Adding release block label, please undo if not the case.

Can you tell me the devicePixelRatio you get at https://bjango.com/articles/min-device-pixel-ratio/? (Make sure you reset the zoom to 100%)

I couldn't reproduce this on ChromeOS. I'll test it on linux soon.

If this indeed caused this, that means that NTP was relying on buggy behavior of image set. I'll look into NTP code then.

I couldn't reproduce this on Linux ToT either (51.0.2684.0). I tested with following device scale factors.

1, 1.25, 1.5 and 2.

Waiting for the answer for the question above.

With respect to comment 4:

Issue is not reproducible on Linus using latest chrome 51.0.2688.0, but issue is reproducible on Windows OS.

What was your devicePixelRatio?

If this only happens on Windows, the root cause could be same as this Windows bug  crbug.com/561183 .

This bug seems to be specific to a change in behavior of the renderer viewport.

561183 applies to backgrounds in the Chrome Frame. On 561183, NTP was noted to have the correct background scale.

It's not about NTP not filling background. It's been this way even before 49. If that's 

 crbug.com/561183  is about a way to handle multiple device scale factors, and this will indeed changes the devicePixelRatio, which can be affected by this bug. (repro step in 561183 does not include that scenario)

My current guess is that when you zoom in, it'll change devicePixelRatio to 1, but chrome still loads 2x assets due to that bug, which causes misalignment.

I'll try to repro this on Linux by applying the same problem on Linux first. If not, I'll investigate on Windows.

I noticed that my comment was incomplete, sorry about that. What I was trying to say is

If NTP not filling the content is not expected behavior, we should file a separate bug.

Clarifying title after offline discussion.

I could reproduce similar behavior on Linux by mimicking Windows device scale factor behavior (similar to what I did in  crbug.com/561183 ). However, the patch that worked on Linux did not seem to work. I'll investigate this on Windows.

rkote@etouch.net, could you please answer my question about devicePixelRatio?

I want to make sure that we're reproducing your issue correctly.

By the way, here is the behavior I'm getting. At ~67% scale, the background gets bigger, probably because it keeps returning 2x background image even the image-set requested one for 1x.


https://drive.google.com/file/d/0B5h6yCnehPpJMkFyNzEyZlpkQ00/view?usp=sharing


and this CL (https://codereview.chromium.org/1825863002/) fixes this behavior.

I was testing on 1.5 scale factor. I could reproduce when I test it on 1x. I'll look into it.

The same issue but I needed one more change to support 1x -> 2x transition.
CL is up for review. https://codereview.chromium.org/1825863002/

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/cb8feac7b8a21a9e63b9d20ad58d5a8121c28fb2

commit cb8feac7b8a21a9e63b9d20ad58d5a8121c28fb2
Author: oshima <oshima@chromium.org>
Date: Wed Mar 30 22:17:02 2016

Allow windows to use both 100P, 200P assets

Theme may use the wrong assets when deviceScaleFactor/devicePixelRatio changes without this change.

BUG= 561183 , 596038 

Review URL: https://codereview.chromium.org/1825863002

Cr-Commit-Position: refs/heads/master@{#384114}

[modify] https://crrev.com/cb8feac7b8a21a9e63b9d20ad58d5a8121c28fb2/ui/base/resource/resource_bundle.cc

I'll wait a bit to fully baked on ToT.

Your change meets the bar and is auto-approved for M50 (branch: 2661)

Please merge your change to M50 branch 2661 by 5:00 PM on Monday, 04/04/16 so we can take it for next week beta. Thank you.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/363ef03169c5bc60d3ac1a53a169dcd49aba1f85

commit 363ef03169c5bc60d3ac1a53a169dcd49aba1f85
Author: Mitsuru Oshima <oshima@chromium.org>
Date: Mon Apr 04 22:08:34 2016

Allow windows to use both 100P, 200P assets

Theme may use the wrong assets when deviceScaleFactor/devicePixelRatio changes without this change.

BUG= 561183 , 596038 

Review URL: https://codereview.chromium.org/1825863002

Cr-Commit-Position: refs/heads/master@{#384114}
(cherry picked from commit cb8feac7b8a21a9e63b9d20ad58d5a8121c28fb2)

Review URL: https://codereview.chromium.org/1853403002 .

Cr-Commit-Position: refs/branch-heads/2661@{#483}
Cr-Branched-From: ef6f6ae5e4c96622286b563658d5cd62a6cf1197-refs/heads/master@{#378081}

[modify] https://crrev.com/363ef03169c5bc60d3ac1a53a169dcd49aba1f85/ui/base/resource/resource_bundle.cc

Rechecked this on chrome version 51.0.2700.0 on Windows 7. Fix is working as intended. No weird behavior is displayed when performed Zoom operation after applying theme.

Adding TE-Verified labels.

Tested the issue on Windows 7 using 50.0.2661.66.Observed that weird behavior is not seen on zoom action after applied theme.Please find attached screencast.

Marking it as TE-Verified.

Deleted: 596038.mp4 5.6 MB

596038.mp4 5.6 MB Download