On Monday, October 12th, UCSC’s Expressive Intelligence Studio will be hosting a symposium on Procedural Content Generation. Please join us to see interesting talks by speakers Julian Togelius (ITU Copenhagen), Mark Riedl (Georgia Tech), and Kate Compton (EA/Maxis). There will also be a panel on “The Future of Procedural Content Generation”, moderated by Michael Mateas (UCSC), and a number of demos by students in EIS. All events will occur on the UC Santa Cruz campus. We hope you will be able to come!
Further information on talks and scheduling information is after the cut.
Talk: Computational Intelligence for Content Creation, Julian Togelius (ITU Copenhagen)
9am – 10am, E2-180 (Simularium)
This talk will outline how various techniques and methods from computational intelligence, such as evolutionary computation and neural networks can be used to procedurally generate game content. The central idea is that content creation is seen as an optimization problem, with models of player behaviour and player experience used as objective / fitness function for the content being optimized. Examples that will be given include track layouts for racing games, rule sets for Pac-Man-like games, and level designs for Super Mario Bros. The talk will conclude with a discussion of how other successful techniques from computational intelligence, such as collaborative filtering, can be subverted to serve the needs of procedural content generation.
Talk: Experience Management through Narrative Content Generation, Mark Riedl (Georgia Tech)
10am – 11am, E2-180 (Simularium)
Storytelling is a pervasive part of the human experience. We as humans tell stories to communicate, inform, entertain, and to educate. In this talk, I will argue for automated narrative generation as an approach to procedural content generation. A computational system capable of narrative generation can automatically produce stories, movies, computer game quests, and agent behaviors that are customized to an individual’s preferences, needs, and desires. Narrative generation also facilitates dynamically created interactive experiences that balance plot coherence against player self-agency. I will describe our approach to narrative content generation, experience management, and potential implications.
Demo Session (Expressive Intelligence Studio)
11am – Noon, E2-392
1. Launchpad: Rhythm-Based Level Generation for Platformers, Gillian Smith
2. BIPED: Computational Support for Playtesting Game Sketches, Adam Smith and Mark Nelson
3. WideRuled: A Simple Interface for Author Goal Based Story Generation, James Skorupski
4. EMPath: An Application of Declarative Optimization-Based Drama Management, Sherol Chen
Talk: Using Procedural Generation for Human-Computer Collaboration, Kate Compton (EA/Maxis)
1:45pm – 2:45pm, E2-180 (Simularium)
There exists an underlying predisposition in procedural generation to treat the generation algorithm as a stand-alone, completely autonomous black box. When the game needs more content, the algorithm supplies it. This ignores one of the greatest advantages of procedural generation: since it runs in real time, the player should be able to engage, modify and direct the system as it is running. In turn, the system can respond to the player’s actions: modifying, correcting, guiding, and judging them. I will discuss several examples of user-guided procedural generation in current games, as well as show examples of how it can be used in the future.
Panel: The Future of Procedural Content Generation, moderated by Michael Mateas (UCSC)
5:30pm – 6:30pm, E2-180 (Simularium)
Panelists are Julian Togelius, Mark Riedl, and Kate Compton.
About the author: Gillian Smith is a 5th year PhD student researching computer supported level design and procedural content generation. Her other interests include women in games and games in education. Read more from this author