The Kodak Lossless True Color Image Suite is an ancient set of reference images from days of film used to make jpeg, and everything else, reproduce colors (and other thing) accurately. The linked GitHub repo is a PCA (Principal Component Analysis) decomposition of those images, which is a bunch of math (linear algebra) you'd use to make compressed representations of those pictures aka how you'd save them as a jpeg. (PCA does a lot of other stuff, but that's what it's doing here.)
I imagine that would be found in the associated paper, but I’m not sure if it’s been published yet. I’m having trouble finding it.
Parent Paper: Baetzel, J. (2026). Statistical Characterization of Inter-Channel Redundancy Structure in the Kodak Lossless True Color Image Suite. Per-Image Principal Component Decomposition of PCD0992.