Expanding Wave Function Collapse with Growing Grids for Procedural Map Generation

Abstract

This paper proposes a method to augment the basic Wave Function Collapse algorithm with the Growing Grid neural network for procedural map generation. The system accept a higher level description of the map, in the form of an image, and the system returns a list of solutions based on learning parameters and module constraints. First, a description of the technical implementation of the system is given. Second, the augmented WFC variant is evaluated from a user experience perspective, in the context of a first person cognitive map formation.