Summary: | SVG filter effect feGaussianBlur unexpectedly lightens image using linearRGB | ||||||||
---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | David Bonnet <david> | ||||||
Component: | SVG | Assignee: | frankhome61 | ||||||
Status: | RESOLVED DUPLICATE | ||||||||
Severity: | Normal | CC: | ap, dino, frankhome61, sabouhallawa, simon.fraser, thorton, zimmermann | ||||||
Priority: | P3 | ||||||||
Version: | 528+ (Nightly build) | ||||||||
Hardware: | Unspecified | ||||||||
OS: | Unspecified | ||||||||
See Also: |
https://bugs.webkit.org/show_bug.cgi?id=202826 https://bugs.webkit.org/show_bug.cgi?id=181267 https://bugs.webkit.org/show_bug.cgi?id=105929 https://bugs.webkit.org/show_bug.cgi?id=212649 |
||||||||
Attachments: |
|
Description
David Bonnet
2014-08-31 10:45:10 PDT
Created attachment 237429 [details]
Rendering comparison between Chrome and WebKit Nightly
>From the comments on the pages mentioned above, it seems to have been introduced since Safari 7.
I can reproduce this with Safari 7.0.6 on OS X 10.9.4, so this doesn't look like a regression despite the above.
I thikn the bug here is that FilterEffect::imageBufferResult() just splats m_premultipliedImageResult into m_imageBufferResult without any linearRGB -> sRGB conversion, and transformResultColorSpace() does nothing for CG (because it assumes that the result will be obtained by painting via CG, which is not accurate). Bug 180565 is the same issue, but on input. (In reply to Simon Fraser (smfr) from comment #4) > Bug 180565 is the same issue, but on input. That's the wrong bug. I meant bug 181267. This bug is fixed in a recent patch https://bugs.webkit.org/show_bug.cgi?id=212649 *** This bug has been marked as a duplicate of bug 212649 *** |