[argyllcms] Re: DeviceLink Profile refining

On 2/12/07, Graeme Gill <graeme@xxxxxxxxxxxxx> wrote:

The refine program is a bit experimental, in that it hasn't been
widely used or evaluated. In spite of my efforts to tame it,
it also doesn't respond very well when faced with out of gamut
target colors. I found in simulations that it tended to bounce
from good to bad each iteration for those colors. Any "noisiness"
of the printer/instrument readings (ie. lack of repeatability)
will make things difficult as well.

Indeed, this is exactly what happens when performing a few iterations.

I'd need to see much more detailed information to comment any further.

I have attached a few measurements.
In the archive ProofingPaper.zip you can find 2 measurement files.
Index 0 means that it was the initial target.

In PremiumGlossyPhoto.zip there are 5 measurement files. As you will
see, with this paper only on step 4 the substrate dE started to

The measurement files contain the following columns:
  - reference CMYK;
  - printed CMYK( device link output );
  - reference Lab;
  - measured Lab;
  - reference-measured dE

Also, at the beginning of each files there are a few statistics about
each measurement. The average substrate dE is computed considering all
the white patches in the target.

Best regards,

Other related posts: