W3C home > Mailing lists > Public > www-style@w3.org > August 2009

Re: [CSS3 gcpm] CMYK colors and color management.

From: Chris Murphy <lists@colorremedies.com>
Date: Sun, 2 Aug 2009 23:52:00 -0400
To: W3C style mailing list <www-style@w3.org>
Message-Id: <94A95251-094C-4A0B-B817-3D093F8826FE@colorremedies.com>
Cc: David Perrell <davidp@hpaa.com>
On Aug 2, 2009, at 9:50 PM, David Perrell wrote:

> Chris Murphy wrote:
> | So you'd color manage images but not color manage CSS content?
> Did I say that?


"I don't see inclusion of CMYK without profiles as disastrous.  
Spec'ing CMYK
colors on paste-ups didn't require printer profiles. But images are  

Seems to say that tagging of CSS content isn't necessary, but you have  
another rule for images. True I am assuming you mean tag images, but  
if you don't then the entire paragraph is unclear.

> | We have
> | that very problem right now with RGB and it's a big sticking point
> | when different kinds of content can and can't be color managed
> | correctly. It causes discontinuity between, say a CSS background and
> | the borders of neighboring images or images with transparency and...
> Not sure how many are stuck. Authors can ensure all content is sRGB or
> untagged.

No they can't. It's sRGB only. See 4.2.1

> But, yes, it seems discontinuity has occurred with PNG due to
> overambitious gamma correction[1]. And now Firefox 3.5 is color- 
> correcting
> PNG and JPG[2] without correcting other content.

It only color manages images that have embedded ICC profiles. There's  
a bunch of info in bugzilla indicating bogus gAMA and cHRM chunks  
rampant in PNG, so my understanding is those chunks are always ignored  
in PNG. (Possibly it is color managed if cHRM primaries are sRGB  

> Will this produce a
> firestorm of discontinuity disgust among color-profile-ignorant users,
> prompting them to besiege the W3C with demands for corrective action?

Given the current implementation, I think it's unlikely many will  
encounter tagged RGB images. If they do, they will more often than  
not, display better than they would have without being tagged. At this  
point, CSS content is not assumed to be sRGB and is still  
uncompensated, and the same applies to untagged images, and of course  
any content managed by a plug-in. Case in point, a big hold up on  
color managing CSS content by default, has been the fact Flash did not  
support color management until version 10. And with version 10, it's  
not the default behavior.

> More
> likely they'll continue to wonder why each application treats the  
> same data
> differently.

Very possible. For the most part there is a rather clear path on how  
all applications should treat the same data.

> And they just might wonder why color correction can't be done
> transparently throughout their entire system, not just on an app-by- 
> app
> basis. Monitors, display adapters, printers...they can all  
> communicate with
> the OS. Can't they just get along?

Bugzilla has examples of ICC display class profiles supplied by  
manufacturers that do not withstand the scrutiny a house plant would  
give it, let alone an engineer, and the subsequent people who should  
have tested the profile. These were found during beta the instant the  
application first starting using the display profile by default, and  
end users reported problems because - WOW they actually have eyeballs  
- apparently unlike some people in charge of creating, testing and  
then posting completely jacked display profiles on unsuspecting users.  
(i.e. background white turning blue or yellow - not subtle stuff.)

A certain well know inkjet printer manufacturer has had a nearly  
impossible time following directions for Mac OS X driver behavior when  
it comes to color management. Product, after product, after product  
continues to be released over the course of YEARS, and they cannot  
seem to get it right. It should be embarrassing. Apparently, it's not  
embarrassing enough.

Nothing surprises me at this point.

> | The point is there is such a thing as RGB output and yet CSS doesn't
> | have a mechanism for that...
> As well it shouldn't.

The fallout of this is that we also can't define intermediate color  
spaces other than sRGB, such as Adobe RGB (1998), or ECI RGB.

And insofar as there is a module for generated content for paged  
media, it seems like you might want to have more than one color space  

Chris Murphy
Color Remedies (TM)
New York, NY
Co-author "Real World Color Management, 2nd Ed"
Received on Monday, 3 August 2009 03:52:42 UTC

This archive was generated by hypermail 2.4.0 : Friday, 25 March 2022 10:07:38 UTC