GDC is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

View, browse and sort the ever-growing list of sessions by pass type, topic, and format. All registered attendees will be able to build their personal schedule in the mobile event app, once live early 2022.

GDC Masterclass courses require a paid pass to attend. Class sizes are kept intentionally small and seats are limited to ensure every student receives one-on-one assistance from the instructor. Pass pricing per course varies, please review the passes and pricing page for more details and secure your spot today!

Machine Learning Summit: 4 Years of Bringing Characters to Life with Computer Brains

Sebastian Starke  (AI Scientist & PhD Student, Electronic Arts, The University of Edinburgh)

Pass Type: All Access Pass, Summits Pass - Get your pass now!

Topic: Programming

Format: Session

Vault Recording: TBD

Audience Level: No

This talk will present a 4-year line of research applying deep learning for character animation and control, and discuss what has been learned creating realistic character movements from motion capture data. It will cover several examples of interactive character control, including quadruped locomotion, character-scene interactions, basketball plays and martial arts movements, and how one piece of work led to the next application, problem and solution. Several video examples of different motion skills will be showcased, and what is key to making AI-powered systems successfully learn and generate high-quality character movements with neural networks, as relevant for developers, artists or researchers working in the games industry.

Takeaway

Attendees will learn about a state-of-the-art series of generating character movements from motion capture data using deep learning. The talk will provide insights into feature and network architecture setups for different motion tasks, how to better model complex and asynchronous character movements via local phases, creating novel motion combinations and variations by neural layering, and how to scale with the growing availability of motion capture data in the coming years.

Intended Audience

Attendees should have basic understanding of machine learning techniques and neural networks, and be familiar with current character animation systems used in industry. Most importantly, the audience should be motivated applying AI-based solutions for character motion tasks in video games.