On 2012-11-23, at 1:12 PM, Iliah Borg wrote: > It has to do with both, and more with raw processing and the way you shoot > than with anything else. But you are not listening. Well, what's got me confused is two-fold. First, the TIFF that I'm feeding to Argyll has no weird color shifts and looks like a typical linear raw camera file -- confirmed by the samples in the .ti3 and by tagging the linear output with something like sRGB. Second, the profile works very well for the chart itself, including the lightest and darkest patches. Running profcheck gives this: [9.578221] I13: 0.005281 0.004926 0.004770 -> 3.851130 7.556675 -4.587199 should be 0.077060 0.022064 -0.034357 [3.318288] N18: 0.530850 0.539870 0.522690 -> 96.132563 0.082416 0.822188 should be 99.354158 -0.236898 0.093899 So, what's got me utterly confused is how the profile can go from R=135 G=138 B=133 => L=96 a=0 b=1 (and a very well-behaved response below that) to R=G=B=255 => L=97 a=108 b=-51. What could possibly account for that *very* hard turn in the profile after L=96, *especially* when everything below that is so close to perfect? It just doesn't make sense to me. What's causing Argyll to extrapolate that, though it left off almost perfectly on the neutral axis with L=96 and almost perfectly balanced channels, and lacking anything beyond that, that that it should suddenly make everything whiter than that not white but cyan? It seems to me that, if you were going to manufacture input data that you would expect to result in a straightforward input profile, you'd manufacture pretty much the data that's in that .ti3. So where's the difference between my .ti3 and a non-screwed-up-.ti3? Which patches are leading Argyll astray? Cheers, b&