New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 788236 link

Starred by 4 users

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Mac
Pri: 2
Type: Bug-Regression



Sign in to add a comment

Videos are displayed with the wrong colormatrix

Reported by 1234andr...@gmail.com, Nov 23 2017

Issue description

Chrome Version       : 64.0.3273.0  (Canary)
URLs (if applicable) : http://hr-a.akamaihd.net/test/colorspaceTest/videoplayer.html
OS version           :  OS X 10.12.6
Audio/Video format (if applicable): Mpeg4 / h.264
Behavior in Safari (if known): OK
Behavior in Firefox (if known):Also wrong.


Video issue

Flash or HTML5?   HTML5

What steps will reproduce the problem?
(1) Open http://hr-a.akamaihd.net/test/colorspaceTest/videoplayer.html
(2) Press Test all

I wrote a small test tool, wich is checking, if the right colorspace is used.
 http://hr-a.akamaihd.net/test/colorspaceTest/videoplayer.html

All Colorbars are from the same Souce, but coded with different colospaces and resolutions.
The used colospaces is signaled in the h.264 SPS/PPS headers

Expected ColorBar-RGB-Values:
White:255,255,255 Yellow:192,192,000 Cyan:000,192,192 Green:000,192,000 Pink:192,000,192 Red:192,000,000 Blue:000,000,192 Black:000,000,000
(+-2 is still ok)

 bt709  is ok:  
White:255,255,255 Yellow:192,190,001 Cyan:001,191,191 Green:001,190,000 Pink:192,000,193 Red:192,000,002 Blue:001,000,192 Black:001,000,001


What went wrong? 
For videos with bt601, the  wrong Colorspace  is used: 
White:255,255,255 Yellow:195,179,000 Cyan:000,172,194 Green:000,161,000 Pink:206,030,199 Red:209,018,000 Blue:000,011,201 Black:001,000,001 



Any other comments?
Chrome Canary on Android does it the other way around wrong. So it uses always the bt601 color matrix
Firefox does it also wrong.
Safari is ok. 
Chrome Canary on Windows works ok.



 
Components: Internals>GPU
Labels: -Type-Bug -Pri-3 hasbisect-per-revision Needs-Triage-M62 Triaged-ET M-62 OS-Mac Pri-1 Type-Bug-Regression
Owner: ccameron@chromium.org
Status: Assigned (was: Unconfirmed)
Able to reproduce the issue on mac 10.12.6 using chrome stable version #62.0.3202.94 and latest canary #64.0.3277.0.
Issue is specific to OS-mac.

Bisect Information:
=====================
Good build: 62.0.3196.0    Revision(497279)
Bad Build : 62.0.3197.0    Revision(497604)

Change Log URL: 
https://chromium.googlesource.com/chromium/src/+log/15a57a6bc51143ccbb44711bfdf2338701b4b4bd..bf7445cbca7f05ad78e941c2f8e04ea9175f97fa

From the above change log suspecting below change
Change-Id: I120440fe2e78d44fb1561028e0f9282b2ccaf6c1
Reviewed-on: https://chromium-review.googlesource.com/634162

ccameron@ - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Thanks...!!
Thanks for the fast response!
Should I file an other Bug for Chrome Canary on Android 7.1.1? 
Cc: hubbe@chromium.org
This one bug should be enough.
hubbe, chris, ping on this one as well. 

Comment 6 by hubbe@chromium.org, Dec 8 2017

How were these files created?
I noticed that the bt601 file is tagged as 240M instead of the more common 170M, but I'm not sure if that's causing any problems or not yet.

Cc: ligim...@chromium.org
Labels: -M-62 RegressedIn-62 ReleaseBlock-Stable M-65 Target-65 FoundIn-64 FoundIn-65 FoundIn-63
This is a recent regression, can we get a fix during M65 time frame?
Adding an RB label for tracking purpose, please change if needed.
Probably an issue with YUVToRGBConverter -- this is reading the video back in a canvas?
Labels: -Pri-1 Pri-2
This isn't a recent regression.
Just to update:

Still we are seeing the same issue on Mac 10.12.6 using chrome latest Canary-65.0.3309.0.As hubbe & Cameron seems OOO,could someone from GPU/Media team please take a look into this issue as it is marked as stable blocker.

Thanks in advance..!
Friendly ping to get an update on this issue.
Thanks..!
Cc: -hubbe@chromium.org ccameron@chromium.org
Labels: -ReleaseBlock-Stable
Owner: hubbe@chromium.org
Dropping RBS since this shipped in M62; hubbe@ can you take a look?
Looking...

Back in office now, let me want to take this on.
Sure, let me know if you want me to do anything, as I actually have a mac now. :)

Cc: -ccameron@chromium.org hubbe@chromium.org
Owner: ccameron@chromium.org
Owner: fs...@chromium.org
This is a <canvas> issue. In canvas elements, we aren't as precise about video color spaces as we are when compositing. I suspect that we hardcode bt709 somewhere. ->fserb

Sign in to add a comment