Light-weight image processor for NodeJS
- Overview
- Installation
- Usage
- Supported formats
- Colors specification
- Note on transparent images
- Note on threading performance
- API
- Open an image from file or buffer
- Create a new blank image
- 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
- 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
- Batch operations
- 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 lwip
Or, clone this repo and cd lwip && npm install
.
You can run tests with npm test
.
Note: Installation of this module involves compiling native code.
If npm install lwip
failes, you probably need to setup your system.
See instructions.
Building on Windows with Visual Studio requires version 2013 or higher.
Usage
Typical workflow:
- Open an image and get an image object.
- Manipulate it.
- Save to disk / Send image buffer over network / etc.
Example (batch operations):
// obtain an image object:
require('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('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]
whereR
,G
andB
are integers between 0 and 255 andA
is an integer between 0 and 100. -
As an object
{r: R, g: G, b: B, a: A}
whereR
,G
andB
are integers between 0 and 255 andA
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
- Transparency is supported through an alpha channel which ranges between 0 and 100. 0 is completely transparent and 100 is completely opaque.
- 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:
- Openning an existing image file or buffer with the
open
method. - Creating a new image object with the
create
method. - Cloning an existing image object with the
image.clone
method. - Extracting a sub-image from an existing image object with the
image.extract
method.
Open an image
lwip.open(source, type, callback)
source {String/Buffer}
: The path to the image on disk or an image buffer.type {String/Object}
: Optional type of the image. If omitted, the type will be inferred from the file extension. Ifsource
is a buffer,type
must be specified. Ifsource
is an encoded image buffer,type
must be a string of the image type (i.e."jpg"
). Ifsource
is a raw pixels buffertype
must be an object withtype.width
andtype.height
properties.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:
- 1 channel is a grayscale image.
- 2 channels is a grayscale image with an alpha channel.
- 3 channels is an RGB image.
- 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
var lwip = require('lwip');
lwip.open('path/to/image.jpg', function(err, image){
// check 'err'. use 'image'.
// image.resize(...), etc.
});
Open buffer example
var fs = require('fs'),
lwip = require('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)
width {Integer>0}
: The width of the new image.height {Integer>0}
: The height of the new image.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}
.callback {Function(err, image)}
Example:
var lwip = require('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)
width {Integer}
: Width in pixels.height {Integer}
: Optional height in pixels. If omitted,width
will be used.inter {String}
: Optional interpolation method. Defaults to"lanczos"
. Possible values:"nearest-neighbor"
"moving-average"
"linear"
"grid"
"cubic"
"lanczos"
callback {Function(err, image)}
Scale
image.scale(wRatio, hRatio, inter, callback)
wRatio {Float}
: Width scale ratio.hRatio {Float}
: Optional height scale ratio. If omitted,wRatio
will be used.inter {String}
: Optional interpolation method. Defaults to"lanczos"
. Possible values:"nearest-neighbor"
"moving-average"
"linear"
"grid"
"cubic"
"lanczos"
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)
width {Integer}
: Canvas' width in pixels.height {Integer}
: Canvas' height in pixels.color {String / Array / Object}
: Optional Color of the canvas. See colors specification.inter {String}
: Optional interpolation method. Defaults to"lanczos"
. Possible values:"nearest-neighbor"
"moving-average"
"linear"
"grid"
"cubic"
"lanczos"
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)
width {Integer}
: Canvas' width in pixels.height {Integer}
: Canvas' height in pixels.inter {String}
: Optional interpolation method. Defaults to"lanczos"
. Possible values:"nearest-neighbor"
"moving-average"
"linear"
"grid"
"cubic"
"lanczos"
callback {Function(err, image)}
Rotate
image.rotate(degs, color, callback)
degs {Float}
: Clockwise rotation degrees.color {String / Array / Object}
: Optional Color of the canvas. See colors specification.callback {Function(err, image)}
Crop
Crop with rectangle coordinates
image.crop(left, top, right, bottom, callback)
left, top, right, bottom {Integer}
: Coordinates of the crop rectangle.callback {Function(err, image)}
Crop a rectangle from center
image.crop(width, height, callback)
width, height {Integer}
: Width and height of the rectangle to crop from the center of the image.callback {Function(err, image)}
Blur
Gaussian blur.
