10169 2010-07-14 20:55:20 +0000 The discontinuity at shadowBlur = 8 here is weird. We have no interop on exact shadow algorithms right now anyway, so maybe it could be fixed in the spec? Clearly authors aren't depending on pixel-perfect canvas shadows. Maybe it should just be the num 2010-10-05 12:58:54 +0000 1 1 1 Unclassified HTML WG pre-LC1 HTML Canvas 2D Context (editor: Ian Hickson) unspecified Other other CLOSED INVALID http://www.whatwg.org/specs/web-apps/current-work/#shadows P3 normal --- 1 contributor ian ayg mike public-html-admin public-html-wg-issue-tracking public-html-bugzilla oldest_to_newest 36866 0 contributor 2010-07-14 20:55:20 +0000 Section: http://www.whatwg.org/specs/web-apps/current-work/complete.html#shadows Comment: The discontinuity at shadowBlur = 8 here is weird. We have no interop on exact shadow algorithms right now anyway, so maybe it could be fixed in the spec? Clearly authors aren't depending on pixel-perfect canvas shadows. Maybe it should just be the number of standard deviations, since that's what SVG does? Posted from: 68.175.61.233 36867 1 ayg 2010-07-14 21:01:17 +0000 Document showing lack of interop: data:text/html,<!doctype html> <canvas height="300" width="300"></canvas> <script> window.addEventListener('load', function () { context = document.getElementsByTagName("canvas")[0].getContext("2d"); context.shadowBlur = 50; context.shadowColor = 'black'; context.fillRect(100, 100, 100, 100); }, false); </script> This looks distinctly different in fairly recent WebKit, Gecko, and Presto (Chrome dev, Firefox 4b1, Opera 10.60). From reading the specs, not knowing anything about Gaussian blurs, I'd think it should look the same as: <!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"> <svg xmlns="http://www.w3.org/2000/svg"><defs> <filter id="Gaussian_Blur"><feGaussianBlur in="SourceGraphic" stdDeviation="5"/></filter> </defs> <rect x="100" y="100" width="100" height="100" style="fill:black;filter:url(#Gaussian_Blur)"/> </svg> which Chrome and Opera render identically to one another, but very different to how they render the canvas. So it should be safe to change the algorithm here substantially. Although probably not to the point of matching SVG, on second thought, since that's very different from what anyone does now for canvas (by an order of magnitude in this case). 36870 2 ayg 2010-07-14 21:14:53 +0000 Er, my canvas example was wrong. This should display identically to the SVG I gave: data:text/html,<!doctype html> <style>* { margin: 0 }</style> <canvas height="300" width="300"></canvas> <script> window.addEventListener('load', function () { context = document.getElementsByTagName("canvas")[0].getContext("2d"); context.shadowBlur = 12.5; context.shadowOffsetX = 400; context.shadowOffsetY = 400; context.shadowColor = 'black'; context.fillRect(-300, -300, 100, 100); }, false); </script> It does in Opera, but Chrome isn't quite right. Either way, they all disagree on the canvas rendering, no interop. 36882 3 ayg 2010-07-14 23:59:13 +0000 Safari 5, IE9PP2, and Opera 10.60 all agree on the canvas rendering on Windows 7, actually, per spec. But Firefox+Chrome is a much bigger market share than that (since IE<9 is out of the picture), so it's still not dependable in practice. 36903 4 ayg 2010-07-15 09:57:36 +0000 Hixie recommended that I bring this up with browser vendors instead of him, so I'll do that.