gl lets you create a WebGL context in Node.js without making a window or loading a full browser environment.
It aspires to fully conform to the WebGL 1.0.3 specification.
// Create context
var width = 64
var height = 64
var gl = require('gl')(width, height, { preserveDrawingBuffer: true })
//Clear screen to red
gl.clearColor(1, 0, 0, 1)
gl.clear(gl.COLOR_BUFFER_BIT)
//Write output as a PPM formatted image
var pixels = new Uint8Array(width * height * 4)
gl.readPixels(0, 0, width, height, gl.RGBA, gl.UNSIGNED_BYTE, pixels)
process.stdout.write(['P3\n# gl.ppm\n', width, " ", height, '\n255\n'].join(''))
for(var i = 0; i < pixels.length; i += 4) {
for(var j = 0; j < 3; ++j) {
process.stdout.write(pixels[i + j] + ' ')
}
}Installing headless-gl on a supported platform is a snap using one of the prebuilt binaries. Using npm run the command,
npm install gl
And you are good to go!
Prebuilt binaries are generally available for LTS node versions (e.g. 18, 20). If your system is not supported, then please see the development section on how to configure your build environment. Patches to improve support are always welcome!
gl runs on Linux, macOS, and Windows.
Node.js versions 16 and up are supported.
headless-gl exports exactly one function which you can use to create a WebGL context,
Creates a new WebGLRenderingContext with the given context attributes.
widthis the width of the drawing bufferheightis the height of the drawing buffercontextAttributesis an optional object whose properties are the context attributes for the WebGLRendering context
Returns A new WebGLRenderingContext object
In addition to all the usual WebGL methods, headless-gl exposes some custom extensions to make it easier to manage WebGL context resources in a server side environment:
This extension provides a mechanism to resize the drawing buffer of a WebGL context once it is created.
In a pure DOM implementation, this method would implemented by resizing the WebGLContext's canvas element by modifying its width/height properties. This canvas manipulation is not possible in headless-gl, since a headless context doesn't have a DOM or a canvas element associated to it.
var assert = require('assert')
var gl = require('gl')(10, 10)
assert(gl.drawingBufferHeight === 10 && gl.drawingBufferWidth === 10)
var ext = gl.getExtension('STACKGL_resize_drawingbuffer')
ext.resize(20, 5)
assert(gl.drawingBufferHeight === 20 && gl.drawingBufferWidth === 5)[NoInterfaceObject]
interface STACKGL_resize_drawingbuffer {
void resize(GLint width, GLint height);
};
Resizes the drawing buffer of a WebGL rendering context
widthis the new width of the drawing buffer for the contextheightis the new height of the drawing buffer for the context
Destroys the WebGL context immediately, reclaiming all resources associated with it.
For long running jobs, garbage collection of contexts is often not fast enough. To prevent the system from becoming overloaded with unused contexts, you can force the system to reclaim a WebGL context immediately by calling .destroy().
var gl = require('gl')(10, 10)
var ext = gl.getExtension('STACKGL_destroy_context')
ext.destroy()[NoInterfaceObject]
interface STACKGL_destroy_context {
void destroy();
};
Immediately destroys the context and all associated resources.
In most cases installing headless-gl from npm should just work. However, if you run into problems you might need to adjust your system configuration and make sure all your dependencies are up to date. For general information on building native modules, see the node-gyp documentation.
- Python 3
- A GNU C++ environment (available via the
build-essentialpackage onapt) - libxi-dev
- Working and up to date OpenGL drivers
- GLEW
- pkg-config
$ sudo apt-get install -y build-essential libxi-dev libglu1-mesa-dev libglew-dev pkg-config
- Python 3
- Microsoft Visual Studio
- d3dcompiler_47.dll should be in c:\windows\system32, but if isn't then you can find another copy in the deps/ folder
headless-gl should work out of the box on most CI systems. Some notes on specific CI systems:
- CircleCI:
headless-glshould just work in the default node environment. - AppVeyor: Again it should just work
- TravisCI: Works out of the box on the OS X image. For Linux VMs, you need to install mesa and xvfb. To do this, create a file in the root of your repo called
.travis.ymland paste the following into it:
language: node_js
os: linux
sudo: required
dist: trusty
addons:
apt:
packages:
- mesa-utils
- xvfb
- libgl1-mesa-dri
- libglapi-mesa
- libosmesa6
node_js:
- '20'
before_script:
- export DISPLAY=:99.0; sh -e /etc/init.d/xvfb start
If you know of a service not listed here, open an issue I'll add it to the list.
If you are running your own minimal Linux server, such as the one one would want to use on Amazon AWS or equivalent, it will likely not provide an X11 nor an OpenGL environment. To setup such an environment you can use those two packages:
- Xvfb is a lightweight X11 server which provides a back buffer for displaying X11 application offscreen and reading back the pixels which were drawn offscreen. It is typically used in Continuous Integration systems. It can be installed on CentOS with
yum install -y Xvfb, and comes preinstalled on Ubuntu. - Mesa is the reference open source software implementation of OpenGL. It can be installed on CentOS with
yum install -y mesa-dri-drivers, orapt-get install libgl1-mesa-dev. Since a cloud Linux instance will typically run on a machine that does not have a GPU, a software implementation of OpenGL will be required.
Interacting with Xvfb requires you to start it on the background and to execute your node program with the DISPLAY environment variable set to whatever was configured when running Xvfb (the default being :99). If you want to do that reliably you'll have to start Xvfb from an init.d script at boot time, which is extra configuration burden. Fortunately there is a wrapper script shipped with Xvfb known as xvfb-run which can start Xvfb on the fly, execute your Node.js program and finally shut Xvfb down. Here's how to run it:
xvfb-run -s "-ac -screen 0 1280x1024x24" <node program>
Yes, with browserify. The STACKGL_destroy_context and STACKGL_resize_drawingbuffer extensions are emulated as well.
They aren't for now. If you want to upload data to a texture, you will need to unpack the pixels into a Uint8Array and feed it into texImage2D. To help reading and saving images, you should check out the following modules:
Only the following for now:
STACKGL_resize_drawingbufferSTACKGL_destroy_contextANGLE_instanced_arraysOES_element_index_uintOES_texture_floatOES_texture_float_linearOES_vertex_array_objectOES_standard_derivativesWEBGL_draw_buffersEXT_blend_minmaxEXT_texture_filter_anisotropic
There is a change log.
Despite the name node-webgl doesn't actually implement WebGL - rather it gives you "WebGL"-flavored bindings to whatever OpenGL driver is configured on your system. If you are starting from an existing WebGL application or library, this means you'll have to do a bunch of work rewriting your WebGL code and shaders to deal with all the idiosyncrasies and bugs present on whatever platforms you try to run on. The upside though is that node-webgl exposes a lot of non-WebGL stuff that might be useful for games like window creation, mouse and keyboard input, requestAnimationFrame emulation, and some native OpenGL features.
headless-gl on the other hand just implements WebGL. It is built on top of ANGLE and passes the full Khronos ARB conformance suite, which means it works exactly the same on all supported systems. This makes it a great choice for running on a server or in a command line tool. You can use it to run tests, generate images or perform GPGPU computations using shaders.
Why use this thing instead of electron?
Electron is fantastic if you are writing a desktop application or if you need a full DOM implementation. On the other hand, because it is a larger dependency electron is more difficult to set up and configure in a server-side/CI environment. headless-gl is more modular in the sense that it just implements WebGL and nothing else. As a result creating a headless-gl context takes just a few milliseconds on most systems, while spawning a full electron instance can take upwards of 15-30 seconds. If you are using WebGL in a command line interface or need to execute WebGL in a service, headless-gl might be a more efficient and simpler choice.
After you have your system dependencies installed, do the following:
- Clone this repo:
git clone [email protected]:stackgl/headless-gl.git - Switch to the headless gl directory:
cd headless-gl - Initialize the angle submodule:
git submodule init - Update the angle submodule:
git submodule update - Install npm dependencies:
npm install - Run node-gyp to generate build scripts:
npm run rebuild
Once this is done, you should be good to go! A few more things
- To run the test cases, use the command
npm test, or execute specific tests by just running them usingnode. - On a Unix-like platform, you can do incremental rebuilds by going into the
build/directory and runningmake. This is way faster runningnpm buildeach time you make a change.
See LICENSES
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent