How can I convert an HTML element to a canvas element?

Take a look at this tutorial on MDN: https://developer.mozilla.org/en/HTML/Canvas/Drawing_DOM_objects_into_a_canvas (archived)

Its key trick was:

var canvas = document.getElementById('canvas');
var ctx = canvas.getContext('2d');

var data = '<svg xmlns="http://www.w3.org/2000/svg" width="200" height="200">' +
           '<foreignObject width="100%" height="100%">' +
           '<div xmlns="http://www.w3.org/1999/xhtml" style="font-size:40px">' +
             '<em>I</em> like ' + 
             '<span style="color:white; text-shadow:0 0 2px blue;">' +
             'cheese</span>' +
           '</div>' +
           '</foreignObject>' +
           '</svg>';

var DOMURL = window.URL || window.webkitURL || window;

var img = new Image();
var svg = new Blob([data], {type: 'image/svg+xml;charset=utf-8'});
var url = DOMURL.createObjectURL(svg);

img.onload = function () {
  ctx.drawImage(img, 0, 0);
  DOMURL.revokeObjectURL(url);
}

img.src = url;

That is, it used a temporary SVG image to include the HTML content as a "foreign element", then renders said SVG image into a canvas element. There are significant restrictions on what you can include in an SVG image in this way, however. (See the "Security" section for details — basically it's a lot more limited than an iframe or AJAX due to privacy and cross-domain concerns.)

There is a library that try to do what you say.

See this examples and get the code

http://hertzen.com/experiments/jsfeedback/

http://html2canvas.hertzen.com/

Reads the DOM, from the html and render it to a canvas, fail on some, but in general works.

Sorry, the browser won't render HTML into a canvas.

It would be a potential security risk if you could, as HTML can include content (in particular images and iframes) from third-party sites. If canvas could turn HTML content into an image and then you read the image data, you could potentially extract privileged content from other sites.

To get a canvas from HTML, you'd have to basically write your own HTML renderer from scratch using drawImage and fillText, which is a potentially huge task. There's one such attempt here but it's a bit dodgy and a long way from complete. (It even attempts to parse the HTML/CSS from scratch, which I think is crazy! It'd be easier to start from a real DOM node with styles applied, and read the styling using getComputedStyle and relative positions of parts of it using offsetTop et al.)