Procedural Content Genration that Optimizes Novelty
Procedural content generation (PCG) promises to reduce production costs and increase the replay value of games. However, to actually engage and retain players, PCG needs to be long-term engaging. Currently, PCG systems deploy automatic difficulty balancing to keep generated content engaging. However, emerging evidence suggests that novelty and variety, spurring curiosity, are independent and probably even more important contributors to player engagement and retention than difficulty.
In this project, we therefore want to see if we can create a PCG system for a simple platformer game (Super Mario Bros.) or strategy game (Polytopia) that can balance the difficulty and optimise the variety of produced level sequences at once.
Specifically, we are interested using Rational Level Design to identify core gameplay beats or mechanics (shooting, jumping) and generate levels that vary the occurrence and sequence of beats relative to the prior play history of a player, using novelty search metrics to assess and filter a generated level pool.
There are 2 places available on this project. Your skills need to fit with one of the following:
- a computer science background and passion for games, ideally also familiarity with PCG systems.
- game design and HCI background for Rational Level Design analysis and playtesting/evaluation.
How to Apply
For more details on the summer school application process (including eligibility and funding) please see the overview page: here