Game Level Blending using a Learned Level Representation
This addresses the limitation of requiring human-annotated representations for procedural content generation in games, making level blending more accessible for unannotated games.
The paper tackles the problem of game level blending by introducing Clustering-based Tile Embeddings (CTE), a learned level representation that eliminates the need for human annotation, and demonstrates comparable or better performance to human-annotated methods in blending levels for Lode Runner and The Legend of Zelda.
Game level blending via machine learning, the process of combining features of game levels to create unique and novel game levels using Procedural Content Generation via Machine Learning (PCGML) techniques, has gained increasing popularity in recent years. However, many existing techniques rely on human-annotated level representations, which limits game level blending to a limited number of annotated games. Even with annotated games, researchers often need to author an additional shared representation to make blending possible. In this paper, we present a novel approach to game level blending that employs Clustering-based Tile Embeddings (CTE), a learned level representation technique that can serve as a level representation for unannotated games and a unified level representation across games without the need for human annotation. CTE represents game level tiles as a continuous vector representation, unifying their visual, contextual, and behavioral information. We apply this approach to two classic Nintendo games, Lode Runner and The Legend of Zelda. We run an evaluation comparing the CTE representation to a common, human-annotated representation in the blending task and find that CTE has comparable or better performance without the need for human annotation.