Roger Breton wrote:
Second observation, I was not expecting that a DOS shell window would open up with Argyll's prompts, to which I have to respond. This is fine. But I have to confess I was expecting to be shielded from this "complexity": not that the system is unworkable as it is, it is remarkable, but I did not think this is what would happen.
Hiding the underlying utilities and the shell window is on my list, but I have yet to find a way to interact with the commandline tools like dispcal programmatically (and cross-platform).
I'll share my settings later but the resulting LUT profile wasn't great. Maybe because I only ask for 127 samples to generate a LUT-base monitor profile? In the skintones of some close-up portraits, in Photoshop, using this first profile, there was some visible, less than flattering color shifts and slight posterizations. Second, using the EyeOnePro, I tried calibrating and create a profile under the exact settings. Not better. Third, I asked for 512 samples. This time, the color was very nice. Very nice indeed.
Yes, it generally benefits from a larger sample count. Reason for the smaller default testchart is that it takes less time, for people who want a quicker profile (I should maybe add to the ReadMe that more samples are recommended for higher quality, especially when building LUT profiles).
Regards, Florian Höch