide perfectly accurate fits (it always errs on the small side tho). You can help these algorithms by providing a starting font size property that is "somewhat close" to what you will think will fit. This way it doesn't have to work so hard.

Also note that when a single word cannot fit on a line, it is broken into pieces. Automatic hyphenation is not supported, nor is "justify".

One final gotcha is that both the wrapper algorithms require that the text rendered at your provided base size must fit at least one character from your text, or else the entire operation aborts.

threshold

Live Demos: Color Threshold, Black & White Threshold

The threshold() method applies a threshold filter to your image, which essentially minimizes or maximizes the value of selected channels in each pixel, if the value is above or below a threshold level. The method accepts an object containing the following properties:

Property Name Type Description
level Integer Threshold level. Range is 0 to 255.
channels String Which channels to apply the filter to, defaults to rgb. See Channels.
clip Object Optional clipping rectangle (see Clipping below).

Example use:

canvas.threshold({
	"level": 128,
	"channels": "rgb"
});

For a more classic black & white threshold effect, simply call desaturate() before thresholding:

canvas.desaturate().threshold({
	"level": 128,
	"channels": "rgb"
});

transform

Live Demos: Rotate 45 Degrees, Custom Matrix

The transform() method applies transformations to your canvas, which can include rotation, horizontal or vertical flipping, or a custom 3x3 transformation matrix. The method accepts an object containing the following properties:

Property Name Type Description
rotate Float Optional degrees to rotate image.
fliph Boolean Optional flip horizontal transformation.
flipv Boolean Optional flip vertical transformation.
matrix Array Optional 3x3 transformation matrix, for custom distortions.
background String Background color to use when expanding canvas, defaults to transparent.
antialias String Image scaling quality, one of best, good, fast or nearest. Defaults to best. See Anti-Aliasing.
fixed Boolean Set this to true to keep the canvas dimensions fixed, instead of auto-expanding them. Only applicable for rotation.

Example use:

canvas.transform({
	"rotate": 45
});

Only one transformation is allowed per call to transform().

Note that when rotating, the canvas dimensions are automatically adjusted so the rotated image fits exactly into the new container. If you do not want this behavior, set the fixed property to true.

The following transform method shortcuts are provided for convenience:

Method Name Example
rotate canvas.rotate( 45 );
flipHorizontal canvas.flipHorizontal();
flipVertical canvas.flipVertical();

trim

Live Demo: Trim Transparent Edges

The trim() method automatically crops the canvas based on a custom color if provided, or the value of the top-left pixel if not. For example, if the top-left corner pixel is pure transparent, then all transparent pixels are cropped from all sides. You can also specify a tolerance to allow similar colors from being trimmed. The method accepts an object containing the following properties:

Property Name Type Description
color String Optional color to trim. If omitted, will use top-left corner pixel.
tolerance Integer How much pixel value difference to allow when trimming, from 0 to 255. Defaults to 0.

Example (will trim based on top-left pixel):

canvas.trim();

Example with specific trim color and tolerance:

canvas.trim({
	color: "rgba(255, 0, 0, 1.0)",
	tolerance: 32
});

This filter will silently fail if no pixels are available for trimming, or if it is about to trim the entire image.

Accessors

getLastError

The getLastError() method returns the last error that occurred, or null if the last operation was a success.

See Errors for details.

getMetrics

The getMetrics() method returns an object that contains performance metrics for all operations that have taken place. This uses the pixl-perf module, and all metrics are in milliseconds. The format is as follows:

{
	"scale": 1000,
	"perf": {
		"total": 217.399,
		"download": 3.7,
		"read": 8.8,
		"resize": 0.6,
		"adjust": 61.299,
		"convolve": 41.6,
		"curves": 15.3,
		"desaturate": 11.3,
		"write": 18.1
	},
	"counters": {
		"bytes_read": 1040704,
		"bytes_written": 82510
	}
}

getPerf

The getPerf() method returns the current performance tracking object, which is from the pixl-perf module. This is so you can add your own metrics to it.

getDimensions

The getDimensions() method returns an object with the current canvas width and height in pixels. Example:

{
	"width": 640,
	"height": 480
}

getEXIF

The getEXIF() method returns an object containing all the parsed EXIF metadata obtained from the image when it was first loaded. This is only applicable if you loaded an image (as opposed to a blank canvas), and it also only works for JPEG images. Example snippet of data:

{
	"Make": "Apple",
	"Model": "iPhone 7",
	"Orientation": 1,
	"XResolution": 72,
	"YResolution": 72,
	"ResolutionUnit": 2,
	"Software": "10.3.1",
	"DateTime": "2017:04:29 16:36:23",
	...
}

We use the exifreader module for parsing this data.

Please note that if you want to capture EXIF data from images, you should do it immediately after loading. Any transforms or filters that rebuild the canvas will scrub the EXIF data.

getCanvas

The getCanvas() method returns the underlying HTML5 Canvas object currently in use by CanvasPlus. Please note that this is only applicable in RGBA mode (see Modes below), and that the canvas is often discarded and rebuilt by many filters and transforms.

getContext

The getContext() method returns the underlying CanvasRenderingContext2D object currently in use by CanvasPlus. Please note that this is only applicable in RGBA mode (see Modes below), and that the context is often discarded and rebuilt by many filters and transforms.

getPixels

The getPixels() method returns a direct pointer to the raw pixels in the current canvas. The format varies depending on the current mode (see Modes below).

In rgba mode, getPixels() returns a Uint8ClampedArray of RGBA pixels, obtained from the data property of the current canvas ImageData. Each element in the array is one 8-bit channel value of one pixel, and every 4 elements is one full 32-bit pixel. Example:

let width = canvas.get('width');
let height = canvas.get('height');
let data = canvas.getPixels();
let offset = 0;

for (let y = 0; y < height; y++) {
	// foreach row
	for (let x = 0; x < width; x++) {
		// for each pixel
		let red = data[ offset + 0 ];
		let green = data[ offset + 1 ];
		let blue = data[ offset + 2 ];
		let alpha = data[ offset + 3 ];
		offset += 4;
	} // x loop
} // y loop

In indexed mode, getPixels() returns a Uint8Array of 8-bit palette indexes. Each value will be between 0 and 255, representing an index in the palette array. See getPalette() below. Example:

canvas.quantize({ colors: 256, dither: true });

let width = canvas.get('width');
let height = canvas.get('height');
let pixels = canvas.getPixels();
let colors = canvas.getPalette();
let offset = 0;

for (let y = 0; y < height; y++) {
	// foreach row
	for (let x = 0; x < width; x++) {
		// for each pixel
		let color = colors[ pixels[offset] ];
		let red = color.r;
		let green = color.g;
		let blue = color.b;
		let alpha = color.a;
		offset++;
	} // x loop
} // y loop

If getPixels() is called while in image mode, the canvas is converted to rgba first.

getPalette

The getPalette() method returns an array of indexed palette colors. This is only applicable in indexed mode (see Modes below). The palette array may contain up to 256 colors, and each array element is an object containing the following properties:

Property Name Description
r Red channel value, from 0 - 255.
g Green channel value, from 0 - 255.
b Blue channel value, from 0 - 255.
a Alpha channel value, from 0 - 255.
uint32 A 32-bit integer value representing the palette color as a "pixel".

To convert your canvas into indexed mode, call quantize() or quantizeFast().

get

The get() method is a generic accessor for fetching named parameters on the CanvasPlus object. Using this you can retrieve the current canvas width, height, image mode, and more. Many are used as defaults for filters. See Parameters below for the full list. Example:

let width = canvas.get('width');
let height = canvas.get('height');

width

The width() method is an alias for get('width'). It returns the current canvas width in pixels. Example:

let width = canvas.width();

height

The height() method is an alias for get('height'). It returns the current canvas height in pixels. Example:

let height = canvas.height();

