Pascal de Bruijn wrote: > The following archive should contain everything of interest: > http://files.pcode.nl/temp/argyll/canon_eos400d_testcase.zip Where are the real images (not only the test chart)? How can we comment naturalness without natural images ? :) Have You check the profiles gamuts? As I can see, lprof profile have maximum gamut comparing the other two. As far as I know, the best profile will be the profile with minimal gamut. So the most balanced, weighted profile will be profilemaker. You have a rather few data patches for 33 CLUT points profile. Moreover, the patches spread of IT8.7/2 in color space isn't optimal for CLUT profiling. IMHO. Of course, lprof gives minimal error for your data, but it's because of excessive tight CLUT optimization. I suspect high errors for other colors than measured in *.ti3. Another problem - the dark patches (particularly gray scale) are very noisy and they have a little difference each from other, at level of 1-2 of 255.