I've worked out where the problem has come from. If I do: ./colprof -v -qm -kp 0 0 1 1 1.2 -cmt -dpp /PSO_LWC_Improved-l310 Then I get this: ./icclu -fb -ip -s100 /PSO_LWC_Improved-l310.icc 0 0 00.000000 0.000000 0.000000 [Lab] -> Lut -> 79.330528 68.579741 51.135219 99.999999 [CMYK]
Oh, right, you were looking at the perceptual table ? (Not that it should differ from the colorimetric table if you didn't trigger gamut mapping.) icclu -ff -ip -s100 PSO_LWC_Improved_eci.icm 79.330528 71.531409 55.250718 99.999999 [CMYK] -> Lut -> 13.933710 0.444178 -0.118185 [Lab] So this (in theory) should look neutral (although see below, it won't match K only).
But I went back through my shell history and found that I'd specified a source profile in colprof:./colprof -v -qm -kp 0 0 1 1 1.2 -S sRGB.icm -cmt -dpp /PSO_LWC_Improved-l310-SsRGBWhich then gives me the odd yellow/cyan result: ./icclu -fb -ip -s100 /PSO_LWC_Improved-l310-S-sRGB.icc 0 0 00.000000 0.000000 0.000000 [Lab] -> Lut -> 56.619898 66.694278 71.962637 99.999999 [CMYK]
This is likely to be quite a different transform, due to the gamut mapping from sRGB.
Why does this happen?
Because a gamut mapping from sRGB is being applied, and the gamut mapping attempts to map the black point of the sRGB to the black point of the printer. Now, one of the subtleties is that the black point of a printer as far as Argyll is concerned, is the darkest color with the same hue as black only. For this printer, we have : icclu -ff -ir -s100 PSO_LWC_Improved_eci.icm 0.000000 0.000000 0.000000 100.000000 [CMYK] -> Lut -> 22.551524 1.132957 2.889973 [Lab] Which compares with the gamut mapped sRGB black: icclu -fb -ip -s100 PSO_LWC_Improved_eci.icm 0.000000 0.000000 0.000000 [Lab] -> Lut -> 56.737462 66.776883 71.959254 99.999999 [CMYK] icclu -ff -ir -s100 PSO_LWC_Improved_eci.icm 56.737462 66.776883 71.959254 99.999999 [CMYK] -> Lut -> 14.666715 1.253770 3.232028 [Lab] Which compares very well for the K only black point extrapolated from L 22 to L 14 (I compute an a* b* aim of about 1.24, 3.18). So, working as designed, but feel free to convince me that this isn't the right way to choose the black point for a CMYK device ! Graeme Gill.