Meta
Making character movement readable
Motivo is an AI model that can move a virtual character through a world it has never seen. What it produced at first was math and wireframe skeletons. The playground gave it controls a person could use: pose the character, have it copy a motion, chain behaviors together, then turn up the gravity or add wind and see if it holds. Those controls came out of animation, where the test of a movement is whether someone watching can read it. We sat down with the researchers and worked out who would use it and what for. It launched at NeurIPS with a public demo anyone could try.



