[HTML5] Context3D.drawToBitmapData produces transparent BitmapData

[mayakwd]

During developing Starling app I've faced with the problem of drawing starling sprites to BitmapData. DisplayObject.drawToBitmapData produces transparent BitmapData. By the help of @peteshand we've discovered that context3D.drawToBitmapData producing same result on HTML5 target.

@peteshand created sample to reproduce this issue: drawToBitmapdata.zip

Issue in starling project: openfl/starling#110

[peteshand]

ok, down grading to Lime 6.4.0 and OpenFL 8.3.0 resolves the issue, so it must be a regression in the latest versions.

[peteshand]

One difference I've noticed within the drawToBitmapData function is the following line:

OpenFL 8.3.0
var image = window.renderer.readPixels ();

OpenFL 8.4.1
var image = #if (lime >= "7.0.0") window.readPixels () #else window.renderer.readPixels () #end;

[peteshand]

This issue has been resolved via this PR:
openfl/lime#1240

[mayakwd]

Thank you, Pete! I'll try to compile with develop branch to check if it's resolved :-)

[mayakwd]

I done testing with latest develop + current starling version. I got total mess with masks and canvas rendering:

1. Quad masks that are not added to displaylist stopped working, their positions attached to stage instead of local position of target, most content with masks is not displaying at all.
2. Rendering of canvas blinks like it's being erased every second frame.

Drawing into bitmapData still produces transparent pixels.
Totally unusable version.

[jgranick]

I just built your sample, it appears to look the same on Flash, Neko and HTML5, using current latest dev of Lime, OpenFL and Starling.

It's tan and white and first, and then clicking any of the other buttons makes it tan instead, and it stays tan over a black background no matter what

[peteshand]

@mayakwd I'm guessing the Masking / Canvas problems you're seeing are a different issue to the drawToBitmapdata issue right? if so it might make sense to create a separate issue.

[mayakwd]

@peteshand, yup, I just need more time, to extract some samples, to illustrate new issues. Closing this.