Re: Is this a scum dot?
Re: Is this a scum dot?
- Subject: Re: Is this a scum dot?
- From: Graeme Gill <email@hidden>
- Date: Thu, 14 Nov 2002 15:58:21 +1100
bruce fraser wrote:
>
If the scanner profile translates RGB 255,255,255, relcol, to
>
something less than L*100, it is by definition broken.
I'd expect it to typically translate it into something
more than L*100 (assuming it wasn't clipped in the
mechanics of the ICC profile) - the paper color of a test chart
is not usually 100% reflective, and the scanners would
normally have some margin to allow for reflectance up
to or beyond 100%.
>
Typically, scanner testcharts record the actual media white of the
>
target in the target description file, and profilers try to reproduce
>
that white in the profile. Different profilers do different
>
normalizations, but it's unusual and generally undesirable to capture
>
the white of the target as L*100, or for that matter to set the
>
scanner to produce RGB 255 from the paper white (or filmbase+fog).
As I mentioned, there are many ways this can be done,
but a profile that is running as a Relative Colorimetric
intent would be expected to do exactly that, render the
color relative to the paper. "the paper" as far
as the ICC profile is concerned is the scanner test
chart paper (what else could it be ?), and "relative to
the paper" means that typically an RGB of less than
255,255,255 will result in an Lab of 100,0,0
( A scanner I am familiar with returns an RGB value
of 249, 244, 255 for the test chart paper, for instance).
>
It's entirely likely that a scanner will, at some point, be called
>
upon to scan something brighter than the white of the calibration
>
target...
Sure, but that's just the way ICC stuff is defined.
Running in Relative Colorimetric mode, it will likely give clipped
results. If you were to run the scanner profile in absolute
colorimetric mode, then this theoretically will give a valid result,
in practice there are problems with trying to use scanner profiles
in absolute mode. One is that the accuracy of the profile will be
compromised outside the gamut exercised by the scanner test chart,
and the other is that scanner profiles with a PCS of Lab don't
have the range to represent relative PCS values above L 100,
which they need to do when used in absolute mode (they get clipped).
To be fair, Profile Maker (as an example) does seem to do things differently
to what I would expect, but then I would regard its scanner profiles
as broken. It seems to make the profile "Relative" by forcing
RGB 255,255,255 to translate to Lab 100,0,0, but this makes
no sense in the context of a scanner. A scanner could
have arbitrary offsets and scale factors applied to its RGB values,
so RGB 255,255,255 doesn't necessarily have any real world
meaning in the scanners RGB space, nor does it
have any logical relationship to the media white
(which is what Relative intent is all about).
Consider a scanner that returns the RGB value 130, 255, 255
for a perfect diffuser. It would be kind of strange, but as long
as the RGB space encompasses a reasonable gamut of
samples colors, it will still be useful as a scanner. What
meaning for it does an RGB value of 255,255,255 have ? (a value
it would never produce).
Graeme Gill.
_______________________________________________
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.