CONSTRUCTIVE PROCEDURAL CONTENT GENERATION USING IMAGE DATABASES

Abstract

Procedural Content Generation (PCG) attempts to define an automated process for designing game content. Angry Birds represents a physics based game in which interesting game play comes from the challenge and/or fun associated with destroying structures using various types of ‘angry birds’. This thesis proposes and demonstrates an approach to automating the design of the structures using a database of source objects. The hypothesis is that the resulting structures will retain some properties that remind the game player of the original object, making it ‘fun’. On the other hand, these structures also tend to be more varied than those produced using current approaches to PCG for the Angry Birds game. The approach is demonstrated using a database of 20 images, resulting in content that we believe improves on the longevity and enjoyment of the game relative to current practice.