Unit 4


Hanjun Kim + Dongyun Kim

GAN City, Dongyun Kim, 2021
Urban Fabric


As we move forward into the third year of our unit, we are now challenging the typical morphology and urban fabric of the city blocks in Seoul, embracing an evolutionary perspective to inspire change and drive innovation in architectural design. The focus of our exploration is the very DNA of urban structures - the city block - and how it contributes to the overall 'Image of the City', a concept deeply elaborated by Kevin Lynch. It's through these building blocks that we'll tap into the essence of Seoul's architectural identity and offer transformative solutions.

Lynch stated, "A good environmental image gives its possessor an important sense of emotional security," and this image is essentially a mental construct that reflects a city's identity. This concept parallels Aldo Rossi's ideology in 'The Architecture of the City', where he emphasizes the influence of urban artifacts in city development and identity. The layers of history embedded within the city, considering their impact on identity, sense of place, and the collective memory. By intertwining the past, present, and future, we envision a Seoul that embraces its cultural heritage while actively engaging with contemporary challenges and aspirations. We shall champion these insights, drawing boundaries and creating space just as Giovanni Battista Piranesi did in his 'Campo Marzio Ichnographia', exploring the simplicity and depth of boundary delineation in a cityscape.

While these theoretical perspectives provide our unit with a robust intellectual backbone, we intend to execute our ideas using AI techniques. We will use 2D GANs as 2D plan generators of block boundaries, and 3D GAN for morphology generation of each block, in an attempt to redefine the city blocks of Seoul. These generated outputs will then be synthesized and rendered through diffusion models, providing an advanced and coherent representation of our theoretical framework.

Our agenda, however, goes beyond merely applying these technological tools. We aim to redefine 'evolution' from our perspective. Evolution is not simply a linear progression but a dynamic process encompassing different pathways and diversities, a concept we've explored in previous workshops. Therefore, our exploration of the city blocks in Seoul will not be a straightforward upgrade, but rather a series of diverse and multiform transformations that give birth to an evolved urban fabric, catering to the multifaceted needs of future generations. 

The final output could be evolution of the cityscape of Seoul, challenging its current morphology and anticipating future mutations through an innovative fusion of architectural theory and artificial intelligence.


The unit will initiate with a discussion on the existing layout and urban fabric of Seoul, referencing the theoretical frameworks mentioned above. After establishing a mutual understanding, the unit will progress through three stages:

1. Research :  

The first stage will involve a thorough analysis of Seoul's city blocks, where students will be grouped and asked to research the morphology, layout, and cultural significance of selected blocks. This will form the foundation for the generation process in the following stages.  

2. Generation

Students will input the data into StyleGAN and 3D GAN, which will generate new plans for block boundaries and suggest fresh morphologies for the city blocks. This stage will be allowing students to experiment with various input parameters and observe the resulting transformations.

3. Synthesis and Application

The final stage involves the combination of the 2D and 3D outputs into a unified architectural proposal for the city block. This synthesis will then be rendered using diffusion models, resulting in a final visual representation of the proposed evolution of the city block.


We will be using Python with 2D GANs, 3D GAN and Stable Diffusion for the generation and rendering process. Additionally, students will be required to familiarise themselves with architectural software such as Rhino 3D and Grasshopper for the design process. 

Expected Learning Outcomes

By the end of the workshop, students will be expected to:

- Develop a comprehensive understanding of urban morphology and its evolutionary characteristics.

- Learn to critically analyze and interpret the relationship between architectural form and urban layout.

- Acquire skills in utilizing AI technology for innovative architectural and urban design.

- Develop the ability to visualize and communicate complex urban design strategies effectively.

- Engage with contemporary architectural discourse on a deeper level, especially around topics of urban evolution and the role of technology in design.

- Construct and present a professional portfolio, including research, design iterations, and final architectural proposal.

AAVS Seoul 2022 Unit4, Moving Parts, Rayoung Kim + Eddy Choi + Isabel Solin


Hanjun Kim

SoomeenHahm Design

Dongyun Kim

TWO Platforms Inc.