Canon 10D profiles

[taob]

First cut at a set of profiles for the Canon 10D available at http://www.risc.org/Images/10D/canon_10d_dnp.zip and hopefully on NI's main site soon. Please let me know how they work for you.

[NITeam]

After taking a look at the target you used to build these profiles, I am afraid they may be not very accurate. The problem with that target is that it contains a lot of gradients. Gradients are NOT featureless areas, gradient is a feature in itself. You could build good profiles with your target if you manually selected gray areas on top of it (they seem to be real flat).

Hope this helps.

Vlad

[taob]

I was thinking about that, so I tried one example (an ISO 3200) one with both auto fine tune and manual fine tune, and although the percentage sliders were slightly different, the end results did not appear to favour one or the other. Does the auto fine-tune not recognize gradient areas and stay away from those?

[NITeam]

It may be close to accurate profiles but using target with gradients is in principle incorrect as gradient are features. AFT may stay away from those or otherwise but the point is that such gradients shouldn't be given to NI as examples of featureless areas. Standard NI calibration target is pretty good for accurate profiling, including doing ATF.

[taob]

I was experimenting with colour patches in addition to the standard greyscale one on your site to see if the noise characteristics differ with chrominance instead of just luminance. The target with the gradients was the only one I had handy at the time, but I can easily turn that into solid colour patches instead (it was originally intended as a print calibration target). Anyway, this will give me an excuse to release a version 1.1 or 2.0 of the profiles, including raw file conversions.