Same here on Windows 10, Chrome version 58.0.3018.0.

Same here, very annoying, can't find a setting/flag/something to correct the rendering. Windows 10,  Version 58.0.3018.0 canary (64-bit)

Same,
58.0.3016.0 - no issues
58.0.3017.0 and newer - desaturated
I thing it`s related to gpu because if i uncheck hardware acceleration in chrome://settings and restart it`s work fine

This worked for me. What GPUs does everyone have? I'm using an AMD Radeon HD 7870 XT. (Also on Windows 7 and Version 58.0.3018.0 canary (64-bit))

Same here. Using an nvidia MSI GTX 760. Tried updating to the latest drivers, no changes.

Google Chrome (58.0.3013.3) works fine.

same thing here. disabling hardware acceleration in settings fixes the problem though

@rpop for triage, this is really bad. I'm seeing this too.

+Needs-Bisect if QA can figure out when this started happening.

siggi@ is seeing this too. Hint: we both remote desktoped into those machines before restarting chrome... not sure if related but could be if hard to repro (not everyone here has a repro..).

I'm seeing this in Version 58.0.3019.1 canary SyzyASan. My Chrome is all dark and ugly too - Gotham City style.

Adding some GPU folks to take a look.

I ran into this on my setup at home, and was able to confirm that this only happens when color management is active.  If I set my color profile to sRGB (under display settings, advanced) and restart Chrome, then things render as expected.  Switching back to the custom profile I use for my display, I get the effect described here.

Interesting, maybe Remote Desktop (re. #10) is triggering color management?

I rebooted my Z840 and the problem persists, so it's likely not related to remote connections.

@#18, funny, just did the same. Confirm it sticks across OS reboot.

This also seems to result in corruption of the autofill flyouts; sometimes (but not always) they fail to paint.

Deleted: AutoFillFlyout.png 9.4 KB

	AutoFillFlyout.png 9.4 KB View Download

 Issue 694335  has been merged into this issue.

I did a manual bisect at home that led to these 2 commits, suspecting the 2nd one:
https://chromium.googlesource.com/chromium/src/+/4ad0a609fea4c17f2b2d71e446bfae18c956b243
https://chromium.googlesource.com/chromium/src/+/ecdfdf5b38b44e99023608dac72afa4f6d293d8a

The merged issue includes an about://gpu dump, more screenshots, and a link to a Reddit thread:
https://www.reddit.com/r/chrome/comments/5v8lvb/chrome_canary_showing_dark_colors_both_in_the/

Possibly relevant CL within the regression range: https://chromium.googlesource.com/chromium/src/+/ecdfdf5b38b44e99023608dac72afa4f6d293d8a

 Issue 694296  has been merged into this issue.

Forum : https://productforums.google.com/forum/#!topic/chrome/KObe42dKcDk

ccameron@, can you please look into the change (https://chromium.googlesource.com/chromium/src/+/ecdfdf5b38b44e99023608dac72afa4f6d293d8a) per c#23? Will update the thread once i find a clear repro.

Thank you!

The code in that patch should not be having any effect yet.

I'll add some additional checks to ensure that it doesn't do anything.

Do we have any other experiments running on Canary that might be causing this (like in  issue 691522 )?

I'll revert that patch in an hour or so if I can't figure out what's going on.

I seem to recall a variations lookup tool somewhere?, but I can't seem to find it at the moment.

Here's a Windows 10 machine with the repro:

Variations	
ea8deb27-3f4a17df
16e0dd70-3f4a17df
a444064c-1410f10
241fff6c-70be023
c68ab9a3-ca7d8d80
1e528f0f-f23d1dea
6025934e-3f4a17df
7c1bc906-ff96eb58
63ac5233-af9fd986
1c752ce9-f23d1dea
6f6207b1-ca7d8d80
cf558fa6-bb299ce2
a5cb8590-3f4a17df
58aac55e-3f4a17df
5ca89f9-f23d1dea
f1ab784a-3d47f4f4
f3499283-ca7d8d80
76b48ab8-a2567007
31362330-3f4a17df
349d561b-3d47f4f4
2d2fd3f5-f23d1dea
9e201a2b-65bced95
5274eb09-3f4a17df
57f575bb-3f4a17df
858ecb94-3f4a17df
684d1cdf-f23d1dea
bcc907f7-870290a7
949c6e63-3f4a17df
b791c1b8-3f4a17df
9773d3bd-1410f10
b22b3d54-65bced95
93731dca-6ba77eb3
2e109477-4f8eb0c8
99144bc3-4da47e09
f79cb77b-1963db0f
23a898eb-3f4a17df
d39326b0-3f4a17df
62d3c5a9-f23d1dea
4ea303a6-da9d8333
3a007b7-46180e7c
447469ba-ca7d8d80
7aa46da5-4995b330
64224f74-1410f10
de03e059-1410f10
f069f10b-65bced95
ad6d27cc-c6d02f41
69bf80fa-3f4a17df
2697ea25-3fc362e4
f3ea30a0-ca7d8d80
b2f0086-b4530445
7fc902e8-f23d1dea
d747916f-e510b2e9
f11cb941-f11cb941
6844d8aa-669a04e0
494d8760-f23d1dea
3ac60855-3ec2a267
f296190c-f794ffd3
4442aae2-4ad60575
ed1d377-e1cc0f14
75f0f0a0-6bdfffe7
e2b18481-4c073154
e7e71889-e1cc0f14
89b3132e-720b026c
644b8345-726d8ace
61b920c1-720b026c
e979f496-33d20d90
6cea8bf2-bda4c67c
828a5926-ca7d8d80
81c6897f-3d47f4f4

Okay, I think I see the issue ... we have a master switch to turn off color conversion by default.

