A Monastic Passage in the Landscape
An exploration in the use of foundation models to address architecture atmospheres
William Kosta (TU Delft - Architecture and the Built Environment)
G. Coumans – Mentor (TU Delft - History, Form & Aesthetics)
S Khademi – Mentor (TU Delft - Building Knowledge)
R. Heykant – Mentor (TU Delft - Teachers of Practice / AE+T)
A. Mulder – Graduation committee member (TU Delft - Education and Student Affairs)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
In my research, I explored how foundation models (artificial intelligence) might be used in architecture. The outcome of my research is a 2D scatter plot of 2500 architectural images, with each point on the plot arranged based on the vector embedding (created by the foundation model). This can be extrapolated and seen as arranging images on 2 axis loosely based on the atmosphere of the space in the image.
With the ability of creating a 2D scatter plot that clusters images based on architectural atmospheres, I wanted to make use of this in an architectural design process.
Specifically I had an interest in designing something that is related to sacred spaces. I always wondered whether for instance: a church spatially feels sacred, mostly because of religion, or because of its architecture? Is there such thing as ‘sacred architecture’? Would a church feel like a sacred space when religion and beliefs are not projected onto it?
This led to my design brief: ‘A Monastic Passage in the Landscape’, which is a space people with a terminal illness diagnosis can visit in order to reflect on mortality. This brief, with the theme of facing mortality, allows me to explore the possibilities and atmospheres that architecture can create when its aim is to address the transcendent and sublime.
Relating to the research, with the 2D scatter plot that clusters images based on atmospheres, the idea is to follow the method of putting in images of the design project into the scatter plot (through the foundation model) to see where the design would land in the 2D plot. From the position in the plot, I hoped to more accurately evaluate and control the resulting atmosphere in my design project.