Open-source ComfyUI Motion Previs Studio brings 3D camera blocking to AI video
navarisun · reddit · 2026-10-31
Developer navarisun released ComfyUI Motion Previs Studio (MIT-licensed), an open-source local-first tool tackling AI video's biggest pain point: precise control over camera movement, character blocking, and choreography.
Core idea: instead of long prompts and blind re-rolls, direct the scene in 3D first, then export control signals to the video model. Workflow: 3D scene → camera & character animation → control passes → ComfyUI → AI video.
Features:
- Dual 3D viewports: free-orbit editor plus live cinematic viewfinder
- Camera rigs: dolly, orbit, tracking, crane, handheld, whip pan, custom trajectories, focal length/DoF/aspect controls
- Procedural character animation (walk, run, combat, gestures) with interactive FK/IK posing
- 24fps timeline with camera and character keyframes
- Control pass exports: MP4 previs, OpenPose skeletons, depth maps, alpha masks, camera matrices
- Optional LLM integration (Ollama/LM Studio or cloud) for natural-language scene direction
ComfyUI integration: local API bridge for a MiniMax H3 Ref2VA workflow, designed for LTX-2.5, Seedance and other reference-capable video models. Author seeks feedback on ControlNet integration, model priorities, and workflow interoperability.
More from coding & agent
- ASC CLI now covers Xcode commands, fastlane migration, and Linux builds via xtool — rudrank · 2026-10-31
- LangChain founder: agent observability is 'Databricks for agents', not Datadog — hwchase17 · 2026-10-31
- A 710k-line PR ports pi-durable and Chord to Rust for crash-safe agent runs — jasonkneen · 2026-10-31
- Strix Halo one-command Docker for vLLM: Qwen3.8-Flash-Next hits 43 tok/s decode — asfbrz96 · 2026-10-31
- Anthropic's VS Code extension on Cursor backdoored for nearly 12 hours, up to 5,000 devs exposed — Ars Technica AI · 2026-10-31
- dhh on coding in the AI era: code is no longer the bottleneck, humans are — NERDDISCO · 2026-10-31