[argyllcms] Re: Printer profile verification?

  • From: Yves Gauvreau <gauvreau-yves@xxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 3 Dec 2018 17:39:23 -0500


My first thought was to use this (colverify) to compare two printing workflow, but from what I see below there are a few more possibilities.

Below in #3 you mention using a different set of test values. I assume we need a "good" set of test values both in quality and in quantity if we want to obtain meaningful results from this kind of test. How do we figure this out?

Is there some documentation on the various algorithm you mention in targen docs?


On 12/1/2018 7:38 AM, Yves Gauvreau wrote:

Sorry for the long delay, my computer crashed.

I have way to many thing to recover so I'll get back some time later on this.



On 11/28/2018 9:58 PM, Graeme Gill wrote:
ygauvreau wrote:


I figured out DisplayCal probably used colverify to generate the report. Any suggestion on
how to proceed for prints? I assume I'll need some meaningful target(s) to begin with, how
do I generate that? Black and white prints either RGB or greyscale anything special for
these? I'd like to compare 2 or more workflows, any suggestion for that?
it depends on what you are trying to do. You could:

1) Verify the profile colorimetric A2B against the test chart used to make it.
    This is a measure of self fit.
    - profcheck can be used for this.

2) Verify the profile colorimetric A2B against the test chart used to make it
    after printing and measuring it again.
    This is a measure of printing and measuring consistency, as well as fit.

3) Verify the profile colorimetric A2B against a test chart with a different set of
    test values.
    This is a measure of profile interpolation accuracy, as well as printing and
    measuring consistency, as well as fit.

4) Test emulation/proofing of another colorspace. The most tricky part of this
    is dealing with colors that are out of gamut of the printer, as (naturally)
    such colors will be printed inaccurately. This is somewhat complex.
    You have to create a set of test values in the emulated space (targen), and then
    convert them to CIE reference values (fakeread ?). You then need to convert those
    reference values into the printer space, for printing and measurement (xicclu then
    manually copy into .ti1 file ?). Print and measure, using colverify with the -L
    parameter to ignore out of gamut values between the reference values and the measured
    values. You may have to also pay proper attention to the intent - typically absolute
    intent is used for the verifications 1-3 above, but this may not be suitable when you
    are doing emulation/proofing, given that the white point may be different.

    Graeme Gill.

Other related posts: