GenEscape: Hierarchical Multi-Agent Generation of Escape Room Puzzles

TL;DR

This paper introduces GenEscape, a hierarchical multi-agent framework that generates visually appealing and functionally coherent escape room puzzle images by decomposing the task into structured stages and enabling agent collaboration.

Contribution

The paper presents a novel multi-agent hierarchical approach that improves the quality and solvability of generated escape room puzzles compared to existing models.

Findings

Agent collaboration enhances puzzle solvability.

The framework reduces shortcuts and improves affordance clarity.

Generated puzzles maintain high visual quality.

Abstract

We challenge text-to-image models with generating escape room puzzle images that are visually appealing, logically solid, and intellectually stimulating. While base image models struggle with spatial relationships and affordance reasoning, we propose a hierarchical multi-agent framework that decomposes this task into structured stages: functional design, symbolic scene graph reasoning, layout synthesis, and local image editing. Specialized agents collaborate through iterative feedback to ensure the scene is visually coherent and functionally solvable. Experiments show that agent collaboration improves output quality in terms of solvability, shortcut avoidance, and affordance clarity, while maintaining visual quality.