Re: Re(2): use of sRGB as a default
Re: Re(2): use of sRGB as a default
- Subject: Re: Re(2): use of sRGB as a default
- From: Andrew Rodney <email@hidden>
- Date: Sun, 20 Jun 2004 11:38:33 -0600
on 6/20/04 11:05 AM, Matt Deatherage wrote:
>
Most discussion I've read (including Apple's
>
developer documentation) calls sRGB a _color space_, but you just
>
called it a "profile" and a "source profile."
It's all three. The profile describes the color space. It's used for
embedding into the image. And that sRGB profile can be the source for
conversions into another space since all such conversions require two
profiles; source and destination. Yes, confusing!
>
What I'm reading is that I should assume that the image is in the sRGB or
>
generic RGB _color space_, but a color space profile is not a source profile
>
for this purpose, is it?
The main argument here is that it should be one or the other (not both). If
we universally assume all untagged RGB images are sRGB, we might be right
sometimes, we might be wrong sometimes but at least the OS or application
can settle on ONE assumption and if the color appearance of the numbers
isn't so hot with this one assumption, we can try another assumption (assign
something else) or alter the numbers to look the way we wish. But having
more than one assumption at the application or OS level greatly confuses the
issue.
Any untagged document presents a big guessing game. Might as well settle on
one initial assumption and I'd suggest sRGB since there's a huge world out
there that has no access to this Apple "Generic RGB" (whatever that's
supposed to be). At least with a lot of capture devices producing untagged
images, sRGB is usually a pretty close guess (even though none of them
actually produce sRGB). What devices are close to producing "Generic RGB"
(whatever that's supposed to be).
Seems like Apple should have made sRGB the "Generic RGB" (whatever that's
supposed to be).
>
When I'm done with the image, I'm supposed to embed a color space profile and
>
not a device profile?
First off, what do you mean by done? If I'm ready to output a document and I
convert to my printer space using a printer profile, I'm done and the image
will be embedded with the printer profile (using Convert to Profile or mode
change). That output space IS a color space. If I convert from Adobe RGB
1998 to Epson 2200 Luster, the numbers are in output space embedded with the
Epson profile (if saved). If I want to keep the document in Adobe RGB 1998
I'm embedding a working space profile which isn't a device profile since all
working spaces are by their nature no based on any real world device.
>
You're telling me I'm *not* supposed to use any of the characteristics of
>
his display by embedding his display profile in the image file?
Right! Because your display is unique to you. Every display is different. So
when a user treats an image as that device, they insure that every other
user will view the image differently and should I convert to a print space,
every resulting conversion will produce a different set of values. Your
display and my display are different. If we both use the same output
profile, we both get different numbers since the source plays as profound a
role in the resulting numbers as the destination. However, if we both use a
working space which isn't based on any device (it's Quasi-Device
Independent), then if we both convert from sRGB to Epson 2200, we both get
identical numbers unlike starting the conversion from your display. So the
idea of Photoshop 5 and later is to divorce the display from the editing of
our images. The display profile is only used for previews.
>
As a programmer, btw, I tune out right about here - I'm not writing Photoshop
>
(theoretically) and how files round-trip between my application and a
>
six-year-old version of Photoshop does not interest me at all.
It should because the Adobe color architecture is clever, flexible and works
really well. The idea of insuring we all edit our images the same, see them
the same, and convert them the same by divorcing the display is key here.
This is how someone viewing my images on a 2.2 calibrated gamma display sees
the same preview as I would see viewing them on a 1.8 calibrated gamma
display or for that matter, all displays which have an accurate display
profile. Prior to PS5, it was total chaos since everyone saw the same
numbers differently (that makes no sense).
>
Programmers want to know what they should do with untagged images by default,
>
both in the simple (invisible) case (like Preview),
Any untagged image presents a big guessing game. We don't know the
definition of the numbers to preview or convert the data. So we have to
guess (unless we know for a fact what the numbers really are and we can
assign the correct profile). So the bottom line here is we need to guess and
what's the best guess? Well sRGB is only because so many devices produce
untagged images in something reasonably close to this color space. If over
night all devices that produced untagged images produced something close to
Adobe RGB 1998 then that would be the best guess and sRGB would not.
>
So, as a programmer, do I *ever* embed a display profile?
There are some very,very rare exceptions but generally speaking, no.
>
Apple often has its own style guidelines, but if the company doesn't have a
>
standard set of terms for these concepts, perhaps the ColorSync community
>
should make one and try to enforce it.
Once we get an idea what the hell their doing (and that's a moving target as
seen between Jaguar and Panther and now we are told, things will change in
Tiger).
Andrew Rodney
http://digitaldog.net/
_______________________________________________
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.