# Procedural Content Generation using Neuroevolution and Novelty Search for Diverse Video Game Levels

arXiv: https://arxiv.org/abs/2204.06934 · HTML: https://mlsanity.com/paper/2204.06934/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 14 Apr 2022.

## Built on by (most cited first, 8 of 8)

- [MarioGPT: Open-Ended Text2Level Generation through Large Language Models](https://mlsanity.com/paper/2302.05981/) (2023)
- [The Procedural Content Generation Benchmark: An Open-source Testbed for Generative Challenges in Games](https://mlsanity.com/paper/2503.21474/) (2025)
- [Measuring Diversity of Game Scenarios](https://mlsanity.com/paper/2404.15192/) (2024)
- [LatticeWorld: A Multimodal Large Language Model-Empowered Framework for Interactive Complex World Generation](https://mlsanity.com/paper/2509.05263/) (2025)
- [Hierarchically Composing Level Generators for the Creation of Complex Structures](https://mlsanity.com/paper/2302.01561/) (2023)
- [Start Small: Training Controllable Game Level Generators without Training Data by Learning at Multiple Sizes](https://mlsanity.com/paper/2209.15052/) (2022)
- [Quality Diversity in the Amorphous Fortress (QD-AF): Evolving for Complexity in 0-Player Games](https://mlsanity.com/paper/2312.02231/) (2023)
- [Expanding Horizons of Level Diversity via Multi-objective Evolutionary Learning](https://mlsanity.com/paper/2509.24341/) (2025)

## Abstract (arXiv)

Procedurally generated video game content has the potential to drastically reduce the content creation budget of game developers and large studios. However, adoption is hindered by limitations such as slow generation, as well as low quality and diversity of content. We introduce an evolutionary search-based approach for evolving level generators using novelty search to procedurally generate diverse levels in real time, without requiring training data or detailed domain-specific knowledge. We test our method on two domains, and our results show an order of magnitude speedup in generation time compared to existing methods while obtaining comparable metric scores. We further demonstrate the ability to generalise to arbitrary-sized levels without retraining.