KimodoUnityBridge

Kimodo adds local AI humanoid animation generation to an existing Unity project. It supports prompt-driven motion, pose and end-effector constraints, analysis, baking, retargeting, Animator content, Timeline authoring, and runtime motion playback.

• Runs from a project-owned local runtime.
• Supports Windows, macOS, and Linux; CUDA is the primary acceleration path and CPU fallback is available.
• Keeps animation generation and assets inside the Unity project workflow.

Requirements

• Unity 2022.3 or newer
• An existing Unity project
• A character with a valid Humanoid Animator for humanoid workflows
• At least 8 GB memory and sufficient disk space for the selected models

Install

In Unity Package Manager, choose Add package from disk and select this repository’s package.json. For a local manifest dependency, use:

"com.unity.kimodo_unity_motion_tools": "file:C:/nvlab/KimodoUnityBridge"

Minimal start

1. Import the package’s Light Sample from Unity Package Manager, or open the FullDemo.
2. Open its Timeline and select a Kimodo clip bound to a character.
3. Enter a motion prompt and choose Generate & Bake.
4. Wait for the project-local runtime and model to become ready, then play the Timeline.

Runtime diagnostics are written to NvlabKimodoQuickServer~/log/setup.log and NvlabKimodoQuickServer~/log/bridge_server.log in the Unity project.

AI agents

Install the package into the target project, then read AGENTS.md, SKILL.md, and TOOLS.md. Discover current commands through KimodoUnityBridge.Command.command_dispatcher and kimodo_help; the live schema and returned values outrank this README.

For asset-backed validation, use the maintained AnimationEval bank. Its active suites are semantic recognition (A), quality comparison (B), and generation/workflow compliance ©; the current B data is not yet blind/anonymous, and suite-specific public/private files define what the evaluator may see.

Links

• Demo video
• FullDemo
• Development and CLI coverage
• Apache License 2.0