Author Identifiers

Andrew Pech
ORCID: 0000-0002-9813-7974

Date of Award


Degree Type


Degree Name

Doctor of Philosophy


School of Science

First Advisor

Dr Martin Masek

Second Advisor

Associate Professor Chiou Peng Lam

Third Advisor

Associate Professor Philip Hingston


With advancements in technology, consumers are expecting higher quality and more detailed video game content. This puts a strain on video game companies and their developers as they are required to manually design and create increasingly complex video game content. Procedural content generation can alleviate this burden by using technology to automatically generate video game content, effectively reducing development time and budget. This thesis presents a novel approach towards procedurally generating video game terrains that meet a set of gameplay requirements as specified by the user. This approach uses a genetic algorithm to evolve a set of modifications that, when applied to a user-specified terrain, incorporates the desired gameplay elements. This approach can aid developers by reducing time spent manually creating and editing video game terrains.

An important aspect of this research involved designing a set of measures capable of characterising gameplay elements. A collection of isovist and graph-connectivity measures were discovered that can characterise different types of gameplay elements so that they can be automatically identified in a video game terrain. This set of measures may be useful in other procedural methods and related fields.

Access Note

Access to Chapter 6 of this thesis is not available.