Colour VS Colorfy

Colour Science is one of the more serious projects I know of, and more or less lets you get as advanced as you want. Used by film professionals among others. https://www.colour-science.org/

How would you define what the perfect color tool is? I would guess like most tools that it depends entirely on the job at hand, and that maybe no one perfect tool can exist. Colour Science might be great at serious color management and perceptual measurements and conversions between standardized color spaces, but not the right tool for a web developer looking for quick & easy way to make an HSV palette generation widget (and not because Colour Science is Python, but because it’s too big and heavy of a hammer).

If you're using a DCI-P3 (wide-gamut HDR) display, then #0000ff (or #aaaaff) in that display's colour space will be emitted as more saturated blue light than what an sRGB display is capable of. I am not sure if you will perceive them as exactly equally bright, brightness is subjective and depends on the reaction of your retina to the light hitting it, which is why #0000ff (blue) appears darker than #00ff00 (green), however the objective energy of the emitted light should be equal between displays of different gamuts, i.e. #ff0000, #00ff00, #0000ff should all result in the same amount of light being emitted in every correctly calibrated display regardless of the RGB primaries (aka gamut) it uses. The whole colour space topic is pretty deep, if you want to learn more about it and how colours are represented and converted between spaces, I encourage you to go and check out https://www.colour-science.org/ and linked resources.

People that work on resolve also work on https://github.com/colour-science/colour which is most REFERENCE code and complete support there is.

I haven’t seen the math for the conversion but the conversion from CCT to xy/uv are given for a particular domain. One of the conversion with the largest domain, i.e. Ohno m, covers domain [1000K, 100000K]: https://github.com/colour-science/colour/blob/develop/colour...

Infinity is very much in extrapolation territory.

I did a little digging and it's this one https://github.com/colour-science/colour/issues/666

Closest I've seen is this https://github.com/davidkrantz/Colorfy

As I already have seen in the comments, someone linked my GitHub repo https://github.com/davidkrantz/Colorfy where I gave this a go. In short, it analyzes the album artwork and computes a suitable background color, which was more or less the same as Spotify set in about 80 % of the cases at the time. The program uses k-means clustering to find k distinct colors in the artwork and then computes a colorfulness index as defined by Hasler and Süsstrunk (2003) https://infoscience.epfl.ch/record/33994/files/HaslerS03.pdf for each of the colors. The color with the highest index is then set to be the optimal background color.