Exposée: Generation and adaption of quest-based virtual environments
Description
Procedural Generation of worlds, levels or other virtual environments has been of great significance in the game development process for a long time. With recent breakthroughs in Machine Learning, Artificial Intelligence and specifically Large Language Models (LLMs) procedurally generating environments depending on their output uncovers a lot of new possibilities.
By defining a quest definition for this use case and processing a quest description through OpenAI’s GPT-4 model, this presentation aims to showcase a concept for generating environments in the Unity Engine, that are cohesive to the input in regards to terrain, locations and tasks, susceptible to quest outcomes and adaptable to them.
