Re: ICC specification bug
Re: ICC specification bug
- Subject: Re: ICC specification bug
- From: Henrik Holmegaard <email@hidden>
- Date: Sat, 9 Dec 2000 07:00:10 +0100
It was pointed out off-line that it is not a bug to name the LUT
AtoB0 (alphanumeric) even if the intent is named say Absolute
Colorimetric (alphabetic).
In device profiles, there are individual LUTs which perform according
to their alphanumeric names. When device profiles are chained on the
fly, a preprocess LUT is built which is the equivalent of a link
profile.
So a link profile may hold the equivalent of more than one (device)
LUT. Which also means that the last digit in the link LUT reflects no
intent, given that the digit isn't used to select a LUT whn there is
only that single one in the fixed conversion structure.
I agree that with a link profile the last digit isn't used to select
the LUT, because there is only that one LUT to select from, but I
don't agree that the LUT in a link should use the same alphanumeric
naming scheme as in device profiles. This is the confusing part.
Given that the LUT in a link profile will represent at least two
device profile LUTs, then it is simpler and much clearer if the
upcoming revisions to the spec quit the three digit system for link
profiles.
The LUT in a link profile might instead be named AtoBn, the letter
'n' to imply 'unspecified number'. At least something other than the
scheme for device profile LUTs, and that in addition implies unknown
multiple sources.
Also, as I seem to have been saying for years and years, the header
must allow details, and the more so with link profiles. When loading
a device link profile, it is crucial that I know what the source
color space and the destination color space were in the original
device profiles.
(Flue season, back to tea and bed ... -:)).
--
------------------------------------------
Henrik Holmegaard, TechWrite
Stationay +45 3880 0721 - +45 3881 0721
Mobile +45 2178 3959
Toelloesevej 69, 2700 Broenshoej, Denmark
------------------------------------------