Vlad Lucaci, Desktop QA (:vlucaci) Reporter
	Description • 6 years ago

[Affected versions]:
-61.0b4 
-62.0a1 (2018-05-10)
-60.0esr


[Affected platforms]:
Windows 10x64
Mac OSX 10.13x64
Ubuntu 16.04x64

[Steps to reproduce]:
1.Launch the Firefox browser.
2.Browse to this .png image : https://tinyurl.com/yd357bk8


[Expected result]:
-Image should be displayed without any rendering issues or texture flaws.

[Actual result]:
-It can be noticed that the image is not displayed correctly and there are rendering issues.

[Regression range]:
This issue is a regression:
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=c00a9c1940c5&tochange=e794cef56df6

	Vlad Lucaci, Desktop QA (:vlucaci) Reporter
	Comment 1 • 6 years ago

Note: 
This issue does not reproduce in Chrome .

	Ryan VanderMeulen [:RyanVM]
	Comment 2 • 6 years ago

huh, I'm not seeing anything super-obvious in that range. Maaaaaybe bug 750334?

	Vlad Lucaci, Desktop QA (:vlucaci) Reporter
	Comment 3 • 6 years ago

Hey Ryan,

The problem here is not as in bug 750334.

For this issue to occur, the user is not required to resize the image in any way.

If you take a closer look to the black marbles , you can see that where the light reflection should be(as per compared with the image in Chrome ) the rendering is actually broken.

	Ryan VanderMeulen [:RyanVM]
	Comment 4 • 6 years ago

Yes, I can see the artifacts this bug got filed for. I'm just looking at what might have changed in our code to have caused them, and that looked like a possible option since it touches imagelayer code.

	Xidorn Quan [:xidorn] UTC+11
	Comment 6 • 6 years ago

So, I can reproduce this issue on Safari and the Preview app on macOS 10.13.4.

Also, I tested in Opera, and this issue is reproducible with Opera 50 and 51, but not reproducible with Opera 53. Opera 51 corresponds to Chrome 64, and 53 corresponds to 66. It looks like there is something specifically got fixed between Chrome 64 and 66.

	Xidorn Quan [:xidorn] UTC+11
	Comment 7 • 6 years ago

I would say this is probably an invalid issue. The artifacts shown in the images are transparent pixels, and in Firefox and many other places, the background color is light color, and thus they show up. However, in Chrome, the background color is black, so they seem to be just black.

If you use Chrome's inspector to change <img>'s background to white, you would see exactly the same problem. And if you use Firefox's inspector to change <img>'s background to black, you would see this issue gets fixed.

Since I don't think the spec specifies what background should it be when displaying a standalone image, I tend to think this bug should be closed INVALID.

	Xidorn Quan [:xidorn] UTC+11
	Comment 8 • 6 years ago

The related code is https://searchfox.org/mozilla-central/rev/a85db9e29eb3f022dbaf8b9a6390ecbacf51e7dd/toolkit/themes/shared/media/TopLevelImageDocument.css#14

So move to Toolkit anyway.

	Xidorn Quan [:xidorn] UTC+11
	Comment 9 • 6 years ago

(I found this because I tried to open this image in GIMP, and observed something different from both Firefox and Chrome. When I zoomed in that image, I saw the transparent checkborad pattern, so I realized the reason here.)

	Xidorn Quan [:xidorn] UTC+11
	Comment 10 • 6 years ago

So in that range, this was likely "regressed" by bug 754133.

	Jared Wein [:jaws] (please needinfo? me)
	Comment 11 • 6 years ago

I'll close this as invalid. It's unfortunate in this case that the color that is behind the transparent pixel clashes with what the user may expect, but in the majority of cases we see images such as graphs and charts that expected a white background. Viewing those graphs and charts against a dark background made them impossible to read and was a regression from before we switched to the dark backgrounds on image documents.