The ray probe that measured bands looked half a metre either side of the
ink and left a piece open where the real face sat further off. Lotus's
defense platform wall had no band below 20.75 m, and Bind's B container
outline had no bands at all.

A check from ~32,000 standing spots compares the painted walls with the
3D scene sliced at each eye. Pieces the scene stops 60%+ of the rays
through rise to the height of what stopped them, cut short where the
scene lets rays through. Long partly-open pieces are cut into one-unit
pieces first. Recorded decisions about openings are left alone.

Spots seeing through a solid painted piece: 11,054 -> 5,905; leak rays
337,158 -> 104,440; false shadows +0.7%. Scripts are in the archive under
scripts/truth.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>