The problem, measured
The same dress. Six shoppers. 69 points of difference nobody is pricing in.
98
medium
warm
79
fair
warm
73
light
neutral
59
fair
neutral
51
rich
neutral
29
rich
cool
Every product page shows one model, so every shopper who does not look like that model is guessing. ToneGrid measures a shopper’s undertone with the YouCam Skin Tone Analysis API, scores the whole catalogue against it, and shows the merchant which customers their colour range quietly fails.
How the score is built
Undertone harmony
Garment hue is weighted on a warm–cool cosine and compared to the skin’s own b*-to-a* bias. Near-grey garments are damped, because they carry almost no hue to clash with.
Contrast fit
Lightness separation against an ideal that shifts with complexion depth, so the garment edge stays readable without tipping into harsh.
Chroma balance
Garment saturation measured as a ratio against the skin’s own chroma, plus a ΔE2000 floor so a shade that disappears into skin cannot score well.