Re: EZ Color and incompatible scanners

[Joe Petolino <petolino@alum.mit.edu>]

wirem@home.com writes:
>>After much work with Monaco EZcolor (1.6 and 2) and two scanners, I've
>>determined that the reflective IT8s (I have worked with three) all render
>>towards the magenta in a way that other materials typically don't. The
>>two scanners I've used are Umax Astra 1200S and HP 7400C.
>>
>>For example, comparing a Macbeth Color Checker chart mid grey with the
>>grey surround of an IT8, I find that when viewed by a 5000K light
>>box they appear similarly neutral. I do not find that the
>>scanners--especially the HP--see
>>them as similar. The scanner sees that Macbeth chart as neutral, but sees
>>the IT8 as strongly magenta. This leads to a green cast when printer
>>profiles made using this scanner.

CDTobie@aol.com
>This is *true* metamerism: meaning two objects that appear similarly colored 
>under one light, and different under another. As long as you only scan color 
>prints, or color film, then this should be a minimal issue, as thats what the 
>IT8 is on... I tend to scan a wide range of natural media art, ink jet 
>prints, and other stray stuff; and find IT8 based scanner profiles to be 
>pretty much useless for that. Its a bit of a joke to use an IT8 in a printer 
>profiling program when the other half of the scan (the patches from the 
>printer) will vary from the IT8 colors to differing degrees with different 
>scanner bulbs.

Perhaps scanner-based profiling would work better if you did it with an
IT8 that was printed on the printer/paper/ink combo that you're
trying to profile, instead of an IT8 printed on Kodak paper.  (You'd
need an accurate colorimetric data file to go with this target, of course).
I think that would eliminate the metamer problem.  It may even be
sufficient just to use the same ink set, even if it was printed on a
different paper with a different printer.  Of course, this approach
probably wouldn't work with commercial profiling packages - I suspect
that they won't let you use a different IT8 than the one that came
with the software.

Anyway, I found an interesting paper which discusses some of the issues 
in scanner-based printer calibration:

    http://graphics.lcs.mit.edu/~victor/publications/pdf/SPIE94_2appraoaches.pdf

    [ yes, that how the link is spelled ]

Besides ink metamerism, other big problems they found were:

    * A scanner will read a different value for a given color patch,
	depending on where you position the patch on the scanner bed;

    * A scanner will read a different value for a given color patch,
	depending on what other colors are nearby.

Perhaps that last point explains why ColorVision switched to an ordered
target (with similar colors near one another) in ProfilerRGB version 2.1.

Anyway, getting back to the topic that started this thread, a while ago 
someone asked if ColorVision had any recommendations for which
scanners to use with ProfilerRGB.  I found the answer, right there on
page 4 of the ProfilerRGB 2.1 manual!

	    System Requirements
	    
	        . . .

	    * A flatbed scanner which can scan, at minimum, 24-bit
		color, 300 dpi.  Any inexpensive scanner will do.
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Maybe you need to buy a cheaper scanner . . .

-Joe
-- 
Joe Petolino                                            petolino@alum.mit.edu
-
Turn off HTML mail features. Keep quoted material short. Use accurate
subject lines. http://www.leben.com/lists for list instructions.