Date of Award


Document Type

Thesis open access


Computer Science

First Advisor

Britton Horn

Second Advisor

Matthew Hibbs

Third Advisor

Sheng Tan


This investigation into the effectiveness of Wave Function Collapse as a Procedural Content Generation Technique (PCG) in Minecraft sets out to determine whether this method can be used easily by players and game designers to generate content that mimics the original content. We also set out to determine whether this technique can be implemented by game designers or community modders easily enough to improve the default generation of settlements in Minecraft. We grade the effectiveness of our output using metrics provided by the Generative Design in Minecraft Competition in order to test whether generated content is effective. Tests were conducted on terrain that was taken from an existing Minecraft world, and featured a mixture of structures ranging from simple to complex in design meant to simulate structures that players would build near the beginning of the game. Unfortunately, our conclusion is that in it’s most basic form, Wave Function Collapse is unsuited as a PCG tool for Minecraft. During the course of our testing, we found that the run times for simple algorithms were too long to be effective, and the algorithm fails to generate content for many of the test cases regularly. In order to make it more suitable, a number of improvements are suggested including global constraints, weight balancing, and layering PCG methods. Overall, this approach has potential, but requires more work before it is a suitable replacement to current PCG methods for Minecraft settlement generation.