Player-Centric Procedural Content Generation: Enhancing Runtime Customization by Integrating Real-Time Player Feedback

Abstract

Game content creation poses significant challenges, particularly for indie developers and small teams, as it presents difficulties in scaling to meet diverse player preferences. Adaptable procedural content generation (PCG) provides a promising solution to this issue. Extensive literature exists on adaptable PCG techniques, encompassing both offline (during development) and online (during gameplay) approaches. Building upon this foundation, we propose a novel extension called Player-Centric Procedural Content Generation (PCPCG) as an additional tool for creating unique game experiences based on player preferences. In contrast to runtime adaptable PCG methods, which adapt or learn based solely on in-game data, PCPCG actively involves players in the learning loop by soliciting their feedback. Additionally, it shifts the focus from designers (as in mixed initiative (MI) and co-creative PCG) to the players as providers of learning data, thus operating during gameplay rather than during development. PCPCG possesses three key qualities: 1) real-time operation during gameplay, 2) active participation of players (not designers) in the learning loop, and 3) online learning from player feedback to create engaging and personalized content. It is important to differentiate PCPCG from content creation aids and in-game data adaptable PCGs. While PCPCG falls under the umbrella of adaptable PCG, it goes beyond relying solely on in-game data by incorporating valuable player feedback as a vital information source for content generation. PCPCG introduces a novel and promising approach to runtime procedural content generation by leveraging player feedback to create adaptive and personalized game content. While our proof of concept demonstrates the viability of PCPCG in a Pac-Man domain, further research is required to explore its limitations as the complexity of the possibility space increases.