Bug 94120
Summary: | [CSS Shaders] css3/filters/custom/filter-fallback-to-software.html and css3/filters/custom/invalid-custom-filter-shader.html fail on Mac build bots | ||
---|---|---|---|
Product: | WebKit | Reporter: | Alexandru Chiculita <achicu> |
Component: | Layout and Rendering | Assignee: | Alexandru Chiculita <achicu> |
Status: | RESOLVED DUPLICATE | ||
Severity: | Normal | CC: | cmuppala, fmalita |
Priority: | P2 | ||
Version: | 528+ (Nightly build) | ||
Hardware: | Unspecified | ||
OS: | Unspecified |
Alexandru Chiculita
The following tests are failing on Mac Build Bots
css3/filters/custom/filter-fallback-to-software.html = IMAGE
css3/filters/custom/invalid-custom-filter-shader.html = IMAGE
Attachments | ||
---|---|---|
Add attachment proposed patch, testcase, etc. |
Alexandru Chiculita
The bug is introduced by http://trac.webkit.org/changeset/125462 . The issue is that copyUnmultipliedImage and copyPremultipliedImage don't do the color conversion on Mac.
There are other filters that are also affected, for example FEComponentTransfer, FEGaussianBlur, FEComposite, FELighting, FEConvolveMatrix.
-webkit-filter: blur(0) has the same issue.
Florin Malita
(In reply to comment #1)
> The bug is introduced by http://trac.webkit.org/changeset/125462 .
Are you sure that the new results are incorrect? After 125462 filters are applied in a different color space (linearRGB, per spec), so the images are expected to be different. The intent was to mark them all for rebaseline, but some were missed.
Alexandru Chiculita
(In reply to comment #2)
> (In reply to comment #1)
> > The bug is introduced by http://trac.webkit.org/changeset/125462 .
>
> Are you sure that the new results are incorrect? After 125462 filters are applied in a different color space (linearRGB, per spec), so the images are expected to be different. The intent was to mark them all for rebaseline, but some were missed.
I think -webkit-filter: blur(0); should not change the colors at all.
Also, adding a grayscale(0) before the blur(0) fixes the issue, that's why I think that is an implementation issue.
Alexandru Chiculita
*** Bug 94169 has been marked as a duplicate of this bug. ***
Alexandru Chiculita
*** This bug has been marked as a duplicate of bug 94372 ***