sRGB and aRGB - Printable and non-Printable Colors

[Doug Gray]

It would be nice to prune the non-colors from the Lab image. The only way I can think of to do that is to calculate the xy values for each lab value (lab2xy if you're using optprop), then somehow figuring out if each xy value is inside the horseshoe. What am I missing here?

Jim

I pruned the Lab values outside the xy horseshoe.

Percent of visible icclab that is included in sRGB: 19.8
Percent of visible icclab that is included in aRGB: 28.7
Percent of visible icclab that is included in ProPhoto RGB: 58.5

Other values are unchanged since they are all in the xy human gamut.

Note that the MacAdam limits are tighter yet. They are the gamut limits imposed by reflecting surfaces.

[Jim Kasson]

I pruned the Lab values outside the xy horseshoe.

Percent of visible icclab that is included in sRGB: 19.8
Percent of visible icclab that is included in aRGB: 28.7
Percent of visible icclab that is included in ProPhoto RGB: 58.5

Other values are unchanged since they are all in the xy human gamut.

Note that the MacAdam limits are tighter yet. They are the gamut limits imposed by reflecting surfaces.

Thanks for doing this. Useful numbers. The MacAdam limits are for non-florescent surfaces, are they not? And they don't take into account lighting variations throughout the scene. I once did a test of representational efficiency, and I dealt with the last point by simulating the lighting of the gamut of surface colors with three different light levels.

Kasson, J.M., and Plouffe, W.E., “An Analysis of Selected Computer Interchange Color Spaces”, ACM Transactions on Graphics, vol. 11, no. 4, October, 1992, pp. 373-405.

https://www.researchgate.net/publication/220183733_An_Analysis_of_Selected_Computer_Interchange_Color_Spaces

Jim

[Doug Gray]

Thanks for doing this. Useful numbers. The MacAdam limits are for non-florescent surfaces, are they not? And they don't take into account lighting variations throughout the scene. I once did a test of representational efficiency, and I dealt with the last point by simulating the lighting of the gamut of surface colors with three different light levels.

Kasson, J.M., and Plouffe, W.E., “An Analysis of Selected Computer Interchange Color Spaces”, ACM Transactions on Graphics, vol. 11, no. 4, October, 1992, pp. 373-405.

https://www.researchgate.net/publication/220183733_An_Analysis_of_Selected_Computer_Interchange_Color_Spaces

Jim

Yes, fluorescence has the potential to increase gamut of reflecting surfaces beyond Macadam boundaries. It's a form of mixed reflective and emissive surfaces.

As for the illuminant, ICCLAB specifies D50 and clips a and b but L*a*b* itself can be used with any illuminant and the MacAdam limits would also change for different illuminants.

[Doug Gray]

The prior color patches posted were all for difficult colors with Baryta semi-gloss on the Canon 9500 II. These are the results for the Canon Matte Photopaper. Excluded were colors with L < 23 since that's the blackest the paper can achieve.

Canon matte has a much smaller gamut. To the point that there were no printable colors that were more than 5 dE2k outside of the Adobe RGB (1998) gamut. There were, however, a rather large number of colors that were within both sRGB and aRGB that, when printed, would be more than 5 dE2k off.

The takeaway is that soft proofing is important because so many colors are not printable in both sRGB and aRGB and this is more critical for typical matte papers with small gamuts.


These percentages are based on a dE2k of 2 or more.

Percent sRGB colors not printable on the 9500 II:  46.2
Percent aRGB colors not printable on the 9500 II:  55.7
Percent of extra colors printable on the 9500 II relative to sRGB:   6.2
Percent of extra colors printable on the 9500 II relative to aRGB:   0.1


These attached image is from colors selected by collecting all patches that print more than 5 dE2k off and were at least 10 dE2k different from each other: