Please provide about:gpu from your system.

I suspect this may be a duplicate of  Issue 648466 .

we see similar issues on macbooks/osx (similar, but not the same since you say v55 is fixed), and we narrowed our bug down to these chromium releases:

  Version 393106: FAIL
  Version 393118-400704: OK
  Version 400715-420612(latest): FAIL

https://commondatastorage.googleapis.com/chromium-browser-snapshots/index.html?prefix=Mac/

I hope this is on-topic, and useful info

andrei.neculau: please provide about:gpu information as a plaintext file from all of the affected machines. Thanks.

The cause is known at this point.

Sounds good - will send when I get back to my Mac this evening

here's my about:gpu

PS: once again, I don't want to hijack jochen's thread, as I'm only certain that the symptoms are the same i.e. see our OK/FAIL webgl canvases (on top of a video stream)

Deleted: gpu.txt 8.2 KB

gpu.txt 8.2 KB View Download

Deleted: Screenshot 2016-09-21 16.08.59.png 198 KB

	Screenshot 2016-09-21 16.08.59.png 198 KB View Download

Deleted: Screenshot 2016-09-21 16.08.18.png 64.8 KB

	Screenshot 2016-09-21 16.08.18.png 64.8 KB View Download

andrei.neculau: if your problem is the same then our workaround isn't yet being applied for your machine type. We thought it was only happening on AMD GPUs, but yours is from Intel.

What machine model is yours? The closest we have is a MacBook Air (13-inch, Early 2015) which has an Intel HD Graphics 6000, not the Intel Iris Graphics 6100. WebGL renders correctly on this machine.

Can you provide a test case?

kbr: my model is MacBook Pro (Retina, 13-inch, Early 2015).

Currently our instance is not available to the public :( that's why I focused on finding the chromium builds (implicitly the git commit ranges) where the regressions happened. I understand this is far from the desirable situation.

andrei.neculau: could you please try running Chrome from the command line like this:

/Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome --user-data-dir=/tmp/t1 --disable_multisampling_color_mask_usage

Does that disable flag make it render correctly?

kbr: nope. tested with chromium builds 420612 and 400715

jochen.hartmann: still waiting for about:gpu from you.

andrei.neculau: please provide a link to some public WebGL page which renders incorrectly. Also, please indicate whether launching Chrome 55 (Canary/Dev) from the Terminal with the command line option:
  --disable-features=WebGLImageChromium
makes the page render correctly.

kbr: yup, confirmed. The "--disable-features=WebGLImageChromium" argument silences the brightness bug.

hi, sorry about the delay, here is my gpu info.. 

Deleted: about-gpu.txt 5.2 KB

about-gpu.txt 5.2 KB View Download

also just FYI, as suggested, starting Chrome with /Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome --user-data-dir=/tmp/t1 --disable_multisampling_color_mask_usage fixes the issue of canvases appearing at 50% opacity for my machine

jochen.hartmann: thanks for your GPU info and confirming. The fix for  Issue 648466  will address your problem. I'm duplicating this into the other bug.

andrei.neculau: we need confirmation that your problem still happens on Chrome Canary (55), and a publicly visible test case. Try several publicly available WebGL pages like the WebGL Aquarium. If you can confirm this, please file a new bug, and include about:gpu gathered from Chrome Canary, as well as the link to the test case. Email me the bug ID once filed. Thanks.

kbr: I confirm that the problem still happens on Chrome Canary (Version 55.0.2871.0 canary). Comment 13 tested specifically build 420612. Both versions react the same to "--disable-features=WebGLImageChromium" -> brightness fixed.

Working on finding a test case and will get back to you.

kbr: I was about to open up a new issue with all the collected GPU info, and also a link to a test case, but I noticed my latest canary (56.0.2885.0) does *NOT* exhibit the brightness bug. At this point, I'm unsure whether I need to do something more, or assume that it will be fixed in the upcoming patch releases. Thanks.

andrei.neculau: I think the fix for your machine may have been accidental but if things continue to work for you then there's nothing more that needs to be done. If you find that Canary regresses again then please file a new bug and email me the bug ID. Also, please test periodically with Chrome Beta especially at the branch points and make sure it's working correctly for you. Thanks.