Open-source mocap pipeline turns any video into Mixamo animation, fully operated by an AI agent
andrew_n_carr · x · 2026-08-19
Dev discovers that the open-source mixamo-llm-mocap pipeline was built and is operated end-to-end by an AI agent: any locked-camera video (filmed or AI-generated) goes through GVHMR pose estimation, spec-driven retargeting, and FK application in Blender via MCP — no mocap suit, no manual keyframing.
- Works with any Mixamo character with real foot contact
- Supports two-performer fights: both tracks split by screen side, retargeted onto Y Bot and Ninja (different proportions), placed at the distance performers actually stood
- Demo: a 10-second clip with 9 punches, a slip and a high side kick processed straight through the pipeline
- Open source (20 commits), designed so an AI agent can run the whole loop
More from coding & agent
- Crusoe releases integrations cookbook for Cursor, Zed, and OpenAI-compatible API — darian314 · 2026-08-19
- Designing Loops for Production-Grade Coding Agents: A Case Study — JosephJacks_ · 2026-08-19
- GPT-5.6-Sol Generates Single-File Real-Time 2D Fluid Sim — scaling01 · 2026-08-19
- RadAgent demonstrates reasoning in medical AI — Michael_D_Moor · 2026-08-19
- LLMs Can't Explain Code Decisions; Excessive Comments Might Be a Feature — idanbeck · 2026-08-19
- RAG design case: Ask LukeW citations link directly to video timestamps — LukeW · 2026-08-19