Bug 109985 - feFlood incorrectly applied color-interpolation-filters
Summary: feFlood incorrectly applied color-interpolation-filters
Status: RESOLVED FIXED
Alias: None
Product: WebKit
Classification: Unclassified
Component: SVG (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P2 Normal
Assignee: Stephen Chenney
URL:
Keywords:
Depends on:
Blocks: 105929
  Show dependency treegraph
 
Reported: 2013-02-15 16:34 PST by Stephen Chenney
Modified: 2013-02-19 07:05 PST (History)
6 users (show)

See Also:


Attachments
Test case. (1.07 KB, image/svg+xml)
2013-02-15 16:34 PST, Stephen Chenney
no flags Details
Patch (16.29 KB, patch)
2013-02-16 08:56 PST, Stephen Chenney
no flags Details | Formatted Diff | Diff
Patch (877.85 KB, patch)
2013-02-16 10:41 PST, Stephen Chenney
no flags Details | Formatted Diff | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Stephen Chenney 2013-02-15 16:34:31 PST
Created attachment 188665 [details]
Test case.

Open the attached file in a WebKit browser (Chrome at least) and note that the greys are two different shades.

In Opera and Firefox they are the same darker shade. That's because we are apparently applying color-interpolation-filters to feFlood, even though it does no interpolation at all. We should just be taking the sRGB specified (50%,50%,50%) color and applying it in sRGB (the default color space for all filter ops).
Comment 1 Stephen Chenney 2013-02-16 08:56:08 PST
Created attachment 188714 [details]
Patch
Comment 2 Stephen White 2013-02-16 08:59:39 PST
Comment on attachment 188714 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=188714&action=review

This looks reasonable to me, but I'll leave for one of the SVG peeps.

> Source/WebCore/platform/graphics/filters/FEFlood.cpp:81
> +    // feFlood does not color interpolation of any kind, so the result is always in ColorSpaceDeviceRGB regardless of

Nit:  This sentence no verb.
Comment 3 Dirk Schulze 2013-02-16 09:24:04 PST
Comment on attachment 188714 [details]
Patch

LGTM. r=me.
Comment 4 Stephen Chenney 2013-02-16 10:41:36 PST
Created attachment 188719 [details]
Patch

Adding linux expectations. Will add Mac expectations too before commit.
Comment 5 WebKit Review Bot 2013-02-18 15:08:45 PST
Comment on attachment 188719 [details]
Patch

Clearing flags on attachment: 188719

Committed r143267: <http://trac.webkit.org/changeset/143267>
Comment 6 WebKit Review Bot 2013-02-18 15:08:48 PST
All reviewed patches have been landed.  Closing bug.
Comment 7 Ryosuke Niwa 2013-02-18 23:50:21 PST
Are you going to rebaseline results on Mac port? If not, then I'm just going to delete ImageOnlyFailures expectations added by this patch.
Comment 8 Stephen Chenney 2013-02-19 07:05:09 PST
(In reply to comment #7)
> Are you going to rebaseline results on Mac port? If not, then I'm just going to delete ImageOnlyFailures expectations added by this patch.

I was just going to delete the expectations, as it seems the tests are passing without any rebaselines for Mac.