Re: Is this a scum dot?
Re: Is this a scum dot?
- Subject: Re: Is this a scum dot?
- From: "Bruce J. Lindbloom" <email@hidden>
- Date: Mon, 18 Nov 2002 16:49:05 -0600
Bruce Fraser wrote:
>
If you map "reference white" (the brightest white on the target) to
>
"device white" you're making the assumption (which experience tells
>
me is unwarranted) that the "reference white" is the whitest white
>
you'll be asked to scan. That is definitely untrue for
>
batch-manufactured IT8 targets.
I think we are having a communication problem between us -- time for a baud
rate check. Let me restate my position again...
There are important issues on both a theoretical side and a practical side
that must be considered. You cannot blindly take one side and ignore the
other. You must use the strengths of both to your advantage. Make an
intelligent compromise. Look at the two sides:
Standards and Specifications
----------------------------
Section 4.5 of IT8.7/1-1993 and IT8.7/2-1993 both say:
"Patch A16 shall be the minimum neutral density. Patch L16 shall be the
maximum neutral density that the product can normally achieve."
Section D.2.6.1 of the latest ICC profile specification says:
"For media-relative colorimetric transforms in the context of ICC profiles
the media XYZ tristimulus values are reproduced relative to the media white
point. Given the media-relative reproduction provided by the media-relative
colorimetric intent, L* = 100 for media white. The PCS tristimulus values
for a media-relative colorimetric transform are also media-relative, that
is, PCS L* = 100 for media white."
From this, it is reasonable (dare I say proper?) to interpret that for
relative colorimetric rendering intent, an input profile should map the RGB
value for patch A16 into Lab 100, 0, 0. Nothing is said about what RGB value
patch A16 should have. Nothing is said dictating that RGB 255 should map
into Lab 100, 0, 0.
Practical Considerations
------------------------
In the real world, a scanner profile is made from an IT8 target. The
particular target used has a certain L* value for patch A16. When actual
originals are scanned, it is quite likely that some may have "whiter whites"
than patch A16 of the IT8 target (let's call this a "special original"). How
is this to be handled?
a) If a scanner is set up during profiling so that its RGB response for
patch A16 is just barely 255, then RGB color clipping will occur when
scanning a special original. This is bad.
b) If the scanner is set up during profiling so that the RGB value of patch
A16 was less than 255 (say 240, but it could be something else), then
scanning the special original would produce RGB greater than 240 (let's say
250), so RGB clipping has not occurred. But since the profile maps 240 into
Lab 100, 0, 0, then our 250 would probably get mapped into Lab 100, 0, 0
also, so we still get clipping. This time on the Lab side of the profile
instead of the RGB from the scanner. This is still bad.
c) The only recourse I can think of (short of custom profile editing for
each original or scanning in AbsCol mode and adjusting the white point
manually afterwards) is to adjust the white point of the scanner when
scanning the special original so that its whitest white has the same RGB
value as patch A16 of the IT8 (namely 240 in our example). This prevents
both RGB clipping and Lab clipping. We are cheating of course, because we
have changed the scanning setup. But we are forced to cheat since our
special original has colors outside the gamut of the IT8 target.
(There are also many other important factors in scanner setup: noise,
quantization, number of bits, tone curve shape, etc.)
Regarding the statement that "by definition" RGB 255 must map into Lab 100,
0, 0, I am not aware of any such definition. That does not mean you
*shouldn't* do it, I just don't think you *must* do it. Perhaps it would
have been better to say "by expert opinion" instead. I can think of a case
where you might want RGB 255 to map into Lab 100, 0, 0: If your profile had
that behavior, then very bright specular highlights (whose detail you don't
care about) would be "caught" at pure white as all three RGB channels
saturate in the scanner. But I don't think any standards control this
particular mapping.
--
Bruce J. Lindbloom
email@hidden
http://www.brucelindbloom.com
_______________________________________________
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.