Building layout generation using site-embedded GAN model

Building layout generation has entered a new era in recent years, leveraging state-of-the-art deep generative methods to learn morphological properties of exiting urban structures and synthesize building alternatives responsive to local context. However, most existing research generally follows an image-to-image translation idea, while overlooking the impact of site/design attributes on building configuration, making their results less performative. Besides, most synthesized layouts are commonly displayed in 2D pixelized images, limiting further performance evaluation and informed decision-making. This study, therefore, proposes a novel GAN-based model, namely site-embedded generative adversarial networks (ESGAN) for automated building layout generation. Both qualitative and quantitative results in New York City indicate ESGAN is capable of synthesizing visually realistic and semantically reasonable layouts. This end-to-end generative system can not only encode a conditional vector to improve performance in different design scenarios but also display synthesized layouts at different levels of detail for human-system interaction. •A novel deep learning-based generative model is proposed for automated building layout generation.•The model incorporates a conditional vector into the generation process to improve performance in different design scenarios.•The model can synthesize visually realistic and semantically reasonable building layouts.•The synthesized layouts are displayed in different levels of detail for human-system interaction.•The model is an end-to-end generative system without much expert knowledge and user intervention.