Engineering, operations & design history
Native limestone canvas representation — November 5, 2026
Full repository source: docs/art-direction/display-cache-history-20261104/arctic-native-canvas-20261105/review.md
Technical documentation includes historical decisions, proposals and verification reports. For playing instructions and practical rules, open Gameplay.
No matching entry. Try a shorter phrase or clear the search.
Native limestone canvas representation — November 5, 2026
Retain the full-native canvas approach as a preferred private representation study for WOG-A-003298. The original approved limestone PNG remains byte-identical. One 1586 × 992 canvas preserves all 1,573,312 decoded source pixels exactly against the previously measured native RGBA reference. The declared retained backing is 6,293,248 bytes per page; this is not a measurement of total memory, GPU storage or renderer cost.
Eight actual normal desktop and portrait-phone Region/Hex views have six exact lifecycle comparisons and two exact foreground pairs. Both independent HTML-image/native-canvas replay pairs are exact across 4,168,400 foreground pixels, and each complete replay hash matches both actual normal-client foregrounds for its viewport. The native limestone comparison crops were directly inspected: fine snow and mineral detail remain identical to the current HTML replay, and the difference panes are black. All submitted original source PNGs and post-capture native input hashes remain unchanged.
This scoped pass does not erase the original HTML path's 24,610-pixel desktop discrepancy. The older phone hash also remains recorded: the current replay reproduces the newer client hash, not that historical hash. No causal account of every image backing or GPU state is proved. Keep the counterexamples and qualify controlled history, cache replacement/ownership, source permission changes, pressure, total cost, other roles and physical devices before activation. The private helper deliberately permits only one prepared backing per page; it is not a production cache manager.
The unchanged full-size ground study WOG-A-008379 participates in these contexts. Its rectangular boundary, oversized full-size grain and snow-apron contact issues remain open. The preferred alpha-domain derivative WOG-A-009281 has not yet been tested in combination with this limestone preparation. Neither original painting is replaced or repainted.
Retain the desktop native comparison limestone-representation-native.png, phone native comparison ../arctic-native-canvas-phone-20261105/limestone-representation-native.png, and phone world context ../arctic-native-canvas-phone-20261105/normal/arctic-native-canvas-context.png. The redundant desktop sheet is the same bytes as the earlier retained full-size witness, so its reference is shared. Five completed-run hints/duplicate artifacts were retired after preservation checks, reclaiming 2,479,571 bytes. Raw readbacks are never persisted.
The library stays at 9,212 catalog files, 9,237 ledger records, 5,640 pending qualifications and 3,597 explicit dispositions. This adds two individual findings and one preferred private rendering study, with zero new artist assets, source admissions or complete game-ready approvals. Preferred artist outputs remain 42. Keep both Citadel hamlets, all source approvals, the production renderer/source contracts, geography and persistent saves.
Evidence: desktop analysis analysis.json, phone analysis ../arctic-native-canvas-phone-20261105/analysis.json, desktop replay replay.json, phone replay ../arctic-native-canvas-phone-20261105/replay.json, individual decisions visual-review.json, registration registration.json, and cleanup temporary-retirement.json.