238022 2022-03-17 07:18:54 -0700 WebGl incorrect rendering on 15.4 using Playcanvas 2022-03-28 17:43:18 -0700 1 1 1 Unclassified WebKit WebGL Safari 15 All iOS 15 RESOLVED FIXED InRadar P2 Major --- 238196 1 mvaligursky webkit-unassigned avpeery dino djg gman kbr kkinnunen kpiddington webkit-bug-importer oldest_to_newest 1852267 0 mvaligursky 2022-03-17 07:18:54 -0700 WebGl rendering on 15.4 has some major problems using Playcanvas engine. This has not been the case on 15.3. Repro steps: https://playcanvas.github.io/#/graphics/area-picker also these additional examples (available on the same url) have similar problems, some don't render at all: - all anim blend trees + few more anim ones - area picker - clustered area lights - grab pass - post effects tested on iphone XR 15.4 and also on Safari 15.4 Mac OS. Both WebGl1 and WebGl2. A possible workaround seems to be disabling of antialiasing (which means passing { antialias: false } to canvas.getContext as an option when initializing WebGl. Most of these examples do some form of rendering to texture and then using the results, but other examples using rendering to texture seems fine. our tracking ticket: https://github.com/playcanvas/engine/issues/4136 Possibly related to: https://bugs.webkit.org/show_bug.cgi?id=237906 https://bugs.webkit.org/show_bug.cgi?id=237006 1852302 1 avpeery 2022-03-17 08:20:08 -0700 We had an issue with webGL rendering with the 15.4 upgrade (on iOS as well), and have also found disabling antialiasing to resolve the rendering problem. https://bugs.webkit.org/show_bug.cgi?id=237918 1852875 2 webkit-bug-importer 2022-03-18 16:17:02 -0700 <rdar://problem/90509311> 1853041 3 kbr 2022-03-19 21:38:41 -0700 Can confirm that on an M1 MacBook Pro with Safari Technology Preview Release 141 (Safari 15.4, WebKit 17614.1.3.8), this example doesn't render correctly with "WebGL on Metal" enabled. It does render correctly with it disabled. It renders correctly on Chrome Canary 102.0.4952.2 (Official Build) canary (arm64) with --use-angle=metal, so this bug has been fixed in top-of-tree ANGLE. 1853586 4 kkinnunen 2022-03-22 02:41:30 -0700 Thanks for the report. I think this is part of the 15.4 msaa regression, linked to in the depends on -section. 1855858 5 dino 2022-03-28 17:43:18 -0700 Fixed in the ANGLE roll https://bugs.webkit.org/show_bug.cgi?id=238171