But ... that switch is halway-side-stepped if your monitor's ICC profile has a non-analytic ICCProfile.

I'm double-checking this locally now. The fix should be very small.

Yup, this was the issue. Adding tests to the fix.

Fix is up at
https://codereview.chromium.org/2705203003/

I had to grab a really big .icc to make the tests actually test the bug in question ... I may want to separate it out into "just the fix" and "just the test".

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1d14a77942910d9f60af73c674809b1f82de3847

commit 1d14a77942910d9f60af73c674809b1f82de3847
Author: Christopher Cameron <ccameron@chromium.org>
Date: Wed Feb 22 00:29:22 2017

color: Fix accidental partial color conversion

For color spaces that can't be represented by an matrix plus analytic
transfer function, we need to ensure that we don't accidentally enable
color conversion when the destination is undefined.

Tests may be disruptive and are in a separate patch.

BUG= 694191 
R=hubbe@chromium.org

Review-Url: https://codereview.chromium.org/2712533002 .
Cr-Commit-Position: refs/heads/master@{#451858}

[modify] https://crrev.com/1d14a77942910d9f60af73c674809b1f82de3847/ui/gfx/color_transform.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1e5ed6bb29f1884453c279a4427a94054bf343f2

commit 1e5ed6bb29f1884453c279a4427a94054bf343f2
Author: Christopher Cameron <ccameron@chromium.org>
Date: Wed Feb 22 00:53:29 2017

color: Fix accidental partial color conversion

For color spaces that can't be represented by an matrix plus analytic
transfer function, we need to ensure that we don't accidentally enable
color conversion when the destination is undefined.

Tests may be disruptive and are in a separate patch.

BUG= 694191 
R=hubbe@chromium.org

Review-Url: https://codereview.chromium.org/2712533002 .
Cr-Commit-Position: refs/heads/master@{#451858}
(cherry picked from commit 1d14a77942910d9f60af73c674809b1f82de3847)

Review-Url: https://codereview.chromium.org/2705223003 .
Cr-Commit-Position: refs/branch-heads/3018@{#3}
Cr-Branched-From: 5e7216844858ad1d08a70ac7aeef88547db2be7f-refs/heads/master@{#451537}

[modify] https://crrev.com/1e5ed6bb29f1884453c279a4427a94054bf343f2/ui/gfx/color_transform.cc

Just to update I am unable to get the experiment(as in  issue 691522 ) when passing chrome from the terminal using 'chrome.exe --force-fieldtrial-params=VideoColorManagement/WithColorMgmt/' or 'chrome.exe --enable-features=video-color-management' on Windows-10, chrome version: 58.0.3019.0. 

The experiment is enabled by default on one of the colleagues canary installation. Will verify this once the canary upgrades to 58.0.3020.0(which should contain fix from C#36).

I had observed the issue on Win 10 using previous canary 58.0.3019.0 and its working fine now on 58.0.3020.0.Hence added respective TE-verified labels for this.

This was reproducible on the Windows-10 system of durga.behera@ as updated in C#39. Checked the same on that system and this repros on both signed/unsigned build of 58.0.3019.0. 

Works fine on Dev RC(58.0.3018.3). Adding the verified label for chrome version: 58.0.3018.3 as well.

 Issue 694041  has been merged into this issue.

I've just updated to 58.0.3020.0 and I'm not experiencing the issue any more.
it's fixed for me
thanks!

I can confirm that 58.0.3020.0 fixed the problem, thank you for your efforts!

Version 58.0.3020.1 canary SyzyASan here, problem fixed.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/86a6cab9976c0355b241d85c30655c2ae51d0005

commit 86a6cab9976c0355b241d85c30655c2ae51d0005
Author: ccameron <ccameron@chromium.org>
Date: Wed Feb 22 22:28:10 2017

color: Tests for fix accidental partial color conversion

Add tests for this issue.

BUG= 694191 

Review-Url: https://codereview.chromium.org/2705203003
Cr-Commit-Position: refs/heads/master@{#452236}

[modify] https://crrev.com/86a6cab9976c0355b241d85c30655c2ae51d0005/ui/gfx/color_transform_unittest.cc
[modify] https://crrev.com/86a6cab9976c0355b241d85c30655c2ae51d0005/ui/gfx/icc_profile.cc
[modify] https://crrev.com/86a6cab9976c0355b241d85c30655c2ae51d0005/ui/gfx/icc_profile.h
[modify] https://crrev.com/86a6cab9976c0355b241d85c30655c2ae51d0005/ui/gfx/test/icc_profiles.cc
[modify] https://crrev.com/86a6cab9976c0355b241d85c30655c2ae51d0005/ui/gfx/test/icc_profiles.h

Added tests, closing.

 Issue 694105  has been merged into this issue.