[argyllcms] Re: Black-/White-Level Observer Dependency?

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sat, 25 Sep 2010 13:37:09 +1000

János wrote:
I can use the "-Q 1964_10" command for dispcal but I can't use it for
dispread. (Or did I miss something? / Is the documentation outdated?)

You save the spectral information in dispread ("-s"), and choose the
observer in colprof etc.

Note there is no problem with calibrating using one observer, and
profiling using another, since (at the end of the day), a color
managed application controls the display by making use of the profile,
and the profile characterizes the calibrated device. (ie, calibration
and characterization are largely independent.)

And (if I use spotread with 1964_10 observer after the FOV10 calibration)
can I feed a CMS software with XYZ.10 values?

I'm not sure what you mean by this.

I need primaries and white coordinates for gamut emulation and I think the
software assumes XYZ.2 coordinates. I doesn't sounds too good for me.

Generally it's good to work in a consistent colorspace (ie. you want to
use a common color "language"). That may be an appearance space, if appearance 
is being
taken into account. Mixing and matching different colorspaces should only 
really be
attempted if you understand exactly what you are doing.

[It would be a slightly unusual situation to want to use a different
angle observer for source and destination profiles for instance, in that
it is typically assumed that matching of the same visual material or image
is desired, and that therefore the images will be viewed from the same
angular distance.
This is one of the problems with the idea of calibrating to a 10 degree
observer - it may make the overall color of the screen match some other
white point better, but if the actual images details are being observer
ate 2 degrees, then 2 degree color matching is what's really desired.]

Or does it effect only the spectral data ---> XYZ coordinates calculations
and I am free to use these XYZ.10 coordinates in any softwares where I can't
set observer types?

Maybe. I'd use the 10 degree adjusted observer for this though, since it's
been optimized to reduce the numerical discrepancy to 2 degree XYZ numbers.

I can't understand how XYZ.10 helps when everything else is based on XYZ.2
but I am not an expert, so...

If you don't understand what the implications are, then don't use it.

Graeme Gill.

Other related posts: