Hi,
What I should have ask maybe, is: would you recommend reading targets a
few times? In theory it should improve accuracy, right? But is it really
necessary?
Would you recommend removing the outliers from the data set?
Any suggestions for the outlier threshold?
What would you recommend to avoid over fitting?
What would you recommend as a more real life test for a profile? I would
assume using patches not used before in the profile creation. How to
evaluate all this as objectively as possible?
All this for RGB printer only.
Thanks,
YG