@mediabox/lwip

Light-weight image processor for NodeJS

This is a branch based off of @randy.tarampi/lwip , customized for my own use.

0. Overview
1. Installation
2. Usage
3. Supported formats
4. Colors specification
5. Note on transparent images
6. Note on threading performance
7. API
8. Open an image from file or buffer
9. Create a new blank image
10. Image operations 0. Resize 0. Scale 0. Contain 0. Cover 0. Rotate 0. Crop 0. Blur 0. Sharpen 0. Mirror 0. Flip 0. Border 0. Pad 0. Adjust saturation 0. Adjust lightness: lighten / darken 0. Adjust hue 0. Fade (adjust transparency) 0. Opacify 0. Paste 0. Set pixel 0. Set metadata
11. Getters 0. Width 0. Height 0. Pixel 0. Clone 0. Extract / Copy 0. Get as a Buffer 0. JPEG 0. PNG 0. GIF 0. Write to file 0. Get metadata 0. Get dominant color
12. Batch operations
13. Copyrights

Overview

This module provides comprehensive, fast, and simple image processing and manipulation capabilities.

There are no external runtime dependencies, which means you don't have to install anything else on your system.

This module is in active development. New features are being added.

Read the background for the development of this module.

Installation

npm install @mediabox/lwip

Or, clone this repo and cd @mediabox/lwip && npm install.

You can run tests with npm test.

NOTE: Installation of this module involves compiling native code

If npm install @mediabox/lwip fails, you probably need to setup your system.

See the instructions for node-gyp.

Building on Windows with Visual Studio requires version 2013 or higher.

Usage

Typical workflow:

0. Open an image and get an image object.
1. Manipulate it.
2. Save to disk / Send image buffer over network / etc.

Example (batch operations):

// obtain an image object:
require('@mediabox/lwip').open('image.jpg', function(err, image){

  // check err...
  // define a batch of manipulations and save to disk as JPEG:
  image.batch()
    .scale(0.75)          // scale to 75%
    .rotate(45, 'white')  // rotate 45degs clockwise (white fill)
    .crop(200, 200)       // crop a 200X200 square from center
    .blur(5)              // Gaussian blur with SD=5
    .writeFile('output.jpg', function(err){
      // check err...
      // done.
    });

});

Example (non-batch):

var lwip = require('@mediabox/lwip');

// obtain an image object:
lwip.open('image.jpg', function(err, image){

  // check err...
  // manipulate image:
  image.scale(0.5, function(err, image){

    // check err...
    // manipulate some more:
    image.rotate(45, 'white', function(err, image){

      // check err...
      // encode to jpeg and get a buffer object:
      image.toBuffer('jpg', function(err, buffer){

        // check err...
        // save buffer to disk / send over network / etc.

      });

    });

  });

});

Supported formats

Decoding (reading):

• JPEG, 1 & 3 channels (grayscale & RGB).
• PNG, transparency supported.
• GIF, transparency supported. Animated GIFs can be read, but only the first frame will be retrieved.

Encoding (writing):

• JPEG, 3 channels (RGB).
• PNG (lossless), 3 channels (RGB) or 4 channels (RGBA).
• GIF (no animations)

Other formats may also be supported in the future, but are probably less urgent. Check the issues to see which formats are planned to be supported. Open an issue if you need support for a format which is not already listed.

Colors specification

In LWIP colors are coded as RGBA values (red, green, blue and an alpha channel).

Colors are specified in one of three ways:

• As a string. possible values:

  "black"    // {r: 0, g: 0, b: 0, a: 100}
  "white"    // {r: 255, g: 255, b: 255, a: 100}
  "gray"     // {r: 128, g: 128, b: 128, a: 100}
  "red"      // {r: 255, g: 0, b: 0, a: 100}
  "green"    // {r: 0, g: 255, b: 0, a: 100}
  "blue"     // {r: 0, g: 0, b: 255, a: 100}
  "yellow"   // {r: 255, g: 255, b: 0, a: 100}
  "cyan"     // {r: 0, g: 255, b: 255, a: 100}
  "magenta"  // {r: 255, g: 0, b: 255, a: 100}

• As an array [R, G, B, A] where R, G and B are integers between 0 and 255 and A is an integer between 0 and 100.

• As an object {r: R, g: G, b: B, a: A} where R, G and B are integers between 0 and 255 and A is an integer between 0 and 100.

Note: The A value (alpha channel) is always optional and defaults to 100 (completely opaque).

Note on transparent images

0. Transparency is supported through an alpha channel which ranges between 0 and 100. 0 is completely transparent and 100 is completely opaque.
1. Not all formats support transparency. If an image with an alpha channel is encoded with a format which does not support transparency, the alpha channel will be ignored (effectively setting it to 100% for all pixels).

Note on threading performance

All operations are asynchronous, and processing takes place in a thread pool managed by libuv which is part of NodeJS. This thread pool is separate from the event loop used to process HTTP requests, so use of lwip should not significantly affect the handling of HTTP requests by a web application. The thread pool is however shared with other threaded native modules such as those providing database and filesystem IO.

The default thread pool size of 4 will be appropriate for most applications. However if your application regularly processes many images concurrently and and you wish to take full advantage of a multicore system or prevent heavy image processing work from delaying database or filesystem IO, you may want to increase the size of the thread pool by setting the UV_THREADPOOL_SIZE environmental variable to the NodeJS process, e.g.:

UV_THREADPOOL_SIZE=8 node your_script.js

API

All operations are done on an image object. An image object can be obtained by:

0. Openning an existing image file or buffer with the open method.
1. Creating a new image object with the create method.
2. Cloning an existing image object with the image.clone method.
3. Extracting a sub-image from an existing image object with the image.extract method.

Open an image

lwip.open(source, type, callback)

0. source {String/Buffer}: The path to the image on disk or an image buffer.
1. type {String/Object}: Optional type of the image. If omitted, the type will be inferred from the file extension. If source is a buffer, type must be specified. If source is an encoded image buffer, type must be a string of the image type (i.e. "jpg"). If source is a raw pixels buffer type must be an object with type.width and type.height properties.
2. callback {Function(err, image)}

Note about raw pixels buffers: source may be a buffer of raw pixels. The buffer may contain pixels of 1-4 channels, where:

0. 1 channel is a grayscale image.
1. 2 channels is a grayscale image with an alpha channel.
2. 3 channels is an RGB image.
3. 4 channels is an RGBA image (with an alpha channel).

In other words, if the image in the buffer has width W and height H, the size of the buffer can be W*H, 2*W*H, 3*W*H or 4*W*H.

The channel values in the buffer must be stored sequentially. I.e. first all the Red values, then all the Green values, etc.

Open file example

const lwip = require('@mediabox/lwip');
lwip.open('path/to/image.jpg', function(err, image){
    // check 'err'. use 'image'.
    // image.resize(...), etc.
});

Open buffer example

const fs = require('fs'),
    lwip = require('@mediabox/lwip');

fs.readFile('path/to/image.png', function(err, buffer){
  // check err
  lwip.open(buffer, 'png', function(err, image){
      // check 'err'. use 'image'.
      // image.resize(...), etc.
  });
});

Create a new image

lwip.create(width, height, color, callback)

0. width {Integer>0}: The width of the new image.
1. height {Integer>0}: The height of the new image.
2. color {String / Array / Object}: Optional Color of the canvas. See colors specification. Defaults to a transparent canvas {r:0, g:0, b:0, a:0}.
3. callback {Function(err, image)}

Example:

const lwip = require('@mediabox/lwip');

lwip.create(500, 500, 'yellow', function(err, image){
  // check err
  // 'image' is a 500X500 solid yellow canvas.
});

Image operations

Resize

image.resize(width, height, inter, callback)

0. width {Integer}: Width in pixels.
1. height {Integer}: Optional height in pixels. If omitted, width will be used.
2. inter {String}: Optional interpolation method. Defaults to "lanczos". Possible values:
   • "nearest-neighbor"
   • "moving-average"
   • "linear"
   • "grid"
   • "cubic"
   • "lanczos"
3. callback {Function(err, image)}

Scale

image.scale(wRatio, hRatio, inter, callback)

0. wRatio {Float}: Width scale ratio.
1. hRatio {Float}: Optional height scale ratio. If omitted, wRatio will be used.
2. inter {String}: Optional interpolation method. Defaults to "lanczos". Possible values:
   • "nearest-neighbor"
   • "moving-average"
   • "linear"
   • "grid"
   • "cubic"
   • "lanczos"
3. callback {Function(err, image)}

Contain

Contain the image in a colored canvas. The image will be resized to the largest possible size such that it's fully contained inside the canvas.

image.contain(width, height, color, inter, callback)

0. width {Integer}: Canvas' width in pixels.
1. height {Integer}: Canvas' height in pixels.
2. color {String / Array / Object}: Optional Color of the canvas. See colors specification.
3. inter {String}: Optional interpolation method. Defaults to "lanczos". Possible values:
   • "nearest-neighbor"
   • "moving-average"
   • "linear"
   • "grid"
   • "cubic"
   • "lanczos"
4. callback {Function(err, image)}

Cover

Cover a canvas with the image. The image will be resized to the smallest possible size such that both its dimensions are bigger than the canvas's dimensions. Margins of the image exceeding the canvas will be discarded.

image.cover(width, height, inter, callback)

0. width {Integer}: Canvas' width in pixels.
1. height {Integer}: Canvas' height in pixels.
2. inter {String}: Optional interpolation method. Defaults to "lanczos". Possible values:
   • "nearest-neighbor"
   • "moving-average"
   • "linear"
   • "grid"
   • "cubic"
   • "lanczos"
3. callback {Function(err, image)}

Rotate

image.rotate(degs, color, callback)

0. degs {Float}: Clockwise rotation degrees.
1. color {String / Array / Object}: Optional Color of the canvas. See colors specification.
2. callback {Function(err, image)}

Crop

Crop with rectangle coordinates

image.crop(left, top, right, bottom, callback)

0. left, top, right, bottom {Integer}: Coordinates of the crop rectangle.
1. callback {Function(err, image)}

Crop a rectangle from center

image.crop(width, height, callback)

0. width, height {Integer}: Width and height of the rectangle to crop from the center of the image.
1. callback {Function(err, image)}

Blur

Gaussian blur.

image.blur(sigma, callback)

0. sigma {Float>=0}: Standard deviation of the Gaussian filter.
1. callback {Function(err, image)}

Sharpen

Inverse diffusion shapren.

image.sharpen(amplitude, callback)

0. amplitude {Float}: Sharpening amplitude.
1. callback {Function(err, image)}

Mirror

Mirror an image along the 'x' axis, 'y' axis or both.

image.mirror(axes, callback)

0. axes {String}: 'x', 'y' or 'xy' (case sensitive).
1. callback {Function(err, image)}

Flip

Alias of mirror.

Border

Add a colored border to the image.

image.border(width, color, callback)

0. width {Integer}: Border width in pixels.
1. color {String / Array / Object}: Optional Color of the border. See colors specification.
2. callback {Function(err, image)}

Pad

Pad image edges with colored pixels.

image.pad(left, top, right, bottom, color, callback)

0. left, top, right, bottom {Integer}: Number of pixels to add to each edge.
1. color {String / Array / Object}: Optional Color of the padding. See colors specification.
2. callback {Function(err, image)}

Saturate

Adjust image saturation.

image.saturate(delta, callback)

0. delta {Float}: By how much to increase / decrease the saturation.
1. callback {Function(err, image)}

Examples:

0. image.saturate(0, ...) will have no effect on the image.
1. image.saturate(0.5, ...) will increase the saturation by 50%.
2. image.saturate(-1, ...) will decrease the saturation by 100%, effectively desaturating the image.

Lighten

Adjust image lightness.

image.lighten(delta, callback)

0. delta {Float}: By how much to increase / decrease the lightness.
1. callback {Function(err, image)}

Examples:

0. image.lighten(0, ...) will have no effect on the image.
1. image.lighten(0.5, ...) will increase the lightness by 50%.
2. image.lighten(-1, ...) will decrease the lightness by 100%, effectively making the image black.

Darken

Adjust image lightness.

image.darken(delta, callback)

Equivalent to image.lighten(-delta, callback).

