[argyllcms] Re: Specifying an algorithm type override for display profiles

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 01 Feb 2013 15:07:49 +0100

Am 01.02.2013 09:25, schrieb Richard Hughes:
Hi all.

GNOME Color Manager used to call colprof with -as, i.e. shaper and matrix, but 
Pascal changed that recently to -aS to keep the grey-scale neutral.

If the display is calibrated, then I'd also tend to use -aS or -aG.

Is there ever an instance where the user would want to create a cLUT (or either 
type) or any other type for _display_ calibration? Is there a way of working 
out when a LUT is a good idea automatically?

Well, depends of course on the "goodness" criterion of your choice. If the criterion is 
"accuracy", then I'd run some kind of cross validation. It is expensive though to build N 
profiles of each type from subsets of the whole measurement data (where N is say 10 or maybe even 
more), so this may stress the user's patience...

Best Regards,

Other related posts: