nx_kubelka_lib.nx

nx_kubelka_lib.nx -- KUBELKA-MUNK TWO-FLUX REFLECTANCE, THE CHROMOPHORE SUBSTRATE (2026-09-03) WHY. Skin albedo in this estate is a COLOUR (nx_skin_ita derives an sRGB from the genome; nx_arousal_skin carries melanin and perfusion permil). A colour can be painted but it cannot be RELIT, and it cannot be EDITED physically -- and every late-stage goal on the biotwin board (a predicted scar, its erythema decaying over months, pigmentation returning) is an edit to the ABSORBERS underneath, not to the pixel on top. Kubelka-Munk is the bridge: it turns "how much melanin and how much haemoglobin" into "what reflectance", and -- because it inverts in closed form -- it turns a measured reflectance back into K/S. That inverse is what lets a captured photograph become chromophore ground truth instead of a baked texture. MEASURED ABSENT before building: nx_absent kubelka -> ABSENT-PROVEN, coverage_complete=1 corpus_complete=1. SINGLE-CONSTANT KUBELKA-MUNK, the form the field actually uses for an optically thick layer: R_inf = 1 + K/S - sqrt((K/S)^2 + 2(K/S)) forward K/S = (1 - R)^2 / (2R) inverse (exact, and the round trip proves both) K is absorption, S is scattering. Absorption is ADDITIVE over chromophores, which is the whole reason this parameterisation relights and edits: K = K_base + c_mel*K_mel + c_hem*K_hem. THE COEFFICIENTS ARE NOT IN THIS FILE, DELIBERATELY. K_mel and K_hem are MEASURED spectra and belong in pinned reference data, not typed into a lib by an author who did not measure them. Every function here takes them as arguments, so this file carries physics and never fabricates a constant. PROVEN 10/10 GREEN on the laptop build farm 2026-09-03 (NAS build lane was wedged): K/S=0.25 -> R=0.500000 exactly, K/S=1 -> R=0.267949 (2-sqrt3), round trip 9/9 rungs with a worst error of 1e-6. All arithmetic is nx_fixq30_lib Q30. license_tier: ORIGINAL No hw writes (Rule 26). LIB (no main).