Learned Motion Matching v1.0 is a game resource pack that generates realistic and smooth transitions for basic humanoid locomotion. Key features include natural, fluid movement with no foot sliding or abrupt animation changes, achieved through a customizable kNN classifier that balances responsiveness and realism. The package supports waypoint navigation using an advanced Distance Matching system powered by a Greedy Algorithm, enabling characters to follow custom paths with precision, loop waypoints, and adjust speed to interact with the environment. With Inverse Kinematics (IK), characters can dynamically hold, grasp, or manipulate objects with realistic hand poses, allowing for customizable hand weights, fine-tuned motions, and complex finger movements. The asset offers effortless setup, supporting all humanoid characters with a one-click configuration and comprehensive documentation.
Dependencies include the Animation Rigging and Sentis packages from the Unity Package Manager. Limitations include support only for basic locomotion, with no included advanced moves such as jumping or vaulting. Custom movements like limping or skipping require external data and are not provided. The package requires Sentis version 1.3, an older version of Sentis. This asset is based on the SSIGRAPH technical paper. Original asset link: https://assetstore.unity.com/packages/tools/animation/learned-motion-matching-312510