Normal biome material observations

This run reuses the sole legitimate disposable QA world. Eleven naturally occurring, unoccupied, feature-free sites were selected with the exact client regional elevation/moisture grading and independently verified against read-only backend natural/effective cells. No terrain, entity or gameplay state was fabricated. The production world was not used.

Four current/private-sampler desktop/phone cohorts exercise all eleven profiles. All 44 cases and 220 normal-client observations complete with the expected exact approved source actually painted; 110 frames use only the private wider-crop sampler. Every tested Region return and Hex reload has exact native canvases and matching geometry/bounds. All 22 cross-cohort Hex landmark/foreground canvas pairs are exact. Region composite canvases differ and do not separately certify foreground equivalence. No source approval, artwork-byte, public-renderer, whitelist or gameplay change occurs.

These are observation and stability results, not visual readiness approvals. Two complete phone wetland scenes have been directly reviewed: broader crop coverage reduces rigid diagonal organization while retaining visible roots/stones and the exact separate foreground tufts, but local contrast/distribution and recurring microstructure remain concerns. Direct visual review of the other 42 recorded cases is still pending. Whole-mesh nonrepetition, perceived fidelity, all physical contact/state, full cost/pressure/lifecycle, runtime promotion and actual production-world appearance remain open.

Evidence: full compact results world-results.json, natural site facts sites.json, cross-cohort comparisons cross-cohort-comparisons.json, source response hashes renderer-http.json, direct visual review visual-review.json, terminal resources owned-resources-terminal.json, catalog registration registration.json. All complete raw canvas comparisons stayed in memory; retain only one decisive Hex witness per case. The redundant current-case checkpoint was removed after exact comparison with the final result; its retirement receipt checkpoint-retirement.json is retained. The earlier policy-blocked QA-world deletion was not retried or bypassed.