[argyllcms] Re: xicclu -g predictability issue

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Tue, 25 Jan 2011 10:15:30 +1100

Elena [service address] wrote:
To summarize: for each main diagonal in the Lab space (or whathever the 3D PCS 
is)
going to absolute black [even those going to edge centers if you compute it
with pos/neg coordinates, so to include red->black, cyan->black, white->black, 
etc.]
you compute a K separation curve, the way xicclu -fif would do if it had already
the suggested -p option available, using the -k or -K parameters given.
All points laying on the main diagonals would then get their proper separation,
optimally chosen and checked by the user with xicclu, if the case.

Sure. I agreed that adjusting K in the whole gamut to meet smoothly
with the K at the gamut surface would be a next step, in an email
some time back. That would help many cases, but doesn't address
the issue of K on the gamut surface being "bumpy" itself.

Those laying between diagonals are constrained to a K curve resulting from
interpolating the K curves of the diagonals around [allowing any colorimetrical
mismatch]. Since the pre-computed diagonals are just those going to absolute 
black,

I wouldn't use a "slices" approach, but would just treat it as an interpolation
volume, since that is a well supported computational paradigm within the code.

Thinking better of what you wrote... a similarly aimed solution could be
generating a device link profile between the desired rgb space and the
CMYK device with -G option and a table resolution of 256. Unless the
source image has >24 bit color resolution no interpolation should happen.
The only matter: I don't think the ICC standard supports 256^3 LUTs...!
I noticed that collink itself limits -r to 100 so I assume that is the standard
imposed.

I don't think that the standard imposes any such limit, it's a matter of
practicality. I doubt that you could invert that many grid points using
the current code, in a reasonable amount of time. If "ultra" with
a res. of 45 takes up to several hours, then a full resolution table
would take about two weeks...

Graeme Gill.

Other related posts: