sharp, jimp, canvas, gm, and image-size are Node.js libraries used for handling image files, but they serve different architectural needs. sharp is a high-performance module built on libvips, optimized for resizing and format conversion in production environments. jimp is a pure JavaScript library with zero system dependencies, making it portable but slower. canvas (node-canvas) implements the HTML5 Canvas API for Node, ideal for dynamic image generation and text rendering. gm wraps GraphicsMagick or ImageMagick, offering powerful features but requiring external system binaries. image-size is a lightweight utility dedicated solely to reading image dimensions without processing the file content.
When building Node.js applications that handle images, choosing the right library impacts performance, deployment complexity, and feature availability. sharp, jimp, canvas, gm, and image-size each take a different approach to solving image problems. Let's compare how they handle common engineering tasks.
The core difference lies in how these packages process pixels. This dictates speed and installation requirements.
sharp uses native C++ bindings to libvips.
// sharp: Native binding usage
const sharp = require('sharp');
// Installs with prebuilt libvips binaries for most OS
jimp is written entirely in JavaScript.
// jimp: Pure JavaScript
const Jimp = require('jimp');
// Zero system dependencies, pure JS implementation
canvas wraps the Cairo graphics library.
// canvas: Cairo binding
const { createCanvas } = require('canvas');
// Needs system libs: libcairo2-dev, libpango1.0-dev, etc.
gm spawns child processes for GraphicsMagick/ImageMagick.
// gm: System binary wrapper
const gm = require('gm');
// Requires 'graphicsmagick' or 'imagemagick' installed on OS
image-size reads file headers directly.
// image-size: Header reader
const sizeOf = require('image-size');
// Reads bytes to determine dimensions, no image decoding
Resizing is the most common operation. The API style varies from promises to callbacks.
sharp uses a chainable promise-based API.
// sharp: Resize with promises
await sharp('input.jpg')
.resize(800, 600)
.toFile('output.jpg');
jimp uses a promise-based chain on the image object.
// jimp: Resize with chains
await Jimp.read('input.jpg')
.then(img => img.resize(800, 600))
.then(img => img.write('output.jpg'));
canvas requires manual drawing onto a new surface.
// canvas: Manual draw and resize
const { createCanvas, loadImage } = require('canvas');
const img = await loadImage('input.jpg');
const canvas = createCanvas(800, 600);
const ctx = canvas.getContext('2d');
ctx.drawImage(img, 0, 0, 800, 600);
// Write buffer to file using fs
gm uses a callback-based or promise-wrapped API.
// gm: Resize with callbacks
const gm = require('gm').subClass({ imageMagick: true });
gm('input.jpg').resize(800, 600).write('output.jpg', (err) => {
if (err) console.error(err);
});
image-size does not support resizing.
// image-size: Resizing NOT supported
// This package only reads dimensions, cannot modify images
const dimensions = sizeOf('input.jpg');
// Cannot resize, use sharp or jimp for modification
Knowing dimensions before processing is vital for validation.
sharp provides metadata via a dedicated method.
// sharp: Get metadata
const metadata = await sharp('input.jpg').metadata();
console.log(metadata.width, metadata.height);
jimp exposes dimensions on the loaded object.
// jimp: Get dimensions
const img = await Jimp.read('input.jpg');
console.log(img.bitmap.width, img.bitmap.height);
canvas exposes dimensions on the loaded image object.
// canvas: Image properties
const img = await loadImage('input.jpg');
console.log(img.width, img.height);
gm includes size in the identify command.
// gm: Identify size
gm('input.jpg').identify((err, features) => {
console.log(features.size);
});
image-size is specialized for this exact task.
// image-size: Dedicated dimension reader
const dimensions = sizeOf('input.jpg');
console.log(dimensions.width, dimensions.height);
Some tasks require creating images from scratch or overlaying text.
sharp has limited drawing capabilities.
// sharp: Composite overlay
await sharp('background.jpg')
.composite([{ input: 'logo.png', top: 10, left: 10 }])
.toFile('output.jpg');
jimp supports basic text and drawing.
// jimp: Print text
const img = await Jimp.read('background.jpg');
const font = await Jimp.loadFont(Jimp.FONT_SANS_16_WHITE);
img.print(font, 10, 10, 'Hello World');
img.write('output.jpg');
canvas excels at text and vector drawing.
// canvas: Draw text and shapes
const ctx = canvas.getContext('2d');
ctx.font = '20px Arial';
ctx.fillText('Hello World', 10, 50);
ctx.fillRect(0, 0, 100, 100);
gm supports text via ImageMagick commands.
// gm: Draw text
const image = gm('background.jpg');
image.draw('text 10,50 "Hello World"').write('output.jpg', cb);
image-size cannot draw or create images.
// image-size: No drawing capabilities
// Cannot create or modify image content
How the package installs affects your CI/CD pipeline and container size.
sharp downloads binaries automatically.
libc6-compat on Alpine Linux.# sharp: Docker example
FROM node:18-alpine
RUN apk add --no-cache libc6-compat
# npm install sharp downloads binaries
jimp installs like any JS package.
# jimp: Docker example
FROM node:18-alpine
# No system deps needed
COPY package*.json .
RUN npm install
canvas requires build tools and system libs.
# canvas: Docker example
FROM node:18-alpine
RUN apk add --no-cache build-base glib-dev libpng-dev jpeg-dev
# Compilation happens during npm install
gm requires the binary installed on the OS.
# gm: Docker example
FROM node:18-alpine
RUN apk add --no-cache graphicsmagick
# Ensure binary matches expected version
image-size has no special requirements.
# image-size: Docker example
FROM node:18-alpine
# Standard node image works fine
Despite their differences, these libraries share some common patterns and goals.
sharp, jimp, canvas, gm) accept file paths or buffers.// Shared pattern: Buffer handling
// sharp
const buffer = await sharp(input).toBuffer();
// jimp
const buffer = await img.getBufferAsync(Jimp.MIME_JPEG);
sharp and jimp use Promises natively.gm traditionally uses callbacks but supports promises via wrappers.// Shared pattern: Async/Await
await processImage('input.jpg'); // Works for sharp, jimp, canvas
// Shared pattern: Try/Catch
try {
await sharp('invalid.jpg').metadata();
} catch (e) {
console.error('Invalid image file');
}
| Feature | sharp | jimp | canvas | gm | image-size |
|---|---|---|---|---|---|
| Primary Use | Processing | Processing | Drawing | Processing | Metadata |
| Language | C++ / JS | Pure JS | C++ / JS | System Bin | Pure JS |
| Resize | ✅ Fast | ✅ Slow | ✅ Manual | ✅ Fast | ❌ No |
| Text/Draw | ⚠️ Limited | ✅ Basic | ✅ Advanced | ✅ Advanced | ❌ No |
| Dependencies | Binaries | None | System Libs | System Bin | None |
| API Style | Promises | Promises | Promises | Callbacks | Sync/Async |
| Package | Speed | Install Effort | Memory Usage | Best For |
|---|---|---|---|---|
sharp | ⚡ Very High | Low (Auto-bin) | Low | Production APIs |
jimp | 🐢 Low | None | High | Simple Scripts |
canvas | ⚡ High | High (Build) | Medium | Dynamic Graphics |
gm | ⚡ High | High (System) | Medium | Legacy Systems |
image-size | ⚡ Instant | None | Low | Validation |
sharp is the industry standard for modern Node.js image processing. It balances speed and ease of use better than any other option. Use it for resizing user uploads, generating thumbnails, or converting formats in production.
jimp is the fallback for when you cannot install native modules. It works in browsers and restricted server environments but will struggle with large files or high concurrency.
canvas is the choice for creative tasks. If you need to draw charts, add dynamic text, or composite layers like a design tool, this provides the most control.
gm is a legacy tool. While powerful, the requirement to manage system binaries and its callback-heavy API makes it less attractive for new greenfield projects.
image-size is a utility, not a processor. Keep it in your toolkit for quick validation checks before handing files off to sharp or jimp.
Final Thought: For most backend image tasks, start with sharp. It solves 90% of problems with the least friction. Reach for canvas only when you need to draw, and jimp only when you must avoid native dependencies.
Choose gm only if you are maintaining a legacy system that already depends on GraphicsMagick or ImageMagick. For new projects, it is generally recommended to avoid this package in favor of sharp due to better performance and easier deployment.
Choose canvas when you need to draw images from scratch, render text onto images, or manipulate pixels using the familiar HTML5 Canvas API. It is ideal for generating charts, memes, or dynamic social media cards server-side.
Choose image-size when you strictly need to read image width and height without modifying the file. It is perfect for validation steps before upload or calculating layout dimensions without the overhead of a full processing library.
Choose jimp if you need a pure JavaScript solution that runs anywhere without compiling native modules or installing system libraries. It is suitable for low-volume tasks, prototyping, or environments where installing native dependencies is impossible.
Choose sharp for production-grade image processing where performance and memory efficiency are critical. It is the best fit for resizing, cropping, and converting formats in high-traffic APIs or serverless functions due to its native bindings and prebuilt binaries.
GraphicsMagick and ImageMagick for node
When reporting bugs please include the version of graphicsmagick/imagemagick you're using (gm -version/convert -version) as well as the version of this module and copies of any images you're having problems with.
First download and install GraphicsMagick or ImageMagick. In Mac OS X, you can simply use Homebrew and do:
brew install imagemagick
brew install graphicsmagick
then either use npm:
npm install gm
or clone the repo:
git clone git://github.com/aheckmann/gm.git
Subclass gm to enable ImageMagick 7+
const fs = require('fs')
const gm = require('gm').subClass({ imageMagick: '7+' });
Or, to enable ImageMagick legacy mode (for ImageMagick version < 7)
const fs = require('fs')
const gm = require('gm').subClass({ imageMagick: true });
Optionally specify the path to the executable.
const fs = require('fs')
const gm = require('gm').subClass({
appPath: String.raw`C:\Program Files\ImageMagick-7.1.0-Q16-HDRI\magick.exe`
});
var fs = require('fs')
, gm = require('gm');
// resize and remove EXIF profile data
gm('/path/to/my/img.jpg')
.resize(240, 240)
.noProfile()
.write('/path/to/resize.png', function (err) {
if (!err) console.log('done');
});
// some files would not be resized appropriately
// http://stackoverflow.com/questions/5870466/imagemagick-incorrect-dimensions
// you have two options:
// use the '!' flag to ignore aspect ratio
gm('/path/to/my/img.jpg')
.resize(240, 240, '!')
.write('/path/to/resize.png', function (err) {
if (!err) console.log('done');
});
// use the .resizeExact with only width and/or height arguments
gm('/path/to/my/img.jpg')
.resizeExact(240, 240)
.write('/path/to/resize.png', function (err) {
if (!err) console.log('done');
});
// obtain the size of an image
gm('/path/to/my/img.jpg')
.size(function (err, size) {
if (!err)
console.log(size.width > size.height ? 'wider' : 'taller than you');
});
// output all available image properties
gm('/path/to/img.png')
.identify(function (err, data) {
if (!err) console.log(data)
});
// pull out the first frame of an animated gif and save as png
gm('/path/to/animated.gif[0]')
.write('/path/to/firstframe.png', function (err) {
if (err) console.log('aaw, shucks');
});
// auto-orient an image
gm('/path/to/img.jpg')
.autoOrient()
.write('/path/to/oriented.jpg', function (err) {
if (err) ...
})
// crazytown
gm('/path/to/my/img.jpg')
.flip()
.magnify()
.rotate('green', 45)
.blur(7, 3)
.crop(300, 300, 150, 130)
.edge(3)
.write('/path/to/crazy.jpg', function (err) {
if (!err) console.log('crazytown has arrived');
})
// annotate an image
gm('/path/to/my/img.jpg')
.stroke("#ffffff")
.drawCircle(10, 10, 20, 10)
.font("Helvetica.ttf", 12)
.drawText(30, 20, "GMagick!")
.write("/path/to/drawing.png", function (err) {
if (!err) console.log('done');
});
// creating an image
gm(200, 400, "#ddff99f3")
.drawText(10, 50, "from scratch")
.write("/path/to/brandNewImg.jpg", function (err) {
// ...
});
// passing a stream
var readStream = fs.createReadStream('/path/to/my/img.jpg');
gm(readStream, 'img.jpg')
.write('/path/to/reformat.png', function (err) {
if (!err) console.log('done');
});
// passing a downloadable image by url
var request = require('request');
var url = "www.abc.com/pic.jpg"
gm(request(url))
.write('/path/to/reformat.png', function (err) {
if (!err) console.log('done');
});
// can also stream output to a ReadableStream
// (can be piped to a local file or remote server)
gm('/path/to/my/img.jpg')
.resize('200', '200')
.stream(function (err, stdout, stderr) {
var writeStream = fs.createWriteStream('/path/to/my/resized.jpg');
stdout.pipe(writeStream);
});
// without a callback, .stream() returns a stream
// this is just a convenience wrapper for above.
var writeStream = fs.createWriteStream('/path/to/my/resized.jpg');
gm('/path/to/my/img.jpg')
.resize('200', '200')
.stream()
.pipe(writeStream);
// pass a format or filename to stream() and
// gm will provide image data in that format
gm('/path/to/my/img.jpg')
.stream('png', function (err, stdout, stderr) {
var writeStream = fs.createWriteStream('/path/to/my/reformatted.png');
stdout.pipe(writeStream);
});
// or without the callback
var writeStream = fs.createWriteStream('/path/to/my/reformatted.png');
gm('/path/to/my/img.jpg')
.stream('png')
.pipe(writeStream);
// combine the two for true streaming image processing
var readStream = fs.createReadStream('/path/to/my/img.jpg');
gm(readStream)
.resize('200', '200')
.stream(function (err, stdout, stderr) {
var writeStream = fs.createWriteStream('/path/to/my/resized.jpg');
stdout.pipe(writeStream);
});
// GOTCHA:
// when working with input streams and any 'identify'
// operation (size, format, etc), you must pass "{bufferStream: true}" if
// you also need to convert (write() or stream()) the image afterwards
// NOTE: this buffers the readStream in memory!
var readStream = fs.createReadStream('/path/to/my/img.jpg');
gm(readStream)
.size({bufferStream: true}, function(err, size) {
this.resize(size.width / 2, size.height / 2)
this.write('/path/to/resized.jpg', function (err) {
if (!err) console.log('done');
});
});
// A buffer can be passed instead of a filepath as well
var buf = require('fs').readFileSync('/path/to/image.jpg');
gm(buf, 'image.jpg')
.noise('laplacian')
.write('/path/to/out.jpg', function (err) {
if (err) return handle(err);
console.log('Created an image from a Buffer!');
});
/*
A buffer can also be returned instead of a stream
The first argument to toBuffer is optional, it specifies the image format
*/
gm('img.jpg')
.resize(100, 100)
.toBuffer('PNG',function (err, buffer) {
if (err) return handle(err);
console.log('done!');
})
If gm does not supply you with a method you need or does not work as you'd like, you can simply use gm().in() or gm().out() to set your own arguments.
gm().command() - Custom command such as identify or convertgm().in() - Custom input argumentsgm().out() - Custom output argumentsThe command will be formatted in the following order:
command - ie convertin - the input argumentssource - stdin or an image fileout - the output argumentsoutput - stdout or the image file to write toFor example, suppose you want the following command:
gm "convert" "label:Offline" "PNG:-"
However, using gm().label() may not work as intended for you:
gm()
.label('Offline')
.stream();
would yield:
gm "convert" "-label" "\"Offline\"" "PNG:-"
Instead, you can use gm().out():
gm()
.out('label:Offline')
.stream();
which correctly yields:
gm "convert" "label:Offline" "PNG:-"
When identifying an image, you may want to use a custom formatting string instead of using -verbose, which is quite slow.
You can use your own formatting string when using gm().identify(format, callback).
For example,
gm('img.png').format(function (err, format) {
})
// is equivalent to
gm('img.png').identify('%m', function (err, format) {
})
since %m is the format option for getting the image file format.
Please document and refer to any platform or ImageMagick/GraphicsMagick issues/differences here.
Check out the examples directory to play around. Also take a look at the extending gm page to see how to customize gm to your own needs.
There are a few ways you can use the gm image constructor.
gm(path) When you pass a string as the first argument it is interpreted as the path to an image you intend to manipulate.gm(stream || buffer, [filename]) You may also pass a ReadableStream or Buffer as the first argument, with an optional file name for format inference.gm(width, height, [color]) When you pass two integer arguments, gm will create a new image on the fly with the provided dimensions and an optional background color. And you can still chain just like you do with pre-existing images too. See here for an example.The links below refer to an older version of gm but everything should still work, if anyone feels like updating them please make a PR
getters
manipulation
drawing primitives
image output
ReadableStream with the processed image dataBuffer instead of a streamGraphicsmagicks compare command is exposed through gm.compare(). This allows us to determine if two images can be considered "equal".
Currently gm.compare only accepts file paths.
gm.compare(path1, path2 [, options], callback)
gm.compare('/path/to/image1.jpg', '/path/to/another.png', function (err, isEqual, equality, raw, path1, path2) {
if (err) return handle(err);
// if the images were considered equal, `isEqual` will be true, otherwise, false.
console.log('The images were equal: %s', isEqual);
// to see the total equality returned by graphicsmagick we can inspect the `equality` argument.
console.log('Actual equality: %d', equality);
// inspect the raw output
console.log(raw);
// print file paths
console.log(path1, path2);
})
You may wish to pass a custom tolerance threshold to increase or decrease the default level of 0.4.
gm.compare('/path/to/image1.jpg', '/path/to/another.png', 1.2, function (err, isEqual) {
...
})
To output a diff image, pass a configuration object to define the diff options and tolerance.
var options = {
file: '/path/to/diff.png',
highlightColor: 'yellow',
tolerance: 0.02
}
gm.compare('/path/to/image1.jpg', '/path/to/another.png', options, function (err, isEqual, equality, raw) {
...
})
GraphicsMagick supports compositing one image on top of another. This is exposed through gm.composite(). Its first argument is an image path with the changes to the base image, and an optional mask image.
Currently, gm.composite() only accepts file paths.
gm.composite(other [, mask])
gm('/path/to/image.jpg')
.composite('/path/to/second_image.jpg')
.geometry('+100+150')
.write('/path/to/composite.png', function(err) {
if(!err) console.log("Written composite image.");
});
GraphicsMagick supports montage for combining images side by side. This is exposed through gm.montage(). Its only argument is an image path with the changes to the base image.
Currently, gm.montage() only accepts file paths.
gm.montage(other)
gm('/path/to/image.jpg')
.montage('/path/to/second_image.jpg')
.geometry('+100+150')
.write('/path/to/montage.png', function(err) {
if(!err) console.log("Written montage image.");
});
https://github.com/aheckmann/gm/contributors
http://github.com/quiiver/magickal-node
https://github.com/aheckmann/gm/wiki
npm test
To run a single test:
npm test -- alpha.js
(The MIT License)
Copyright (c) 2010 Aaron Heckmann
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.