Title
Remastering games with Deep Generative Models.
Subject
This thesis aimed to explore the concept of using deep generative models for remastering games and evaluate the idea of AI remastering. There are many old games that, with a remaster they can get a new life. Our goal in this project is to utilize an existing 2D game, identify the optimal model and architecture for remastering its assets (Textures and images), and remake the assets with AI. There are a few key questions we aim to answer, such as whether the model can accurately preserve the style of the game for all assets, whether the game’s theme can be altered, and whether players can distinguish between real textures and AI-generated ones in the new version.
Your Tasks
You’ll need to familiarize yourself with generative models, explore them, and identify the best candidate models for generating assets. Then, experiments will be conducted to evaluate this project using both qualitative and quantitative methods.
Requirements
- Experience in Python.
- Familiarity with generative models.
