Place visibility diagnosis — November 4, 2026

Five unchanged sources contributed20 repeated normal frames,followed by12 keep-specific method frames. These do not add distinct-source coverage.

Squarekeep’s initial town-occlusion attribution is withdrawn. Matching the actual source-canvas context attributes and replaying over the real before-draw background gives98,306 exact source-alpha128 pixels across desktop/phone Region/Hex. The keep’s72-world-unit source canvas remains a physical-scale/readability question. CSS,other canvases,soft edges and full-world acceptance remain separate.

Phone Region source/control hit areas intersect2,347 Desert Gate pixels,804 Cohesive Ward pixels and1,599 Cohesive Gate pixels. These are hit-area overlap counts,not opaque UI-alpha measurements. No hits occur in the sampled phone Hex views or floating-stone views;proximity alone is not counted as overlap.

The initial default-context Hex replay differed at64,678desktop and13,924phone pixels,max21/12. Matching willReadFrequently and the remaining actual context attributes removes those differences in this study. The alpha255-only attempt measured0 source-body pixels and cannot prove visibility. All these method results are retained;none resolves earlier terrain/Hex failures.

No source pixels,renderer,geography,save or approval changed. Owned browsers and server closed. Raw pixels remained in memory;byte-identical context sheets reuse existing witnesses. See validation-summary.json and attribution-correction.json.