Hue

Adjust image hue.

image.hue(shift, callback)

0. shift {Float}: By how many degrees to shift each pixel's hue.
1. callback {Function(err, image)}

Examples:

0. image.lighten(0, ...) will have no effect on the image.
1. image.lighten(100, ...) will shift pixels' hue by 100 degrees.

Note: The hue is shifted in a circular manner in the range [0,360] for each pixel individually.

Fade

Adjust image transperancy.

image.fade(delta, callback)

0. delta {Float}: By how much to increase / decrease the transperancy.
1. callback {Function(err, image)}

Note: The transparency is adjusted independently for each pixel.

Examples:

0. image.fade(0, ...) will have no effect on the image.
1. image.fade(0.5, ...) will increase the transparency by 50%.
2. image.fade(1, ...) will make the image completely transparent.

Opacify

Make image completely opaque.

image.opacify(callback)

0. callback {Function(err, image)}

Paste

Paste an image on top of this image.

image.paste(left, top, img, callback)

0. left, top {Integer}: Coordinates of the top-left corner of the pasted image.
1. img {Image object}: The image to paste.
2. callback {Function(err, image)}

Notes:

0. If the pasted image exceeds the bounds of the base image, an exception is thrown.
1. img is pasted in the state it was at the time image.paste( ... ) was called, eventhough callback is called asynchronously.
2. For transparent images, alpha blending is done according to the equations described here.
3. Extra caution is required when using this method in batch mode, as the images may change by the time this operation is called.

Set Pixel

Set the color of a pixel.

image.setPixel(left, top, color, callback)

0. left, top {Integer}: Coordinates of the pixel from the left-top corner of the image.
1. color {String / Array / Object}: Color of the pixel to set. See colors specification.
2. callback {Function(err, image)}

Notes:

0. If the coordinates exceed the bounds of the image, an exception is thrown.
1. Extra caution is required when using this method in batch mode, as the dimensions of the image may change by the time this operation is called.

Set metadata

Set the metadata in an image. This is currently only supported for PNG files. Sets a tEXt chunk with the key lwip_data and comment as the given string. If called with a null parameter, removes existing metadata from the image, if present.

image.setMetadata(metadata)

0. metadata {String}: a string of arbitrary length, or null.

Getters

Width

image.width() returns the image's width in pixels.

Height

image.height() returns the image's height in pixels.

Get Pixel

image.getPixel(left, top) returns the color of the pixel at the (left, top) coordinate.

0. left {Integer>=0}
1. top {Integer>=0}

Color is returned as an object. See colors specification.

Clone

Clone the image into a new image object.

image.clone(callback)

0. callback {Function(err, newImage)}

Example: See examples/clone.js

Note: The image is cloned to the state it was at the time image.clone( ... ) was called, eventhough callback is called asynchronously.

image.width(); // 500
image.clone(function(err, clone){
    clone.width(); // 500
});
image.resize(100, 100, function(err, image){
    image.width(); //100
});

Extract

Copy an area of the image into a new image object.

image.extract(left, top, right, bottom, callback)

0. left, top, right, bottom {Integer}: Coordinates of the area to copy.
1. callback {Function(err, newImage)}

Example: See examples/extract.js

Note: The sub-image is extracted from the original image in the state it was at the time image.extract( ... ) was called, eventhough callback is called asynchronously.

Get as a Buffer

Get encoded binary image data as a NodeJS Buffer.

When opening an image, it is decoded and stored in memory as an uncompressed image. All manipulations are done on the uncompressed data in memory. This method allows to encode the image to one of the specified formats and get the encoded data as a NodeJS Buffer object.

image.toBuffer(format, params, callback)

0. format {String}: Encoding format. Possible values:

• "jpg"
• "png"
• "gif"

0. params {Object}: Optional Format-specific parameters (See below).
1. callback {Function(err, buffer)}

Supported encoding formats:

JPEG

The params object should have the following fields:

• quality {Integer}: Defaults to 100.
• progressive {Boolean}: Defaults to false.

Note that when encoding to JPEG the alpha channel is discarded.

PNG

The params object should have the following fields:

• compression {String}: Defaults to "fast". Possible values:
  • "none" - No compression. Fastest.
  • "fast" - Basic compression. Fast.
  • "high" - High compression. Slowest.
• interlaced {Boolean}: Defaults to false.
• transparency {true/false/'auto'}: Preserve transparency? Defaults to 'auto'. Determines if the encoded image will have 3 or 4 channels. If 'auto', the image will be encoded with 4 channels if it has transparent components, and 3 channels otherwise.

