Reproducible on 54.0.2792.0/8581.0.0 (Official Build) dev-channel Parrot.

@Sindhu: Could you please update the M-53 or M-52 behavior as well. Marking this Beta for M-54 as of now.

Issue is working fine in 52.0.2743.57/8350.46.0 beta channel parrot. i.e; No red color is seen in fish bowl.

This looks like parrot specific? In that case we should not make this a beta blocker since it is only blocking one board. vsuley@/dhaddock@ can you please confirm this is still reproable on M53 latest build?

Song, could you verify this issue in M53?

If this is happening just on Parrot, this should be still happening on all SNB devices.

songsuk@ Can we please validate and confirm this is still happening and on all devices?

Not reproduce the issue on Chrome 53.0.2785.23/CrOS 8530.24.0 - Candy, Daisy.  
I'll check the issue on Parrot/m53( issue629136 ) & m54. 

The issue happens on Chrome 53.0.2785.24/ CrOS 8530.25.0 - Parrot

Adding kbr@. kbr@, Does this look like a graphics issue for parrot? Removing RBB since this issue only happens on this site and does not block any functionality.

This is a question for Canvas 2D TL junov@. My expertise lies mainly in WebGL.

Moving this nonessential bug to the next milestone.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

I've seen similar symptom before. We get a fully red background when the codec required for decoding the video of the wave pattern is missing.  In this case, the wave pattern is present but it is faint, and the red is also faint, which suggests that this video is getting properly decoded, but the video element is not compositing correctly.

The video used by the video element is: https://testdrive-archive.azurewebsites.net/performance/fishbowl/Video/Water.mp4

Perhaps you can use it to produce a reduced test case.

Why would this be a video bug? :-)

IIRC fishbowl is a video file, but it's possible I'm remembering incorrectly :)

Yes the water is a video, but I think the bug is that we import it in the wrong pixel format. I don't think the bug will be a video bug in the end...

Adding junov@ again. Any idea how we might verify #18 or fix, Justin?

+cc ccameron, hubbe
Looking for insight regarding the compositing of video elements.

Does the video file look good on it's own, or is it pink too?

https://testdrive-archive.azurewebsites.net/performance/fishbowl/Video/Water.mp4

#22, the video file on Water.mp4 looks good. Water color is blue. 
Chrome 53.0.2785.55/CrOS 8530.49.0- Parrot

If translate the video into YUV, and replace the V plane with a duplicated U plane, it becomes pink. Not sure if that's what's going on here, but it could be.
I've seen similar bugs when working with NV12 before, does this device use NV12?

this doesn't seems like a media>video bug. remove Media>Video

> this doesn't seems like a media>video bug. remove Media>Video

Based on what?

This P1 issue has been dropped on the floor.  marcheu@ could you please help find an owner on CrOS side?

I just retested, and the background is solid red where the water should be. So according to comment #12 it shoudl be a video bug...

It reproduces on lumpy too btw.

posciak: ping.  This is a P1 that has been languishing.  Please look at it as soon as you can.

This reproduces for me on Linux desktop Chrome 54.0.2840.90, so it doesn't seem to be CrOS- or HW acceleration-specific...

Reproduces on my pixel, the water video is completely gone and the background is more red than shown in the screenshot above. 

In fact the video layer is 100% red on the pixel.

Reproduce the comment 32&33 on Firefox (Ubuntu 14.04) with  https://developer.microsoft.com/en-us/microsoft-edge/testdrive/demos/fishbowlie/ 


The fishbowl water/color is properly displayed on chrome 56.0.2905.0/CrOS 8953.0.0-Parrot with https://people-mozilla.org/~jmuizelaar/fishbowl/fishbowl-gl.html

Looks like there are some significant differences between these two fishbowl implementations. On my linux machine, the first one (microsoft) turns red, the second one (mozilla) doesn't.

(Tested on both stable and ToT with the same results.)

Wait a minute.....
It just fails to fetch the video! (404)

Re #36 - hubbe@, so the video missing issue is just a 404 on all your devices (linux, pixel)?

It also sounds like you were seeing a brighter red than the original repro - so this may just be the 404 having a different symptom.

posciak@, can you confirm whether your Linux repro is the brighter-red 404 issue? Or the original light-red issue from Parrot?

I am seeing a 404 for Water.mp4 in the developer tools, yes.

This issue is seen in Linux with latest 57.0.2937.0 dev build. Attaching screenshot for reference.

Deleted: Issue 627707.png 368 KB

	Issue 627707.png 368 KB View Download

I think the 404 issue is different from the original issue, but it makes it hard to reproduce the original issue....

Yeah I'm 90% sure that we have a second issue now, which is just hiding the original issue. I certainly remember seeing red water (with the little waves etc.) and not just plain red.

The video is 404, so the full red now on the page is just background colour of page.

A version of the page is on github, it contains a valid version of the video, should help with repro the original issue: https://github.com/MicrosoftEdge/Demos/tree/77ea2fe7fbe5b17c0f39a74ebce9502f4e1e1b43/fishbowlie

It looks like the Linux 404 issue was a red herring.  It's necessary to test with the GitHub version in #42.

Could someone please re-test on a Parrot device?

When testing with the github version, it's still broken on ToT.

Testing this on a parrot device, it was broken in M53. (worked in M52)
If I disable hardware video decoding in about:flags, the pink hue goes away.
Could it be that M52 didn't have hardware decoding on this device?

We've had hw video decode for much longer than that on parrot (it was the first intel device where we enabled it, probably circa 2012/2013).

This seems to be some sort of alpha problem, because when if I press F12 and find the video element and change it's background color from "red" to "black", the demo looks normal. (And if I change it to "white", the water disappears completely for some reason...)

I'm afraid I'm not enough of a chromeos expert to triage this further.
Assigning to posciak@ to see what he can do.

Back to vsuley@ to see if the problem persists.