Closed Bug 1153923 Opened 9 years ago Closed 6 months ago

Font rendering is extremely poor over certain backgrounds.

Categories

(Core :: Graphics: Layers, defect, P3)

37 Branch
x86
macOS
defect

Tracking

()

RESOLVED INCOMPLETE

People

(Reporter: ccurzio, Unassigned)

References

()

Details

(Whiteboard: [gfx-noted])

Attachments

(2 files)

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:37.0) Gecko/20100101 Firefox/37.0
Build ID: 20150402191859

Steps to reproduce:

This issue has been around for a long time. Fonts render extremely poorly when displayed over certain kinds of backgrounds. Once you've scrolled past the background, font rendering returns to normal.


Actual results:

Fonts display very poorly when displayed over a background, with extremely jagged aliasing. Please see the attached screenshot. Compare the top three instances of "(Self.DaystromInstitute)" to the bottom two. You can observe the behavior here: http://www.reddit.com/r/DaystromInstitute/

This is readily reproducible on my 2010 iMac running OS X 10.10. This does not happen using Safari. I am also seeing a similar issue while filling out this Bugzilla form, however I am seeing odd behavior. In the "What happened?" field, whenever text scrolls the aliasing gets completely destroyed - but it corrects itself after exactly five seconds. 



Expected results:

Fonts need to remain consistently displayed.
Build ID: 20151208030212
User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:45.0) Gecko/20100101 Firefox/45.0

Hi, 
I tested this on Mac OS X 10.10 and I can reproduce this, fonts are displayed very poorly. On Chrome works ok. 
Go to this link: http://www.reddit.com/r/DaystromInstitute/
Fonts are displayed very poorly, see the attachment.
Status: UNCONFIRMED → NEW
Component: Untriaged → Layout: Text
Ever confirmed: true
Product: Firefox → Core
Presumably a matter of subpixel vs grayscale antialiasing....
(In reply to Boris Zbarsky [:bz] from comment #3)
> Presumably a matter of subpixel vs grayscale antialiasing....

I'm not sure it's quite as simple as that. In attachment 8591756 [details], both the "bad" and "good" renderings of "(Self.DaystromInstitute)" show subpixel AA; but the bad ones look like they've been through some sort of traumatic filtering or gamma adjustment or something, leaving them a poor shadow of their former selves.

(In reply to ccurzio from comment #0)
> I am also seeing a similar issue while filling out
> this Bugzilla form, however I am seeing odd behavior. In the "What
> happened?" field, whenever text scrolls the aliasing gets completely
> destroyed - but it corrects itself after exactly five seconds. 

I think this *is* grayscale vs subpixel AA, and it's linked to the difference between an active and inactive layer, or something. I see the same thing here (in Nightly) in a bugzilla comment field: scrolling causes the text to render with grayscale AA, and then after a few seconds of inactivity, it re-renders with subpixel AA. On a retina screen, I can barely tell the difference (magnifying the screen greatly makes it easier to see), but with a non-retina display it would be much more glaring.

Moving this to Graphics, as I think it's all tied in with rasterization/layers/compositing/gfx-ish stuff like that.
Component: Layout: Text → Graphics: Layers
Does this still reproduce?
Flags: needinfo?(ccurzio)
Whiteboard: [gfx-noted]
Tested on Mac OS X 10.10 with FF 47 and FF Nightly 50.0a1 and I can't reproduce it. Ccurzio please retest with the latest FF version and see if you can reproduce it?
Severity: normal → S3
Flags: needinfo?(ccurzio)

Reporter, are you still experiencing this issue?

Flags: needinfo?(ccurzio)

Redirect a needinfo that is pending on an inactive user to the triage owner.
:bhood, since the bug has recent activity, could you have a look please?

For more information, please visit BugBot documentation.

Flags: needinfo?(ccurzio) → needinfo?(bhood)
Status: NEW → RESOLVED
Closed: 6 months ago
Flags: needinfo?(bhood)
Resolution: --- → INCOMPLETE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: