Dear Graeme, Thank you for your response. >> 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. > > Hmm. Note that calibration in the above context is a device space table (ie. > it is to linearise the device values before mapping them to CIE using > a matrix or cLUT). Working with film, the calibration can be done against density (or L*, or Y); because the values can be measured directly from the film and compared to the target. I usually have a characteristic curve that I apply "manually" to the scan results before profiling. That takes a lot of stress from working with colour film, and solves b/w film completely - no profile is needed for b/w film anymore, refine works the trick in a very decent manner. I understand that I'm in a small niche here, most of the users do not have access to film recorders, and pre-exposed control strips are not readily available these days. With digital there is no simple way to get density or L* values, and linearization in RGGB domain does not work too well. This seems to call for calibration after converting for CIE (by applying a matrix). However G1 channel is always closely enough related to Y, and linearization for just G1 (or, at least, G) takes care of light un-evenness, flare, and such. -- Iliah Borg ib@xxxxxxxxxxx