some thoughts on CIELAB was Dry Time for UltraChrome inks/Calibration
some thoughts on CIELAB was Dry Time for UltraChrome inks/Calibration
- Subject: some thoughts on CIELAB was Dry Time for UltraChrome inks/Calibration
- From: "tlianza" <email@hidden>
- Date: Wed, 5 Feb 2003 16:21:34 -0500
Robin Myers wrote:
"This is one of the reasons that the L*a*b*
was never intended for color matching calculations. It was designed to
be used for representing color differences only. There are a number of
color science volumes that discuss this and I would invite you to check
the reference list on my website for them."
Robin does a great service here. Many years ago, not too long after I was
starting Sequel, I was asked to publically review the first Revision of the
ICC spec at a forum at Stanford University. The Key Speaker was Dr. Hunt
(author of the book "The Reproduction of Colour") and he and I had a chance
to go to dinner. I told him that I was a little suprised about the
selection of Lab especially for hardcopy profiling. None of the supporting
work of the space supported this use, PARTICULARLY with respect to inter
media comparisons at mismatched luminances.
1. The space didn't have a visual correlate to saturation. (Although a lot
of people mistakenly thought that "colorfullness" was somehow the same as
saturation....)
2. It used a bizarre Chromatic normalization scheme that resembled the
recommendations of Von Kries, but actually would later come to be called the
"Wrong Von Kries" transform. Of course, that didn't matter because the ICC
Spec recommended the same transformation to "adapt" to the PCS.They added
insult to injury by calling this "normalized" CIEXYZ colorimetry.
3. The lightness transformation was well understood to work over a very
limited range of white differences and ambient conditions. That was well
understood in the 1950's and 60's,.
4. The shape of the space, with respect to the gridding process resulted in
very inefficient use of the cubic space that it was specified to occupy.
This implied that you had to use very fine sampling or very careful sampling
and fine interpolation.
5. Sampling in the native colorspaces, using either log or linear mappings,
didn't seem to map well to Lab so sophisticated 3 and 4 dimensional
interpolations would have to be used.
6. Finally, when plotted in Lab space, most hardcopy devices, "hooked" which
implied that numerical mappings would not be single valued....very "uncool"
mathematically. If the devices included a pre linearization and limiting
step, then the hooking was minimized.
Hunt smiled a bit and said something like, "Well, you understand the
problem, now let's get that bottle of wine....". Hunt had been doing a lot
of publishing of his work in appearance and he was well aware of the
politics in the color community.
Many knowledgable people raised objections to utilizing that space but there
was no mechanism to vote your concerns because the intial club was closed to
all but the original founding companies. When it finally opened up, the cost
of entry was a real barrier for many small companies. It was also obvious to
me that it had been essentially hijacked by a ten pound guerilla in the
printing sector who had already embraced and marketed the CIELAB political
mess.
What I find interesting and often frustrating is how "embedded" this
thinking has become in the ICC community. A generation of "experts" have
grown up that have basically accepted and promoted the dogma that was forced
into the process for reasons that were more political and economic than
technical.
The densitometer is very useful because it makes a measurement that better
relates to the machine dependencies than a spectro. It costs less so they
can be deployed in greater numbers than spectros.It was suggested that one
might want to include linearization information in the profile so that the
profile could be checked with a minimum number of measurements. Had the
historical elements been stored in the profile, we could easily see if the
process has drifted. . The politics of the time dictated that everyone would
soon own a spectro so why bother putting the quality control features in the
system?
We have a framework of a system that works in many cases. It certainly
hasn't met the hype and the implementation issues are very hard on the
hardware guys. The pace of change is glacial and I would argue that the
whole process was compromised from day one. As I read many of the posts
on this group, we see that all is not a bed of roses in the ICC world. It's
great that we have folks like Robin, Graeme, and Bruce L. who point us in
the directions of the theory and application of color science. They are
voices of experience with good theoretical backgrounds.
Take care,
Tom Lianza
Technical Director
Sequel Imaging Inc.
25 Nashua Rd.
Londonderry, NH 03053
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.