Tokenization Allows Multimodal Large Language Models to Understand, Generate and Edit Architectural Floor Plans

TL;DR

HouseMind is a multimodal large language model designed for understanding, generating, and editing architectural floor plans, effectively combining geometry, semantics, and spatial reasoning for improved control and validity.

Contribution

The paper introduces HouseMind, a unified framework with discrete room tokens and multimodal alignment that advances controllable and coherent floor plan generation and editing.

Findings

Achieves superior geometric validity in generated layouts

Provides controllable floor plan synthesis from text instructions

Remains efficient and suitable for local deployment

Abstract

Architectural floor plan design demands joint reasoning over geometry, semantics, and spatial hierarchy, which remains a major challenge for current AI systems. Although recent diffusion and language models improve visual fidelity, they still struggle with coherent spatial reasoning and controllable generation. We present HouseMind, a multimodal large language model that unifies floor plan understanding, generation, and editing in one framework. We introduce discrete room-instance tokens to construct a unified vocabulary that bridges layouts and symbolic reasoning. With multimodal alignment and instruction tuning, the model synthesizes coherent, controllable layouts from text instructions. Experiments show how the framework achieves superior geometric validity and controllability while remaining efficient and locally deployable.