nino loss wrote: > Just changing colprof to -un and -r0 got the values down to peak 0.6 > and 0.28! Also I set -al as C1 is expecting LUT with LAB PCS. Yes, but this doesn't prove much with such a small number of test patches. By setting -r0 the interpolation between the test patches has low stiffness, and therefore may model the underlying device behaviour in a _worse_ fashion than what you had before without -r0. Remember that the real measure of accuracy is to the actual underlying device behaviour, not the self fit to the test patches. Graeme Gill.