[argyllcms] Re: Profiling between two images

  • From: Gerhard Fuernkranz <nospam456@xxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 01 Feb 2009 22:35:03 +0100

Pascal de Bruijn wrote:
> Well, I'm not betting my money on the fact it might work with different 
> camera's as well. I'm not sure how subject to change the spectral properties 
> of sensors are between generations...

Well, for a device independent solution I'd rather suggest a model which
splits the camera's overall color transformation into two parts: 1) the
colorimetric transformation from the raw color space to the processed
color space [which preserves the colors, but does not alter them] and 2)
an "artistic component" which additionally _alters_ the colors. The
latter part of the transformation could then be treated as the "camera's
style" and could be described by an abstract profile in a device
independent way [i.e. this "style" could then be applied to arbitrary
images in arbitrary color spaces, including of course raw images from
other cameras, granted that you have a colorimetric profile for the raw
color space of the other camera available].

I.e. the overall model for the camera's transformation would then look like:

raw color space -> colorimetric transformation -> PCS ->
-> "camera style" -> PCS -> colorimetric transformation ->
-> output color space (jpeg)


Other related posts: