This post uses OSA-UCS for its perceptual colour space. This colour space is too expensive to use for many applications, since there is no closed-form equation for converting between OSA-UCS and other colour spaces. Instead, you need to use an iterative root finder algorithm.
The main issue with perceptual colour spaces is that no perfect Euclidean 3D perceptual colour space exists, because the underlying domain is non-Euclidean. So anything that you do use in practice will be an approximation, and you make tradeoffs in performance and in choosing where to hide the distortions created by flattening a non-Euclidean space into a Euclidean coordinate system.
Oklab is a high performance perceptual colour space that does a reasonable job of managing the distortions. It’s good for computer graphics on a GPU. https://bottosson.github.io/posts/oklab/
@raphlinus does a good job of critiquing Oklab https://raphlinus.github.io/color/2021/01/18/oklab-critique.html
It’s interesting how we (mis)understand color. This post has 27 upvotes and 1 comments. The subject is popular but no / just one understands it.
Posts like this pop up every month or quarter. I’m reading them, bookmarking them, and planning to revisit ‘my own theory’ (https://github.com/metamn/color) based on them.
However none of them gives an Aha! moment, as, for example, Continuous Typography (https://maxkoehler.com/posts/continuous-typography/) does for typography.
So far, the search for the holy grail of colors is on.