Using Model Counting for Game Development: Quantifying Difficulty of 2D Platformer Levels for Diverse Playable Characters

Game development is a field that benefits immensely from using Generative Artificial Intelligence. In particular, level design is a labor-intensive task that can be assisted by AI. However, one challenge in level design is ensuring the balance of the level. To further support level designers, we propose a technique using model counting for the automatic analysis of a level for multiple different playable characters, in order to quantify the difficulty of the level for each character. By using model counting we can analyze a level without the need for human play testing. We implement a prototype of our tool and show the viability of this approach for a 2D platformer.