Arctic foreground stability diagnosis — November 4, 2026

The existing native arctic limestone source WOG-A-003298 remains unchanged and retains its individual artwork approval. Whole-renderer qualification remains pending. Repeated desktop context captures differ at 24,610 native foreground pixels within [185, 1399, 775, 1632], including 7,905 alpha differences and no opaque-core differences. The existing landmark glyph is outside the differing area. Recorded native image identities, transforms, geometry and styles match; that evidence does not prove equality of every raster input or GPU state.

The original full-size arctic ground retest produces the same foreground failure. Within each case, immediate and stable-observed foreground pixels match exactly, so waiting does not resolve the cross-case mismatch. Eight isolated context/transform variants and 48 warmup/repaint states reproduce only one of the two historical foreground hashes. Requested context preferences do not prove hardware backend selection. No causal raster fix or public renderer change is justified by these results.

Retain the original limestone PNG SHA-256 f88dae17ed12e5f99af8b869e3c67bca16cda26c8da4734161fa828451449ad7, the bounded comparison witness and compact diagnostic results. Continue attribution of the missing raster input and quantify actual visible error before accepting foreground stability. There is no new artwork approval, source admission or production-world evidence.

See full-size normal-client results normal/results.json, isolated replay native-raster-preferences.json, warmup matrix raster-warmup.json, and the preferred native domain study.