> While I'm still learning color management, I have some doubts about
> this. AFAIK two color spaces are commonly supported in digital
> camreas: sRGB and AdobeRGB. If I set a camera to XY color space, I
> would not expect to get colors which are outside of that color space.
> Maybe with some very wild image manipulations one could manage to hit
> the limits of AdobeRGB, but I doubt that such an image would print
> well. So does this article have any valid points?
>
        Careful with that learning.... it's a bottomless pit as far as I'm 
concerned.  I went down that road a year or so back and shored up some of 
my ignorance, but there's still plenty left.

        There are a few issues:
- XYZ colorspace does not represent all colors... just all the colors that 
people can see.  Camera sensors can capture more (think infrared for 
starters).
- Many trichromatic (e.g. sRGB, AdobeRGB) colorspaces do not adequately 
cover the visible spectrum.  Ones with wider (e.g ProPhotoRGB) use colors 
that are imaginary but can also represent more of the real colors as well:
http://en.wikipedia.org/wiki/ProPhoto_RGB_color_space
- Most printing techniques are grossly inadequate to represent much of the 
visible spectrum.

        IMO, if you want to keep full color spectrum, you'll likely want a 
very wide space:
http://en.wikipedia.org/wiki/Adobe_Wide_Gamut_RGB_color_space
http://www.adobe.com/products/adobemag/archive/pdfs/98auhtbf.pdf
        ... but to prevent posterization you'll also likely have to keep 
16-bit throught all the processing.

-Cory

-- 

*************************************************************************
* Cory Papenfuss, Ph.D., PPSEL-IA                                       *
* Electrical Engineering                                                *
* Virginia Polytechnic Institute and State University                   *
*************************************************************************


-- 
PDML Pentax-Discuss Mail List
[email protected]
http://pdml.net/mailman/listinfo/pdml_pdml.net

Reply via email to