Show HN: Simple algorithm and color space to generate diverse skin tones
Points and comments are a snapshot, not live.
Constructs a 3D color space and equations to generate diverse digital skin tones.
The project manually labels RGB colors as plausible skin tones, then applies PCA to transform the dataset into a workable shape. The author manually fits spherical equations to the transformed data, defining a "TUV" space that maps to RGB. The result is a color picker and Python procedures for sampling diverse skin tones, with adjustable R² to control plausibility. Limitations include manual bias, simplified single-color representation, and display variability; the work is offered as "good enough" for character creators and digital art.
What commenters are saying
Commenters are broadly positive, praising the methodology and applications for games and digital art. Several note that the picker initially shows unrealistic colors (green, blue, purple) and that reducing R² excludes those. One commenter points out the underrepresentation of pale skin tones, another mentions redhead translucency. A doctoral thesis is cited for parameterizing skin color via melanin and hemoglobin concentration. A few express discomfort with the introspection on race and colorism, calling it distasteful.