Procedural Level Generation For 2D Platformer Games Based On Graph And Chunk

Abstract

Level design is one of the key components in determining the gameplay experience in 2D platformer games. However, manual level design is time-consuming and difficult to scale. Graph theory can be used to represent the abstract structure of a level, with nodes as spatial units and edges as paths connecting those spaces. The problem is that this graph structure cannot always be directly translated into a valid and playable physical layout due to platformer constraints, such as jump height, safe distance, and gravity. This research proposes a chunk-based transformation approach, in which each graph node is represented as a fixed-size level chunk containing a platform layout, obstacles, and entry and exit points. The transformation of the graph into a physical level is carried out through the selection, arrangement, and alignment of chunks based on the graph’s topology. This study aims to design a framework for transforming an abstract graph G=(V,E) into a chunk-based 2D platformer level layout. The method used is qualitative-descriptive, involving the formulation of a graph model, rules for mapping and arranging chunks, and playability criteria. The results of the study reveal three main stages: mapping vertices to corresponding chunks, arranging chunks according to the graph’s structure, and validating reachability using an automatic path-finding agent.