Diagnostic Colorimétrique

Shade Diagnostic · Where real colour-difference figures do the talking

Choose the skin tone closest to yours and every shade is converted into CIELAB, then compared with the complete CIEDE2000 formula. The figures are real, and you are welcome to check them yourself.

Step one · Choose your skin tone

About these reference swatches These are reference steps, not measured skin. Screens render colour differently on every device, so we still suggest trying a shade in daylight — the value of this tool is that it turns “how far off” into a figure you can check, not that it replaces a swatch on your own skin.

Ivoire Naturel

Neither noticeably yellow nor noticeably pink; gold and silver jewellery both sit well.

CIELAB: L* 79.1 a* 8.1 b* 13.7
Hue angle h_ab: 59.3° Chroma C* 15.9
Undertone: Neutral Neutral · balanced
ITA° Depth: 64.8° Very light · Chardon 1991

Complexion · Teint View the piece

A complexion shade is meant to match, so these are ranked by ΔE00 colour difference, smallest first — a quantity that can actually be measured.

#03 Beige Nu Neutral · balanced 1.70 ΔE00 Very close ΔE00 1–2 — a difference you are likely to notice only when the two are held directly side by side.

#02 Rose Ivoire Cool · rosy base 2.43 ΔE00

#01 Porcelaine Froide Cool · rosy base 6.33 ΔE00

#04 Sable Chaud Warm · golden base 7.91 ΔE00

#05 Caramel Miel Warm · golden base 15.82 ΔE00

#06 Espresso Couture Neutral · balanced 33.69 ΔE00

Lips · Lèvres View the piece

A lip colour is meant to contrast, not to match. The order below comes from a colour heuristic the maison sets itself (hue direction plus lightness gap); it is not an industry standard. The ΔE00 figures are still real measurements, given for reference only.

#405 Cannelle Nue Coral · warm 27 ΔL* · Hue 33° Hue angle 33° against a suggested 20°; lightness gap ΔL* 27 against a target of 28. The closer both are, the higher this shade ranks. (Reference ΔE00 25.0 — lip and cheek colour is meant to sit apart from your skin, so a larger figure here is not a worse result.)

#518 Rose Givrée Blue-leaning · cool 21 ΔL* · Hue 12°

#520 Rose Mythique True red · neutral 39 ΔL* · Hue 22°

#312 Pêche Couchant Coral · warm 23 ΔL* · Hue 38°

#610 Moka Profond Coral · warm 41 ΔL* · Hue 34°

#808 Cassis Noir Blue-leaning · cool 51 ΔL* · Hue 8°

Cheeks & highlight · Joues View the piece

Cheek colour works to a smaller lightness gap, and is ordered by the same colour rule the maison sets itself rather than by any measurable standard.

#05 Terracotta Feuille Coral · warm 21 ΔL* · Hue 44° Hue angle 44° against a suggested 20°; lightness gap ΔL* 21 against a target of 12. The closer both are, the higher this shade ranks. (Reference ΔE00 19.7 — lip and cheek colour is meant to sit apart from your skin, so a larger figure here is not a worse result.)

#02 Pivoine Aurore Coral · warm -1 ΔL* · Hue 41°

#03 Abricot Ambré Golden · warm 5 ΔL* · Hue 54°

#04 Or Chaud Golden · warm 9 ΔL* · Hue 68°

#01 Perle Champagne Golden · warm -9 ΔL* · Hue 72°

This CIEDE2000 implementation is checked pair by pair against the 34 reference values published by Sharma, Wu & Dalal (2005), and agrees with every one of them to four decimal places.

How this tool does its arithmetic

Each shade’s hex value is first linearised with the piecewise transfer function from IEC 61966-2-1, converted to CIEXYZ against the D65 white point, and then into CIELAB. The difference between two colours is computed with the complete CIEDE2000 formula — all three weighting terms for lightness, chroma and hue, plus the hue rotation term R_T that is the one most often left out. Drop it and the blue region, along with any pair that swaps chroma for hue, comes out wrong.

The implementation is checked line by line against the 34 reference values published by Sharma, Wu & Dalal (2005), and agrees with all of them to four decimal places. That set is the industry test precisely because several of its pairs straddle the 0/360° hue discontinuity — the place where most incorrect implementations give themselves away. tools/verify-ciede2000.php in the theme folder re-runs the check whenever you want.

Depth uses ITA° (Individual Typology Angle, Chardon et al., 1991), the open classification used across dermatology and the cosmetics industry; the band boundaries are taken straight from the literature. Undertone uses the CIELAB hue angle, with thresholds set by this theme and written out in the code comments — that is a design judgement you can tune, not an industry standard, and the interface does not pretend otherwise.

The lip and cheek rankings are not a measurement but a house heuristic (hue direction plus lightness gap), and the interface labels them as exactly that. The ΔE00 figures themselves are still genuinely computed; it is simply that for lip colour, “smaller is better” does not hold.