nx_creatorfit.nx

nx_creatorfit.nx -- ★★★CREATOR: THE INVERSE LOOP. Everything else in this programme runs FORWARD -- parameters and a seed produce a body. This runs BACKWARD: given a body, recover the parameters that would have produced it. That inversion is the whole difference between generating *a* person and twinning *this* person, and it is the step a scan-based model cannot offer at all, because a scan has no parameters to solve for -- only vertices. ★★WHY THE PROCEDURAL BASIS IS THE RIGHT THING TO FIT, and it is not a consolation prize. The search space is a few dozen anatomically meaningful integers, and **it cannot express an anatomically impossible human** -- so the fit physically cannot chase noise into a body that could not exist. Free-form mesh deformation can, and does: it will happily fit a scanning artifact by growing a spike. Every point this optimiser can reach is a body the generator would have been willing to emit unprompted. The measured stakes (SOTA, July 2026): a scan-free parametric model fits a real subject to ~2.4mm, against ~1.8mm for a model trained on six hundred thousand scans. A ~0.6mm penalty for owning your own generator. ★★★COMPOSITION, NOT REIMPLEMENTATION -- the design constraint that shaped this organ. A fitter needs to (a) build a candidate body and (b) measure it. Both already exist as proven organs, so this one FORKS them: nx_body_proc -> canon, nx_body_gen -> mesh, nx_twinbench rms -> one integer of error. It carries NO copy of the canon rules and NO copy of the distance metric. A fitter with its own metric is a fitter that can score beautifully against a ruler nobody else agrees with, and the two copies drift the moment either is improved. ★★THE TOOTH THAT GATES EVERYTHING: SYNTHETIC RECOVERY. Before this may be pointed at a person it must be pointed at a body whose parameters are already KNOWN -- generate a target from a chosen vector, hide the vector, and demand the optimiser find it. Anything else is unfalsifiable: on a real subject there is no answer key, so a broken optimiser and a working one both produce "a number that went down". nx_creatorfit probe <target.nxmesh> <sex> <build> <musc> -> rms error of one candidate, micrometres nx_creatorfit fit <target.nxmesh> -> recovered vector, coarse-to-fine nx_creatorfit selftest license_tier: ORIGINAL expect_exit: 0 No hw writes (Rule 26).