Open-Source ComfyUI Motion Previs Studio Blocks 3D Shots Before Video Generation

navarisun · reddit · 2026-10-31

The author released ComfyUI Motion Previs Studio, an MIT-licensed, local-first 3D previsualization tool built to fix a recurring pain point: video models are getting stronger, but precise control over camera movement, character blocking and spatial composition remains hard. The idea is to direct the scene in 3D first, then hand control passes to a video model.

Key features

ComfyUI integration: exports reference and control assets for LTX-2.5, MiniMax H3, Seedance and other reference-input video models, plus a direct local ComfyUI API bridge for a MiniMax H3 Ref2VA workflow that submits previs video and reference images and displays the result. The author is soliciting feedback on ControlNet integration, which video model to prioritize, workflow interoperability (JSON import/export vs API-driven) and production usability.

Related event: Open-Source ComfyUI Motion Previs Studio Directs AI Video via 3D Storyboards(2 posts)→

Original post →

More from coding & agent

coding & agent channel →