Misc

set

The set() method sets an arbitrary named parameter on the CanvasPlus object. Using this you can disable features such as automatic EXIF rotation correction, enable debug mode, and more. See Parameters below for the full list. Example:

canvas.set( 'autoOrient', false ); // do not auto-rotate based on EXIF data
canvas.set( 'debug', true ); // output debugging data to console

clearLastError

The clearLastError() method clears the previous error state, if any. Afterwards, getLastError() will return null. This method is called internally by every filter and transform just prior to executing.

attachLogAgent

The attachLogAgent() method allows you to attach your own debug and error logging agent (object). The object should have debug and error methods defined, which accept three arguments (code, message, and data). See Logging above for more details and an example implementation.

importImage

The importImage() method allows you to import your own native HTML Image object into CanvasPlus. Please note that the image must be completely loaded before importing, otherwise an error will be raised. Example use:

let img = new Image();
img.onload = function() {
	canvas.importImage( img );
};
img.src = "my-image.jpg";

The canvas will be completely replaced with the image, and inherit its exact width, height, and content. This also sets the internal mode to image.

importCanvas

The importCanvas() method allows you to import your own native HTML5 Canvas object into CanvasPlus. Example:

canvas.importCanvas( document.getElementById('mycanvas') );

The canvas used by CanvasPlus will be completely replaced with the new canvas, and inherit its exact width, height, and content. This also sets the internal mode to rgba.

reset

The reset() method completely resets CanvasPlus, removing all pixels, data and reverting settings to defaults. It is effectively the same as destroying and recreating the object. Example:

canvas.reset();

render

The render() method converts CanvasPlus into rgba mode, rendering the internal image or indexed pixels onto the canvas. This is only applicable when in image or indexed mode (see Modes below), and is called automatically by most filters and transforms. You should rarely need to call it directly, but it is provided just in case. Example:

canvas.render();

histogram

The histogram() method generates an image histogram of the current canvas pixels, returning an object with the following properties:

Property Name Type Description
red Array An array of all the red histogram counts per value.
green Array An array of all the green histogram counts per value.
blue Array An array of all the blue histogram counts per value.
alpha Array An array of all the alpha histogram counts per value.
redMax Integer The maximum red value across all the array elements.
greenMax Integer The maximum green value across all the array elements.
blueMax Integer The maximum blue value across all the array elements.
alphaMax Integer The maximum alpha value across all the array elements.

Example use:

let histo = canvas.histogram();

For example, let's say you had an image that contained a large amount of pure black (rgb(0, 0, 0)). Your histogram would therefore contain high numbers for the first (zeroth) array indexes in each color channel. The array indexes are the color channel brightnesses, and the values are the counts of pixels in that specific brightness slice. In this case histo.red[0], histo.green[0] and histo.blue[0] would all contain high numbers, because those elements are literally "counts" of the darkest pixels.

You can use the redMax, greenMax and blueMax properties to determine the maximum slice counts across the arrays. These maximums can be used to normalize the data and scale it down to fit into a 2D histogram display (see the playground for an example of this).

The histogram feature only works in rgba mode (see Modes below), and will convert the canvas into RGBA if necessary.

hash

The hash() method generates a perceptual hash of the current canvas pixels, using the fast Blockhash algorithm. This is basically a unique ID or thumbprint for the image, represented as a 16-character hexadecimal string, which is almost always the same when compared to similar images. Example use:

let hash = canvas.hash();
// "02fe3c3c7ca2471b"

The basic idea with these hash IDs is that you can compare them using a Hamming distance, and see exactly how similar two images are. CanvasPlus provides a hammingDistance() method for you to use for this purpose. Here is an example:

// compute initial hash
let hash1 = canvas.hash();

// filter image
canvas.adjust({
	brightness: -40,
	contrast: 40,
	hue: 180,
	saturation: 50
});

// compute new hash
let hash2 = canvas.hash();

// get hamming distance
let dist = canvas.hammingDistance( hash1, hash2 );

console.log("Initial Image Hash: " + hash1);
console.log("Final Image Hash:   " + hash2);
console.log("hash Distance: " + dist);

Example output:

Initial Image Hash: 02fe3c3c7c92471b
Final Image Hash:   02fe3c3c7ca2471b
Hash Distance: 2

Note that the Blockhash algorithm has some limitations:

For images in general, the algorithm generates the same blockhash value for two different images in 1% of the cases (data based on a random sampling of 100,000 images).

