Description
anymove is a GitHub repository dedicated to advanced motion capture technology, enabling users to capture motion for any creature from a single AI video. The project leverages a sophisticated setup that integrates two camera views within one frame, enhancing the accuracy and detail of the motion capture process.
Read More
At its core, anymove employs a DINOv2 joint reader, a powerful component that interprets and tracks the movements of creatures within the video. This allows for precise articulation and replication of motion, making it suitable for a wide range of applications in animation, game development, and research. The system is designed to fit the creature's rig accurately in both camera views, ensuring consistent and high-quality motion data.
The output of the anymove process is a GLB file, a widely supported 3D model format that includes geometry, materials, and animations. This makes the captured motion easily transferable and usable across various 3D software and platforms. The project's open-source nature encourages community contributions and allows for continuous improvement and adaptation to new challenges in motion capture.
Recent updates to the repository indicate ongoing development, with commits addressing features like per-joint weights, plant hysteresis, and rerolled walk and run prompts, suggesting a focus on refining the realism and flexibility of the motion capture. The project also includes tools for building a creature's final pack and has seen efforts to ensure compatibility with Windows operating systems, broadening its accessibility to a wider user base. The emphasis on framing each creature with its own camera further highlights the project's commitment to detailed and customizable motion capture solutions.
- CategoryComputer Vision Tools
- Pricing ModelOpen source
- DeploymentCloud
- PlatformsWindows
anymove's Core Features
Motion capture from a single AI video
Supports any creature type
Utilizes two cameras in one frame for enhanced data
Integrates DINOv2 joint reader for precise tracking
Rig fitting in both camera views
Outputs motion data in GLB format
Per-joint weights for detailed control
Plant hysteresis for realistic ground interaction
Rerolled walk and run prompts for varied animations
Tools for building creature's final pack
Windows operating system compatibility
Individual camera framing for each creature
Getting Started with anymove
Clone the repository: Obtain the project files by cloning the anymove GitHub repository to your local machine.
Install dependencies: Set up the required software and libraries by following the instructions in the `requirements.txt` file.
Configure environment: Adjust environment variables as needed, potentially using the `.env.example` file as a guide.
Prepare creature rig: Use the provided tools in the `blender` directory to build and prepare your creature's final pack and rig.
Process AI video: Input your AI video into the system for motion capture, leveraging the DINOv2 joint reader.
Generate GLB output: Receive the motion-captured data as a GLB file, ready for use in 3D applications.
Optimize motion: Apply per-joint weights, plant hysteresis, and other prompts to refine and customize the captured motion.
anymove's Use Cases
- 3D Animation Production
- Game Development
- Virtual Reality Experiences
- Character Rigging
- Biomechanical Research
- Educational Content Creation






