The fact of whether or not a chemical reaction takes place in a certain context is a
chemical property of a substance. Any fact about a substance
on its own, not interacting with other molecules (or at least interacting with them a way that causes no chemical reactions to take place, e.g. being cooled by convection of Neon gas, or scratching in a Moh’s hardness test, or piezoelectric effects), is a
physical property.
Sweetness is, as you’ve defined it, a chemical property, so it’s not one of the things I was talking about being “easy” to predict.
But pigmentation is a physical property, as it is a reaction between a molecule and inbound photons†. Likewise, the output of a “pigmentation calculation”—if we know of one—should just be more photons. Wavefunction in, wavefunction out.
We know how photons work. We know how our eyes perceive them. As long as we have a framework for figuring out how a material physically manipulates the photons absorbed by it, we should be able to calculate the pigmentation of every simple molecule by brute force.
——
† The pigmented molecule can be in the chemical form or physical state that gives it that pigment very ephemerally, due to e.g. a clock reaction. But at any given instant in time, we can say “the molecule is in this state, so what function does it apply to inbound photons now.”
Note also that I’m defining pigmentation here, not total appearance, so this ignores any fluorescence, phosphorescence, triboluminescence, Cherenkov radiation, etc. that the molecule also emits and which we perceive as light. All those purely-physical processes are able to be admitted into this model without actually making it untenable, but it becomes harder to understand, since it becomes a Feynman diagram thing, rather than simple transformation over wavefunctions.