Dear Graeme, Graeme Gill wrote: > Argyll currently creates a colorimetric B2A table with a > colorimetric intent. This means that any colors outside the gamut > are clipped with minimum delta E - that's what colorimetric means. > (Actually, beyond a certain distance they are clipped with a minimum > CIECAM02 Delta Jab, although this transition is what caused the reversal > problem mentioned above). The gamut surface of your device is quite > flat in Jab space near black, and so the clipped Jab 0,0,0 value doesn't > quite correspond to RGB 0,0,0. > > When you simply link (say) an idealised RGB source to your profile, > the source black falls outside the destination gamut, so it gets clipped > by the cLUT approximation of the colorimetric (that is minimum delta E) > clipping. This may not be quite what you want in the shadows, since > delta hue & saturation is given equal weight to delta luminance. ie., > you'd really like a perceptual clipping, but you're using the colorimetric > table. [ie. I would say that you were getting good results previously > using the colorimetric B2A table, more by accident than design.] Thanks for the explanation. > So what I suggest is: Don't use colorimetric if you want a perceptual > mapping between your source and destination. Use a gamut mapping that > deals properly with mapping white and black. You can do this either > by using "colprof -s source.icc", colprof -S source.icc", > or by creating a device link using "collink -G". You can control what > intent the perceptual table or device link is created with using the > appropriate intent flags. This is not really clear to me: what is the source.icc profile supposed to be? If I understood well, I should use a monitor profile. So, I should create a first ICC profile and then use it as source.icc... Is that correct? Best, Alberto Invita i tuoi amici e Tiscali ti premia! Il consiglio di un amico vale più di uno spot in TV. Per ogni nuovo abbonato 30 € di premio per te e per lui! Un amico al mese e parli e navighi sempre gratis: http://freelosophy.tiscali.it/