[argyllcms] Re: colprof and proofing purposes

  • From: Nikolay Pokhilchenko <nikolay_po@xxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 12 Oct 2009 18:39:34 +0400

Michael Schulz wrote:

> I would like to ask you and other people who have already experiences using
> targen to create test target values for proofing purposes in practice, which
> targen options you normally use.

For RGB printers:
At first, I check the printer linearity. It may be checked by preliminary 
profile after printing several hundred default OFPS-spreaded patches. For 
example:

targen -v -d3 -f384 Preliminary
printtarg...
chartread...
colprof -v Preliminary
xicclu -ir -pl -fif -b -g Preliminary.icc

If the curves on plot a quate straight from they start to they end, the printer 
is OK. If the curves are significantly devate from straight line between start 
and end (they may be like a gamma-curves), the printer is non-linear.
If the printer is linearized well, I use only default OFPS. It gives the best 
results in the case. For example "targen -v -d3 -w -f2304 RGBTarget".
If the printer is rather nonlinear, I generate target with "perceptual space 
body centered cubic grid spread" and with preliminary profile. Example:

targen -v -d3 -f2304 -c"Preliminary.icc" -A0.95 -I -w RGBTarget

For CMYK printers:
The "perceptual space body centered cubic grid spread" is computing on inverted 
profile table. The inverted profile table is depend of black generation 
selected. But black generation may be computed only on OFPS or "device space 
body centered cubic grid spread" data. So it is necessary to print OFPS (the 
best way) or "device space body centered cubic grid spread" target in any case.
When the desired black generation selected and normal profile is built, the 
additional black-generation-depended target may be printed with "perceptual 
space body centered cubic grid spread". The advantage of perceptual space 
spreaded target (based on normal profile) will be the patch placement only by 
selected black generation, not in all device space. Being combined with 
previous *.ti3 data it may significant improve the next profile quality (with 
the same black generation) exact in needed device space areas.

> I'm talking about the default option (OFPS) - with the result of mostly dark
> patches, the -R option which gives patches with a perceptual space
> distribution and the -c option using a previous profile.

OFPS, -r and -i will produce the close results. They generally dark because of 
subtractive CMYK print principle. The device space spread will produce many ink 
combinations and many of combinations really must be dark. It is necessary to 
print all of device space combinations for correct black generation computing.
The one of a problems may be the printer linearity problem. Try to build 
preliminary profile (about 600 patches) and evaluate CMY linearity with zero 
and default black generation:

xicclu -ir -fif -pl -b -g -kz Preliminary.icc
xicclu -ir -fif -pl -b -g -kr Preliminary.icc

Alalyze the graphs and conclude if the printer linear. It may be very 
important. In several non-linear cases the targets with OFPS may be extremely 
dark and even they reading by instrument may became a problem. The instrument 
may don't distinct the margins between patches on such targets.
So the best results will be with linearized printer.

Good Luck!

Other related posts: