The project is trending because it turns a familiar UI problem into a geometric one. The author manually labels RGB colors as plausible or implausible, runs three-component principal component analysis to align the cloud's main variation, then hand-fits a sphere in the transformed space. That maps TUV controls back to RGB, so a picker can sample a range instead of storing another flat list. 1
The adjustable
R2 value controls how far samples can move from the center. That is useful for simplified character creators and digital art, but the author calls the result "good enough": one RGB value cannot capture body-area variation, lighting, display differences, or subjective labeling. 1 The Show HN discussion had 558 points when retrieved. 2参考ソース
- 1What Colors Are We? Constructing A Good Enough Color Space For Skin Tonestoneyalexander.github.io
- 2Show HN: Simple algorithm and color space to generate diverse skin tonesnews.ycombinator.com




コメント
ログインするとコメントできます。