Procedural Content Generation in Games

This course gives an overview of the relatively new, quite disparate and very exciting field of procedural content generation. In this field, we study methods for generating game content (levels, maps, rules, textures, vegetation, quests etc) with no or limited human input. The course content will be closely informed by recent and ongoing research around the world, and also by recent and not-so-recent developments in commercial and indie games. While most of the course will consist of lectures describing algorithms, example systems and design considerations, there will also be much room for discussion, and an exercise where course participants will be able to implement their own algorithms in the context of generating levels for a platform game. The following topics will be discussed: What PCG is and isn't, and why you might want to use it; The search-based approach; Agents and cellular automata; Fractals and noise; Grammars and L-systems; Rules and mechanics; Constraint satisfaction approaches; The experience-driven perspective; Mixed-initiative generation; evaluating content generators.

Registration has ended -Thank you!

Time and location

Wednesday 6th of November - from 9.15 -17.15 in room TC103 and

Thursday 7th of November - from 9.15 -17.15 in room TC103 (Tietotalo)

Tampere University of Technology (Korkeakoulunkatu 1)

Map of the TUT campus, parking and wireless network

Speaker

Associate Professor Julian Togelius, IT University of Copenhagen, Denmark

Organizers: Teaching associate Timo Kellomäki and Professor Mikko Tiusanen

Note: The reading for the course is the book at

http://pcgbook.com/

You are advised to have a look at the first two chapters of the book
before the event.

You should bring your own laptop to the event if at all possible.

Please also install Java and some integrated development environment (IntelliJ IDEA is recommended). You can also download the Mario AI framework which will be used for the practical exercise:

http://www.marioai.org/LevelGeneration/getting-started

Tentative Schedule

Day 1, 6th of November (TC103)

9:15-10:45 Introduction. The search-based perspective.
10:45-11:15 Break
11:15-12:45 Constructive generation methods for dungeons and levels.
Fractals and noise with applications to landscape generation.
12:45-14:00 Lunch*
14:15-15:45 Grammars and L-systems with applications to plants and
levels. Introduction to the Mario AI framework..
15:45-16:15 Break
16:15-17:15 Practical exercise with creating level generators for the
Mario AI framework.

Informal get-together at the evening

Day 2, 7th of November (TC103)

9:15-10:45 Rules and mechanics. Introduction to the Video Game
Description Language.
10:45-11:15 Break
11:15-12:45 Generating quests and narrative. Experience-driven PCG.
12:45-14:00 Lunch*
14:15-15:45 Mixed-initative PCG. Evaluating content generators.
15:45-16:15 Break
16:15-17:15 Further work on Mario AI level generators. Course
participants present their content generators to each other.

* Lunch can be bought at own expense from Restaurant Zip (Tietotalo).

REGISTRATIONS

This seminar is free-of-charge for INFORTE.fi member organization's staff and their PhD-students. For others the participation fee is 750 €. The participation fee includes access to the event and the event materials. Lunch and dinner are not included.

CREDIT POINTS FOR PhD STUDENTS

Doctoral students participating in the seminar can obtain two (2) credit points. This requires participating on both days and completing the assignments given at the seminar.