Mickael Profeta wrote:
Yes you are right, the change coincide with each "repeat" in the process. So I
> imagine this is the wanted behaviour and there is no problem with that. > We understood wrongly that each patch correspond to one color and should not change. The calibration process is attempting to discover the RGB value that most accurately produces the target grey level for that step of the calibrated response curve. This is a numerical inversion problem.
About the profile, I still got bad banding when I try to use the generated profile as a
> screen profile in gqview for exemple. Here is the result obtains with a shaper+matrix > profile applied to the hsv image with cctiff. I also attach tha calibration and the > profile file. Do you see an obvious mistake and the way the correct this. I try to > follow the doc and always get this behaviour on different screens. There doesn't seem anything particularly notable about the calibration or the profile, except that there seems to be a fair level of blue at the black point. This is reflected in the black points color, and the blue channel response curve. Without knowing how you are using the profile (ie. how your RGB test file is being transformed), it's not possible to say much more about it. Graeme Gill.