Microdaf’s corrected window study is published for internal review, with the lowest row retained and the four upper rows enlarged.

Stage 386 publishes the banked Stage 385 geometry without further architectural changes. Both Blender and GLB are byte-identical to the banked versions. The comparison baseline is Stage 326. 335 meshes, 1,562 triangles, zero textures; 163 surfaces changed in Stage 385 and 172 preserved.

Upper openings are 1.30 by 3.90 approximate study units; lowest openings remain 0.92 by 3.90. Frames, reveals and surrounding facade moved together. The top study cut rose from 13.12 to 14.64; this is not a building-height measurement. The viewer offers complete street, elevated and frontal views, whole retained photographs and an unchanged baseline.

The new study is available from the city review menu. City geometry is unchanged; this partial surface study has no accepted attachment transform. Hidden entrance-wall junction and photographer follow-up remain open. Pending cards S217-01, S332-01, S337-01, S347-01 and S383-01 remain untouched. No new identity question or completed building.

Deployment uses a fresh complete live clone, explicit file patch, separate complete backup and rehearsed preserving rollback. Full local deployment records and REPORT-FREEHAND343.md are retained in C:/pepper-codex/v4live/stage386 and its parent.

INTERNAL-REVIEW-ONLY / EXTERNAL_DISTRIBUTION_BLOCKED. First-party field photographs retain provenance; inherited Google-derived geometry and historical sources retain the internal-review restriction. No pixels generated or altered; photographs are references, never textures.