For instance the target ECI2002CMYKi1(A3)_1_2.tif
contains - according to the reference file - CMYK
numbers by steps of 10, some values by steps of 5
and some single values 3 and 7. The maximum is 100.
All tests by CS2.
Let's consider for a patch with one value 70.0 the
Photoshop 8bpc: 70%
Photoshop 16bpc: 70%
Photoshop 32bpc: 0.698
Aha, it's a problem of mapping 0..100 to 0..255
and eventually of wrong rounding.
This leads to the conclusion, that CMYK values
in the TIFF should be defined as integer numbers
0..255 and that the reference file should contain
decimal numbers with sufficient accuracy in this
format: ###.## .
It's perhaps not really essential for the accuracy
of a calibration, but why should everything be based
on a wrong concept ?
By the way:
I think I've solved the problem, how to modify targets
for a guaranteed inklimit, if this should be still