[argyllcms] Re: targen options

Gerhard Fuernkranz wrote:
> I've the impression, for subjective judging they are even more critical 
> than e.g. a dE94 weighting would suggest.

My experience has been that to get an acceptable match for the paper
white (for instance) a match of < 1 delta E is needed (CIE94 and
straight delta E are about the same on the neutral axis).

> Unfortunately, a combined linearization of the channels cannot be done 
> with 1D color ramps, but also requires 3D patches near the grayscale. 
> And maybe, also an iterative approach might be necessary, which adds 
> more points at each iteration, taking the errors of the previous 
> iteration into account.

Yes, those were my thoughts as well. Easier on a display, where the
test patches can be shown and measured one by one.

> I'm even wondering whether it may be necessary to do this separately for 
> the B2A tables for colorimetric and perceptual intent (since the 
> perceptual gray axis may differ from the colorimetric one).

In theory I suppose so, although if the device is that badly behaved
that it is noticeably inaccurate a few delta E away from a given neutral
axis, I'm not sure there is much hope for it.

Graeme Gill.

Other related posts: