The Rust engine: quiddity for what a part's features are, haecceity for where its surfaces are.

Every answer here is either a flat binary — points, normals, face ids, depths, triangle indices — or a JSON string. That is deliberate. A 64 × 64 frame is 4 096 points, and building it as nested Erlang terms one at a time in Rust costs more than the ray casting that found them; quiddity's own recognition document is already JSON, and quiddity serve speaks JSON, so one representation means nothing is modelled twice.

A NIF is a library the BEAM loads into itself, so its absence is a raised ErlangError rather than a returned error: loaded?/0 is the question a caller asks once.