Skip to content

Latest commit

 

History

History
executable file
·
431 lines (287 loc) · 16.5 KB

README.md

File metadata and controls

executable file
·
431 lines (287 loc) · 16.5 KB
Raw

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.

Memory usage is kept to a minimum, no child processes are spawned, everything remains non-blocking thanks to libuv and Promises/A+ are supported.

This module supports reading and writing JPEG, PNG and WebP images to and from Streams, 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

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

A missing or incorrectly configured Xcode Command Line Tools installation can lead to a library not found for -ljpeg error. If so, please try:

xcode-select --install

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 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 --without-fftw
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.

Install libvips on Heroku

Alessandro Tagliapietra maintains an Heroku buildpack for libvips and its dependencies.

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
});
var transformer = sharp().resize(300, 200).crop(sharp.gravity.north);
readableStream.pipe(transformer).pipe(writableStream);
// Read image data from readableStream, resize and write image data to writableStream
var image = sharp(inputJpg);
image.metadata(function(err, metadata) {
  image.resize(metadata.width / 2).webp().toBuffer(function(err, outputBuffer, info) {
    // outputBuffer contains a WebP image half the width and height of the original JPEG
  });
});
var pipeline = sharp()
  .rotate()
  .resize(null, 200)
  .progressive()
  .toBuffer(function(err, outputBuffer, info) {
    if (err) {
      throw err;
    }
    // outputBuffer contains 200px high progressive JPEG image data,
    // auto-rotated using EXIF Orientation tag
    // info.width and info.height contain the dimensions of the resized image
  });
readableStream.pipe(pipeline);
sharp('input.png')
  .rotate(180)
  .resize(300)
  .sharpen()
  .withMetadata()
  .quality(90)
  .webp()
  .toBuffer()
  .then(function(outputBuffer) {
    // outputBuffer contains 300px wide, upside down, sharpened,
    // with metadata, 90% quality WebP image data
  });
http.createServer(function(request, response) {
  response.writeHead(200, {'Content-Type': 'image/webp'});
  sharp('input.jpg').rotate().resize(200).webp().pipe(response);
}).listen(8000);
// Create HTTP server that always returns auto-rotated 'input.jpg',
// resized to 200 pixels wide, in WebP format
sharp(inputBuffer)
  .resize(200, 300)
  .interpolateWith(sharp.interpolator.nohalo)
  .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 inputBuffer
  });
sharp('input.gif').resize(200, 300).embedBlack().webp().toBuffer(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().toBuffer().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

Input methods

sharp([input])

Constructor to which further methods are chained. input, if present, 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.

The object returned implements the stream.Duplex class.

JPEG, PNG or WebP format image data can be streamed into the object when input is not provided.

JPEG, PNG or WebP format image data can be streamed out from this object.

metadata([callback])

Fast access to image metadata without decoding any compressed image data.

callback, if present, gets the arguments (err, metadata) where metadata has the attributes:

  • format: Name of decoder to be used to decompress image data e.g. jpeg, png, webp (for file-based input additionally tiff and magick)
  • width: Number of pixels wide
  • height: Number of pixels high
  • space: Name of colour space interpretation e.g. srgb, rgb, scrgb, cmyk, lab, xyz, b-w ...
  • channels: Number of bands e.g. 3 for sRGB, 4 for CMYK
  • orientation: Number value of the EXIF Orientation header, if present

A Promises/A+ promise is returned when callback is not provided.

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.

Image transformation options

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([gravity])

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

gravity, if present, is an attribute of the sharp.gravity Object e.g. sharp.gravity.north.

Possible values are north, east, south, west, center and centre. The default gravity is center/centre.

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.

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. Mirroring is currently unsupported.

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%.

interpolateWith(interpolator)

Use the given interpolator for image resizing, where interpolator is an attribute of the sharp.interpolator Object e.g. sharp.interpolator.bicubic.

Possible interpolators, in order of performance, are:

  • bilinear: Use bilinear interpolation, the default (and fastest) interpolation.
  • bicubic: Use bicubic interpolation, which typically reduces performance by 5%.
  • vertexSplitQuadraticBasisSpline: Use VSQBS interpolation, which prevents "staircasing" when enlarging and typically reduces performance by 5%.
  • locallyBoundedBicubic: Use LBB interpolation, which prevents some "acutance" and typically reduces performance by a factor of 2.
  • nohalo: Use Nohalo interpolation, which prevents acutance and typically reduces performance by a factor of 3.

Output options

withMetadata([boolean])

Specifies if the output image should contains the original metadata or not.

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.

jpeg()

Use JPEG format for the output image.

png()

Use PNG format for the output image.

webp()

Use WebP format for the output image.

Output methods

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 two arguments (err, info) where:

  • err contains an error message, if any
  • info contains the final resized image dimensions in its width and height properties

A Promises/A+ promise is returned when callback is not provided.

toBuffer([callback])

Write image data to a Buffer, the format of which will match the input image by default. JPEG, PNG and WebP are supported.

callback, if present, gets three arguments (err, buffer, info) where:

  • err is an error message, if any
  • buffer is the resultant image data
  • info contains the final resized image dimensions in its width and height properties

A Promises/A+ promise is returned when callback is not provided.

Utility methods

sharp.cache([memory], [items])

If memory or items are provided, set the limits of libvips' operation cache.

  • memory is the maximum memory in MB to use for this cache, with a default value of 100
  • items is the maximum number of operations to cache, with a default value of 500

This method always returns cache statistics, useful for determining how much working memory is required for a particular task.

var stats = sharp.cache(); // { current: 75, high: 99, memory: 100, items: 500 }
sharp.cache(200); // { current: 75, high: 99, memory: 200, items: 500 }
sharp.cache(50, 200); // { current: 49, high: 99, memory: 50, items: 200}

sharp.counters()

Provides access to internal task counters.

  • queue is the number of tasks this module has queued waiting for libuv to provide a worker thread from its pool.
  • process is the number of resize tasks currently being processed.
var counters = sharp.counters(); // { queue: 2, process: 4 }

Testing

Build Status

npm test

Running the tests requires both ImageMagick and GraphicsMagick plus one of either libmagick++-dev or libgraphicsmagick++.

brew install imagemagick
brew install graphicsmagick

sudo apt-get install imagemagick graphicsmagick libmagick++-dev

Performance

Test environment

  • Intel Xeon L5520 2.27GHz 8MB cache
  • Ubuntu 13.10
  • libvips 7.38.5

The contenders

  • imagemagick-native - Supports Buffers only and blocks main V8 thread whilst processing.
  • imagemagick - Supports filesystem only and "has been unmaintained for a long time".
  • gm - Fully featured wrapper around GraphicsMagick.
  • sharp - Caching within libvips disabled to ensure a fair comparison.

The task

Decompress a 2725x2225 JPEG image, resize and crop to 720x480, then compress to JPEG.

Results

Module Input Output Ops/sec Speed-up
imagemagick-native buffer buffer 0.97 1
imagemagick file file 2.49 2.6
gm buffer file 3.72 3.8
gm buffer buffer 3.80 3.9
gm file file 3.67 3.8
gm file buffer 3.67 3.8
sharp buffer file 13.62 14.0
sharp buffer buffer 12.43 12.8
sharp file file 13.02 13.4
sharp file buffer 11.15 11.5
sharp +sharpen file buffer 10.26 10.6
sharp +progressive file buffer 9.44 9.7
sharp +sequentialRead file buffer 11.94 12.3

You can expect much greater performance with caching enabled (default) and using 16+ core machines.

Licence

Copyright 2013, 2014 Lovell Fuller, Pierre Inglebert, Jonathan Ong and Chanon Sajjamanochai

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.