Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models

The urban road spatial structure is a crucial and complex component of urban design. Generative design models, such as the Stable Diffusion model, can rapidly and massively produce designs. However, the opacity of their internal architecture and the uncertainty of their outcomes mean that the result...

Full description

Saved in:
Bibliographic Details
Main Authors: Dian Yu, Bo Wan, Qiang Sheng
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Buildings
Subjects:
Online Access:https://www.mdpi.com/2075-5309/14/12/3720
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850242067052101632
author Dian Yu
Bo Wan
Qiang Sheng
author_facet Dian Yu
Bo Wan
Qiang Sheng
author_sort Dian Yu
collection DOAJ
description The urban road spatial structure is a crucial and complex component of urban design. Generative design models, such as the Stable Diffusion model, can rapidly and massively produce designs. However, the opacity of their internal architecture and the uncertainty of their outcomes mean that the results generated do not meet specific disciplinary assessment criteria, thus limiting their widespread application in planar design and planning. Additionally, traditional software processes targeting specific indicators are time-consuming and do not allow for rapid evaluation. To address these challenges, we utilized several areas of the road spatial structures in six cities and their corresponding four space-syntax parameters as training samples. We simultaneously trained two models: one is a LoRA Model based on the Stable Diffusion architecture used for generating road networks similar to those of various city road spatial structures; the other is a CoAtNet Model (Convolution + Transformer) used as an evaluation model to predict the space-syntax parameters of road structures and calculate the Mean Absolute Percentage Error (MAPE) relative to real urban samples. Subsequently, by linking these two models end-to-end, we were able to filter out generated samples with the smallest MAPE, thereby enhancing the structural similarity between the generated results and the actual urban road spatial structures. This process of rapid generation and swift evaluation of network configurations marks a critical advancement towards better performance and more customized design solutions.
format Article
id doaj-art-be8449ed56bb4609bbf366dbdd434afe
institution OA Journals
issn 2075-5309
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Buildings
spelling doaj-art-be8449ed56bb4609bbf366dbdd434afe2025-08-20T02:00:24ZengMDPI AGBuildings2075-53092024-11-011412372010.3390/buildings14123720Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet ModelsDian Yu0Bo Wan1Qiang Sheng2School of Architecture and Art, Beijing Jiaotong University, Beijing 100044, ChinaSchool of Architecture and Art, Beijing Jiaotong University, Beijing 100044, ChinaSchool of Architecture and Art, Beijing Jiaotong University, Beijing 100044, ChinaThe urban road spatial structure is a crucial and complex component of urban design. Generative design models, such as the Stable Diffusion model, can rapidly and massively produce designs. However, the opacity of their internal architecture and the uncertainty of their outcomes mean that the results generated do not meet specific disciplinary assessment criteria, thus limiting their widespread application in planar design and planning. Additionally, traditional software processes targeting specific indicators are time-consuming and do not allow for rapid evaluation. To address these challenges, we utilized several areas of the road spatial structures in six cities and their corresponding four space-syntax parameters as training samples. We simultaneously trained two models: one is a LoRA Model based on the Stable Diffusion architecture used for generating road networks similar to those of various city road spatial structures; the other is a CoAtNet Model (Convolution + Transformer) used as an evaluation model to predict the space-syntax parameters of road structures and calculate the Mean Absolute Percentage Error (MAPE) relative to real urban samples. Subsequently, by linking these two models end-to-end, we were able to filter out generated samples with the smallest MAPE, thereby enhancing the structural similarity between the generated results and the actual urban road spatial structures. This process of rapid generation and swift evaluation of network configurations marks a critical advancement towards better performance and more customized design solutions.https://www.mdpi.com/2075-5309/14/12/3720generative road network designspace syntaxDiffusion ModelconvolutionTransformerCoAtNet
spellingShingle Dian Yu
Bo Wan
Qiang Sheng
Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
Buildings
generative road network design
space syntax
Diffusion Model
convolution
Transformer
CoAtNet
title Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
title_full Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
title_fullStr Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
title_full_unstemmed Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
title_short Automated Generation of Urban Spatial Structures Based on Stable Diffusion and CoAtNet Models
title_sort automated generation of urban spatial structures based on stable diffusion and coatnet models
topic generative road network design
space syntax
Diffusion Model
convolution
Transformer
CoAtNet
url https://www.mdpi.com/2075-5309/14/12/3720
work_keys_str_mv AT dianyu automatedgenerationofurbanspatialstructuresbasedonstablediffusionandcoatnetmodels
AT bowan automatedgenerationofurbanspatialstructuresbasedonstablediffusionandcoatnetmodels
AT qiangsheng automatedgenerationofurbanspatialstructuresbasedonstablediffusionandcoatnetmodels