Automated Difficulty Assessment Model for Platformer Games: A Comprehensive Approach
Yannick Francillette, Hugo Tremblay, Bruno Bouchard, Simon Lescieux, Mathis Rozon, Jules Linard · 2023
In general, a video game offers a seamless progression in gameplay difficulty, starting with easy levels that allow players to grasp the basic mechanics of the game, and gradually introducing more challenging obstacles as they progress. The success of a game title heavily relies on its ability to provide a wellbalanced difficulty curve and a satisfying sense of progression. Designing a game entails a complex and time-consuming process that involves extensive playtesting. One promising approach to address this challenge is the utilization of software tools capable of automatically evaluating the difficulty of game levels. In this paper, we present a comprehensive model for automatically assessing the difficulty levels of platformer games. Our model is based on the formal calculation of static danger zones within levels and the analysis of enemy movement patterns using simulated pheromones. To validate our model, we implemented it and conducted tests using the complete set of levels from the original Super Mario Bros. game. The paper includes a detailed presentation of the model, the tools developed, and a comparative analysis showcasing the computed results.