| Session Name: | Machine Learning Summit: 'Wobbledoll': ML-Powered Self-Balancing Synthesis Ragdoll |
| Speaker(s): | Nan Ma |
| Company Name(s): | LIGHTSPEED STUDIOS |
| Track / Format: | Machine Learning Summit |
Did you know free users get access to 30% of content from the last 2 years?
Get your team full access to the most up to date GDC content
| Overview: | Realistic full-body physical interactions have long been a major challenge for the gaming industry. Numerous solutions have been proposed, from basic animation and IK-based solution to more intricate middleware such as Euphoria by Natural Motion. However, many development teams find these approaches are either designed to only function under specific conditions or overwhelmingly complicated and impractical to implement. Machine learning can be utilized as an aid in the execution of realistic full-body physical interactions. Our methods have demonstrated the potential for generating high-fidelity motions for a variety of physical interactions, whilst being substantially more accessible in terms of engineering complexity. |
Recommended
