>> Yellow cast might be the result of wrong setting of the clipping in the >> converter as Nikolay suggested. > > I don't think this is the problem. I checked saturation with dcraw and > the value set in the dcraw code is lower than the one of my camera. Close to the saturation the sensor can be non-linear. That is especially the case when the cameramaker is after maximum well capacity. Camera needs to be linearized before profiling, especially when using matrix-type profiles with single gamma. A question to Graeme if I may: Is there a way to incorporate linearization curves into input profiles? colprof with a refine result as "-p" seems to produce good results only for output profiles. Inverting the sequence of ti3 files for refine does not help. The reason I ask is not only digital cameras, but scanning negatives as well. For negatives I obtain an L* -> L* table by measuring the densities on the film shot of the Kodak N2 target, and the reference L* values of the target are also known (and measured). The profile composed by refine is correct if used as a stand-alone. -- Iliah Borg ib@xxxxxxxxxxx