Illuminating Mario Scenes in the Latent Space of a Generative Adversarial Network

TL;DR

This paper introduces a method for exploring GAN latent spaces in game level design using quality diversity algorithms, enabling designers to generate diverse, high-quality Mario levels aligned with specific gameplay measures.

Contribution

It applies quality diversity algorithms to GAN latent spaces for controllable, diverse level generation, a novel approach in procedural game content creation.

Findings

Generated levels are playable and stylistically similar to human designs.

Designers can specify gameplay measures to control level diversity.

User study shows levels influence perceived difficulty and appearance.

Abstract

Generative adversarial networks (GANs) are quickly becoming a ubiquitous approach to procedurally generating video game levels. While GAN generated levels are stylistically similar to human-authored examples, human designers often want to explore the generative design space of GANs to extract interesting levels. However, human designers find latent vectors opaque and would rather explore along dimensions the designer specifies, such as number of enemies or obstacles. We propose using state-of-the-art quality diversity algorithms designed to optimize continuous spaces, i.e. MAP-Elites with a directional variation operator and Covariance Matrix Adaptation MAP-Elites, to efficiently explore the latent space of a GAN to extract levels that vary across a set of specified gameplay measures. In the benchmark domain of Super Mario Bros, we demonstrate how designers may specify gameplay measures to our system and extract high-quality (playable) levels with a diverse range of level mechanics, while still maintaining stylistic similarity to human authored examples. An online user study shows how the different mechanics of the automatically generated levels affect subjective ratings of their perceived difficulty and appearance.