You might use to argyll to do the gamut mapping of a target
from source space and print the resulting file with ACPU or whatever it is
called, read it back in and use some utility to compare desired with mapped
values. There are a bunch of commercial programs out there, with graphical
interfaces, that will compute deltas. Whatvtarget and wjat source space is
left as an exercise for the reader.
On Feb 20, 2017 12:50, "Graeme Gill" <graeme@xxxxxxxxxxxxx> wrote:
Henrik Olsen wrote:
What I would like now is a procedure for printing a chart for a measuredcontrol test
from physical print using the profile (no simulation) - confirming howclose or not I
am to achieving color matching. It is unclear to me how best toaccomplish this. Please
I'm not clear what you want to do. You say "control test" and "no
but then say "color matching", which implies simulation (i.e. proofing or
of a source colorspace.
Background: I have previously trying to determine which target printingmethod was
correct, as I saw obvious visual differences between two recommendedapproaches (Adobe
Color Printer Utility and ColorSync vs Canon Print Studio Pro with colormanagement).
But who was right?
Sorry, that depends very much on your operating system, program and printer
driver detail. It can be quite difficult to figure out, since there are so
may hidden parts, and on Apple Mac. systems, a string of bugs and
over many years have caused issues with trying to do this reliably.
I though Canon might have their own profiles roughly right, so Ivarious methods,
preconditioned the target with their profile, and after printing with
read them all with chartread. Then I did a colverify between the .ti2target and the
.ti3 measurements - and the Canon method was showing many and bigerrors, while the
others were within reason. That’s the closest I’ve been to validatingsomething - but
not validating created profiles as of yet.
It depends what aspect of a profile you want validate. A2B tables ? B2A
In some ways it's kind of pointless, since the A2B is meant to reflect the
behavior of the device, so verifying is just like making another profile.
And if the characterization is accurate, then you can verify a lot about
by checking against the A2B (i.e. invprofcheck).
But in principle, here's an outline of how to do these two things:
To check A2B: Create another printer test chart with a different number
of test patches. Print it (no color management) and measure it.
Either use profcheck, or run the .ti1 through fakeread and the the
profile to create a check .ti3, and then run colverify on the two .ti3
To check the B2A: Create another printer test chart. Convert to
CIE values using fakeread and the printer profile to make a .ti3 file.
(This makes sure that all the test values will be in gamut.)
Convert back from CIE to device values using fakeread -U,
rename the .ti3 file to .ti1 and edit the header to be a .ti1 file,
then print a test chart from the .ti1 file using printtarg with no
color measurement and measure it. Use colverify to compare the first
.ti3 with the values read from the test chart.
I’d like to be able to evaluate two different kinds of patch sets:under /ref, but
1) Fixed set of patches Like ColorChecker Classic or SG. I see files
haven’t figured out how to make them into a .ti1 for use with printtarg,or making my
own. Have seen image files with attempts to match the reference CIEvalues with a given
color workspace. But would like to have a .ti1 and .ti2 generated withthose values.
You would have to make your own using targen.
Question is then, how exactly I should print and compare. I assign theprinttarg with
the new profile, so RGB values get the right CIE meaning (matching thepreconditioned
data using same profile). I now see multiple routes and need feedback. Ican print
without color management.
As soon as you print with color management, then a source profile enters
and you would be verifying a proofing or rendering workflow involving two
ICC device profiles. This is possible, but has it's complexities, including
being able to pin down what the system or graphics program color management
is exactly doing.
I can print with the profile, but then absolute or relativeworking space
colorimetric and no BPC? Should I after profile assignment convert to
(ProPhoto) and then print with profile, again abs or rel intent? Shouldcolverify have
any options about whitepoint matching or anything else of interest?
That's up to you. In order to do useful verification, you need to be
clear on exactly what you are setting the different systems up to do, and
how they are operating.
It comes down to what you are really trying to figure out :- are you
debug your workflow, or are you trying to evaluate the profile creation ?
You can't do both at once - there are too may variables.
It's also perfectly possible to evaluate the profiles without involving
the normal print workflow - do all the color management using ArgyllCMS,
and always print with no color management. This is one way of pinning down
the color transformation details.