Hello Graeme On 17-Jan-2011, Graeme Gill wrote: > One of the "tricks" with creating multi-dimensional grid interpolations > is to try and create per channel input and output curves so as to > minimize the discontinuities created at grid points due to the linear > nature of the interpolation within the grid points. If the per channel > curves can make the device characteristic look smoother, then artefacts > due to the cLUT interpolation will be less. With this regard, what about calibrating the device to the per-channel curves ? I try to explain better. Frankly I never used the calibration capability of argyll since (for too technical reasons) I have to make the channels calibration for every paper at a very low level stage with proprietary methods, it's something very precise and difficult, however the target I always aimed so far is for linear density. I always wondered, if I chosed as calibration goal having linear per-channel profile curves, could it improve results, or even just being more sensate ? It's no problem for me to iterate my calibration process until obtaining linear per-channel curves rather than a linear per channel density response, if that would lead to better results for the profiler. /&