Color management myths and misinformation video

[alain]

Erik,

I'm not an expert either, and perhaps Andrew can comment. However, I looked up the Rec 2020 gamut on Wikipedia (shown below, first image). It does not appear to be much wider than Adobe RGB (second image below, which is taken from a paper by Prof Dr Hoffmann). What he describes as real world surface colors are those colors that occur in nature from non-emissve (reflected light) sources. These are outlined by the dashed line. Some of the greens in his diagram appear to be out of the 2020 gamut. Some knowledgeable sources have said that this real world surface color gamut is too restrictive. Personally, I don't see any need for a new color space. Properly used, ProPhotoRGB works fine.

Bill

Hi the Rec 2020 gamut is quite a lot  larger than adobe and contains most of Pointers gamut (which seems to be one off the sets of real live colors).  A nice artikel, where the diagrams are rather clear http://www.tftcentral.co.uk/articles/pointers_gamut.htm and very instructive about sRGB and adobeRGB.

[Robert Ardill]

OK, here is an example of converting Andrew's photo from ProPhoto to sRGB using both a Perceptual mapping and a Relative Colorimetric mapping.  If you don't believe that it is possible to do a Perceptual mapping from Workspace to Workspace then look here: http://www.luminous-landscape.com/forum/index.php?topic=93997.0, where I explain two methods of doing so (the second method is certainly the one I would prefer at this stage as the first method relies on a beta v4 profile from the ICC).



[Right-click on the image to see it full-size]

The left-most image is Andrew's original ProPhoto image: so what you see will depend on your display, but even with an AdobeRGB-size display the colors are being mapped with a Relative Colorimetric intent, so some of the colors will be clipped and shifted.

The middle image is converted from the original ProPhoto image to sRGB using a Perceptual mapping and the rightmost image is converted to sRGB using the normal Relative Colorimetric mapping.

I have indicated areas that I think are better in one image compared to the other.  In general I would say that for this particular mapping from an extremely wide image gamut to sRGB that the Perceptual mapping works best, but there are areas where the Relative mapping seems better to me.  At any rate, I am not advocating one mapping over another: but I am suggesting that having the ability to choose the rendering intent, especially when converting to sRGB (for the web) is a really big advantage.

Of course, needless to say (if you have read my previous posts here), I think that converting this image to sRGB is doing it a serious injustice as the image gamut is way outside any output device or display profile that is currently available.

Robert