Students will brainstorm and explore potential applications that enable AI generation of meaningful geometry for Architecture and Engineering.


Methods for Encoding and AI Generation of Architectural Geometry


“The promise of deep learning is to discover rich, hierarchical models that represent probability distributions [and] map a high-dimensional, rich sensory input to a class label”  (Ian Goodfellow et all – GANs paper)

While in the previous decades, predictive and discriminative AI became evermore embedded in our daily lives – computer vision is used in most consumer and industrial devices, predictive analysis guides your daily choices, etc. – it is clear that generative AI is the big bet for the upcoming years. As this branch of machine learning is finally reaching the wider public with its hyper accessible interfaces and emotionally triggering results, it’s important to address its potential impact in Architecture, and more generally in the AEC. Even with the limited applicability (2d images), stable diffusion models were responsible for a considerable portion of the visual content posted under the #architecture in 2022. 

 Source: Post Digital Temples – Mahmoud Ramdane | Naitik Sharma | Daniyal Tariq –  Macad 2022

With equally passionate supporters as critics, these images drew attention to the need for AI methods that solve AEC problems, as well as to the complexity that future AI tools for AEC need to address.

Source: Post Digital Temples – Mahmoud Ramdane | Naitik Sharma | Daniyal Tariq – Macad 2022

In this course, we will go from studying the fundamentals of the AI paradigm, to learning and applying key generative AI models, to a more in depth study of the concepts pertinent to the application of AI for the generation of sparse 3d objects.

Source: Ch-AI-r convergence –  Maria Papadimitraki | Michal Gryko –  Macad 2022

The overarching goal of this course is to brainstorm and explore potential applications that enable AI generation of meaningful geometry for Architecture and Engineering. For this we will investigate topics of data representation, synthetic data generation, and conditional and unconditional models, in an intent to understand the end to end pipeline and parameters that come into play in constructing the optimal model for such problems.

A more expanded discussion on the topic of AI in architecture can be found here.

Learning Objectives 

  • Get introduced to fundamental concepts of AI and its applications in Architecture.
  • Understand Generative AI concepts and Models: methods, layer types,  hyperparameters
  • Construct and manipulate Generative Adversarial Neural Networks models
  • Learn to explore and manipulate the latent space 
  • Set up a synthetic data generation algorithm and run experiments generate and recreate new data  


Faculty Assistants

Projects from this course

Facade Evolutions

By employing [input_output] + [pix2pix] combination of machine learning strategy, this exercise explores the opportunity to facilitate facade developments. [over 2000 variations] Inspirations [Input_Output] + [pix2pix] combination is a powerful machine learning tool for innovative architecture design. A meaningful and robust dataset is critical for a successful project. Pilot-test and extensive back-and-forth adjustments are also … Read more


Can we find a systemized approach for extracting the shape of cities? City.Style.GAN is a small research on the possible generative networks capables of generating new buildings massings according to the urban shape fo the city. Methodology: In a first aproximation to the urban generation, large portions of three urban tissues are extracted in OSMnx … Read more

Generative Urban Farming

The experiment parts from a self organizing model that iterates based in the incident sun hours, to later be evaluated in 2 methods: DCGAN’s and PIX2PIX. Problem Statement: The objective is to utilize urban farming strategically in central Budapest to enhance the affordability of student life. The key questions revolve around the potential yield, geometric … Read more


RESEARCH QUESTION Custom facade design has high costs,currently these designs are reserved to high end residential design. What if AI could give a valid variety of proposals for balconies customized arrangements that take into account microclimate and privacy of the outdoor space? METHODOLOGY DESIGN PROCESS PIX TO PIX GAN Allows to explore several options It is quicker … Read more


l INTRODUCTION “According to the United Nations, 2.5 billion people are projected to migrate to an urban center in the next 30 years” “[micro units] will increase by 75% in the next ten years, appealing to a wide demographic including students, young professionals, singles, seniors, weekend commuters, and even tourists.” mmmmm … Read more

Greening Vienna

Today we have a challenge to reduce our waste. As we know construction is responsible for one third of all waste that is produced annually. That is why it is important to look into reusing the buildings instead of demolishing. The process of giving new life to existing buildings is called «adaptive reuse For this … Read more

All Nighter vs. 3DGAN

What kind of architecture is the most suitable for AI? The architecture of the Superdutch is already reduced to diagrams and blue foam models allowing for brute force prototyping. Ten ideas are created everyday per person on teams of ten people in firms like OMA, JDS, and BIG etc. That’s a hundred ideas per day … Read more


I PROCESS I PROBLEM STATEMENT The model will consider factors such as building orientation, facade modifications, and roof area expansion and tilt, ensuring optimal solar panel placement and alignment to maximize energy generation. By developing this AI generative model, we aim to overcome the limitations posed by Seattle’s climate, accelerate the adoption of solar energy, … Read more


Embarking on a project set within the bustling cityscape of New York, we aimed to unveil latent development potentials in urban spaces. Bulk, or the capacity for expansion or extension of a structure within the constraints of zoning laws and local regulations, harbors unexplored opportunities. Often, significant development potential lies hidden within existing building envelopes, … Read more


PROBLEM Maximizing energy efficiency and comfort in buildings with strategically designed and placed sun shades for optimal solar heat reduction is becoming crucial in cities especially in arid zones. This raises an important question how can we use new technologies in achieving such goals. ML MODEL SELECTION in 2017 a very interesting model using Image-to-Image … Read more