Graeme Gill schrieb:
Some comments spring to mind:
This probably won't work for narrow band scanner sensitivities.
Thats' what I suspect as well.
How do you calibrate your spectral scanner ?
What maths is needed to extract spectral info from RGB+filter output ?
I was basically thinking of methods like here
http://www.cis.rit.edu/mcsl/research/PDFs/Camera02.pdf
But as you say, there's a bunch of literature, describing several
possible methods.
Likely we do not even need to reconstruct the reflectance spectra, but a
mapping from the N device channels of the "modified scanner" to XYZ
(under D50) is sufficient. Basically this could be established like a
RGB to XYZ mapping. However, a CLUT with 9, 12, or even 15 input
dimensions is no longer practicable. With a large number of device
channels, maybe even a Nx3 matrix matrix is sufficient? The more device
channels (with linearly independent CMFs) are captured, the closer the
device to XYZ mapping can be described by a matrix (assuming linear raw
digital counts). http://www.cis.rit.edu/mcsl/research/PDFs/Camera05a.pdf
reports pretty good numbers with only 6 or 9 device channels and a
linear mapping. But they don't use a scanner, but a camera which likely
behaves a priori more colorimetric. So if a Nx3 matrix does'n suffice,
then maybe a matrix using 2nd order mutivariate polynomial terms of the
N device values?
Regarding calibration, I've no evidence whether the spectra found on an
IT8 target are sufficient to establish the mapping. If not, then maybe
additional targets with inkjet spectra might be required.
Is it all worth it?
Right, that's probably the main question.
For that much effort, you may be better of buying
a spectrometer, if your aim is to read test patches, rather than
do a spectral scan of artwork etc.
I do have one anyway, so that't not the issue. But obviously many people
are seeking rather for a $100 and not a $1000 solution.
Thogh the art reproduction folks have proven that these methods
basically do work, I have no evidence, whehter it may or may not work
eventually in practice with a consumer scanner, cheap filter foils, etc.
(due to various possible limitations I can imagine).
But granted that it would work reasonably, wouldn't it be worth a few
dollars for the end user to buy some filter foils, and the effort to
make e.g. four or five scans instead of one?
Regards,
Gerhard
There is a bunch of literature about this sort of thing, mostly
from the art preservation/reproduction folks, who are most active
in pursuing spectral reproduction.
Graeme Gill.
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________
Lcms-user mailing list
Lcms-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/lcms-user
