Retrieval-Augmented Sketch-Guided 3D Building Generation

2026-03-17

AI Summary

This work proposes a multi-stage sketch-to-3D generation framework to address low fidelity in 3D reconstruction, lack of component-level editability, and misalignment in multi-stakeholder communication during the early design phase of Japanese detached houses—issues often caused by information loss in hand-drawn sketches. The framework integrates generative and retrieval-based mechanisms, leveraging multimodal representations, component segmentation, and unified mesh decomposition to enable modular customization and replacement of architectural elements such as windows and doors. The study introduces the first facade-and-component dataset tailored to Japanese detached residences and incorporates retrieval-augmented generation to support sketch-driven fine-grained editing. Experimental results demonstrate that the proposed method significantly outperforms existing approaches in both design fidelity and modular customizability, thereby facilitating an efficient and personalized residential design workflow.

Technology Category

• Computer Vision: Multi-modal Vision
• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

In the early design stage of Japanese detached houses, the lack of a unified design representation among clients, sales representatives, and designers leads to design drift and inefficient feedback. Usually, sketches handed off by sales representatives may lose details for quick drawing, which reduces the fidelity of subsequent 3D generation using generative AI models. The generated 3D model typically takes the form of a single unified mesh, preventing component-level editing. To solve these issues, we propose a multi-stage 3D generative design framework capable of producing architectural models from rough design sketches. The framework combines generative and retrieval-based methods to enable component-level editing and personalized customization. It adopts a multimodal representation for 3D model generation and applies component segmentation to localize architectural components such as windows and doors and uses retrieval to support targeted replacement of components. Experiments show that the work enables modular customization which is thought to be suitable for personalized architectural design. This work introduces a multi-stage sketch-to-3D framework for Japanese detached houses, provides facade and component datasets, and shows effectiveness through quantitative and expert evaluations.

Problem

• design drift
• sketch-to-3D generation
• component-level editing
• architectural design representation
• 3D model fidelity

Innovation

• retrieval-augmented generation
• sketch-to-3D
• component-level editing
• multimodal representation
• modular customization