Current WebKit nightlies fail to support the various opacity properties properly, resulting in the following incorrect behaviors: - a <rect> with opacity="0.5" is actually implemented as fill-opacity="0.5" and stroke-opacity="0.5", instead of compositing fill and stroke together first and then applying opacity on the whole painted shape - gradient fills and strokes do not take into account fill-opacity and stroke-opacity values. This may be related to bug <a href="http://bugs.webkit.org/show_bug.cgi?id=12501">12501</a>.
Created attachment 14910 [details] Reference image for test
Created attachment 14911 [details] Reference image for test
Created attachment 14912 [details] Reference image for test
Created attachment 14913 [details] Reference image for test
Created attachment 14914 [details] Testcase for fill-opacity, stroke-opacity, opacity as applied to solid and gradient fills and strokes This test case works fine in both Mozilla trunk and Opera 9.21. The first row is SVG and the second row are reference PNG images with the correct rendering.
I'm not surprised by this bug at all. Our opacity handling has always been a bit shady... particularly due to the total lack of good test cases on the subject. ;)
(In reply to comment #5) > Created an attachment (id=14914) [edit] > Testcase for fill-opacity, stroke-opacity, opacity as applied to solid and > gradient fills and strokes > > This test case works fine in both Mozilla trunk and Opera 9.21. The first row > is SVG and the second row are reference PNG images with the correct rendering. > Hi Antoine, I'm finally on this bug report again. Could you also create a testcase for patterns, that would be excellent.. Greetings, Niko
*** This bug has been marked as a duplicate of 12501 ***