Graeme Gill schrieb:
Ahhh, but metamerism still exists! Metamerism due to different ink
will be elminated but there may still be metamerism which causes the
scanner to think that two colors are very similar when they are
actually not.
This can be one issue. Another more certain one when profiling CMYK
devices is that there are spectrally different combinations that have
the same same appearance due to the amount of K used.
Recently, I've actually evaluated an example for metamerism regarding
black generation. For the worst case I encountered, I computed a color
difference of 7-8 dE94 units under F10 illuminant, for two patches,
printed with different black generation, which have approximately the
same color under D50. That's pretty much and by far not neglectible! The
evaluation was based on measurement from a professional inkjet LFP.
Using a printed scanner calibration chart can't compensate for this
(in fact you have to choose the level of GCR to print the scanner test
chart). You would have to print several scanner test charts with a
range of GCRs, and then use the appropriate scanner profile for
measuring printer test patches with the matching GCR to hope to
compensate for this effect.
... or probably restrict to RGB printer profiling (with all its
disadvatages, but under Windows it's in many cases the only option anyway).
Btw, I'm actually wondering whether well-known spectral imaging
approaches might also work for turning a scanner into a more
colorimetric device:
The idea is to scan the same image a) without, and b) with M different
color filter foils between the glass and the paper. So we finally end up
not only with three captured RGB channels, but with M+1 scans, yielding
a total of N=(M+1)*3 (for instance 12 or 15) captured color channels.
Now it should be possible to characterize either the N-channel to
spectral mappping (e.g. based on the first few (e.g. 10-15) principal
components of typical printed reflectance spectra), or a N-channel to
XYZ mapping, which should suffer less from metamerism, than the RGB to
XYZ mapping does.
Of course there are several issues to consider:
- the optimal choice of the filters
- spatial uniformity of the filters
- general and cheap availability of suitable filter foils
- spatial image registration for the M+1 images is necessary
- I'm not sure, if it works as well with narrow-band scanners
- What kind of target to use for characterization - is an
IT8.7 target with spectral reference data sufficient?
- other issues I missed so far ???
Any comments or objections?
Regards,
Gerhard
-------------------------------------------------------
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
