Jianing Luo and Boyuan Yu's project, Generative Building Facades with Pix2Pix GAN in Hong Kong, aims to harmonize new developments with existing architectural styles in the city.
They constructed a dataset of 160 facade images from old neighborhoods in Hong Kong and used Pix2Pix GANs, a generative model, to automatically generate contextually aligned building facades.
The AI-generated facades demonstrate high fidelity to Hong Kong's vernacular forms by successfully reconstructing elements like window arrangements, floor divisions, and shopfronts.
This approach offers a way to streamline facade design in dense urban areas, preserving architectural heritage, reducing subjectivity in analysis, and providing cleaner, stylized design alternatives.