nx_curvebench.nx

nx_curvebench.nx -- ★★★OUTLINE CURVATURE: the metric that replaces waist-to-hip ratio. WHY WHR IS GONE, and it was not an aesthetic judgement -- it was arithmetic. Parsing ANSUR II, the US Army anthropometric survey: **0.0% of 1,986 women are at or below WHR 0.70. The minimum in the entire sample is 0.701.** A target that no member of a 2,000-person population reaches is not a target, it is a number that got repeated. Independently: Hubner & Ufken 2024 measured outline CURVATURE explaining ~65% of attractiveness variance against WHR's ~28%, and Lidborg & Boothroyd 2025 (N=125,062) found most significant WHR-fertility associations running the WRONG WAY, which removes the mechanism WHR was supposed to have. ★★THE POINT IS NOT THAT CURVATURE IS A *LOWER* TARGET -- it is a BETTER-MEASURING one. It captures the same visual quality more faithfully, so aiming at an idealised figure becomes a NUMBER on this axis rather than an argument about a ratio nobody meets. ★★★AND THAT REFRAMES WHAT A "MUSE" IS. What a generator needs from a reference figure is not pixels, it is AN AIM POINT ON A METRIC. Extract the number, aim at the number: it composes with the fitter, it survives the reference going away, and it is not a likeness of anybody. A stored image is a liability with an expiry date; a curvature target is a parameter. WHAT IT MEASURES: the front-view silhouette as a half-width profile w(y), then the BEND in that profile. ★DIMENSIONLESS BY CONSTRUCTION -- both w and y are normalised by stature before differencing, so the index describes SHAPE and not SIZE. A metric that grows with the body is measuring the body, not its form, and T3 exists to prove this one does not. nx_curvebench measure <body.nxmesh> [stature_mm] -> curvature index + WHR, side by side nx_curvebench selftest license_tier: ORIGINAL expect_exit: 0 No hw writes (Rule 26).