Yes, and as discussed on Twitter, sRGB is defined for the CIE 1931 2 Degree Standard Observer, not other observers. It is not defined spectrally but as a set of chromaticity coordinates (in the CIE 1931 Chromaticity Diagram). Thus, strictly speaking, sRGB values can only be computed for the CIE 1931 2 Degree Standard Observer. I tried four different observers, using strict integration:
- CIE 1931 2 Degree Standard Observer: [ 153.9026317 , 180.75617631, 255. ]
- CIE 1964 10 Degree Standard Observer: [ 150.32288938, 184.67113624, 255. ]
- CIE 2012 2 Degree Standard Observer: [ 145.14899585, 180.69807338, 255. ]
- CIE 2012 10 Degree Standard Observer: [ 148.44893332, 185.66849549, 255. ]
- Article: (148,177,255)Perhaps think of it as two steps? One of physics and biology, using minimally-flawed spectra, CMFs, and math, to get a plausible chromaticity. And a separate step of communication, using standards of sRGB and image ICC tags (rendering intent), to get browsers to convey that chromaticity to the user as a minimally-misleading (for the use case) color.
Thus the 10 vs 2 deg CMF choice might depend on the physical angular size of the emitting area. And before browsers supported rendering intent, for use cases where users were eyeball comparing the color with screen white, one might calculate using a very non-standard D58 white-point, as the blue-ish D65 would make white chromaticity render as a pink color, with users misled to think the chromaticity was pink.