PNGs render with yellowish tint
Reported by
das.x...@gmail.com,
Oct 31 2016
|
|||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/56.0.2902.0 Safari/537.36 Example URL: https://commons.wikimedia.org/wiki/File:Screenshot_of_MassMessage_MediaWiki_extension.png Steps to reproduce the problem: 1. Open any PNG image file What is the expected behavior? Untinted image, in true colors expected. What went wrong? The colors of the image are being rendered in a warmer colorspace and/or with an added yellow tint. Does it occur on multiple sites: Yes Is it a problem with a plugin? No Did this work before? Yes Does this work in other browsers? N/A Chrome version: 56.0.2902.0 Channel: dev OS Version: 10.0 Flash Version: Shockwave Flash 23.0 r0
,
Oct 31 2016
Of course, here is a screenshot with the expected result in Firefox (Left) and the problem in Chrome (Right). I even changed to incognito mode and reset all flags to make sure it isn't anything experimental or plugin related.
,
Oct 31 2016
Ouch. I'm definitely not seeing that. Does this still occur in the latest Chrome Canary? I would also try disabling all extensions as well to make sure that's not causing any issues. Interestingly, the other images on the page are also PNGs and they look fine (the logo in the top left and the icons). Does this occur on other pages? In any case, I'll leave this for triage with the image team.
,
Oct 31 2016
,
Oct 31 2016
Due to recent changes to color correction?
,
Oct 31 2016
That seems very likely. I'll take this. das.xwcg@gmail.com, can you please provide color space configuration of your monitor as an ICC profile (this is how it is communicated from Windows to Chrome)? For this image, Chrome will be performing a color correction step to the monitor color space - this is likely where the problem is.
,
Oct 31 2016
msarett@chromium.org, I have attached the ICC profile used by my monitor at work. I'll try to send you the one I have at home too when I get home, I experience the same problem there!
,
Oct 31 2016
Thanks a lot! That would be very helpful.
,
Oct 31 2016
The following revision refers to this bug: https://skia.googlesource.com/skia.git/+/0186661e85737ac2f4805f876c8d2d4157126f68 commit 0186661e85737ac2f4805f876c8d2d4157126f68 Author: Matt Sarett <msarett@google.com> Date: Mon Oct 31 17:41:57 2016 Reject non-D50 matrices from ICC profiles BUG:660838 GOLD_TRYBOT_URL= https://gold.skia.org/search?issue=4200 Change-Id: Ib57eb3705d6fe638e3a9cb56788937fc7e282847 Reviewed-on: https://skia-review.googlesource.com/4200 Commit-Queue: Matt Sarett <msarett@google.com> Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Leon Scroggins <scroggo@google.com> [add] https://crrev.com/0186661e85737ac2f4805f876c8d2d4157126f68/resources/icc_profiles/SM2333SW.icc [modify] https://crrev.com/0186661e85737ac2f4805f876c8d2d4157126f68/src/core/SkColorSpace_ICC.cpp [modify] https://crrev.com/0186661e85737ac2f4805f876c8d2d4157126f68/tests/ColorSpaceTest.cpp
,
Oct 31 2016
When the above change rolls into Chrome, it should fix the "yellow" issues. I will leave this bug opened though. It appears that the monitor ICC profile that you've shared is invalid because the matrix is not specified in D50. However, I think the intent of the profile is perfectly clear - it specifies the matrix in terms of D65. I want to make sure I'm not missing something here. Maybe it's ok to use D65 matrices as long as they are clearly marked as D65? Or maybe it's technically invalid, but we can still respect the intent of the profile and do a useful transformation anyway?
,
Oct 31 2016
,
Nov 11 2016
Marking this as fixed. It is the correct behavior to reject the invalid profile. |
|||||||||||
►
Sign in to add a comment |
|||||||||||
Comment 1 by bokan@chromium.org
, Oct 31 2016Components: -Blink Blink>Image
Labels: Needs-Feedback
NextAction: 2016-11-14