This is focused on the illuminant, treating the surfaces as having a spectral response that’s simply a per-frequency multiplicative factor. But to me the more interesting part of spectral rendering is being able to deal with surfaces that are non-trivial functions of the incident illumination, like phosphorescence/day-glo effects, where energy incident at one wavelength is re-emitted at a different one. It’s quite hard to capture this accurately in anything based on RGB.
I wonder, for monochromatic cases like the sodium vapor lamp and the cyan light, if you could get approximate good results by roasting the primaries, doing RGB, then rotating back
So if the problem with sodium vapor is that RGB thinks it's red plus green, rotate it until it's pure red, rotate the materials too, light everything, then rotate back.
Maybe that is mathematically equivalent to existing real time color correction.
