DanieleLoiacono / DoomGAN

DOOM level generation using Generative Adversarial Networks

DOOM level generation using GAN

This is a project developed at Politecnico di Milano by Edoardo Giacomello, Pier Luca Lanzi, and Daniele Loiacono. The project aims at investigating the generation of DOOM levels using Generative Adversarial Networks (GAN). We trained two types of GAN on more than 1000 DOOM levels created by the community (available here) and, then, used them to generate novel levels.

Report

Video

An example of playable level (from the video):

Download WAD

Sets of raw levels:

Set 1 (32 Levels)

Set 2 (32 Levels)

Code

See DoomGAN Repository