Avaliação comparativa em termos de memória e tempo de execução do algoritmo WFC aplicado à geração de mapas de jogos.

Abstract

The Wave Function Collapse (WFC) algorithm plays a prominent role in the gaming industry, commonly used for the automatic generation of textures and maps. Such usage is part of the field of Procedural Content Generation (PCG), which aims to enhance replayability and alleviate the workload of developers in manual content creation. The use of PCG hás experienced a notable increase in recent years, driven by the constant growth in the size and complexity of produced games. Although several PCG techniques have already been extensively documented and tested, WFC is a relatively recent approach that lacks comprehensive evaluations of its effectiveness compared to other techniques. This thesis aims to fill this gap by conducting an analysis of the complexity of the WFC algorithm in terms of time and hardware resources, comparing it to alternatives in the context of game map generation. PCG algorithms were executed on maps of various sizes, ranging from 100px to 1000px. During the executions, memory consumption and execution time data were recorded. The results showed that, among the evaluated algorithms, WFC excels in terms of memory consumption, surpassing the other algorithms in this aspect. On the other hand, the Binary Space Partitioning Room Placement (BSPRP) proved to be the most efficient in terms of time, significantly outperforming the performance of WFC. Finally, the Random Room Placement (RRP) was the least efficient, both in terms of memory consumption and execution time.