[argyllcms] Re: Profile input white not mapping to output white
- From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Mon, 26 Nov 2012 11:15:09 -0700
Iliah,
Perhaps an iterative or integrated (in the Calculus sense) approach could
permit doing color profiling on the raw channel data before demosaicing?
Granted, it would require more computing horsepower, but modern hardware has
already caught up to the point that we could tolerate another step back in
processing speed in exchange for better quality.
Alternatively...RPP (as I understand it) currently does white balancing (before
demosaicing) by simple linear multiplication of the four channels, if I'm not
mistraken. It seems to me that applying a shaper / matrix to the channels
instead of (or in addition to) doing the linear multiplication would
conceptually accomplish the same end result but with potentially better
fidelity.
Of course, determining what non-linear adjustments should be made might require
a profiling mechanism other than the familiar ICC workflow....
b&
Other related posts:
