Reading your results
What the numbers mean.
| Term | Meaning |
|---|---|
| L* | lightness, 0 (black) to 100 (white); the depth of a color |
| ITA° | skin depth band computed from L* and b*: how light/deep your skin is; says NOTHING about warm vs cool |
| hue angle h° | the undertone axis: skin lives ~25–80°; lower = pinker/cooler, higher = more golden/warmer |
| chroma C* | distance from gray: how saturated a color is; skin sits ~9–30 |
| ΔE2000 | perceptual color difference; <1 invisible, <3 'instrument grade', >6 obvious |
| the 4 axes | temperature (warm↔cool), value (light↔deep), chroma (bright↔muted), contrast (high↔low). Every season is a point in this space, and so are you |
| entropy (bits) | how uncertain the tournament still is; 3.58 bits = all 12 seasons equally likely, 0 = certain. Each vote burns some down |
| posterior | the live probability for each season given everything so far |
| confidence | capped at 95% on purpose. We expect photo-based analysis to land around 75–85% agreement with an in-person analyst (our own estimate — no published figure exists); certainty would be a lie |
Sister seasons share a border in axis space (True Winter ↔ Bright Winter). Nearly all real confusion is between sisters. That's why the tournament serves sister head-to-heads when the race is close, using the colors that DIFFER between them.
The agreement check (home page, once you have two measured sessions) is the actual verdict: same season from photos shot on different days, with skin retest ΔE under 3, is what makes the diagnosis trustworthy. One session is a hypothesis.
Your document
Everything lands in one place: the PDF dossier from the Verdict page. Your season and palette, your face broken down (shape, thirds, fifths, canons, profile angles if you shot one), the skin-surface read, and a full appendix showing how the color was computed, every formula and correction with your actual numbers. That document is the deliverable; the bench is just how you get there.