Abstract

Procedural content generation (PCG) is a powerful and convenient tool that is used to algorithmically generate content instead of handcrafting it. Widely used in video game development, procedural content generation automates certain parts of the development process that can otherwise be time-consuming, such as creating game levels and building terrain. Roguelike games are a video game genre known for having procedurally generated dungeons. In this thesis, we have created a procedural content generator for two-dimensional roguelike dungeons and evaluated its performance. We present two different PCG algorithms, one built using breadth-first search and one using the concepts of depth-first search with backtracking. Additionally, we then use the generated layout to create a three-dimensional representation of the dungeons using the Unity game engine. We attempt to evaluate the “enjoyability” of the generated dungeons objectively, by creating a new metric called the “Choice heuristic” and modifying an existing metric that Gellel and Sweetser proposed. Based on the information from our evaluation, we conclude that the generator using the breadth-first algorithm generates dungeons that are more enjoyable to explore compared to the generator using the depth-first algorithm.

Advisor

Palmer, Daniel

Department

Computer Science

Disciplines

Computer Sciences

Keywords

Procedural Content Generation, PCG, video games, backtracking, depth-first search, breadth-first search, roguelike

Publication Date

2023

Degree Granted

Bachelor of Arts

Document Type

Senior Independent Study Thesis

Download

Share

COinS
 

© Copyright 2023 Manan Shahi