Re: use of sRGB as a default
Re: use of sRGB as a default
- Subject: Re: use of sRGB as a default
- From: bruce fraser <email@hidden>
- Date: Sun, 20 Jun 2004 13:38:50 -0700
At 12:57 PM -0700 6/20/04, John Zimmerer wrote:
Bruce,
Let me know which apps, and I'll file the bugs. Remember, 10.3.4 for
proper behavior in Preview.
I'll try to get you a comprehensive list next week.
Let's be honest: The choice of color space to use for untagged data
is arbitrary. There is never a right answer, there's just the
prevailing practice.
Yup.
There is no such thing as an sRGB device. Some may have sRGB modes,
where they try to simulate sRGB, but we're still left with
device-dependent color.
Yup.
If I were to speculate how much web content was intentionally in
sRGB, I'd be willing to bet that only 10% or less of web content has
intentionally been converted to sRGB. Most is simply untaggged, so
we'll never know for sure.
All true. But it's a fair bet that the vast majority of such content
was created and viewed on a gamma 2.2 display with sRGB-like
primaries, and it's a given that most online print services, and most
other things that interact with this untagged content, assume sRGB
absent any better information.
Still the broader point is without a profile, no one will ever know
for sure which color space the content creator had in mind. Tag all
data, and the point is moot.
You must live in a very different universe from the one I do. My Macs
are more than fast enough to do on-the-fly display compensation-heck,
they've been doing it since Photoshop 5-but my DSL line sure as hell
isn't fast enough to make me enjoy 3,144 bytes of extra data in every
JPEG. Talk to the ImageReady team sometime about what their product
actually gets used for-it's mostly about shaving every last K
possible off images.
Now, as to the question of Safari, I'll grant that the reining
standards body does have clear recommendations here with respect to
untagged data. I won't argue the merit of their recommendation,
since the audience is obviously pro sRGB here. So if the group feels
so strongly that we should treat all untagged data as sRGB, then
I'll pass that along to the Safari product manager.
To be pellucidly clear, my feeling is that the only excuse for
producing untagged RGB color data is when it's destined for the web,
and (until we all move to a HDR, 32-bit floating-point imaging model
with overrange encoding) that data should always be sRGB because
a) sRGB is the space that comes closest to representing the "average"
viewing condition
b) most consumer-level cameras and printers, most online printing
services, and a very large number of digital minilabs assume sRGB and
at least attempt to simulate it
c) if people want to produce web content that looks decent on 97% of
the machines used to view the web, converting to sRGB is really the
only option. No Windows browsers currently in use pay any attention
to embedded profiles-they just throw RGB counts at the display,
resulting in a gamma closer to 2.2 than not (and certainly closer to
2.2 than to 1.8), with primaries that resemble those of sRGB.
d.) It IS the recommendation of the reigning standards body.
It feels weird to be characterized as "pro-sRGB"-it'll be big news to
Michael Stokes-but on the Web, sRGB is a fait accompli, and building
a system that ensures that the Web looks different (not better, just
different) on Macs from the way it does on the vast majority of
computers seems a misimplementation of Think Different.
Yes, the idea of scanners, cameras, and printers producing sRGB is of
course a damned lie, but that's how people view, evaluate, edit, and
on rare occasions, tag said RGB. Arbitrarily remapping all that color
to something else seems distinctly unhelpful.
I think we both want the same thing, which is to make color
management disappear as a concern for normal human beings. I just
don't see the current design furthering that end, though I remain
open to being convinced.
Best,
Bruce
JZ
On Jun 20, 2004, at 11:46 AM, bruce fraser wrote:
At 12:41 AM -0700 6/20/04, John Zimmerer wrote:
Bruce,
Generic RGB Profile.icc is the default profile for untagged RGB
images on Panther. Its primaries allow for blue skies and red
apples. By using this as the default, all untagged color will be
consistent across all Panther machines.
Two problems with that.
1.) I've yet to find an application that actually uses Generic RGB
rather than usermonitorRGB for untagged images.
2.) Even if apps behaved as expected, unless the image was created
on a Panther machine, you've simply guaranteed that the image will
look consistently wrong on all Panther machines.
I've been as vocal about the shortcomings of sRGB as anyone else on
the planet, but
a) it's based on real device behavior
b) it's perfectly possible to represent both blue skies and red
apples in sRGB
c) if I were to speculate as to how much web content is sRGB vs how
much web content is originated as untagged RGB on Panther machines,
the former number would be very much larger than the latter one.
I still fail to see how this behavior, even if it worked, would be useful.
Bruce
--
email@hidden
--
email@hidden
_______________________________________________
colorsync-users mailing list | email@hidden
Help/Unsubscribe/Archives:
http://www.lists.apple.com/mailman/listinfo/colorsync-users
Do not post admin requests to the list. They will be ignored.