Martin Weberg wrote:
Any suggestions on how to find the true TAC?
It depends what you're after. The exact behaviour will depend on the CMM used - different CMM's may use different interpolation algorithms within a grid cell. Putting this aside, to narrow down the worst case would probably involve searching the cells for worst case candidates, and then using an analytic or numerical approach to computing the worst case within each cell. A cruder approach is to sample the the profile behaviour, and hope that fine enough sampling returns an accurate enough value.
(you didn't do a new profile with the FOGRA39L-data with your development code then running icclu?)
Yes, the results are a little different. Possibly this is due to not using CIE delta E 94 in computing the per channel curves in the V0.60 release. The differences in the underlying profile (as opposed to the B2A table) are smaller: xicclu -fif -ir FOGRA39L_v60.5.icm 0.000000 0.000000 0.000000 [Lab] -> Lut -> 1.000000 0.857005 0.475548 1.000000 [CMYK] (clip) xicclu -fif -ir FOGRA39L_vcurr.icm 0.000000 0.000000 0.000000 [Lab] -> Lut -> 1.000000 0.856197 0.457543 1.000000 [CMYK] (clip) Graeme Gill.