Open
Bug 1513155
Opened 7 years ago
Updated 4 years ago
Investigate why 1-value difference on each RGB components appears on the difference between flattened opacity into background-color and non-flattened cases
Categories
(Core :: Layout, enhancement, P3)
Core
Layout
Tracking
()
NEW
| Tracking | Status | |
|---|---|---|
| firefox66 | --- | affected |
People
(Reporter: hiro, Unassigned)
Details
In bug 1512754, I added a reftest that the test image has a background-color animation in an opacity element and the reference image uses static background-color with the same DOM tree structures. In the test where background-color is animated, we don't flatten the opacity so that we composite the opacity value on the compositor. Unfortunately it results 1-value differences on R and B components on every pixel on a certain platforms (e.g. MacOSX). This bug is for investigating how the difference happens and find out whether we can avoid it or not.
Below link is the result of the reftest on MacOSX.
https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://queue.taskcluster.net/v1/task/Nq4OiYoiT56uB8xGQDWaoA/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1
Initially I was thinking the difference is caused by GPU drivers or some such, but interestingly the difference doesn't appear on gpu reftests on Windows 7 whereas it appears on normal reftest on Windows 7.
Updated•4 years ago
|
Severity: normal → S3
You need to log in
before you can comment on or make changes to this bug.
Description
•