For photographs, the algorithm generates practically unique blockhashes, but for icons, clipart, maps and other images, the algorithm generates less unique blockhashses. Larger areas of the same color in an image, either as a background or borders, result in hashes that collide more frequently.

The CanvasPlus hash() method only works in rgba mode (see Modes below), and will convert the canvas into RGBA if necessary.

Modes

CanvasPlus has three internal "modes" which specify how your image data is represented in memory:

Mode Description
image This mode is set when an image is loaded, but not yet rendered onto a canvas. load() sets this mode.
rgba This mode means the pixels are stored in a RGBA canvas. This is the primary mode of operation. create() sets this mode.
indexed This mode is only set when the image is quantized, and is only used for outputting a GIF or indexed PNG image.

Modes are automatically changed according to how the library is used. For example, all filters require an RGBA canvas, so the mode is automatically converted to rgba if necessary. Only quantize() and quantizeFast() convert the mode to indexed.

If you request access to the underlying canvas object via getCanvas() or getContext(), and the canvas is in either image or indexed mode, it is converted back to rgba. getPixels() works with both rgba and indexed modes.

The reason why image mode exists is that the first filter applied after loading is very often a resize(), and it is much faster to delay rendering of the image until resize time, so it can be done in one single operation.

You can use get() to query the current mode, but you should rarely need to.

Channels

CanvasPlus works in "RGBA" mode, which stands for Red, Green, Blue, and Alpha. Many filters can customize which channels they are applied to, and take a single channels property. This property is a string, which expects the abbreviated form of the channels to act upon, e.g. rgba for all channels including alpha. Here are some more examples:

Channels Description
rgba Apply to all color channels including the alpha channel (red, green, blue and alpha).
rgb Apply to only the RGB color channels (red, green and blue) excluding the alpha channel.
r Apply to only the red channel.
g Apply to only the green channel.
b Apply to only the blue channel.
a Apply to only the alpha channel.
gb Apply to the green and blue channels only.
ra Apply to the red and alpha channels only.

Gravity

Here are all the possible gravity parameter values, and what they mean. Note that all gravity strings are case-insensitive.

Gravity Value Alias Description
center - Center both horizontally and vertically. This is typically the default.
northwest topleft Pin to top-left corner.
north top Pin to top, center horizontally.
northeast topright Pin to top-right corner.
east right Pin to right side, center vertically.
southeast bottomright Pin to bottom-right corner.
south bottom Pin to bottom, center horizontally.
southwest bottomleft Pin to bottom-left corner.
west left Pin to left side, center vertically.

Anti-Aliasing

The antialias parameter is used by several filters including resize, transform, composite, mask, and draw. Here are all the possible antialias parameter values, and what they mean. Note that this is highly implementation-dependent, as in your results may vary between browsers and Node.js canvas. All strings are case-insensitive. The default is always best.

Value Description
best Highest quality setting, but also the slowest. In the browser imageSmoothingEnabled is enabled and imageSmoothingQuality is set to high.
good Medium quality setting. In the browser imageSmoothingEnabled is enabled and imageSmoothingQuality is set to medium.
fast Low quality setting. In the browser imageSmoothingEnabled is enabled and imageSmoothingQuality is set to low.
nearest Lowest quality setting, but also the fastest. In the browser imageSmoothingEnabled is disabled.

The word nearest refers to nearest-neighbor interpolation, and results in pixelated transforms (no anti-aliasing).

Clipping

Many methods accept a clip property, which allows you to optionally restrict the filter to a rectangle inside the canvas. This is similar to the marquee rectangle selection tool in many paint apps. If specified, clip should be an object with the following properties:

Property Name Type Description
x Integer Coordinate of the left side of the clipping rectangle.
y Integer Coordinate of the top side of the clipping rectangle.
width Integer Width of the clipping rectangle, in pixels.
height Integer Height of the clipping rectangle, in pixels.

