[Webkit-unassigned] [Bug 46493] [chromium] Add mipmap support for ImageLayerChromium

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=46493





--- Comment #18 from Vangelis Kokkevis <vangelis at chromium.org>  2010-09-29 13:43:49 PST ---
(In reply to comment #17)
> (In reply to comment #16)
> > (From update of attachment 69182 [details] [details])
> > Code looks good.  Can you please also add a layout test that demonstrates that this works? What I'm after is making sure that we don't get hardware which fails to create mipmaps or reports that it supports mipmapping but doesn't. The test could for example create an image layer with a colored checkerboard texture, apply a perspective 3D transform on it and read back the pixels to make sure they are not black. Thanks.
> 
> Sure, not a problem.
> 
> I have a question: if hardware claims to support mip-mapping, and doesn't, are we guaranteed to get black pixels? Might it possibly fall-back to linear/nearest interpolation of the raw texture, in which case the pixels won't be black?

If the drivers are faulty, they could really be doing anything.  If it does end up falling back to nearest, it would be somewhat fortunate. It would be nice to have a layout test in place that we can run on a number of machines to make sure there are no unexpected failures.  Probably worthwhile having both at POT and an NPOT layer.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.