148619 - CSS color wrong in 16-bit display

Status: RESOLVED WONTFIX

(Core Graveyard :: GFX, defect, P3)

Description

[Boris 'pi' Piwinger]

On http://piology.org/ I have the following defined in my CSS:
a:hover
{
  background-color:#fff0df;
}

It basically does what it should, but making a screenshot and checking it, I see
that #fff1df is used for display. I chose the color to match the background
image, so this is pretty obvious (surprisingly I can see it with the naked eye).
I tested with IE6 which matches the colors exactly, so this is not a problem of
making the screenshot or in my graphics program.

pi

Comment 1

[Boris 'pi' Piwinger]

I missed to say: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.0+) Gecko/20020530

I have seen that effect (but not verified) with Linux. I'll check tomorrow and
report that.

pi

Comment 2

[Boris 'pi' Piwinger]

Attachment: Screenshot using IE6

Colors match here.

Comment 3

[Boris 'pi' Piwinger]

Attachment: Screenshot using Mozilla

Colors are off by one.

pi

Comment 4

[David Baron :dbaron:]

Could you give more details about your graphics configuration (color depth,
video card, etc.), and perhaps a screenshot using Mozilla?

Any chance this is related to bug 1759?  Do you have a side-by-side screenshot
of Mozilla and IE?

Comment 5

[David Baron :dbaron:]

->GFX.

Comment 6

[Boris 'pi' Piwinger]

There are no transparancies AFAICS, so not bug 1759.

My graphics card is a Hercules 3D Prophet 4000XT.

My screen settings are 1024*768 with 16-bit colors.

Anything more?

I don't think, this is important, but all my pages validate (HTML and CSS).

pi

Comment 7

[Chu Alan #2]

it seems that i cannot reproduce it with my old 2002052621/Linux build ...

By the way, is this bug related to bug 114701, there is a link to testpage in
comment #3, you may try that.

Comment 8

[Chu Alan #2]

Attachment: is there any problem in this shot?

pi, can you see any problem in this shot?
by the way, the link to testpage should be in bug 114701 comment #3.

Comment 9

[Boris 'pi' Piwinger]

That might indeed be related. I see a problem in that testpage. I don't really
understand this stuff, so I cannot tell, if this is the same thing.

Your attachment is correct, the colors match.

BTW: There was a mistake in my original report. Displayed is ffefdf, i.e., one
smaller, not bigger, than defined in CSS.

pi

Comment 10

[Boris 'pi' Piwinger]

Changing to 32-bit colors solves the problem. Still, IE could do it for 16-bit.

pi

Comment 11

[Christian :Biesinger (don't email me, ping me on IRC)]

Which mozilla version are you using?
Is the background image a PNG image?

Comment 12

[Boris 'pi' Piwinger]

From comment 1: Mozilla/5.0 (Windows; U; Win98; en-US; rv:1.0.0+) Gecko/20020530

The background is http://piology.org/img/backpi.gif (and it actually is a GIF,
2-colored).

So the point of this bug is, that two identical RGB colors are displayed
differently, depending on where the show up. That should be consistent.

pi

Comment 13

[Boris 'pi' Piwinger]

Attachment: Screenshot with Linux

Very strange. The color in the pi (from background image) should be one colore,
but I see two there (difference four). One matches the hover background. Maybe
this is a mistake in ksnapshot.

pi

Comment 14

[Boris 'pi' Piwinger]

The longer I look at it, on Linux I also see it.

pi

Comment 15

[Kevin McCluskey (gone)]

-> dcone

Comment 16

[Chu Alan #2]

here is some of my thought about this bug and bug 114701 which made me think
these two bugs are related :-

1. Both bugs are about inconsistent colour display (the appearant colour of
image is lighter) from same colour value from an image and from other sources
(CSS, attributes etc.)

2. Both bugs only occurs on 16-bit colour settings (see comment 10 and bug
114701 comment 5)

3. pi did not pass the testcase (comment 9) :)

Comment 17

[Boris 'pi' Piwinger]

I really don't understand that stuff. If one of you experts want to mark it as a
dupe, I'm totally happy. Assuming that other bug gets fixed;-)

pi

Comment 18

[dcone (gone)]

There can be a few things that make a difference here.  A 16 bit display means
you are gonna use 5 bits per color for each color.. and have on bit left over. 
Depending on the graphics card that one bit will be used for one of the colors..
so graphics card to graphics card those colors can change.  

Now.. and application can choose to use only 5 bits per color on a display..
that makes it more consistent.  Explorer could be doing this.  I also think but
have not found a way to detect how a color card uses these bits and can make an
adjustment.  As far as Mozilla is concerned.. unless there is something I am
missing.. we don't adjust these colors, we just pass them off to the card..
telling it to draw using these colors.  I will double check this.. but I am
almost positive.  The color differences you see..may be some gamma correction,
card correction, etc.  Explorer may deal with things by an internal adjustment.

Comment 19

[Gordon P. Hemsley [:GPHemsley]]

This bug has been buried in the graveyard and has not been updated in over 5 years. It is probably safe to assume that it will never be fixed, so resolving as WONTFIX.

[Mass-change filter: graveyard-wontfix-2014-09-24]