Example use:

canvas.sepia({
	clip: { x: 50, y: 50, width: 100, height: 100 }
});

Parameters

Here are all the general parameters you can access using get() and change using set():

Parameter Name Type Description
readEXIF Boolean Automatically read image EXIF data on load. Defaults to true (enabled).
autoOrient Boolean Automatically correct image rotation on load, using EXIF data. Defaults to true (enabled).
throw Boolean Enables try/catch mode (will throw on error). Defaults to false (disabled).
debug Boolean Enables debug logging mode using console.log and console.error. Defaults to false (disabled).
gravity String Default gravity (alignment) for filters such as composite(), expand(), resize() and text(). Defaults to center.
colors Integer Default number of palette colors for quantization. Defaults to 256.
dither Boolean Default dither option for quantization. Defaults to false (disabled).
ditherType String Default dither algorithm for quantization. Defaults to FloydSteinberg.
alpha Boolean Flag indicating whether canvas contains alpha (transparent) pixels or not. Used by GIF and PNG output.
mode String Current internal mode of the canvas. Do not change this manually.
width Number Current canvas width in pixels. Do not change this manually (use resize()).
height Number Current canvas height in pixels. Do not change this manually (use resize()).
maxArea Number Maximum allowed canvas area in square pixels (designed to limit memory usage). Defaults to 0 (no limit).

Node Parameters

The following parameters are only applicable in Node.js, when using the canvas module:

Parameter Name Type Description
progressive Boolean Optionally save JPEGs in progressive scan mode. Defaults to false.
chromaSubsampling Boolean Optionally save JPEGs with chroma subsampling. Defaults to true.
compressionLevel Integer Compression level for PNG images. Defaults to 6.
pngFilter String PNG filter algorithm, for 32-bit PNG images. Defaults to PNG_ALL_FILTERS. See below.
webp Object WebP encoding settings. See webp-wasm for details.

The pngFilter parameter specifies the algorithm for preparing the PNG data for compression. The available values are: PNG_ALL_FILTERS, PNG_FILTER_SUB, PNG_FILTER_UP, PNG_FILTER_AVG, PNG_FILTER_PAETH, and PNG_FILTER_NONE. This is only applicable for writing 32-bit PNGs. See the PNG Filter Spec for more information.

Browser Parameters

The following parameters are only applicable when running in the browser:

Parameter Name Type Description
fontWeight String Set the font weight, e.g. bold (defaults to normal).
fontStyle String Set the font style, e.g. italic (defaults to normal).
useDataURLs Boolean Use Data URLs instead of Blobs when loading and saving images. This may improve performance when using headless Chrome on Linux.

Development

To compile the browser version of the library from the Node.js source, you will need to have browserify installed. Then change into the main package directory and type:

npm run browserify

This should generate the canvas-plus.js standalone file.

Acknowledgments

CanvasPlus would not be possible without these awesome open-source packages:

Module Name Description License
canvas Cairo backed Canvas implementation for Node.js. MIT
image-q Complete Image Quantization Library MIT
omggif GIF 89a encoder and decoder MIT
exifreader Read EXIF meta data from image files. MPL 2.0
browserify Browser-side require() the node.js way MIT
squoosh Make images smaller using best-in-class codecs, right in the browser. Apache 2.0

Also, special thanks go to:

License

The MIT License (MIT)

Copyright (c) 2017 - 2023 Joseph Huckaby.

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

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

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

Please note that the exifreader package, one of our dependencies, is MPL 2.0 licensed. This license is a bit more restrictive than the MIT license, so please use with caution. If you don't need any EXIF features, it would be relatively easy to fork canvas-plus and remove the dependency.

Please note that the WebP implementation we use in Node.js is borrowed from Squoosh, which is Apache 2.0 licensed. This license is a bit more restrictive than the MIT license, so please use with caution. If you don't need any WebP features, it would be relatively easy to fork canvas-plus and remove the dependency.