Procedural Generation in Games: Focusing on Dungeons

School of Computing and Mathematical Sciences, University of Greenwich, London, SE10 9LS, UK

^* Corresponding author. Email: zs7008d@gre.ac.uk

Abstract

This report is talking about the applications of procedural generation in games. There are many AI-generated games on the market, such as No Man’s Sky. Due to the huge curiosity about it and the purpose of studying how procedural generation works, the research delved into the principles of procedural generation of dungeons. To start with the research, we first focused on a dungeon, analyzing how the dungeon is created and then we consider how the procedural generation works. Typically, all the resources are from the internet, including some academic tutorials. By separating the whole problem into three parts (terrain generation, content generation and object generation), we finally realized each part of the procedural generation, and how it works to build a simple dungeon. First, in terrain generation, we focus on a different algorithm to generate the terrain, such as fractal terrain generation, bitmap terrain generation and generating landscape by Perlin noise. In the second part, content generation usually generates loots, enemies and creates a dungeon system with various rooms and paths. And the last one is object generation, for example, which holds creatures, trees, weapons, vehicles and so on. Through these methods, we can gain a deeper understanding of the significant role of procedural generation. All in all, through the above-mentioned aspects to illustrate their respective effects on procedural generation, and guide people’s preliminary cognition and understanding of procedural generation.