Player-Oriented Procedural Generation: Producing Desired Game Content by Natural Language

Procedural Content Generation (PCG) plays a vital role in digital games and interactive media, using algorithms and rules to automatically generate the core elements of a game, aiming to provide a rich and unique experience for the player. This study proposes an innovative player-oriented PCG approach to integrate PCG into the player’s gaming experience to enhance game interactivity. We developed a Quick Custom Map Generation System (QMBS) using Large Language Modelling (LLM), allowing players to describe and create game content directly through natural language. This approach allows players to engage in game design intuitively and makes the gaming experience more personalized and diverse. In addition, this research highlights the great potential of LLM in facilitating human-computer interaction and creative tasks, opening up the possibility of new areas of game design.