How Procedural Content Generation affects player experience in 2D platformer games

Abstract

Procedural content generation (PCG) has become an established technique in game development, offering increased content variety with reduced manual design effort. However, its impact on player experience remains an active area of study, particularly in genres that demand precise spatial design such as 2D platformers. This thesis investigates how PCG level design affects perceived flow, how constraint complexity shapes player experience, and whether prior platformer experience moderates players’ perception of procedurally generated content. A 2D side-view platformer, Gone Fishing, was developed in Unity, featuring a constraint-based PCG system inspired by the room-template approach used in Spelunky. Levels are assembled at runtime from a library of 82 handcrafted room templates distributed across 15 opening categories, connected via a random walk path generation algorithm. Two configurations were evaluated, a 3-fish (small) and a 6-fish (large) condition, differing in generated level size and complexity. A user study was conducted with 20 participants of varying platformer experience, who each played both configurations. Player experience was measured using the Game Experience Questionnaire (GEQ) alongside completion time data and observational notes. Results indicate that the PCG system consistently produced structurally valid and broadly enjoyable levels. Immersion and flow subscales averaged close to 4 out of 5 across all experience groups, and negative affect remained low throughout. The fish counter parameter successfully produced a perceptible difficulty gradient, but uncontrolled enemy spawn probability introduced within-configuration variability that affected perceived fairness. Prior platformer experience significantly moderated challenge and tension scores, with novice players struggling primarily at the mechanical level before layout complexity became a factor. The findings suggest that even a lightweight constraint-based PCG system can produce engaging platformer content, but that consistent player-centered quality requires constraints that account for individual skill, not only structural validity.