Procedurally Generating Game Level with Specified Difficulty
Zong-Han Wu, Kevin Lai, Li-An Lin, Ming-Han Huang, Wen‐Kai Tai · 2018
With the development of procedural content generation, more and more games use it to generate game contents so that each play can have a different experience. However, the generated game content does not guarantee enough gameplay to keep the player's freshness. In this paper, we generate Pac-Man mazes with specified difficulty with respect to the number of intersections and elements given. In order to verify whether the generated maze can retain the original Pac-Man style or not, we evaluate the appearance of the maze by the esthetic score based on distributions and sizes of various attributes in the generated maze and the original maze. We have divided the generated mazes into nine different difficulty levels, and for each difficulty level, the maze can nearly be generated in 40ms and it is playable! Therefore, our approach is practical and viable for real-time game applications.