MaxLiu22/Motion2MixamoRig

What it solves

Motion2MixamoRig automates the process of transferring human motion from a video or a static pose from a photo onto a 3D character rigged for Adobe Mixamo. It eliminates the manual effort of keyframing or complex motion capture by using AI to extract skeletal movement and retarget it to a standard 3D character format.

How it works

The project uses a command-line pipeline (or a Blender extension) to process input media. It first extracts 3D human motion data from a video or image using inference models (such as GVHMR). It then maps this motion onto the skeleton of a provided Mixamo-rigged FBX file. The final output is a skinned .glb file ready for use in game engines like Unity or 3D software like Blender, along with preview videos or images showing the results.

Who it’s for

This tool is designed for game developers, 3D animators, and digital artists who want to quickly apply real-world human movement to their 3D characters without needing professional motion capture hardware.

Highlights

• Dual Interface: Offers both a command-line tool for developers and a user-friendly Blender extension for non-coders.
• Versatile Input: Supports both video clips and static photos to generate 3D character animations.
• Efficient Workflow: Allows users to reuse extracted skeletal motion data (.npz files) to quickly apply the same movement to different character models without re-running the heavy inference step.
• Ready-to-Use Output: Generates standard .glb files that are compatible with major 3D software and game engines.