MITHRIL: A Multi-modal and Immersive Generative AI Approach for Roleplaying Tabletop Games

Michael Atkins, William Denman, Tuna Han Salih Meral, Mustafa Doga Dogan, Pinar Yanardag

Track: Creative demo

Keywords: table top games, personalization, content generation, d&d

TL;DR: We present MITHRIL, a multi-modal generative system for table top games such as Dungeons & Dragons.

Abstract: Tabletop Role-Playing Games (TTRPGs) like Dungeons \& Dragons (D\&D) offer rich, collaborative storytelling experiences where Dungeon Masters (DMs) play a crucial role in creating narrative arcs, designing encounters, and bringing the game world to life. However, the demands of crafting engaging stories, developing complex characters, and managing the intricate details of gameplay can be overwhelming for DMs, often requiring significant effort and time. This paper introduces a novel system that assists DMs in generating multimodal content, including both text descriptions and images for characters, campaigns, and story arcs, with advanced personalization features. Leveraging state-of-the-art AI capabilities, the system enables DMs to automatically generate detailed character backstories, skill sets, and high-quality images, creating a more immersive and visually enriched narrative experience. Unlike existing tools that focus solely on text generation, our system integrates both textual and visual modalities and allows for a high degree of customization, such as embedding players' faces onto character avatars. This unique approach addresses the growing demand for personalized and dynamic content in TTRPGs, reducing the DM's workload while enhancing creativity and player engagement. We detail the design and implementation of our system, evaluate its usability and effectiveness, and discuss how it sets a new standard for multimodal content generation and personalized storytelling in TTRPGs.

Submission Number: 40