[argyllcms] Re: Striving for excessive accuracy on RGB printers like Epson 9900
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Thu, 04 Jul 2013 19:17:38 +1000
Mike Rozier wrote:
> Did you had good results using calibration charts, or find just jumping
> into creating targets was better or the same? I understand the
> complexities and unknowns of the RGB->CMYKLcLmLkLlkOG transformation, and
> that RGB 0/255/255 doesn't translate into just cyan ink being used.
Hi,
This is very device dependent. Given the built in, Epson tuned separation
within the Epson RGB driver, calibration may not give much benefit. The
thing to look at is how progressive R, G & B ramps are. Calibration
is typically or more benefit when dealing with a raw driver where excessive
dot gain is present and/or where day to day density variations need to be
compensated for, without re-profiling.
> Wondering if it makes sense to go through a few iterations of chaining
> pre-conditioning profiles, and if that would continue "honing in" on the
> best areas to explore. Or if it doesn't quite work like that.
Maybe. If you have critical colors to reproduce, you may be better off
adding them to the test set. The way to do that is to choose the colors
(which may be in a proofing device colorspace) look them up in terms
of CIE values, then inverse lookup those CIE values in your current
printer profile (xicclu -fif etc.). Plug the resulting device values into
the .ti1 file and re-measure. Or even create a test set of just those
colors and merge them with your existing .ti3 and re-create the profile.
[I toyed with the idea of creating a tool to do this from a target
image - clustering the colors to create a test set tilted towards that
particular image.]
Graeme Gill.
Other related posts:
