Note: There are a few cases of duplicates in user autocompletion which are being worked on.

Subpixel antialiasing too strong/incorrect everywhere

VERIFIED FIXED in mozilla7

Status

()

Core
Graphics
VERIFIED FIXED
6 years ago
6 years ago

People

(Reporter: echogene.alpha, Assigned: Joe Drew (not getting mail))

Tracking

unspecified
mozilla7
x86
Linux
Points:
---
Dependency tree / graph

Firefox Tracking Flags

(Not tracked)

Details

Attachments

(4 attachments, 1 obsolete attachment)

(Reporter)

Description

6 years ago
User-Agent:       Mozilla/5.0 (X11; Linux i686; rv:7.0a1) Gecko/20110526 Firefox/7.0a1
Build Identifier: Mozilla/5.0 (X11; Linux i686; rv:7.0a1) Gecko/20110526 Firefox/7.0a1

Subpixel antialiasing has become too strong/wrong in the latest builds of Nightly.  That is, where it is red and blue, it is too red and too blue.  This happens on both web pages and the entire user interface (menu bar and below).

Reproducible: Always

Steps to Reproduce:
1. Open Firefox
2. Look at text

Actual Results:  
Text looks weird.  Upon zooming in, it's easy to see that subpixel antialiasing is wrong.

Expected Results:  
Text should look like it did in previous builds.
(Reporter)

Comment 1

6 years ago
Created attachment 535848 [details]
Comparison of antialiasing

A comparison of subpixel antialiasing in Nightly and Firefox 4.
That's a pretty glaring difference. Jeff, might this possibly be related to the recent cairo update (bug 562746)?
Status: UNCONFIRMED → NEW
Ever confirmed: true
Component: General → Graphics
Product: Firefox → Core
QA Contact: general → thebes
The cairo update alone is definitely responsible for a change in text rendering on my Linux machine.
Blocks: 562746
tracking-firefox7: --- → ?

Comment 4

6 years ago
Since cairo 1.9.10, this change is in effect http://cgit.freedesktop.org/cairo/commit/?id=7a023a62f7517ad0d54f4d59c99909fadcc05e82

I think it maybe related, https://bugs.freedesktop.org/show_bug.cgi?id=34886#c3
(disclaimer: I am not an expert in cairo, freetype, fontconfig or mozilla; so don't take what I say to heart... :)).
(Assignee)

Updated

6 years ago
Status: NEW → RESOLVED
Last Resolved: 6 years ago
Resolution: --- → DUPLICATE
Duplicate of bug: 456448
(Assignee)

Comment 6

6 years ago
OK, I'm going to undupe this for a short-term patch. The long-term fix is bug 456448, but we're blocked on releng for that.
Status: RESOLVED → REOPENED
Resolution: DUPLICATE → ---
(Assignee)

Comment 7

6 years ago
Created attachment 538345 [details] [diff] [review]
dlsym FT_Library_SetLcdFilter for now

OK, let's just dlsym() FT_Library_SetLcdFilter for now. This is not a long-term fix, because on computers that don't have this symbol we'll be a rainbowy mess. (We have to stop supporting distributions with old freetypes.) But until all the infrastructural work to upgrade Freetype and stop supporting old versions of it is done, this'll do.
Assignee: nobody → joe
Status: REOPENED → NEW
Attachment #538345 - Flags: review?(karlt)
Attachment #538345 - Flags: review?(bjacob)
(Assignee)

Comment 8

6 years ago
Created attachment 538346 [details] [diff] [review]
dlsym FT_Library_SetLcdFilter for now

Forgot to qrefresh.
Attachment #538345 - Attachment is obsolete: true
Attachment #538345 - Flags: review?(karlt)
Attachment #538345 - Flags: review?(bjacob)
Attachment #538346 - Flags: review?(karlt)
Attachment #538346 - Flags: review?(bjacob)
Attachment #538346 - Flags: review?(bjacob) → review+
Comment on attachment 538346 [details] [diff] [review]
dlsym FT_Library_SetLcdFilter for now

I probably would have replaced initialized_setLcdFilter with

  static setLcdFilterFunc setLcdFilter =
      (setLcdFilterFunc) dlsym (RTLD_DEFAULT, "FT_Library_SetLcdFilter");

in the code block scope but I haven't checked the C standard to make sure that is called on first pass, and what you have does the right thing.
Attachment #538346 - Flags: review?(karlt) → review+
(Assignee)

Comment 10

6 years ago
That's how I started out, but it seems you can't initialize statics with a non-constant.
Oh, interesting, thanks.

Comment 12

6 years ago
Mozilla/5.0 (X11; Linux i686; rv:7.0a1) Gecko/20110610 Firefox/7.0a1

Using a range of Unicode box-drawing characters and mathematical operators, you can upgrade to Bugzilla Color Edition™!

││││││││
││││││││
││││││││
││││││││

║║║║║║║║
║║║║║║║║
║║║║║║║║
║║║║║║║║

||||||||
||||||||
||||||||
||||||||

∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣
∣∣∣∣∣∣∣∣

⊪⊪⊪⊪⊪⊪⊪⊪
⊪⊪⊪⊪⊪⊪⊪⊪
⊪⊪⊪⊪⊪⊪⊪⊪
⊪⊪⊪⊪⊪⊪⊪⊪

⟒⟒⟒⟒⟒⟒⟒⟒
⟒⟒⟒⟒⟒⟒⟒⟒
⟒⟒⟒⟒⟒⟒⟒⟒
⟒⟒⟒⟒⟒⟒⟒⟒

∐∐∐∐∐∐∐∐
∐∐∐∐∐∐∐∐
∐∐∐∐∐∐∐∐
∐∐∐∐∐∐∐∐

This comment works as a testcase for me. These are all quite distinctly coloured on my screen. It's almost likeable.

Comment 13

6 years ago
Created attachment 538565 [details]
Pretty!

Note that I have my monospace font size set to 16 in Firefox as the default is too small for me. Zooming in and out changes the colours :-)
(Assignee)

Comment 14

6 years ago
http://hg.mozilla.org/integration/mozilla-inbound/rev/76d7b28ca1ba
Whiteboard: [inbound]
http://hg.mozilla.org/mozilla-central/rev/76d7b28ca1ba
Status: NEW → RESOLVED
Last Resolved: 6 years ago6 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla7
(Assignee)

Updated

6 years ago
Whiteboard: [inbound]
Duplicate of this bug: 663698

Comment 17

6 years ago
Created attachment 538765 [details]
Fixed

Confirming that this is fixed for me (compare with attachment 538565 [details]).

Updated

6 years ago
Status: RESOLVED → VERIFIED
(Assignee)

Updated

6 years ago
Depends on: 666732

Updated

6 years ago
tracking-firefox7: ? → ---
Blocks: 404637
You need to log in before you can comment on or make changes to this bug.