GDC + VRDC 2019 Session Scheduler

View, browse and sort the ever-growing list of GDC sessions by pass type, track, format, and more! With Session Scheduler, create an account to build your own GDC schedule and access it anywhere, including on the GDC app, once live.

If you've registered for GDC, you can use the same login credentials. Adding a session to your schedule does not guarantee you a seat. Sessions do fill up, so please arrive early to sessions that you would like to attend.

Rethinking Your Performance Capture Pipeline with Machine Learning

Stephen Shaw (Technical Lead, ALICE Studios, Ubisoft)

Location: Room 2001, West Hall

Date: Wednesday, March 20

Time: 11:30am - 12:00pm

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

Topic: Visual Arts

Format: Session

Vault Recording: Video

Traditional motion capture and post-processing has evolved over the years with improved software and practices, but the steps involved to produce clean animation are still accomplished through a hybrid of automated tasks and manual intervention. This talk will explore some of the latest advancements in performance capture at Ubisoft's ALICE studios and focus on improvements to their shoot floor and process time. With the research coming from Ubisoft La Forge into developing neural networks to automatically clean up captured data, they are close to providing a fully automated solution to deliver animation to the client. The presentation will look at the implementation and challenges involved in transforming the traditional performance capture pipeline into one that encapsulates these new research endeavors and the speaker will provide practical advice on how to use machine learning research in a data pipeline through relating the studio's experiences with the process. The talk will also explore some new technologies and approaches that are improving their capture studio that all contribute to a fully automated motion capture processing solution.

Takeaway

Attendees will learn how to transform their existing processing pipelines with a more modern approach to allow them to focus less effort on data processing, complete with the potential benefits and challenges. Attendees who are more interested in understanding how to adopt machine learning into production should also benefit from the presentation.

Intended Audience

Anybody with interest in performance capture and understanding the process of performance capture and with interest in how to industrialize a machine learning solution into a live pipeline. The presentation will provide some brief on the motion capture process and will not be targeted toward only those within the industry.