Skip to content

GoogleChromeLabs/imagecapture-polyfill

Repository files navigation

ImageCapture polyfill

Build Status Dependency Status devDependency Status

ImageCapture is a polyfill for the MediaStream Image Capture API.

Status

As of June 2017, the ImageCapture spec is relatively stable. Chrome supports the API starting with M59 (earlier versions require setting a flag) and Firefox has partial support behind a flag. See the ImageCapture browser support page for details.

Prior art

Prior to this API, in order to take a still picture from the device camera, two approaches have been used:

  1. Set the source of a <video> element to a stream obtained via navigator[.mediaDevices].getUserMedia, then use a 2D canvas context to drawImage from that video. The canvas can return a URL to be used as the src attribute of an <img> element, via .toDataURL('image/<format>'). (1, 2)
  2. Use the HTML Media Capture API, i.e. <input type="file" name="image" accept="image/*" capture>

Demo

The demo currently shows grabFrame() and takePhoto().

Quick start

yarn add image-capture

Or, with npm:

npm install --save image-capture

In your JS code:

let videoDevice;
let canvas = document.getElementById('canvas');
let photo = document.getElementById('photo');

navigator.mediaDevices.getUserMedia({video: true}).then(gotMedia).catch(failedToGetMedia);

function gotMedia(mediaStream) {
  // Extract video track.
  videoDevice = mediaStream.getVideoTracks()[0];
  // Check if this device supports a picture mode...
  let captureDevice = new ImageCapture(videoDevice);
  if (captureDevice) {
    captureDevice.takePhoto().then(processPhoto).catch(stopCamera);
    captureDevice.grabFrame().then(processFrame).catch(stopCamera);
  }
}

function processPhoto(blob) {
  photo.src = window.URL.createObjectURL(blob);
}

function processFrame(imageBitmap) {
  canvas.width = imageBitmap.width;
  canvas.height = imageBitmap.height;
  canvas.getContext('2d').drawImage(imageBitmap, 0, 0);
}

function stopCamera(error) {
  console.error(error);
  if (videoDevice) videoDevice.stop();  // turn off the camera
}

photo.addEventListener('load', function () {
  // After the image loads, discard the image object to release the memory
  window.URL.revokeObjectURL(this.src);
});

Methods

Start by constructing a new ImageCapture object:

let captureDevice;

navigator.mediaDevices.getUserMedia({video: true}).then(mediaStream => {
  captureDevice = new ImageCapture(mediaStream.getVideoTracks()[0]);
}).catch(...)

Please consult the spec for full detail on the methods.

constructor(videoStreamTrack)

Takes a video track and returns an ImageCapture object.

getPhotoCapabilities

TBD

setOptions

TBD

takePhoto

Capture the video stream into a Blob containing a single still image.

Returns a Promise that resolves to a Blob on success, or is rejected with DOMException on failure.

captureDevice.takePhoto().then(blob => {
  
}).catch(error => ...);

grabFrame

Gather data from the video stream into an ImageBitmap object. The width and height of the ImageBitmap object are derived from the constraints of the video stream track passed to the constructor.

Returns a Promise that resolves to an ImageBitmap on success, or is rejected with DOMException on failure.

captureDevice.grabFrame().then(imageBitmap => {
  
}).catch(error => ...);

Compatibility

The polyfill has been tested to work in current browsers:

  • Chrome 55+
  • Firefox 49+
  • Chrome 52+ for Android
  • Firefox 48+ for Android

For the widest compatibility, you can additionally load the WebRTC adapter. That will expand support to:

  • Chrome 53

For older browsers that don't support navigator.getUserMedia, you can additionally load Addy Osmani's shim with optional fallback to Flash - getUserMedia.js. Alternatively, the getUserMedia wrapper normalizes error handling and gives an error-first API with cross-browser support.

Development

yarn
yarn run dev

npm

npm install
npm run dev

To make your server accessible outside of localhost, run npm/yarn run lt.

Before committing, make sure you pass yarn/npm run lint without errors, and run yarn/npm run docs to generate the demo.