O.K., I've tried the -aS and -aG options to colprof, using the original large measurements file for the original calibration. The result with colprof -aS -qh is good: the darkest discernible textures correspond to RGB numbers around (10,10,10) and above in the file. And greys remain grey--well, there's just a slightest hint toward the warm side, only observable on large areas of the darkest greys surrounded by pitch black, e.g. black bears cavorting on moonless nights, something only the most dedicated pixel-peepers photograph. And the brightest 5-7 RGB levels actually do show very slight discernible differences from white, which the LUT profile did not. Great for noticing blown highlights. Profile testing does show larger errors than a LUT profile, but the difference isn't really visible. On Mon, Nov 23, 2009 at 7:04 PM, Gerhard Fuernkranz <nospam456@xxxxxx> wrote: > Michel Joly de Lotbiničre wrote: >> [...] (matrix profiles were not good on the neutral grayscale quality >> measure) [...] > > Did you use the -as/-ag or the -aS/-aG options? The latter create equal > shapers for the R, G and B channels, avoiding that the profile can ruin > the good neutral gray axis again that was established by the calibration. > > Regards, > Gerhard > > >