GIF

The params object should have the following fields:

• colors {Integer}: Defaults to 256. Number of colors in the color table (at most). Must be between 2 and 256.
• interlaced {Boolean}: Defaults to false.
• transparency {true/false/'auto'}: Preserve transparency? Defaults to 'auto'. Determines if the encoded image will have 3 or 4 channels. If 'auto', the image will be encoded with 4 channels if it has transparent components, and 3 channels otherwise.
• threshold {Integer} - Between 0 and 100. Pixels in a gif image are either fully transparent or fully opaque. This value sets the alpha channel threshold to determine if a pixel is opaque or transparent. If the alpha channel of the pixel is above this threshold, this pixel will be considered as opaque; otherwise it will be transparent.

Write to file

Write encoded binary image data directly to a file.

image.writeFile(path, format, params, callback)

0. path {String}: Path of file to write.
1. format {String}: Optional Encoding format. If omitted, will be inferred from path extension. Possible values are specified in Get as a Buffer section.
2. params {Object}: Optional Format-specific parameters.
3. callback {Function(err)}

Get Metadata

Get the textual metadata from an image. This is currently only supported for tEXt chunks in PNG images, and will get the first tEXt chunk found with the key lwip_data. If none is found, returns null.

image.getMetadata()

Get dominant color

Get the pixel color that occurs most frequently in a picture.

image.dominantColor(pixels_to_skip)

0. pixels_to_skip {Int}: the number of pixels to skip while iterating through the image. Ex. supplying 1 will skip every other pixel, 0 will skip none. The greater the number the less accuracy the result will have.

Batch operations

Each of the image operations above can be done as part of a batch of operations. Operations can be queued, and executed as a batch at any time.

Each one of the image operations has a batch equivalent which takes the same arguments, except the callback, which is not needed.

When all batch operations had been queued, they can be executed in one of several methods, as explained below.

Obtaining a batch object

In order to start queueing operations, a batch object first needs to be obtained from the image.

// obtain a batch object from the image:
const batch = image.batch();

Using a batch object

Use the batch object to queue image operations. Each of the operations above has a batch equivalent. Operations can be chained.

Remember, the batch manipulation methods do not take a callback.

Example:

batch.rotate(45, 'white').scale(0.5).blur(5);

Executing a batch

There are several methods which start the execution of a batch. Once a batch finishes an execution, it becomes empty and can be resued to queue additional operations.

Execute batch and obtain the manipulated image object

When all desired operations had been queued, execute the batch with the exec() method. exec takes a callback argument; callback is a function which receives an error object and the manipulated image object:

batch.exec(callback)

• callback {Function(err, image)}:
  • err: An error object or null when no error.
  • image: An image object of the manipulated image.

batch.exec(function(err, image){
  // check err, use image
});

Execute batch and obtain a Buffer object

Batch objects have a toBuffer convenience method.

batch.toBuffer(format, params, callback)

See parameters of image.toBuffer().

Execute batch and write to file

Batch objects have a writeFile convenience method.

batch.writeFile(path, format, params, callback)

See parameters of image.writeFile().

Notes on batch operations

An image can have more than one batch object, but all batch objects modify the same underlying image. This means the order of execution matters.

const batch1 = image.batch().rotate('45', 'black');
const batch2 = image.batch().border(15, 'black');

This will rotate the image 45degs and then add a black border:

batch1.exec(function(err, image){
    batch2.exec(function(err, image){
        // ...
    });
});

While this will add a black border and then rotate the image 45degs:

batch2.exec(function(err, image){
    batch1.exec(function(err, image){
        // ...
    });
});

Copyrights

The native part of this module is compiled from source which uses either your natively installed libraries, or the bundled versions of the following:

• Independent JPEG Group's free JPEG software:
  • Website
  • Readme
• libpng:
  • Website
  • Readme
• zlib:
  • Website
  • Readme
• The CImg Library
  • Website
  • Readme
• giflib
  • Website
  • Readme

License