Argue with the palette

[Aejkatappaja]

Every hue in cendre came out of a measurement: Planck's law for the coals, published emission wavelengths for everything else, pushed through the CIE 1931 colour matching functions into OKLCH. Lightness and chroma were chosen, because a spectral line sits far outside sRGB and has to be brought into gamut anyway.

So there are two kinds of disagreement, and only one of them is interesting.

Not interesting: the hue. brass is 61.4° because sodium emits at 589 nm. Moving it is not an opinion about the theme, it is deleting the theme.

Interesting: everything that was chosen rather than measured.

• Lightness. brass at L 0.838 is the brightest thing on screen and frost at 0.600 the dimmest, because cinder, ember and brass sit within 36° of each other and separation had to be bought somewhere. Is the ordering right? Is frost too far down?
• Chroma. Pigments run 0.072 to 0.115, semantics 0.125 to 0.160. Unbounded, the optimiser that placed the semantics produced neon. Is the ceiling too low?
• Role assignment. One role, one pigment. Keywords take cinder, types take frost, every literal takes sap. A declared name is not a role, so variable and constant names stay in fg and only the value takes a pigment. This is the rule people push back on most.
• comment at 3.32:1. Deliberately under AA. Most dark themes people keep for years land near there, and a comment as loud as the code it explains is noise. Say if you disagree, but say it with the ratio you would use instead.
• The three depths. hard, medium, soft. frost and error fall under 4.5:1 on soft, which is stated rather than hidden.

Bring a screenshot and a language. "Too dim" in Go and "too dim" in TypeScript are different claims, because the two hit different rules.

Full derivation, hue by hue: https://cendretheme.com/#derivation