[argyllcms] Re: Colprof question

Fabrizio Levati wrote:
I'm sorry but I was in a hurry and didn't post the correct images nor explained what I am trying to understand. 

This should be clearer:

image1 -> xicclu -g -fb -ir -kp 0 .1 1 .95 0 -l330 -L95 FOGRA39.icm


Sorry, but this command doesn't make sense. If you're using -fb, then
you're using the B2A table of the profile (PCS->CMYK), and the black
generation is already set in the table, so the -k parameters
are being ignored. It's only if you are using -fif that the
inking parameters (total and black ink limits, black generation) will
be used in the plot, since -fif looks up PCS->CMYK by inverting the A2B table.

So, to explore what grey curve you would get by setting certain
inking parameters during the profile creation, you use something
like:

xicclu -g -fif -ir -kp 0 .1 1 .95 1 -l330 -L95 FOGRA39.icm

Having re-built the profile from the .ti3 data with those
inking parameters, you would verify the resulting neutral
axis behaviour:

xicclu -g -fif -ir  FOGRA39_new.icm


image2 -> xicclu -g -fb -ir ISOcoated_v2_eci.icc
I'm trying to understand how to make my Black start from .1 instead of 0 and getting the wanted shape ending at 1 with a level of .95 I know that I'm doing something wrong but I want to learn how to take advantage of colprof in making the profiles I need. Sadly I am not able to help myself just using the documentation... 

Unfortunately the black generation controls are rather hard
to use, and it is quite difficult to get a smooth looking result
using them. This is because the controls let you set a certain
shape curve as the K target, but the actual K value created in
the B2A table will be slightly different due to the limitations
of the gamut and maximum ink, and these affects are not smoothed
out. Creating a target curve shape with numbers is a non-intuitive
exercise in the first place. [This isn't easily fixed - a different
approach is needed to address these problems.]

One thing to keep in mind though, is that the plot created using
xicclu -fif will look rougher in detail than the actual resulting
B2A table response, because the B2A table has a limited cLUT resolution,
and this tends to filter and smooth the curves.

The parameters you have above, look like the right place to start
from, except that you have also set the shape parameter to a very strange
value - start with 1 rather than an extreme convex shape of 0,
but you need to use -fif to see the effects of them.

Graeme Gill.

Other related posts: