Xiong’an: An Imagined City

Jiawei Yao, Zhaoxuan Yang, Nishang Wang


A{AI} - Architecture and Artificial Intelligence


Matias del Campo + Sandra Manninger


The ambitious new city planning of Xiong’an provides us with an opportunity to explore AI and neural networks as developing tools to create a workflow that is applicable across from regional planning to small-scale architecture. Our goal is to design a decentralized city and several charming pavilions scattered throughout the city without planned cities' typical image. The proposal pictures Xiong’an in different scales, from the large regional planning, to the block-scale urban design and specific architecture designs, with the help of AI and neural networks.

In order to explore the possibilities of applying AI and neural networks in designing the new city of Xiong’an, we selected two other cities, Venice and Tokyo, as test objects for our thesis. Venice and Tokyo are two typical examples of historical and walkable cities from the west and the east. The first step we did is to show how the images of these two cities could be projected onto Xiong’an by using the 2D-to-2D style transfer technology. The technology helps us to generate imagined images of Xiong’an as a city that doesn’t take time or historical context to fully grow like Venice and Tokyo. Then we also tried to merge more urban networks from these two cities into Xiong’an, like the water system from Venice and the modern public transportation system from Tokyo.

After learning the possibilities and abilities of using AI and neural networks, for the regional scale planning, we trained the AI to study the urban context of Venice and Tokyo, then merged characteristics of these two cities and combined them with the existing Xiong’an’s satellite map using the 2D-to-2D style transfer technology, while keeping its natural geography at the same time. For the block scale design, by reading the new generated satellite map, we applied those ideal public space paradigms from Venice and Tokyo onto the urban areas within the map, to create a series of imagined public spaces. For the architecture scale, each of these urban areas has a pavilion that works as an art anchor. The form and space of these art anchors are defined by the prototypes generated from the imagined public space image, and the texture generated from the imagined street views that surround the art anchors.