Dynamic Environment Generation in Classic Video Games
Members
- Nuo Wen Lei (nlei)
- Haiyang Wang (hwang330)
- Jingbo Yang (jyang113)
Introduction
This project aims to revolutionize the way we experience classic video games by introducing dynamically generated environments that evolve in real-time as the players interact with the game. Our focus will be on utilizing a state-of-the-art video diffusion model, trained exclusively on game video footage, to generate "out-of-sample" scenes that maintain the essence of the original game while offering new, unpredictable gameplay experiences. By incorporating a latent action model, we plan to enable the game environment to learn and adapt to interactive actions, further enhancing the dynamism and replayability of classic titles.
Deliverables
Previous Reflections
Built With
- keras
- python
- tensorflow
Log in or sign up for Devpost to join the conversation.