Hello again Argyll CMS users, I've made some matrix profiles for my cameras and wonder if the model-fitting algorithm in Argyll CMS is robust. I have an X-Rite Digital ColorChecker SG (CCSG) chart. Here are some different results (where R = red patch E4, G = green patch F4, B = blue patch G4, av24 = average of 24 ColorChecker Classic patches within the CCSG chart, pk24 = peak of 24 ColorChecker Classic patches). All errors are dE2000 reported by colprof. Note the large blue errors: 140 patches: R = 4.1, G = 3.9, B = 8.7, av24 = 2.8, pk24 = 8.7 (blue) 139 patches: R = 3.8, G = 3.1, B = 8.1, av24 = 2.7, pk24 = 8.1 (blue) 24 patches: R = 3.5, G = 2.9, B = 4.1, av24 = 2.7, pk24 = 6.8 (orange-yellow) In the second case, I removed the worst patch L7, dE00 = 14.6, and profiled the remaining 139 patches. In the second case, I profiled only the 24 ColorChecker Classic patches within the CCSG chart. The blue error makes a noticeable difference in real photos with blue skies. It seems to me that the distribution of patches, both numerically and in colour, results in some weighting of the model-fitting algorithm. For example, the CCSG has lots of middle-grey and white patches (which is a good thing). Secondly, patches that fit rather poorly (outliers) may have undue leverage on the results. Perhaps the CCSG chart is designed for CLUT profiles and includes some strange colours and deep blacks that can only be modeled well with a CLUT "correction" profile? A simpler chart might actually be better suited for matrix profiles? For matrix profiles with the CCSG, is it sensible to trim patches that fit poorly, but always retaining RGB and CMYK primary colours of course? I think it's important to get blue colours right. Stephen.