RFD 1050: Model image for lingbot_map_environment_scan
Problem
LingBot-Map turns a walk video into a 1:1 metric twin of a room. It runs two phases, and RFD 1026 marks its parameter count unknown.
The scale is the product. A twin that is 3 percent small is not a twin. Every other model in the catalog may be wrong by a scale factor and still be useful. This one may not.
Decision
Do not package this model image. Abandon this line of work.
RFD 1064 turns the roadmap toward character concepts, and away from environment scanning. The parameter-count measurement this RFD waited on is no longer needed, because the model itself is out of scope.
See DETAILS.md for the two phases, the gravity-alignment rule, and the metric gate this RFD sketched before the pivot.
THE LICENCE BLOCKER IS ANSWERED, AND WAS NEVER LOOKED AT. This RFD listed it open and abandoned the work before asking. Robbyant/lingbot-map is Apache-2.0, weights included.
The scope inverted too: RFD 1171’s presence loop makes the scan a character tool, because the static room is what a moving person is separated from. RFD 1171 takes it and the manifest places it.