Re: untagged RGB data
Re: untagged RGB data
- Subject: Re: untagged RGB data
- From: John Zimmerer <email@hidden>
- Date: Sat, 20 Dec 2003 12:33:51 -0800
I'm not sure I buy the need for "OS working spaces", since I truly
believe all data should be tagged. And ultimately, this is the domain
of the user space applications (including AppleScript and sips),
since they manipulate color directly. The OS should simply render
source color accurately, and have a single definition (across
workflows and continents) of how untagged source data is treated, per
color space.
John,
There are many reasons for working with untagged data.
RGB profiles are small, certainly, but you need to get off your T1
line and back on dialup to understand the impact of adding 4k to every
single JPEG on the web. It's not going to happen anyway, but I'll
argue strongly that it makes bugger-all sense to load the web with
tons of redundant data in the form of zillions of copies of the sRGB
profile.
For the web, there is already a standard that says you don't have to
embed profiles (sRGB). I'll grant that there is work to do in Safari
(at least) to better handle this assumption. However, if you're happy
to treat all untagged RGB data as sRGB, this would be the first I've
heard of it!
Also, Apple proposed some changes to the CSS spec that would allow
profile references. The recommendations were made several years ago,
but the CSS spec hasn't been finalized (not the version with the
recommendations). This would allow you to reference a profile per
image, or for an entire page. In theory, only one 4K RGB profile would
have to be downloaded for an entire page of images.
For CMYK, the same argument applies in spades. Real World Photoshop CS
will contain about a thousand images, most of them under a megabyte,
the very largest ones maybe 5 megabytes. It will be a cold day in hell
before I contemplate adding 1.5MB of profile data to every image.
You don't have to embed the same profile you used to separate the
images, only a profile based on the one used for separation. Since
you've already used a PCS-Device table to separate the image, all you
really need is one Device->PCS table to take care of proofing the data.
And that table doesn't have to be 16-bit with 33 grid points -- it can
be 8-bit with 11 grid points.
I think I can fairly claim to be as big an evangelist of color
management as anyone else on the planet, but the "all data should be
tagged" argument is not one I'll buy. Not only is it simply not going
to happen, there are plenty of very good reasons why it shouldn't
happen.
If you're going to leave everything to the apps, scrap ColorSync. If
you want a system-level OS, it needs to behave in a way that the
majority of its users will find useful. Right now just about everyone
on this list is telling you that it currently is not doing so, and
giving you detailed reasons why it is not doing so.
There is a single issue at hand in this thread: whether the operating
system should provide the equivalent of working spaces. I'm not yet
convinced that this is necessary. But I'm interesting in hearing the
arguments in favor. Tell me why the OS needs this -- how have we lived
so long without it?
BTW, re:"The OS should simply render source color accurately, and have
a single definition (across workflows and continents) of how untagged
source data is treated, per color space."
Tagging it with the local monitor profile BY DEFINITION guarantees a
unique definition of untagged color for each workstation, possibly
even for each user on a single one...
In Panther, we don't tag with the monitor profile, for the same reason
that Photoshop turned away from using the display as the definition of
RGB. We tag untagged RGB data when printing or saving as PDF with
Generic RGB... this insures that the resulting PDF file will be viewed
the same, and will print the same, on every machine. What it appears
you and others are agreeing on is that having a single definition of
RGB for untagged data is OK (some would prefer sRGB to Generic RGB),
but that the OS should match all untagged data from this singular space
to the display, which we currently don't do. I happen to agree, and as
soon as this can be done without slowing your machine to a crawl, we'll
do it.
JZ
_______________________________________________
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.