wassermanproductions /motion-previs-studio

About motion-previs-studio

Motion Previs Studio reads a reference shot and pulls out control layers for AI video. It finds body pose, depth, edges, and the camera move, then saves them as files. Pick a shot range and a mode, like camera only or full scene. Builds are unsigned, so you'll see security warnings, and there's no Linux package yet.

Best for: Filmmakers who want measured pose, depth and camera-move references from an existing shot before generating new footage.

• Imports local clips or web video URLs, and lets you pick a shot range
• Finds 2D and 3D body pose landmarks on your machine with MediaPipe
• Solves camera pan, tilt, zoom and roll from the background, not the actor
• Exports OpenPose BODY_25 skeleton video plus per-frame keypoints JSON
• Bundles depth, edges, lineart, masks, animatic and prompt files
• Lets an AI agent drive a running app over a localhost MCP server

1. 1
   On macOS, install with the command below.

   curl -fsSL https://raw.githubusercontent.com/wassermanproductions/motion-previs-studio/main/install.sh | bash

   Copy
2. 2
   On Windows 11, download the win-x64.exe installer from GitHub Releases.
3. 3
   If you downloaded the macOS .dmg in a browser, run the command below once.

   xattr -cr "/Applications/Motion Previs Studio v4.app"

   Copy
4. 4
   Import a local clip or paste a compatible video URL, then choose the shot range.
5. 5
   Pick a reference mode and control layers, run analysis, then export the Production Pack.