Re: Gamma correction

On 19 Feb., 20:45, "Gernot Hoffmann" <hoffm...@xxxxxxxxxxxx> wrote:
As long as you're not working in a color managed environment,
you can use any linear or nonlinear coding for your gradient
pattern.

So far, the 'correc't grayscale by appearance is nowhere defined.
But the correct grayscale by measurable luminance is defined.

Color management tries to map scene luminance to visual
luminance as good as possible. Essentially, this is a linear
mapping (e.g. for a scanner or a digital camera) .
There are really big difficulties, concerning the compression
of the dynamical range, but this has nothing to do with the
trivial power function by Gamma.

Best regards --Gernot Hoffmann

Thanks for your help, I will maybe spend some money on calibrating my
monitor correctly.

I'm now trying to verify my implementations of various color
conversions by comparing my results to some of your data:
http://www.fho-emden.de/~hoffmann/colcalc03022006.pdf
(In the following passage I'm referring to page 25 "3.3.3
SpectroCalc / Examples / sRGB with CAT / Illumin. A+ D65")

I was able to reproduce the XYZ values (0.9077, 0.7956, 0.2896) for
the A illuminant. However when transforming these values to the sRGB
color space I noticed that I was not doing any chromatic adaptation.

My current approach with my raytracer is:
1) Run a physical simulation, gathering a spectral power distribution
2) Converting the spectrum to XYZ values by integration (emissive
case, no illuminant assumed)
3) Transforming the XYZ values to sRGB values (gamma corrected)

From what I've read I could deduce that chromatic adaptation has to be
performed if the "reference white" does not match the "media white".
But what actually is the "reference white" in the case of my computer
generated image? (The "media white" I guess is D65 because that's the
reference white of sRGB)
Does the XYZ color space always have a D50 reference white?

.