World-GAN: a Generative Model for Minecraft Worlds

TL;DR

World-GAN is a novel 3D GAN-based method that generates large Minecraft worlds from a single example, using block2vec representations to handle diverse blocks and style transfer.

Contribution

Introduces World-GAN, the first data-driven Minecraft world generator using a 3D GAN and block2vec for style and size flexibility.

Findings

Successfully generates large Minecraft world snippets from a single example

Uses block2vec to handle diverse block types and enable style transfer

Demonstrates style modification in generated worlds

Abstract

This work introduces World-GAN, the first method to perform data-driven Procedural Content Generation via Machine Learning in Minecraft from a single example. Based on a 3D Generative Adversarial Network (GAN) architecture, we are able to create arbitrarily sized world snippets from a given sample. We evaluate our approach on creations from the community as well as structures generated with the Minecraft World Generator. Our method is motivated by the dense representations used in Natural Language Processing (NLP) introduced with word2vec [1]. The proposed block2vec representations make World-GAN independent from the number of different blocks, which can vary a lot in Minecraft, and enable the generation of larger levels. Finally, we demonstrate that changing this new representation space allows us to change the generated style of an already trained generator. World-GAN enables its users to generate Minecraft worlds based on parts of their creations.