> > Actually I'm still a bit sceptic regarding the shaper/matrix/shaper > > optimized curves. In many cases they seem to work very well, but > > I also encountered data sets, where "-ni" or "-ni -no" did result in > > better accuracy. Graeme, btw, the epson_760_glossy.ti3 data set posted by Mike Russel is an example for a profile whose quality seems to be rather spoiled by the shapers, wrt. the self-fit error, and also wrt. the estimated generalization error (cross validation). Shapers + CLUT (-qh): profile check complete, peak err = 5.922979, avg err = 1.845065 GCV: 4.6 RMS CLUT only w/o shapers (-qh -ni -no): profile check complete, peak err = 3.598966, avg err = 1.270233 GCV: 2.5 RMS Regards, Gerhard