[argyllcms] Re: Calibration issues using argyll

Steffen wrote:
What concerns me is the following: when I calibrate my display with argyll, neutral gradients seem to have a slight greenish color cast in the dark tones compared to the results produced by iColor. This is consistent through a few months of using argyll, the results are always the same, so it has to be systematic. 

I think I've seen this sort of effect, and it was on an Eizo display.
I haven't seen it on other LCD displays. My impression though, was that
it was a viewing angle effect. Viewed directly in front of the
display the effect wasn't obvious, but viewed at an angle above
the display it became evident.

The only theory I can come up with, is that this type of display
really isn't capable of maintaining the target neutral axis hue
down to so close to the black point, due to viewing direction color
inconsistency near black.

Typically it's not a good idea to try and set the black point on
an LCD to the white point target hue, since LCD's have such poor
black levels, and the hue can only be changed by adding brightness to it.
This is why the -k parameter is defaulted to 0 for LCD displays.
With a -k of 0, the target white point hue is maintained
down most of the neutral axis, and then blended over
quite near the black point. Perhaps if the blend was a bit
slower, this would set a target hue near black that is
closer to the native black, and the viewing angle effects
would be less objectionable.
The next thing is the profile created. Argyll measures quite different color rendering abilities than iColor does. With argyll, the gamut seems to shift quite a bit towards green and blue, covering less of red to yellow tones. These results also are very consistent. For the creation of the profiles, Argyll scanned in alot more test patches (2000) than iColor did (roughly 80-100?). These were also the profiles used in the comparision I link to at the end of this message. 

It is quite possible that iColor (if it came with the Eizo) is using
a display specific instrument calibration matrix, while Argyll uses
the default instrument LCD calibration matrix. This would explain
some differences in the profile and apparent gamut.
While I certainly would put more confidence in Argyll regarding the profile, the calibration seems to be a bit of a problem. Now, I have tried several approaches when calibrating, and the slight colour cast also remains if I do not calibrate at all but just use targen, dispread and colprof to create the ICC. 

If you tell me which platform you are running on, I can make available
a version of dispcal with an extra option that would let you alter
the blend rate used to the native black point target, to see
if this improves the appearance of the calibration.

Graeme Gill.

Other related posts: