Hello Roger On 5-Jan-2011, Roger Breton wrote: > Elena, > > I admire your persistence. But I believe you should start with a better > quality paper because the color will behave more predictably than on the > cheap paper. Then, when you are convinced that profiling works on the better > quality paper, you can try on the cheaper paper. But if I was you, I would > take the easier path first. I know mine are just "academic" obsessions, don't worry :-) But I'm made this way. I'm also a theoretical, a researcher, and so on... if the used approach/algorithm fails just because the paper is bad, I suspect there's something basically wrong, so I'm concerned that the very same wrong principle will someway affect good papers also, even to lesser extents. I would be very disappointed seeing that a resulting profile made on quality glossy paper, for example, computed from 4000-5000 patches, shows a black ramping down at the end, even slightly, when I STATED that it must go from 0 to 100% (if the case), with only slight deviations from a linear state (I accept that because it would make a sense) but I expect that the ending points are there, and the C,M,Y channels are computed as a function of the K curve I choose to reflect a neutral grey, not that at some point K is made bending down unnecessarily. In addition, the theoretical point of view: what, if I would really like to have also a good profile for plain paper ? One could be wanting it, of course. Think of newspaper paper, for example. Is there around a more ugly and bad paper than that ? But nowadays, web news printers too are very concerned of color quality. I don't think they would accept such a behaviour from a profile for their press. And they perhaps also use a very low total coverage, of 200 maybe. I forced even 300 for plain paper, with some printing trick. I wonder what would happen here if I decided to try such a low ink limit... maybe that would even be a solution, paradoxally !? I just have to try... > Just beware of paper filled with optical > brighteners, if you can avoid them. I think that the -f option + usage of spectral data should be there for that, if even not making miracles. I don't think my problem is related to those, however. /&