DIAMBRA Game Painter

This project is an experiment that applies in real-time the style of famous paintings to popular fighting retro games, which are provided as Reinforcement Learning environments by DIAMBRA (GitHub - Website).

It is based on this implementation of the paper Perceptual Losses for Real-Time Style Transfer and Super-Resolution.

How to run it

• Create a virtual environment using your preferred tool (Conda, VirtualEnv, etc)
• Install required python packages: pip install -r requirements.txt
• Run the script on a random game: diambra run python gameEmulationStyleTransf.py
• Run the script on a specific game: diambra run python gameEmulationStyleTransf.py --gameId game-ID where available games and their game-IDs can be found in DIAMBRA documentation