Evolutionary Wave Function Collapse

2026-07-02

AI Summary

Traditional Wave Function Collapse (WFC) struggles to optimize global objective properties of generated content, particularly in procedural content generation tasks requiring functional or structural constraints. This work proposes a novel approach that integrates evolutionary algorithms at the WFC input stage: small exemplars (genotypes) are evolved and subsequently expanded into full levels (phenotypes) via WFC, guided by domain-specific fitness functions. The method significantly improves generation quality in tasks where local relationships determine global properties—such as maze connectivity—demonstrating its efficacy. However, it remains challenged by scenarios involving strong global constraints, such as Zelda-like dungeon generation, thereby delineating the method’s current applicability boundaries.

Abstract

Wave Function Collapse (WFC) is a widely used procedural content generation method that learns local adjacency constraints from example inputs to generate larger outputs. In this paper, we explore combining WFC with evolutionary search by evolving the small input examples used by WFC rather than directly evolving complete levels. In this approach, WFC acts as a genotype-to-phenotype mapping. The generated levels are then evaluated through domain-specific fitness functions. We evaluate the method in two domains with different relationships between local and global structure: Maze connectivity maps and Zelda-style dungeon layouts. Our results show that evolutionary optimization over WFC inputs improves generation quality in domains where properties emerge from local relationships, while domains requiring global constraints remain challenging. These findings suggest that evolutionary search can effectively guide WFC generation when target objectives align with local structure.

Problem

• Wave Function Collapse
• procedural content generation
• evolutionary search
• local constraints
• global structure

Innovation

• Evolutionary Search
• Wave Function Collapse
• Procedural Content Generation
• Genotype-to-Phenotype Mapping
• Local Constraints