Digital Gemology Cut Performance Index

Ray-traced light-performance scores (brilliance, fire, scintillation, tilt, symmetry, spread, shape-norm, leakage, overall) for 45 gem cuts, in diamond and in moissanite. Computed by Digital Gemology's light-census ray tracer from each cut's facet geometry. Source and live page: https://digitalgemology.com/learn/cut-performance-index Same data as JSON: https://digitalgemology.com/data/cut-performance-index.json and through the API at https://digitalgemology.com/api/cut-index Version: 2026-10-31 (census-ba47e2e1.13d5809b) License: CC BY 4.0. Use and share freely, with credit to Digital Gemology and a link to https://digitalgemology.com/learn/cut-performance-index. Also on Kaggle: https://www.kaggle.com/datasets/digitalgemology/digital-gemology-cut-performance-index Also on Hugging Face: https://huggingface.co/datasets/DigitalGemology/cut-performance-index Contents cut-performance-index.csv and cut-performance-index.json hold the same 90 rows: 45 cuts × 2 materials. 37 cuts are scored in each material; the others are listed as "not yet scored" with the reason. Figures are unrounded; the live page shows one decimal. Top five overall: Diamond: DG-Miliano (92.7), DG-Rafaello (92.6), DG-Vitruvio (91.7), T105 Round Brilliant (90.6), DG-Broken Hearts and Arrows (90.4) Moissanite: DG-M5 (95.1), DG-Vitruvio (92.9), DG-Miliano (92.0), Portuguese Bird's Nest (91.8), DG-Broken Hearts and Arrows (91.7) Highest overall score in each family: Brilliants — diamond: DG-Vitruvio (91.7); moissanite: DG-Vitruvio (92.9) Jewelry cuts — diamond: Portuguese Bird's Nest (86.4); moissanite: Portuguese Bird's Nest (91.8) Historic cuts — diamond: T57 Round Brilliant (88.6); moissanite: T57 Round Brilliant (89.8) House cuts — diamond: DG-Miliano (92.7); moissanite: DG-M5 (95.1) Cuts covered Brilliants: T73 Round Brilliant, T105 Round Brilliant, DG-Vitruvio, DG-Anteros, DG-Broken Hearts and Arrows Jewelry cuts: Asscher, Baguette, Cushion, Emerald Cut, Emerald Square, Heart, Hexagon, Marquise, Octagon, Oval, Pear, Portuguese Bird's Nest, Princess, Radiant, Radiant Square, Rectangle, Square, Teardrop, Triangle, Trilliant, Trillion Historic cuts: Old Single, Mazarin, Peruzzi, Old Mine, Old European, T57 Round Brilliant House cuts: DG-Bowtie, DG-Cabochon, DG-Elongated Hexagon, DG-M5, DG-Miliano, DG-Octagonal Kite, DG-Rafaello, Elongated Hexagonal Bipyramid Platonic solids: Dodecahedron, Hexahedron, Icosahedron, Octahedron, Tetrahedron Each cut's page on digitalgemology.com sells its 3D model files (STEP, 3DM, STL, OBJ, FBX, GLB), and most also sell a GemCad faceting diagram. The free gemstone simulator on the site shows the cuts in diamond, moissanite, cubic zirconia, ruby, sapphire, emerald and amethyst. Columns All scores are out of 100 except leakage (per cent, lower is better). Spread and shape-norm can be above 100. Scintillation, symmetry and spread depend only on the shape, so they are the same in both materials. ColumnMeaningslugCut identifier, as in the cut's URL.nameCut name.categoryCut family on the site (Brilliants, Jewelry cuts, Historic cuts, House cuts, Platonic solids).materialdiamond (refractive index 2.417) or moissanite (2.65). Never compare or rank across materials.status'scored', or 'not yet scored' (then the scores are empty and reason says why).rankPlace within the material by the overall score shown to one decimal; equal shown figures share a place.overallThe combined score: 0.34 × brilliance + 0.20 × fire + 0.16 × tilt + 0.10 × scintillation + 0.10 × symmetry + 0.10 × (100 − leakage).brillianceThe share of the light that enters the stone and comes back out of the crown towards the viewer, in per cent.fireThe share of that returned light whose colors spread apart by more than 0.6° as it leaves, in per cent. It does not say how much light returns.scintillationA sparkle score from the number of facets the stone really shows: 100 × (1 − e^(−facets / 45)). It depends only on the geometry.tiltHow much of its face-up brilliance the stone keeps when tilted 20°, capped at 100.symmetryHow evenly the pavilion facets of each design family are cut: 100 − 6 × their scatter in degrees.spreadFace-up size for the weight compared with a stone 61.5% deep: 100 × ∛(61.5 / depth %), capped at 130. Above 100 means larger than that stone.shape_normBrilliance compared with what the shape family can reach (93 for round brilliants, 86 for fancy brilliants, 76 for step cuts, 64 for others), capped at 125.leakageThe share of the light that enters and leaks out through the pavilion, in per cent. Lower is better.geometryWhich facet set was traced (the census plane set, or our own 3D model).reasonWhy a cut is not scored yet; empty for scored cuts.urlThe cut's page on digitalgemology.com.engine_versionVersion of the scoring code; it changes whenever that code changes.generatedDate the figures were last changed. Method Each cut is a set of flat facets that close into a convex solid, scaled so that its longest side is 1, table up. 3,200 rays fall on the stone from above, at most about 7° from vertical (cosine-weighted), over the whole face-up outline, with a fixed random seed, so every run gives the same scores. At each surface a ray refracts by Snell's law or reflects totally beyond the critical angle, for up to 12 reflections inside the stone. A ray counts as returned when it leaves through the crown within 64° of vertical after at least one reflection inside the pavilion; a ray that leaves through the pavilion counts as leakage. For fire, the exit is traced again at the refractive index minus and plus the dispersion; when the two directions differ by more than 0.6°, the returned ray counts as fire. Tilt reruns the trace with the stone turned 20° about its long axis (2,200 rays). The trace is all or nothing: no Fresnel splitting, and no absorption, color, inclusions or polish. The light cone, viewing limit, fire threshold, weights and constants are fixed settings of the method, the same for every cut, so the scores compare cuts with each other; they are not absolute measures. Full method: https://digitalgemology.com/learn/cut-performance-index Citation Digital Gemology (2026). Digital Gemology Cut Performance Index, version 2026-10-31 (engine census-ba47e2e1.13d5809b). https://digitalgemology.com/learn/cut-performance-index