The arch compositional agency has historically fluctuated
between order and subversion. Architectural imagery of this compositional range
has been extensively documented on multiple mediums throughout history, which
makes for a good case study for machine learning. This project undertakes the
challenge of tridimensionalizing 2D GAN-produced latent imagery through a
series of workflows that engage datasets as an animated computable matter from
which to choose specific cross-sections or moments, ultimately defying or
questioning the designer’s role in an AI-based design process. The exhibits
search to operate within the most trending digitized mediums, embracing the
instrumentality of social media and NFT culture in AI and current architectural
representation.
Made using machine learning workflows.
Type: Ongoing Research
Research: Machine Learning, 2D & 3D GAN.
1. Generative StyleGAN.
2. New-Found 3D GAN arch facade typologies.
3. Morphing 3D GANs.