Abstract:This paper proposes a two-phase text-to-floorplan generation method, which guides a Large Language Model (LLM) to generate an initial layout (Layout-LLM) and refines them into the final floorplans through conditional diffusion model. We incorporate a Chain-of-Thought approach to prompt the LLM based on user text specifications, enabling a more user-friendly and intuitive house layout design. This method allows users to describe their needs in natural language, enhancing accessibility and providing clearer geometric constraints. The final floorplans generated by Layout-LLM through conditional diffusion refinement are more accurate and better meet user requirements. Experimental results demonstrate that our approach achieves state-of-the-art performance across all metrics, validating its effectiveness in practical home design applications. We plan to release our code for public use.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in architectural design, creating floor plans that meet user requirements remains a challenge. Traditional design methods not only rely on professional knowledge, but also require designers to make multiple iterative adjustments to meet specific user needs, which makes personalized design difficult. Although existing training models have made certain progress in improving the accuracy, interactivity, and efficiency of floor plan generation, these methods have not yet reached the friendliness and accuracy that can be widely adopted by ordinary users. Specifically, this paper aims to solve the following problems: 1. **Lack of user - friendliness**: Existing methods require users to specify room layouts through complex diagrams (such as bubble diagrams), which has a high threshold for non - professional users. 2. **Insufficient fine - grained control of geometric and numerical constraints**: End - to - end generation methods directly from text to floor plans have difficulty meeting the fine - grained geometric and numerical constraints specified in the text. 3. **Lack of large - scale datasets**: Successfully training a model that directly maps text to house layouts requires at least a large amount of data, and the currently available datasets are not large enough to support such training. To solve these problems, the authors propose a two - stage text - to - floor - plan generation method (HouseLLM), which uses a large - language model (LLM) to generate an initial layout and further optimizes it through a conditional diffusion model to generate the final house floor plan. This method allows users to describe their needs in natural language, improving accessibility and accuracy and better meeting user needs. ### Specific problem summary: - **How to make floor plan generation more user - friendly?** - **How to ensure that the generated floor plan can accurately reflect the fine - grained geometric and numerical constraints described by the user in the text?** - **How to achieve high - quality floor plan generation without a large - scale dataset?** By introducing HouseLLM, the authors hope to provide a more intuitive and accurate method, enabling ordinary users to easily design house floor plans that meet their own needs.

HouseLLM: LLM-Assisted Two-Phase Text-to-Floorplan Generation

Web3D-based Automatic Furniture Layout System Using Recursive Case-Based Reasoning and Floor Field.

ChatHouseDiffusion: Prompt-Guided Generation and Editing of Floor Plans

LLplace: The 3D Indoor Scene Layout Generation and Editing via Large Language Model

Chat2Layout: Interactive 3D Furniture Layout with a Multimodal LLM

LayoutLLM: Layout Instruction Tuning with Large Language Models for Document Understanding

CoLay: Controllable Layout Generation through Multi-conditional Latent Diffusion

TextLap: Customizing Language Models for Text-to-Layout Planning

Intelligent Home 3D: Automatic 3D-House Design from Linguistic Descriptions Only

Reason out Your Layout: Evoking the Layout Master from Large Language Models for Text-to-Image Synthesis

Graph2Plan: Learning Floorplan Generation from Layout Graphs

Low-code LLM: Graphical User Interface over Large Language Models

Layout Generation for Various Scenarios in Mobile Shopping Applications.

Large Language Model for Participatory Urban Planning

LayoutCopilot: An LLM-powered Multi-agent Collaborative Framework for Interactive Analog Layout Design

DStruct2Design: Data and Benchmarks for Data Structure Driven Generative Floor Plan Design

PosterLLaVa: Constructing a Unified Multi-modal Layout Generator with LLM

Generating layout designs from high-level specifications

LLM4DESIGN: An Automated Multi-Modal System for Architectural and Environmental Design

Formal-LLM: Integrating Formal Language and Natural Language for Controllable LLM-based Agents

LayoutGPT: Compositional Visual Planning and Generation with Large Language Models