Color scientists: is there a color sampler that is nontrivial in the following certain way?

This is of mild interest in reflection spectrometry. Are there available sets of color swatches that each get their reflection spectra from different compounds, as opposed to a mix ratio of a small number of pigments? So, for example, this set would include examples of metamerism. “Color” samples might be the wrong term, as we are talking about a level of detail beyond color as perceived by humans, but I don’t know what the right term would be.

What I mean is, the paint store can mix hundreds or thousands of different colors for you, but they have a machine to do it that only uses perhaps 8 pigments that get metered in different doses into a white base to make all the colors. If you collect 1000 different colors from that paint store, and analye their spectra, you will find that there are 8 different spectra you can multiply by changing factors and add together to reproduce any of the 1000 spectra (I oversimplify a little in saying “add” but you get the drift).

I wish I had samples each of which used a different chemical compound as its colorant. 1000 of these spectra would not be able to be combined out of a smaller number of source spectra.

Perhaps the artist’s colors in oil paints would work, as they are “chromium yellow” or “cobalt blue” because of the substances that color them. But perhaps not - for all I know, artist’s paint makers have long ago done away with troublesome obscure ingredients and reduced these products to a few source colorants mixed in certain ways.

Anyway - are there such color sample collections? Is there a particular name for them?

Interesting question. At first I thought it was a pointless exercise–most products are designed to be seen by a human in the end, and which of countless metamers you see is irrelevant.

But it does actually matter, because metamers are relative to a light source with a fixed spectrum. Two patches of brown that look the same under a CFL may not look the same under an LED, even if the light from the CFL and LED are metamers.

I’d suppose that the spectral qualities of any given paint are either known or easily acquired, and you could create a database of paints, using that to feed a program which tries to match a given spectrum as closely as possible. Or use a programmable light source instead of a sample patch.

But I’m no color scientist; hopefully someone else has a better answer.

I don’t know if nicely-packaged sets of these things exist, but depending on what you want them for, you could probably make them. There exist tunable filters that are close to being monochromatic delta functions, and of course from a large set of delta functions you could reconstruct whatever spectrum you wanted.