image.blur(sigma, callback)
sigma {Float>=0}
: Standard deviation of the Gaussian filter.callback {Function(err, image)}
Sharpen
Inverse diffusion shapren.
image.sharpen(amplitude, callback)
amplitude {Float}
: Sharpening amplitude.callback {Function(err, image)}
Mirror
Mirror an image along the 'x' axis, 'y' axis or both.
image.mirror(axes, callback)
axes {String}
:'x'
,'y'
or'xy'
(case sensitive).callback {Function(err, image)}
Flip
Alias of mirror
.
Border
Add a colored border to the image.
image.border(width, color, callback)
width {Integer}
: Border width in pixels.color {String / Array / Object}
: Optional Color of the border. See colors specification.callback {Function(err, image)}
Pad
Pad image edges with colored pixels.
image.pad(left, top, right, bottom, color, callback)
left, top, right, bottom {Integer}
: Number of pixels to add to each edge.color {String / Array / Object}
: Optional Color of the padding. See colors specification.callback {Function(err, image)}
Saturate
Adjust image saturation.
image.saturate(delta, callback)
delta {Float}
: By how much to increase / decrease the saturation.callback {Function(err, image)}
Examples:
image.saturate(0, ...)
will have no effect on the image.image.saturate(0.5, ...)
will increase the saturation by 50%.image.saturate(-1, ...)
will decrease the saturation by 100%, effectively desaturating the image.
Lighten
Adjust image lightness.
image.lighten(delta, callback)
delta {Float}
: By how much to increase / decrease the lightness.callback {Function(err, image)}
Examples:
image.lighten(0, ...)
will have no effect on the image.image.lighten(0.5, ...)
will increase the lightness by 50%.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)
shift {Float}
: By how many degrees to shift each pixel's hue.callback {Function(err, image)}
Examples:
image.lighten(0, ...)
will have no effect on the image.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)
delta {Float}
: By how much to increase / decrease the transperancy.callback {Function(err, image)}
Note: The transparency is adjusted independently for each pixel.
Examples:
image.fade(0, ...)
will have no effect on the image.image.fade(0.5, ...)
will increase the transparency by 50%.image.fade(1, ...)
will make the image completely transparent.
Opacify
Make image completely opaque.
image.opacify(callback)
callback {Function(err, image)}
Paste
Paste an image on top of this image.
image.paste(left, top, img, callback)
left, top {Integer}
: Coordinates of the top-left corner of the pasted image.img {Image object}
: The image to paste.callback {Function(err, image)}
Notes:
- If the pasted image exceeds the bounds of the base image, an exception is thrown.
img
is pasted in the state it was at the timeimage.paste( ... )
was called, eventhoughcallback
is called asynchronously.- For transparent images, alpha blending is done according to the equations described here.
- 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)
left, top {Integer}
: Coordinates of the pixel from the left-top corner of the image.color {String / Array / Object}
: Color of the pixel to set. See colors specification.callback {Function(err, image)}
Notes:
- If the coordinates exceed the bounds of the image, an exception is thrown.
- 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)
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.
left {Integer>=0}
top {Integer>=0}
Color is returned as an object. See colors specification.
Clone
Clone the image into a new image object.
image.clone(callback)
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)
left, top, right, bottom {Integer}
: Coordinates of the area to copy.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)
format {String}
: Encoding format. Possible values:
"jpg"
"png"
"gif"
params {Object}
: Optional Format-specific parameters (See below).callback {Function(err, buffer)}
Supported encoding formats:
JPEG
The params
object should have the following fields:
quality {Integer}
: Defaults to100
.
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 tofalse
.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 to256
. Number of colors in the color table (at most). Must be between 2 and 256.interlaced {Boolean}
: Defaults tofalse
.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)
path {String}
: Path of file to write.format {String}
: Optional Encoding format. If omitted, will be inferred frompath
extension. Possible values are specified in Get as a Buffer section.params {Object}
: Optional Format-specific parameters.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()
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:
var 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 ornull
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.
var batch1 = image.batch().rotate('45', 'black');
var 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 the following: