There is a huge demand in simulating fast and complex interactions that involve multiple contacts between a character and objects, an environment, and other characters, especially in computer games and films. For example, for basketball games, the players need to dribble the ball while making various movements with different foot-fall patterns to compete with the opponent characters.
To solve this problem, an AI project has been developed by researchers from University of Edinburgh and Electronic Arts. This AI can created animated characters that play basketball very naturally.
The project explores the opportunities of deep learning for character animation and control. It has been gaining popularity and has now become a stable & modular framework for data-driven character animation, including data processing, network training and runtime control.
The project can be used for animating:
Developed using Unity 3D, Tensorflow, PyTorch
This research paper focuses on physically-based character animation, where the characters are controlled kinematically using physical rules as constraints, or controlled by torques under physical environments, can be applied for synthesizing motion that involves fast, dynamic contacts.
Methods such as spacetime constraints [Witkin and Kass 1988] let users provide the contact pattern as conditions, and then optimize the motion using physically-based constraints.
Read in-depth about this project here: Local Motion Phases for Learning Multi-Contact Character Movements Research Paper
Code Coming Soon
Fill out this Google Form and get a chance to be featured in this growing AI community
Q blocks is currently in private beta stage and access is provided by invitation only.
Please fill out the form below to get a chance to be one of the first few people to access Qblocks' revolutionary technology.
Thank you for signing up!
You will receive updates from us on email. Please check your Inbox as well as Spam Folder.
Copyright © Qblocks - Distributed Supercomputer for HPC