[argyllcms] Re: Targen options

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Wed, 29 Aug 2007 15:42:50 +1000

Martin Weberg wrote:

Hi list!

I'm using a DTP20 for profiling inkjets mainly. This is what I use building
targets for 3 pages A4 layout:

targen -d3 -g21 -m5 -f1680

Do you people think adding the -g21 will make b/w(RGB) images print better
than without the -g21 option?

It may do. It can help to locate test points along the grey balance
axis, but this isn't easy to do when generating output profiles.

It often works well with input profiles, in which some of the critical
colors to be reproduced (ie. proofing) happen to be R=G=B greys.

There is enough flexibility to do this more deliberately though, if
you can be bothered. What you would do is to generate a profile
as per usual, and then use xicclu -fif -ir to lookup the device
values for the greys you'd like to improve. This might be L*a*b*
values with L* from 0 to 100 in steps, with a* = b* = 0 for instance.
You can create a text file and batch them into xicclu.
Add these RGB values to your .ti1 file manually (a spreadsheet
program like OpenOffice Calc or Excel is often useful for cutting
and pasting columns together). You can either fill the XYZ columns in
with zero's (they are just used for patch recognition), or use
xicclu -f -ir -px to generate values from the RGB's.
Stick them at the end with suitably SAMPLE_ID's, and adjust
the NUMBER_OF_SETS appropriately. Re-profile using the new
.ti1 that has test patches that are expected to land pretty close
to the output grey you are after.

Will -m5 make any difference (I'm thinking of printing gradients with pure
and secondary colors)? Or are the patches better spent in the -f?

It's hard to be sure about these things without simply testing it out.
If you expect to be doing a lot of "powerpoint" like, fully saturated
graphics using primaries and secondaries, there may be a slight advantage
in using -m. Generally I'd recommend spending the patches on the default
far-point-optimised distribution (-f).

Do you think the number of patches are ok for high quality prints or should
I add one more A4 page full of patches?

The best thing is to test it (if you can be bothered). 1680 patches is a 
number for a high quality RGB profile, and it depends on how well behaved
the device is. If it has regions where the color changes rapidly with small
changes in RGB, then more patches may help (but there are many other
factors too - the resolution of the A2B and B2A tables for instance.)

Graeme Gill.

Other related posts: