sharp

The typical use case for this high speed Node.js module is to convert large images of many formats to smaller, web-friendly JPEG, PNG and WebP images of varying dimensions.

The performance of JPEG resizing is typically 8x faster than ImageMagick and GraphicsMagick, based mainly on the number of CPU cores available. Everything remains non-blocking thanks to libuv and Promises/A+ are supported.

This module supports reading and writing images of JPEG, PNG and WebP to and from both Buffer objects and the filesystem. It also supports reading images of many other types from the filesystem via libmagick++ or libgraphicsmagick++ if present.

When generating JPEG output all metadata is removed and Huffman tables optimised without having to use separate command line tools like jpegoptim and jpegtran.

Anyone who has used the Node.js bindings for GraphicsMagick will find the API similarly fluent.

This module is powered by the blazingly fast libvips image processing library, originally created in 1989 at Birkbeck College and currently maintained by John Cupitt.

Installation

npm install sharp

Prerequisites

Node.js v0.10+
libvips v7.38.5+

libvips can take advantage of liborc if present. Warning: versions of liborc prior to 0.4.19 suffer memory leaks and version 0.4.19 suffers buffer overflows.

Install libvips on Mac OS

brew install homebrew/science/vips --with-webp --with-graphicsmagick

The gettext dependency of libvips can lead to a library not found for -lintl error. If so, please try:

brew link gettext --force

Install libvips on Ubuntu Linux

Ubuntu 14.x

sudo apt-get install libvips-dev

Ubuntu 13.x

Compiling from source is recommended:

sudo apt-get install automake build-essential git gobject-introspection gtk-doc-tools libfftw3-dev libglib2.0-dev libjpeg-turbo8-dev libpng12-dev libwebp-dev libtiff5-dev libexif-dev libxml2-dev swig
git clone https://github.com/jcupitt/libvips.git
cd libvips
git checkout 7.38
./bootstrap.sh
./configure --enable-debug=no --enable-cxx=yes --without-python --without-orc
make
sudo make install
sudo ldconfig

Ubuntu 12.x

Requires libtiff4-dev instead of libtiff5-dev and has a bug in the libwebp package. Work around these problems by running these commands first:

sudo add-apt-repository ppa:lyrasis/precise-backports
sudo apt-get update
sudo apt-get install libtiff4-dev

Then follow Ubuntu 13.x instructions.

Usage examples

var sharp = require('sharp');

sharp('input.jpg').resize(300, 200).toFile('output.jpg', function(err) {
  if (err) {
    throw err;
  }
  // output.jpg is a 300 pixels wide and 200 pixels high image
  // containing a scaled and cropped version of input.jpg
});

sharp('input.jpg').rotate().resize(null, 200).progressive().toBuffer(function(err, outputBuffer) {
  if (err) {
    throw err;
  }
  // outputBuffer contains 200px high progressive JPEG image data, auto-rotated using EXIF Orientation tag
});

sharp('input.png').rotate(180).resize(300).sharpen().quality(90).webp().then(function(outputBuffer) {
  // outputBuffer contains 300px wide, upside down, sharpened, 90% quality WebP image data
});

sharp(inputBuffer).resize(200, 300).bicubicInterpolation().embedWhite().toFile('output.tiff').then(function() {
  // output.tiff is a 200 pixels wide and 300 pixels high image containing a bicubic scaled
  // version, embedded on a white canvas, of the image data in buffer
});

sharp('input.gif').resize(200, 300).embedBlack().webp(function(err, outputBuffer) {
  if (err) {
    throw err;
  }
  // outputBuffer contains WebP image data of a 200 pixels wide and 300 pixels high
  // containing a scaled version, embedded on a black canvas, of input.gif
});

sharp(inputBuffer).resize(200, 200).max().jpeg().then(function(outputBuffer) {
  // outputBuffer contains JPEG image data no wider than 200 pixels and no higher
  // than 200 pixels regardless of the inputBuffer image dimensions
});

API

sharp(input)

Constructor to which further methods are chained. input can be one of:

Buffer containing JPEG, PNG or WebP image data, or
String containing the filename of an image, with most major formats supported.

resize(width, [height])

Scale output to width x height. By default, the resized image is cropped to the exact size specified.

width is the Number of pixels wide the resultant image should be. Use null or undefined to auto-scale the width to match the height.

height is the Number of pixels high the resultant image should be. Use null or undefined to auto-scale the height to match the width.

crop()

Crop the resized image to the exact size specified, the default behaviour.

max()

Preserving aspect ratio, resize the image to the maximum width or height specified.

Both width and height must be provided via resize otherwise the behaviour will default to crop.

embedWhite()

Embed the resized image on a white background of the exact size specified.

embedBlack()

Embed the resized image on a black background of the exact size specified.

rotate([angle])

Rotate the output image by either an explicit angle or auto-orient based on the EXIF Orientation tag. Mirroring is not supported.

angle, if present, is a Number with a value of 0, 90, 180 or 270.

Use this method without angle to determine the angle from EXIF data.

withoutEnlargement()

Do not enlarge the output image if the input image width or height are already less than the required dimensions.

This is equivalent to GraphicsMagick's > geometry option: "change the dimensions of the image only if its width or height exceeds the geometry specification".

sharpen()

Perform a mild sharpen of the resultant image. This typically reduces performance by 30%.

bilinearInterpolation()

Use bilinear interpolation for image resizing, the default (and fastest) interpolation if none is specified.

bicubicInterpolation()

Use bicubic interpolation for image resizing. This typically reduces performance by 5%.

nohaloInterpolation()

Use Nohalo interpolation for image resizing. This typically reduces performance by a factor of 2.

progressive()

Use progressive (interlace) scan for JPEG and PNG output. This typically reduces compression performance by 30% but results in an image that can be rendered sooner when decompressed.

quality(quality)

The output quality to use for lossy JPEG, WebP and TIFF output formats. The default quality is 80.

quality is a Number between 1 and 100.

compressionLevel(compressionLevel)

An advanced setting for the zlib compression level of the lossless PNG output format. The default level is 6.

compressionLevel is a Number between -1 and 9.

sequentialRead()

An advanced setting that switches the libvips access method to VIPS_ACCESS_SEQUENTIAL. This will reduce memory usage and can improve performance on some systems.

toFile(filename, [callback])

filename is a String containing the filename to write the image data to. The format is inferred from the extension, with JPEG, PNG, WebP and TIFF supported.

callback, if present, is called with a single argument (err) containing an error message, if any.