Diffusing Functional Morphologies
Dongyun Kim + Hanjun Kim + Lloyd Sukgyo Lee
The unit aims to explore a design method that utilises latest technologies to create a culture specific design that defies design approach based on periodic and stylistic understandings of architecture. Students will be given a typology specific to Korea which they have to incorporate into a design manifestation. The design should integrate newly developed elements generated through GANs(Generative Adversary Network—in this case, StyleGAN and 3D GAN).
Various architectural elements that were used in Korea throughout different eras will be first divided according to their functions. Elements in each function category, all from different eras, merge and give birth to new elements that incorporate the Korean understanding of functional morphologies. These new GAN-driven elements transcends periodic definition and will lead to a design that is culture specific.
The unit will initiate by discussing the current state of Korean architecture in correspondence to the statement above. During discussion, the concept and the elements of the Korean Architecture will be defined. After a general understanding is reached, the unit will progress through 3 stages to produce different outputs.
1. Research :
The first stage will see students divided into two groups. One group will focus on architectural elements of Korean architectures while the other group will initiate research on its morphological definitions. The students will be asked to analyse the organisation, composition & decomposition, and the logic of different precedents, while modeling them into 3D format.
In the second stage, multiple sets of 3D objects produced in the research stage will be diffused through machine learning libraries in Python. The output of this stage should be able to shine a light on different ways of morphological understanding of Korean architectural elements. A degree of randomness is expected in the diffusion process. This stage will act as a testbed to observe numerous applications of different compositional logics to existing examples, while allowing students to imagine beyond the original composition of architecture. The output of this stage will be unified in format to allow for comprehensiveness between disparate ways of application utilised.
The third stage will ask students to play with the possibilities of re-composed elements. Different scenario will be given and the students will choose from the produced morphology from the Diffusion stage to further develop their designs. Each team will develop 3 designs from 3 different elemental morphology compositions and one will be chosen at the end, after the discussions with the tutors. Each team will then design the chosen object until the end of the workshop and should aim to produce final render.
Rhino 3D & Grasshopper, Python (StyleGAN, 3D GAN, etc.)
Expected Learning Outcomes
At the end of the workshop, Students will be able to:
- have a daily basis project portfolio
- learn How to design and utilise a machine learning process to produce different compositions in an architectural context
- understand and build a rigorous methodology that could result in architectural output
- learn how to outline a design process using a machine learning technology
- have a design design proposal using elements derived from machine learning process
- investigate functional applications of architectural elements, decompose them and reassemble these elements analyzed by machine perspective