Graeme Gill wrote: > The problem is that there are real world colorants with color values > outside the ICC L*a*b* PCS range, Hi Graeme, thanks for your analysis. So I need to send my monitor back, because it is out of gamut ;-) [no, of course I certainly won't do that - but indeed, subjectively I was not prepared that any of my monitor's colorants might be beyond +/-128] > so your source profile is in XYZ PCS and in trying to create colorant > tables in the link in L*a*b* PCS, the XYZ can't be encoded. In this > case it's the green primary that is out of encoding range. I was still wondering why there is any need to use CIELAB and not XYZ (since both, src and dst profile have XYZ PCS) but a brief look into the ICC spec reveals that for a device link only CIELAB numbers are permitted in the colorant tables :-( > Of course the implication is that you can't create an accurate LUT > based L*a*b* PCS profile for the XL20 either. You're right, obviously that's yet another implication. Unfortunately the device is not perfectly additive either (it's still an LCD display), limiting the accuracy of a matrix/trc model as well. Best Regards, Gerhard