Endless tower

Introduction:

There are many ways of implementing 2d procedural map generation systems these days. Well-described approaches in the literature include grammar-based systems and cellular automata, but also more recent methods that use machine learning to generate a level based on previously seen map data. All of these methods can create a playable world, but they all have their own problems and none of them are perfect. Because level generation is tied to the specific game it is used for, it’s hard to evaluate a level generation process by itself only – it is necessary to also take into account the game’s design. In order to investigate this relationship, I first analyze six popular dungeon generation methods and form my own map generation idea. In order to bring my idea alive, I designed a new game on top of this map generation idea and keeps grinding it according to my game design and tests. Through this process, I investigate how my level generation algorithm influences my game design, and keep iterating it to see how my design choices can influence my level generation methods, in order to understand the interplay between the two. Thus, my thesis represents an inquiry into designing with and for procedural level generation.

Paper:

Interactions Between Game Design and Procedural Level Generation